Dokumen - Pub An Introduction To Symmetric Functions and Their Combinatorics

STUDENT MATHEMATICAL LIBRARY
Volume 91
An Introduction
to Symmetric
Functions
and Their
Combinatorics
Eric S. Egge
An Introduction
to Symmetric
Functions
and Their
Combinatorics
STUDENT MATHEMATICAL LIBRARY
Volume 91
An Introduction
to Symmetric
Functions
and Their
Combinatorics
Eric S. Egge
Editorial Board
Satyan L. Devadoss John Stillwell (Chair)
Rosa Orellana Serge Tabachnikov
2010 Mathematics Subject Classification. Primary 05E05, 05A05, 05A19.
For additional information and updates on this book, visit

www.ams.org/bookpages/stml-91
Library of Congress Cataloging-in-Publication Data

Cataloging-in-Publication Data has been applied for by the AMS.
See http://www.loc.gov/publish/cip/.
Copying and reprinting. Individual readers of this publication, and nonprofit li-
braries acting for them, are permitted to make fair use of the material, such as to
copy select pages for use in teaching or research. Permission is granted to quote brief
passages from this publication in reviews, provided the customary acknowledgment of
the source is given.
Republication, systematic copying, or multiple reproduction of any material in this
publication is permitted only under license from the American Mathematical Society.
Requests for permission to reuse portions of AMS publication content are handled
by the Copyright Clearance Center. For more information, please visit www.ams.org/
publications/pubpermissions.
Send requests for translation rights and licensed reprints to reprint-permission
@ams.org.
2019
c by the author. All rights reserved.
Printed in the United States of America.

∞ The paper used in this book is acid-free and falls within the guidelines
established to ensure permanence and durability.
Visit the AMS home page at https://www.ams.org/
10 9 8 7 6 5 4 3 2 1 24 23 22 21 20 19
Contents
Preface ix
Chapter 1. Symmetric Polynomials, the Monomial Symmetric

Polynomials, and Symmetric Functions 1
§1.1. Symmetric Polynomials 2
§1.2. The Monomial Symmetric Polynomials 7
§1.3. Symmetric Functions 10
§1.4. Problems 19
§1.5. Notes 21
Chapter 2. The Elementary, Complete Homogeneous, and

Power Sum Symmetric Functions 23
§2.1. The Elementary Symmetric Functions 23
§2.2. The Complete Homogeneous Symmetric Functions 38
§2.3. The Power Sum Symmetric Functions 44
§2.4. Problems 49
Chapter 3. Interlude: Evaluations of Symmetric Functions 53

§3.1. Symmetric Function Identities 53
§3.2. Binomial Coefficients 57
§3.3. Stirling Numbers of the First and Second Kinds 60
v
vi Contents
§3.4. q-Binomial Coefficients 64

§3.5. Problems 71
§3.6. Notes 73
Chapter 4. Schur Polynomials and Schur Functions 75

§4.1. Schur Functions and Semistandard Tableaux 75
§4.2. Schur Polynomials as Ratios of Determinants 89
§4.3. Problems 111
§4.4. Notes 116
Chapter 5. Interlude: A Rogues’ Gallery of Symmetric

Functions 119
§5.1. Skew Schur Functions 119
§5.2. Stable Grothendieck Polynomials 129
§5.3. Dual Stable Grothendieck Polynomials 137
§5.4. The Chromatic Symmetric Function 144
§5.5. Problems 153
§5.6. Notes 156
Chapter 6. The Jacobi–Trudi Identities and an Involution

on Λ 157
§6.1. The First Jacobi–Trudi Identity 157
§6.2. The Second Jacobi–Trudi Identity 171
§6.3. The Involution ω 178
§6.4. Problems 183
§6.5. Notes 189
Chapter 7. The Hall Inner Product 191

§7.1. Inner Products on Λk 191
§7.2. The Hall Inner Product and Cauchy’s Formula 196
§7.3. The Hall Inner Product on the Power Sum
Symmetric Functions 201
§7.4. Problems 206
§7.5. Notes 207
Contents vii
Chapter 8. The Robinson–Schensted–Knuth Correspondence 209

§8.1. RSK Insertion: Constructing P (π) 210
§8.2. Constructing Q(π) 223
§8.3. Implementing RSK with Growth Diagrams 232
§8.4. Problems 242
§8.5. Notes 245
Chapter 9. Special Products Involving Schur Functions 247

§9.1. The Pieri Rules 248
§9.2. The Murnaghan–Nakayama Rule 256
§9.3. Problems 269
Chapter 10. The Littlewood–Richardson Rule 271

§10.1. Products of Tableaux 272
§10.2. Knuth Equivalence 278
§10.3. The Relationship Between P and word 285
§10.4. The Littlewood–Richardson Rule 291
§10.5. Problems 303
§10.6. Notes 307
Appendix A. Linear Algebra 309

§A.1. Fields and Vector Spaces 309
§A.2. Bases and Linear Transformations 312
§A.3. Inner Products and Dual Bases 316
§A.4. Problems 320
Appendix B. Partitions 323

§B.1. Partitions and a Generating Function 323
§B.2. Problems 325
Appendix C. Permutations 327

§C.1. Permutations as Bijections 327
§C.2. Determinants and Permutations 331
§C.3. Problems 334
viii Contents
Bibliography 337
Index 341
Preface
Many excellent books are devoted entirely or in part to the founda-

tions of the theory of symmetric functions, including the books of
Loehr [Loe17], Macdonald [Mac15], Mendes and Remmel [MR15],
Sagan [Sag01], and Stanley [Sta99]. These books approach symmet-
ric functions from several different directions, assume various amounts
of preparation on the parts of their readers, and are written for a va-
riety of audiences. This book’s primary aim is to make the theory
of symmetric functions more accessible to undergraduates, taking a
combinatorial approach to the subject wherever feasible, and to show
the reader both the core of the subject and some of the areas that are
active today.
We assume students reading this book have taken an introduc-
tory course in linear algebra, where they will have seen bases of vector
spaces, transition matrices between bases, linear transformations, and
determinants. We also assume readers have taken an introductory
course in combinatorics, where they will have seen (integer) parti-
tions and their Ferrers diagrams, binomial coefficients, and ordinary
generating functions. For those who would like to refresh their mem-
ories of some of these ideas or who think there might be gaps in
their linear algebraic or combinatorial background, we have included
summaries of the ideas from these areas we will use most often in
the appendices. In particular, we have included explanations of dual
ix
x Preface
bases and the relationship between determinants and permutations,

since these will play important roles at key moments in our study of
symmetric functions, and they may not appear in a first course in
linear algebra.
Symmetric functions have deep connections with abstract alge-
bra and, in particular, with the representation theory of the sym-
metric group, so it is notable that we do not assume the reader
has any familiarity with groups. Indeed, we develop the one alge-
braic fact we need—that each permutation is a product of adjacent
transpositions—from scratch in Appendix C. Leaving out the re-
lationship between symmetric functions and representations of the
symmetric group makes the book accessible to a broader audience.
But to see the subject whole, one must also explore the relationship
between symmetric functions and representation theory. So we en-
courage interested students, after reading this book, to learn about
representations of the symmetric group, and how they are related to
symmetric functions. Two sources for this material are the books of
James [Jam09] and Sagan [Sag01].
As in other areas of mathematics, over the past several decades
the study of symmetric functions has benefited from the use of com-
puters and from the dramatic increase in the amount of available
computing power. Indeed, many contemporary symmetric functions
researchers use software, such as Maple, Mathematica, and SageMath,
for large symmetric functions computations. These computations, in
turn, often lead to new conjectures, new ideas, and new directions for
exploration.
We take a dual approach to the use of technology in the study of
symmetric functions. On the one hand, we do not assume the reader
has any computer technology available at all as they read: a patient
reader will be able to work out all of the examples, solve all of the
problems, and follow all of the proofs without computer assistance.
On the other hand, we encourage readers to become proficient with
some kind of symmetric functions software. Specific programs and
platforms come and go, so we do not recommend any platform or
software in particular. But we do recommend that you find a way to
have your computer do your symmetric functions computations for
you and to use it to explore the subject.
Preface xi
Many of the main results in the theory of symmetric functions

can be understood in several ways, and they have a variety of proofs.
Whenever possible, we have given a proof using combinatorial ideas.
In particular, we use families of lattice paths and a tail-swapping
involution to prove the Jacobi–Trudi identities, we use the Robinson–
Schensted–Knuth (RSK) correspondence to prove Cauchy’s formula,
and we use RSK insertion, jeu de taquin, and Knuth equivalence
to prove the Littlewood–Richardson rule. The study of symmetric
functions can motivate these ideas and constructions, but we find
(and we think the reader will find) they are rich and elegant, and
they will reward study for its own sake.
The study of symmetric functions is old, dating back at least
to the study of certain determinants in the mid- to late-nineteenth
century, but it remains an active area of research today. While our
primary goal is to introduce readers to the core results in the field,
we also want to convey a sense of some of the more recent activity
in the area. To do this, we spend Chapter 5 looking at three areas
of contemporary research interest. In Section 5.1 we introduce skew
Schur functions, and we discuss the problem of finding pairs of skew
tableaux with the same skew Schur function. In Sections 5.2 and
5.3 we introduce the stable and dual stable Grothendieck functions,
which are analogues of the Schur functions. Finding analogues of re-
sults about Schur functions for these symmetric functions is a broad
and lively area of current research. Finally, in Section 5.4 we discuss
Stanley’s chromatic symmetric functions. These symmetric functions,
which are defined for graphs, are the subject of at least two longstand-
ing open questions first raised by Stanley. We introduce one of these
questions, which is whether there are two nonisomorphic trees with
the same chromatic symmetric function.
One of the best ways to learn mathematics is to do mathematics,
so in many cases we have tried to describe not only what a result says
and how we prove it, but also how we might find it. In particular,
we introduce several topics by raising a natural question, looking at
some small examples, and then using the results of those examples
to formulate a conjecture. This process often starts with questions
arising from linear algebra, which we then use combinatorial ideas
to answer, highlighting the way the two subjects interact to produce
new mathematics. We could use a different expository approach to
xii Preface
cover the material more efficiently, but we hope this approach brings
the subject to life in a way a more concise treatment might not.
To get the most out of this book, we suggest reading actively, with
a pen and paper at hand. When we generate data to answer a ques-
tion, try to guess the answer yourself before reading ours. Generate
additional data of your own to support (or refute) your conjecture,
and to verify patterns you’ve observed. Similarly, we intend the ex-
amples to be practice problems. Trying to solve them yourself before
reading our solutions will strengthen your grasp of the core ideas, and
prepare you for the ideas to come.
Speaking of doing mathematics, we have also included a variety
of problems at the end of each chapter. Some of these problems are
designed to test and deepen the reader’s understanding of the ideas,
objects, and methods introduced in the chapter. Others give the
reader a chance to explore subjects related to those in the chapter,
that we didn’t have enough space to cover in detail. A few of the
problems of these types ask the reader to prove results which will be
used later in the book. Finally, some of the problems are there to
tell the reader about bigger results and ideas related to those in the
chapter. A creative and persistent reader will be able to solve many
of the problems, but those of this last type might require inventing
or reproducing entirely new methods and approaches.
This book has benefitted throughout its development from the
thoughtful and careful attention, ideas, and suggestions of a variety
of readers. First among these are the Carleton students who have
used versions of this book as part of a course or a senior capstone
project. The first of these students were Amy Becker ’11, Lilly Betke-
Brunswick ’11, Mary Bushman ’11, Gabe Davis ’11, Alex Evangelides
’11, Nate King ’11, Aaron Maurer ’11, Julie Michelman ’11, Sam
Tucker ’11, and Anna Zink ’11, who used this book as part of a senior
capstone project in 2010–11. Back then it wasn’t really a book; it
was just a skeletal set of lecture notes. Based on their feedback I
wrote an updated and more detailed version, which I used as a text
for a seminar in the combinatorics of symmetric functions in the fall
of 2013. The students in this seminar were Leo Betthauser ’14, Ben
Breen ’14, Cora Brown ’14, Greg Michel ’14, Dylan Peifer ’14, Kailee
Rubin ’14, Alissa Severson ’14, Aaron Suiter ’15, Jon Ver Steegh
Preface xiii
’14, and Tessa Whalen-Wagner ’15. Their interest and enthusiasm

encouraged me to add even more material and detail, which led to a
nearly complete version of the book in late 2018. In the winter and
spring of early 2019, Patty Commins ’19, Josh Gerstein ’19, Kiran
Tomlinson ’19, and Nick Vetterli ’19 used this version as the basis
for a senior capstone project. All three groups of students pointed
out various typographical errors, generously shared their comments,
criticisms, corrections, and suggestions, and suggested several ways
in which the material could be presented more clearly and efficiently.
I am grateful to all of them for their care, interest, and enthusiasm.
Also crucial in the development of this book were several other
readers who shared their ideas, corrections, and suggestions. Jeff
Remmel showed me the combinatorial approach I use to prove the
Murnaghan–Nakayama rule. Becky Patrias suggested including the
results involving stable and dual stable Grothendieck polynomials and
elegant tableaux. And several anonymous reviewers provided very
thorough and detailed comments, corrections, and suggestions. I am
grateful for the time and effort all of these people put in to improve
this book.
Finally, Eko Hironaka, senior editor of the AMS book program,
has been patiently but persistently encouraging me to finish this book
for longer than I care to admit. Thank you, Eko, for not giving up
on it.
In spite of everyone’s best efforts, it is likely some errors remain.
These are all mine. There are also undoubtedly still many ways this
book could be improved. You can find more information related to
the book, along with a list of known errors, at www.ericegge.net/cofsf.
I would very much like to hear from you: if you have comments or
suggestions, or find an error which does not already appear on the
list, please email me at eegge@carleton.edu. And thanks for reading!
Eric S. Egge
July 2019
Chapter 1
Symmetric Polynomials,
the Monomial
Symmetric Polynomials,
and Symmetric
Functions
Symmetric polynomials and symmetric functions arise naturally in

a variety of settings. Certain symmetric functions appear as char-
acters of polynomial irreducible representations of the general linear
groups. The algebra of symmetric polynomials in n variables is iso-
morphic to the algebra generated by the characters of the irreducible
representations of the symmetric group Sn . Certain graph invariants
related to proper colorings of graphs are symmetric functions, and
recently knot invariants like the Jones polynomial have been found to
be related to symmetric functions. A variety of symmetric functions
arise as generating functions for various families of combinatorial ob-
jects, and certain generalizations of symmetric polynomials represent
cohomology classes of Schubert cycles in flag varieties.
In short, symmetric polynomials and symmetric functions are
worth studying for the many places they turn up in algebra, geometry,
1
2 1. Monomial Symmetric Polynomials
topology, graph theory, combinatorics, and elsewhere. But symmet-

ric functions and symmetric polynomials are also worth studying for
their own structure, and for the rich way the algebraic questions they
generate have natural and elegant combinatorial answers. Our goal
will be to reveal as much of this structure as we can, without wor-
rying too much about the rich connections symmetric functions have
elsewhere.
In this chapter we start our study of symmetric functions and
symmetric polynomials with the symmetric polynomials. We explain
what it means for a polynomial to be symmetric, we describe how to
construct symmetric functions, and we explain how symmetric func-
tions are related to symmetric polynomials. We will see that the set of
all symmetric functions with coefficients in Q is a finite-dimensional
vector space over Q, and we will use a natural construction to give
the first of several bases we will see for this space.
1.1. Symmetric Polynomials

We often think of each permutation π ∈ Sn as a bijection from [n] ∶=
{1, 2, . . . , n} to [n], meaning π permutes the elements of [n]. (See
Appendix C for background on permutations.) If instead of [n] we
have a set of variables x1 , . . . , xn , then we can also view π as a per-
mutation of these variables, so that π(xj ) = xπ(j) for 1 ≤ j ≤ n.
Pushing this a bit further, we might eventually want to combine our
variables into polynomials, and then try to view π as a function on
these polynomials. This turns out to be reasonably easy to do: if
f (x1 , x2 , . . . , xn ) is a polynomial in x1 , . . . , xn , then we define π(f )
by setting π(f ) ∶= f (xπ(1) , xπ(2) , . . . , xπ(n) ). That is, π acts on f by
permuting (the subscripts of) its variables.
Example 1.1. Suppose that in one-line notation we have π = 2413

and σ = 4312. If we also have f (x1 , x2 , x3 , x4 ) = x21 + 3x2 x4 − 2x23 x4
and g(x1 , x2 , x3 , x4 ) = 3x3 x24 , then compute π(f ), π(g), π(f + g),
π(f ) + π(g), π(f g), π(f )π(g), σ(f ), π(σ(f )), and (πσ)(f ).
1.1. Symmetric Polynomials 3
Solution. By definition we have
π(f ) = f (xπ(1) , xπ(2) , xπ(3) , xπ(4) )

= x2π(1) + 3xπ(2) xπ(4) − 2x2π(3) xπ(4) .
Since π(1) = 2, π(2) = 4, π(3) = 1, and π(4) = 3, we find
π(f ) = x22 + 3x4 x3 − 2x21 x3 .
Similarly, π(g) = 3x1 x23 .

To compute π(f + g), first note that f + g = x21 + 3x2 x4 − 2x23 x4 +
3x3 x24 , so π(f + g) = x22 + 3x4 x3 − 2x21 x3 + 3x1 x23 . On the other hand,
we have π(f ) = x22 + 3x4 x3 − 2x21 x3 and π(g) = 3x1 x23 , so π(f ) + π(g) =
x22 + 3x4 x3 − 2x21 x3 + 3x1 x23 .
To compute π(f g), first note that f g = 3x21 x3 x24 +9x2 x3 x34 −6x33 x34 ,
so π(f g) = 3x1 x22 x23 + 9x1 x33 x4 − 6x31 x33 . On the other hand, from our
previous computations we find π(f )π(g) = 3x1 x22 x23 +9x1 x33 x4 −6x31 x33 .
Computing as we did for π, we find σ(f ) = x24 + 3x2 x3 − 2x21 x2 . If
we now apply π to σ(f ), we find π(σ(f )) = x23 + 3x1 x4 − 2x22 x4 . On
the other hand, πσ = 3124, so (πσ)(f ) = x23 + 3x1 x4 − 2x22 x4 . □
As we can already see in Example 1.1, we will be considering

polynomials in many variables. To make our notation less cumber-
some, for all n ≥ 1 we will write Xn to denote the set of variables
x1 , . . . , x n .
The computations in Example 1.1 suggest that the action of the
permutations on polynomials is compatible with our usual arithmetic
operations on polynomials, as well as with composition of permuta-
tions. In our next result we make this observation precise.
Proposition 1.2. Suppose f (Xn ) and g(Xn ) are polynomials, c is

a constant, and π, σ ∈ Sn are permutations. Then
(i) π(cf ) = cπ(f );

(ii) π(f + g) = π(f ) + π(g);
(iii) π(f g) = π(f )π(g);
(iv) (πσ)(f ) = π(σ(f )).
Proof. (i) This follows from the fact that π permutes the subscripts
of the variables without changing any coefficients.
(ii) This follows from the fact that if f is a sum of monomials tj ,
then π(f ) is the sum of the monomials π(tj ).
(iii) Suppose first that f and g are each a single term, so that
f (Xn ) = axa1 1 ⋅ ⋅ ⋅ xann and g(Xn ) = bxb11 ⋅ ⋅ ⋅ xbnn for constants a,b,a1 , . . . ,
an , b1 , . . . , bn . Then we have
π(f g) = π(abxa1 1 +b1 ⋅ ⋅ ⋅ xnan +bn )

1 +b1 an +bn
= abxaπ(1) ⋅ ⋅ ⋅ xπ(n)
= axaπ(1)
1
⋅ ⋅ ⋅ xaπ(n)
n
bxbπ(1)
1
⋅ ⋅ ⋅ xbπ(n)
n
= π(f )π(g),
and the result holds in this case. To show the result holds in general,
suppose f (Xn ) = Y1 + ⋅ ⋅ ⋅ + Yk and g(Xn ) = Z1 + ⋅ ⋅ ⋅ + Zm , where each
Yj and each Zj is a monomial in x1 , . . . , xn . Using (ii) and the fact
that (iii) holds for monomials, we find
⎛ k m ⎞
π(f g) = π ( ∑ Yj )(∑ Zl )
⎝ j=1 l=1 ⎠
k m
= π( ∑ ∑ Yj Zl )
j=1 l=1
k m
= ∑ ∑ π(Yj Zl )
j=1 l=1
k m
= ∑ ∑ π(Yj )π(Zl )
j=1 l=1
k m
= ( ∑ π(Yj ))(∑ π(Zl ))
j=1 l=1
k m
= π( ∑ Yj )π(∑ Zl )
j=1 l=1
= π(f )π(g),
which is what we wanted to prove.

1.1. Symmetric Polynomials 5
(iv) By definition we have

(πσ)(f ) = f (x(πσ)(1) , . . . , x(πσ)(n) )
= f (xπ(σ(1)) , . . . , xπ(σ(n)) )
= π(f (xσ(1) , . . . , xσ(n) ))
= π(σ(f )),
which is what we wanted to prove. □
After computing π(f ) for various permutations π and polynomi-

als f , we eventually notice f and π(f ) are usually different, but we
sometimes get π(f ) = f . When π(f ) = f , we say f is invariant under
π. If f is invariant under π, then we might also say f is fixed by π,
or π fixes f .
Example 1.3. Find all permutations σ ∈ S3 which fix x1 x22 x3 +
x21 x2 x3 .
Solution. Every polynomial is invariant under the identity permuta-

tion. If we check the other five permutations in S3 , then we find
x1 x22 x3 + x21 x2 x3 is also invariant under (12), but not under any other
permutation. □
Every permutation has its own set of invariant polynomials, but

some polynomials are invariant under every permutation. These are
the polynomials we plan to study.
Definition 1.4. We say a polynomial f (Xn ) is a symmetric poly-
nomial in x1 , . . . , xn whenever π(f ) = f for all π ∈ Sn . We write
Λ(Xn ) to denote the set of all symmetric polynomials in x1 , . . . , xn
with coefficients in Q.
We leave it as an exercise to use Proposition 1.2 to show Λ(Xn )

is a vector space over Q for all n, and that in fact Λ(Xn ) is infinite
dimensional for all n ≥ 1.
Example 1.5. Which of the following are symmetric polynomials in
x1 , x2 , x3 ? Which are symmetric polynomials in x1 , x2 , x3 , x4 ?
(a) 3x1 x2 x3 + x1 x2 + x2 x3 + x1 x3 + 5.
(b) x21 x2 + x22 x3 + x1 x23 .
Solution. (a) This polynomial is a symmetric polynomial in x1 , x2 , x3 ,

but it is not invariant under the permutation (14), so it is not a
symmetric polynomial in x1 , x2 , x3 , x4 .
(b) The image of this polynomial under the permutation (12) is
x22 x1 + x21 x3 + x23 x2 , which is not the original polynomial, so it is not
a symmetric polynomial in either x1 , x2 , x3 or x1 , x2 , x3 , x4 . □
When we apply a permutation π to a polynomial f , it may alter

the terms of f dramatically. But some properties will always be
preserved. For example, no matter how much a term is changed by
π, its total degree remains the same. As a result, it is useful to break
our symmetric polynomials into pieces in which every term has the
same total degree.
Definition 1.6. We say a polynomial f (Xn ) in which every term has

total degree exactly k is homogeneous of degree k. We write Λk (Xn )
to denote the set of all symmetric polynomials in x1 , . . . , xn which are
homogeneous of degree k. Note that for every nonnegative integer k,
every term in the polynomial 0 has degree k (since 0 has no terms),
so 0 ∈ Λk (Xn ) for all k ≥ 0 and all n ≥ 1.
Note that if f is any polynomial in x1 , . . . , xn , then we can group

the terms of f by their total degree, so f can be written uniquely
as a sum f0 + f1 + ⋅ ⋅ ⋅, where fk is homogeneous of degree k for each
k ≥ 0. If f happens to be a symmetric polynomial in x1 , . . . , xn , then
for every π ∈ Sn , we have
f0 + f1 + ⋅ ⋅ ⋅ = f
= π(f )
= π(f0 + f1 + ⋅ ⋅ ⋅)
= π(f0 ) + π(f1 ) + ⋅ ⋅ ⋅ .
Since π does not change the total degree of any monomial, we must
have π(fj ) = fj for every j ≥ 0. That is, if f is a symmetric polynomial
in x1 , . . . , xn and f = f0 + f1 + ⋅ ⋅ ⋅ is the decomposition of f into its
homogeneous parts fj , then each fj is also a symmetric polynomial
in x1 , . . . , xn . With this in mind, we often restrict our attention to
the homogeneous symmetric polynomials of a given total degree.
1.2. The Monomial Symmetric Polynomials 7
Example 1.7. We saw in Example 1.5 that 3x1 x2 x3 + x1 x2 + x2 x3 +

x1 x3 + 5 ∈ Λ(X3 ). Write this symmetric polynomial as a sum f0 + f1 +
⋅ ⋅ ⋅, where fk ∈ Λk (X3 ) for all k ≥ 0.
Solution. In general fk is the sum of all of the terms of total degree

k, so in this case we have f0 = 5, f1 = 0, f2 = x1 x2 + x2 x3 + x1 x3 ,
f3 = 3x1 x2 x3 , and fk = 0 for k ≥ 4. □
As we did for Λ(Xn ), we leave it as an exercise to show Λk (Xn )

is a vector space over Q for all k ≥ 0 and all n ≥ 1. In contrast with
Λ(Xn ), we will see the space Λk (Xn ) is finite dimensional.
1.2. The Monomial Symmetric Polynomials

So far we have seen exactly one symmetric polynomial, so it is nat-
ural to ask for more examples. One simple way to construct more
symmetric polynomials is to be even more demanding: pick a set of
variables x1 , . . . , xn , pick a monomial in those variables, and try to
construct a symmetric polynomial that includes your monomial as a
term.
Example 1.8. Find a symmetric polynomial f ∈ Λ(X3 ) that includes
x31 x2 as one of its terms. Similarly, find a symmetric polynomial
g ∈ Λ(X3 ) that includes 3x21 x2 x23 as one of its terms. Use as few other
monomials as possible in both f and g.
Solution. If f is a symmetric polynomial, then it must be invariant

under (12). So if one of its terms is x31 x2 , then another of its terms
must be (12)(x31 x2 ) = x1 x32 . Similarly, f must also include the terms
x2 x33 , x31 x3 , x1 x33 , and x32 x3 . On the other hand, x31 x2 + x1 x32 + x2 x33 +
x31 x3 + x1 x33 + x32 x3 is a symmetric polynomial, so it must be the one
we are looking for.
When we reason in the same way for g as we did for f , we find
we get some duplicate terms. Since we do not need these duplicates,
we find g = 3x21 x2 x23 + 3x1 x22 x23 + 3x21 x22 x3 . □
When we look more closely at our work in Example 1.8, we see we

can construct a symmetric polynomial in x1 , . . . , xn by starting with
a monomial in these variables, and then adding the distinct images of
this monomial under the permutations in Sn . If one monomial is the

image of another under some permutation, then both monomials will
give us the same symmetric polynomial under this construction. This
means we can rearrange the factors in our monomial to ensure that
when we list the variables in the order x1 , . . . , xn , their exponents
form a partition. (See Appendix B for background on partitions.)
With this in mind, we set some notation and terminology for the
symmetric polynomials we’ve found.
Definition 1.9. Suppose n ≥ 1 and λ is a partition. Then the mono-

mial symmetric polynomial mλ (Xn ) indexed by λ is the sum of the
l(λ) λ
monomial ∏j=1 xj j and all of its distinct images under the elements of
Sn . Here we take xj = 0 for all j > n, so if l(λ) > n, then mλ (Xn ) = 0.
We will often have partitions as subscripts, as we do for the mono-

mial symmetric polynomials. When all of the parts of these partitions
are less than 10, we will save some space by omitting the commas
and parentheses. So, for example, we will write m4431 (Xn ) instead of
m(4,4,3,1) (Xn ).
Example 1.10. Compute the four monomial symmetric polynomials

m21 (X2 ), m21 (X3 ), m3311 (X3 ), and m3311 (X4 ).
Solution. The monomial x21 x2 has only one image (other than itself)
under the elements of S2 , namely x1 x22 . Therefore,
m21 (X2 ) = x21 x2 + x1 x22 .
Notice that if we hold λ constant and increase the number of variables,
then we get more images of x21 x2 , and therefore a new monomial
symmetric polynomial, namely
m21 (X3 ) = x21 x2 + x21 x3 + x1 x22 + x1 x23 + x22 x3 + x2 x23 .
The length of the partition (32 , 12 ) is greater than three, so m3311 (X3 )
is 0. The requirement that we take only distinct images of our mono-
mial comes into play when we compute m3311 (X4 ), which turns out
to be
x31 x32 x3 x4 + x31 x2 x33 x4 + x31 x2 x3 x34
+ x1 x32 x33 x4 + x1 x32 x3 x34 + x1 x2 x33 x34 . □
1.2. The Monomial Symmetric Polynomials 9
We saw in our solution to Example 1.10 that the polynomials

m21 (X2 ), m21 (X3 ), m3311 (X3 ), and m3311 (X4 ) are all symmetric
polynomials in their respective variables, and it may already be clear
that mλ (Xn ) is a symmetric polynomial in x1 , . . . , xn for all n and
all λ. Nevertheless, the following proof of this fact uses ideas and
techniques which will be useful later on, so we include it here.
Proposition 1.11. For all positive integers n and all partitions λ,

the polynomial mλ (Xn ) is a symmetric polynomial in x1 , . . . , xn .
Proof. When n < l(λ), we have mλ (Xn ) = 0, in which case the result
is clear, so we assume n ≥ l(λ).
In view of Proposition C.5, it is sufficient to show σj (mλ (Xn )) =
mλ (Xn ) for all j with 1 ≤ j ≤ n − 1, where σj is the adjacent trans-
position (j, j + 1). To do this, it is sufficient to show that every
term xµ1 1 ⋅ ⋅ ⋅ xµnn has the same coefficient in σj (mλ (Xn )) as it has in
mλ (Xn ). Note that the coefficient of xµ1 1 ⋅ ⋅ ⋅ xµnn in mλ (Xn ) is 1 if
µ1 , . . . , µn is a reordering of λ1 , . . . , λn and 0 otherwise. On the other
hand, the coefficient of xµ1 1 ⋅ ⋅ ⋅ xµnn in σj (mλ (Xn )) is the coefficient
µ µj
of xµ1 1 ⋅ ⋅ ⋅ xj j+1 xj+1 ⋅ ⋅ ⋅ xµnn in mλ (Xn ). Furthermore, this coefficient
is 1 if µ1 , . . . , µj+1 , µj , . . . , µn is a reordering of λ1 , . . . , λn and 0 oth-
erwise. But µ1 , . . . , µn is a reordering of λ1 , . . . , λn if and only if
µ1 , . . . , µj+1 , µj , . . . , µn is a reordering of λ1 , . . . , λn , and the result
follows. □
Note that if λ ⊢ k, then the monomial symmetric polynomial

mλ (Xn ) is homogeneous of degree k, so mλ (Xn ) ∈ Λk (Xn ). In
fact, immediately after Example 1.7 we promised we would see that
Λk (Xn ) is finite dimensional. We conclude this section by keeping
that promise, showing that if n ≥ k, then the monomial symmetric
polynomials form a basis for this space.
Proposition 1.12. If n ≥ 1, k ≥ 0, and n ≥ k, then the set
{mλ (Xn ) ∣ λ ⊢ k}
of monomial symmetric polynomials is a basis for Λk (Xn ). In par-

ticular, dim Λk (Xn ) = p(k), the number of partitions of k.
Proof. First observe that for any monomial xα 1 ⋅ ⋅ ⋅ xn , there is only

1 αn
one partition λ whose parts are a rearrangement of α1 , . . . , αn , and

xα1 ⋅ ⋅ ⋅ xn is a term in mλ (Xn ). In particular, if λ and µ are parti-
1 αn
tions with λ ≠ µ, then mλ (Xn ) and mµ (Xn ) have no terms in com-

mon. Therefore, since each mλ (Xn ) is nonzero, ∑λ⊢k aλ mλ (Xn ) = 0
can only occur if aλ = 0 for all λ ⊢ k. In other words, {mλ (Xn ) ∣ λ ⊢ k}
is linearly independent.
To see that this set also spans Λk (Xn ), suppose f ∈ Λk (Xn ). If
f = 0, then f = ∑λ⊢k 0 ⋅ mλ (Xn ), so suppose f ≠ 0. We argue by
induction on the number of terms in f . Since f is symmetric, f has
µ
a term of the form α ∏nj=1 xj j for some µ ⊢ k and some constant
α. Moreover, by the symmetry of f all of the distinct images of
this monomial under the action of Sn also appear in f . Therefore
f − αmµ (Xn ) ∈ Λk (Xn ) and f − αmµ (Xn ) has fewer terms than f .
By induction f − αmµ (Xn ) is a linear combination of the elements of
{mλ (Xn ) ∣ λ ⊢ k}, and thus f is as well.
The fact that dim Λk (Xn ) = p(k), the number of partitions of k,
now follows from the fact that there is exactly one monomial sym-
metric polynomial in Λk (Xn ) for each partition of k. □
We note that Proposition 1.12 also tells us the monomial sym-

metric polynomials form a basis for Λ(Xn ).
1.3. Symmetric Functions

In Proposition 1.12 we saw that if n ≥ k, then the dimension of
Λk (Xn ) is independent of n. This fact is an example of a more general
phenomenon: if we have enough variables, then the algebraic prop-
erties of Λk (Xn ) do not depend on exactly how many variables we
have.
To see another illustration of this central general principle, sup-
pose n ≥ 2 and consider the product m11 (Xn )m21 (Xn ). When n = 2,
we have
m11 (X2 )m21 (X2 ) = x31 x22 + x21 x32 .

1.3. Symmetric Functions 11
Table 1.1. The product m11 (Xn )m21 (Xn ) for 2 ≤ n ≤ 7, as

a linear combination of monomial symmetric polynomials
n m11 (Xn )m21 (Xn )

2 m32
3 m32 + 2m311 + 2m221
4 m32 + 2m311 + 2m221 + 3m2111
5 m32 + 2m311 + 2m221 + 3m2111
6 m32 + 2m311 + 2m221 + 3m2111
7 m32 + 2m311 + 2m221 + 3m2111
When n = 3 we have
m11 (X3 )m21 (X3 ) = x31 x22 + x21 x32 + x31 x23 + x21 x33 + x32 x23 + x22 x33
+ 2x3 x2 x3 + 2x1 x32 x3 + 2x1 x2 x33
+ 2x21 x22 x3 + 2x21 x2 x23 + 2x1 x22 x23 .
Our products are all homogeneous symmetric polynomials of de-

gree 5, so we can write them as linear combinations of the monomial
symmetric polynomials. When we do this for n = 2 and n = 3, we find
m11 (X2 )m21 (X2 ) = m32 (X2 )
and
m11 (X3 )m21 (X3 ) = m32 (X3 ) + 2m311 (X3 ) + 2m221 (X3 ).
In Table 1.1 we write m11 (Xn )m21 (Xn ) as a linear combination

of the monomial symmetric polynomials, suppressing the arguments
Xn in the answers. It appears from these data that if n ≥ 4, then
m11 (Xn )m21 (Xn ) = m32 (Xn ) + 2m311 (Xn ) + 2m221 + 3m2111 (Xn ). In
addition, it seems that if n < 4, then we have the same sum, except
we remove those monomial symmetric polynomials whose indexing
partition has more than n parts. We can prove this happens for
any product mλ (Xn )mµ (Xn ) by using a simple relationship between
mλ (x1 , . . . , xn ) and mλ (x1 , . . . , xn , 0).
Proposition 1.13. For any partition λ if n ≥ l(λ), then

mλ (x1 , . . . , xn , 0) = mλ (x1 , . . . , xn ).
More generally, if n ≥ l(λ), then

mλ (x1 , . . . , xn , 0, . . . , 0) = mλ (x1 , . . . , xn )
´¹¹ ¹ ¹ ¹ ¹¸ ¹ ¹ ¹ ¹ ¹¶
j
for any j ≥ 1.
Proof. Since n ≥ l(λ), both of the polynomials mλ (x1 , . . . , xn , 0) and

mλ (x1 , . . . , xn ) include the term xλ1 1 ⋅ ⋅ ⋅ xλl l , where l = l(λ). In par-
ticular, neither mλ (x1 , . . . , xn , 0) nor mλ (x1 , . . . , xn ) is 0. In fact,
the terms in mλ (x1 , . . . , xn , 0) are exactly those for which the ex-
ponents of x1 , . . . , xn are some permutation of λ1 , . . . , λl and n − l
zeros. These are exactly the terms in mλ (x1 , . . . , xn ) as well, so
mλ (x1 , . . . , xn , 0) = mλ (x1 , . . . , xn ).
The fact that mλ (x1 , . . . , xn , 0, . . . , 0) = mλ (x1 , . . . , xn ) for any
´¹¹ ¹ ¹ ¹ ¹¸ ¹ ¹ ¹ ¹ ¹¶
j
j ≥ 1 follows by induction on j, since setting xn+j = 0 and then
setting xn+j−1 = 0 in the resulting polynomial gives us the same result
as setting xn+j = 0 and xn+j−1 = 0 all at once. □
Now suppose we have n ≥ 1 and partitions λ and µ. In Problem

1.15 you will have a chance to show that for any n ≥ 1, the product
mλ (Xn )mµ (Xn ) is in Λ(Xn ). Since the monomial symmetric poly-
nomials are a basis for Λ(Xn ), we know there are rational numbers
aνλ,µ (n) such that
(1.1) mλ (Xn )mµ (Xn ) = ∑ aνλ,µ (n)mν (Xn ).

ν
In principle ν could be any partition, but all of the terms in mλ (Xn )

have total degree ∣λ∣, and all of the terms in mµ (Xn ) have total degree
∣µ∣, so all of the terms in their product have degree ∣λ∣+∣µ∣. This means
we can assume ν is a partition of ∣λ∣ + ∣µ∣.
We would like to say aνλ,µ (n) is independent of n for all n ≥ 1,
but this is not quite true. In particular, if n = l(λ) + l(µ), then
λl(λ)
one term in mλ (Xn ) is xλ1 1 ⋅ ⋅ ⋅ xl(λ) and one term in mµ (Xn ) is
µ
xµl(λ)+1
1
⋅ ⋅ ⋅ xl(λ)+l(µ)
l(µ)
. This means one term in their product is
λ l(λ) µ1 µ
xλ1 1 ⋅ ⋅ ⋅ xl(λ) xl(λ)+1 ⋅ ⋅ ⋅ xl(λ)+l(µ)
l(µ)
.
Therefore, if we write λ ∪ µ to denote the partition we obtain by

sorting λ1 , . . . , λl(λ) , µ1 , . . . , µl(µ) into weakly decreasing order, then
we see aλ∪µ λ∪µ
λ,µ (l(λ) + l(µ)) ≠ 0, even though aλ,µ (l(λ) + l(µ) − 1) = 0. In
other words, the best we can hope for is that aνλ,µ (n) is independent
of n for all n ≥ l(λ) + l(µ).
Corollary 1.14. For any partitions λ, µ, and ν ⊢ ∣λ∣ + ∣µ∣ and any
n ≥ 1, let the numbers aνλ,µ (n) be defined by
mλ (Xn )mµ (Xn ) = ∑ aνλ,µ (n)mν (Xn ).

ν
If n ≥ l(λ) + l(µ), then aνλ,µ (n) does not depend on n.
Proof. Suppose n = l(λ) + l(µ) and fix j ≥ 1. By definition,
mλ (Xn+j )mµ (Xn+j ) = ∑ aνλ,µ (n + j)mν (Xn+j ).

ν
If we set xn+1 = xn+2 = ⋅ ⋅ ⋅ = xn+j = 0 and use Proposition 1.13, then

we find
mλ (Xn )mµ (Xn ) = ∑ aνλ,µ (n + j)mν (Xn ).
ν
Comparing this last line with equation (1.1) and using the fact that
the monomial symmetric polynomials are a basis for Λ(Xn ), we see
aνλ,µ (n + j) = aνλ,µ (n), which is what we wanted to prove. □
As we mentioned above, the moral of Corollary 1.14 is that if

we have enough variables (that is, if n is large enough), then the al-
gebraic properties of Λk (Xn ) and our basis of monomial symmetric
polynomials do not depend on exactly how many variables we have.
However, we also saw that “enough” means different things in differ-
ent contexts. If we only cared about the product m11 (Xn )m21 (Xn ),
then enough would be four. But if we are actually interested in the
product m4321 (Xn )m77221 (Xn ), then enough is nine.
Instead of worrying about how much is enough in every new situ-
ation, we would like to just assume we have infinitely many variables.
To do this, we need to lay some formal groundwork involving “poly-
nomials” which are allowed to have infinitely many terms. We start
with a formal definition.
Definition 1.15. Let N be the set of nonnegative integers and set

N∞ = N × N × N × ⋅ ⋅ ⋅. A formal power series with coefficients in Q is a
function f ∶ N∞ → Q such that if f (a1 , a2 , . . .) ≠ 0, then only finitely
many of a1 , a2 , . . . are nonzero.
At first glance a formal power series does not seem to be related to

a polynomial which may have infinitely many terms. But we connect
these ideas by identifying each tuple (a1 , a2 , . . .) with the monomial
xa1 1 xa2 2 ⋅ ⋅ ⋅, viewing f (a1 , a2 , . . .) as the coefficient of xa1 1 xa2 2 ⋅ ⋅ ⋅, and
identifying the function f with the sum
(1.2) ∑ f (a1 , a2 , . . .)xa1 1 xa2 2 ⋅ ⋅ ⋅ .
(a1 ,a2 ,...)∈N∞
For example, if f is given by

⎧
⎪
⎪a1 if a2 = a3 = ⋅ ⋅ ⋅ = 0,
f (a1 , a2 , . . .) = ⎨
⎪
⎪
⎩0 otherwise,
then we identify f with the series x1 + 2x21 + 3x31 + ⋅ ⋅ ⋅ = ∑∞ j

j=1 jx1 . For
a formal power series f , the sum in (1.2) is “formal” in the sense that
we generally do not assign it meaning in the usual sense of addition.
A formal power series can have finitely many terms or infinitely
many terms, so every polynomial is also a formal power series. In
fact, formal power series also inherit some terminology from polyno-
mials. For instance, the total degree (or degree, for short) of a term
f (a1 , a2 , . . .)xa1 1 xa2 2 ⋅ ⋅ ⋅ is a1 + a2 ⋅ ⋅ ⋅, which is finite by the last condi-
tion in Definition 1.15. We say a formal power series is homogeneous
of degree k whenever each of its terms has total degree k, and we note
that every formal power series f can be written uniquely as a sum
f = f0 + f1 + ⋅ ⋅ ⋅, where fk is homogeneous of degree k for all k ≥ 0.
In some contexts it is useful to consider convergence properties of
formal power series, since this opens up the possibility of using tools
from complex analysis to draw conclusions about the coefficients in
a given series. However, we will not be concerned with questions of
convergence. Instead, it is the algebra of formal power series which
will be of most interest to us. For instance, we can add formal power
series and multiply them by scalars, just as we do for polynomials, so
the set of formal power series in X is a vector space over Q. In fact,
as the next two examples suggest, we can also multiply formal power
series.
Example 1.16. Compute the product of the formal power series

f (x) = ∑∞ ∞
j=0 x and g(x) = ∑j=0 jx .
j j
Solution. To express this product as a formal power series, we need

to determine, for each j, the coefficient of xj in the product
(1 + x + x2 + x3 + ⋅ ⋅ ⋅)(x + 2x2 + 3x3 + ⋅ ⋅ ⋅).
Since every term in the second factor has a factor of x, the constant
term in f (x)g(x) will be 0. Before we combine like terms, the terms
in f (x)g(x) are exactly the products of one term from f (x) and one
term from g(x). The only way such a term can have the form ax
is if we choose 1 from f (x) and x from g(x), so the coefficient of x
is 1. There are two ways such a term can have the form ax2 : we
can choose 1 from f (x) and 2x2 from g(x), or we can choose x from
f (x) and x from g(x). Therefore the coefficient of x2 in f (x)g(x) is
2 + 1 = 3. In general, there are j ways a term in f (x)g(x) can have
the form axj : for each m with 1 ≤ m ≤ j, we can choose xm−1 from
f (x) and (j − m + 1)xj−m+1 from g(x). Therefore the coefficient of xj
in f (x)g(x) is 1 + 2 + ⋅ ⋅ ⋅ + j = (j+1
2
), and we have
∞
j+1 j
f (x)g(x) = ∑ ( )x . □
j=1 2
Example 1.17. Compute the product of the formal power series

f = ∑∞ ∞
j=1 xj xj+1 and g = ∑j=1 xj .
Solution. Working formally, we could write f g as

∞ ∞
f g = ∑ ∑ xj xj+1 xm .
j=1 m=1
Although this expresses f g correctly, some terms appear more than

once on the right side, so it gives us less insight into f g as a for-
mal power series than we would like. To gain more insight, we need
to determine the coefficient of every monomial in x1 , x2 , . . . in the
product
(x1 x2 + x2 x3 + x3 x4 + ⋅ ⋅ ⋅)(x1 + x2 + x3 + ⋅ ⋅ ⋅).
Before we combine like terms, the terms in f g are exactly the products
of one term from f and one term from g. As a result, the only terms
in f g with nonzero coefficients are those of the form xm xj xj+1 , where
m < j − 1 or m > j + 2, those of the form x2j xj+1 , those of the form
xj x2j+1 , and those of the form xj xj+1 xj+2 . We can only obtain terms
of the first form in one way: choose xj xj+1 from f and xm from g.
Therefore, each of these terms has coefficient 1. Similarly, we can
only obtain terms of the second and third forms in one way, so each
of these also has coefficient 1. But we can obtain a term of the form
xj xj+1 xj+2 in two ways: choose xj xj+1 from f and xj+2 from g, or
choose xj+1 xj+2 from f and xj from g. Therefore, each of these terms
has coefficient 2. Combining these observations, we can express f g as
∞ ∞ ∞
f g = ∑ ∑ xj xj+1 xm + 2 ∑ xj xj+1 xj+2 . □
j=1 m=1 j=1
m≠j−1
m≠j+2
In each of these examples we have a product in which each factor

has infinitely many terms. However, there are only finitely many ways
to obtain any particular monomial in the result, so it is still possible to
express each product as a formal power series. Moreover, this is true
in general. That is, if X is any set of variables, then any monomial
in the variables in X can be expressed as a product of monomials in
only finitely many ways. Therefore, for any formal power series f and
g in X, the product f g is a well-defined formal power series.
Now that we have our usual algebraic operations on formal power
series, we would like to extend the action of permutations on polyno-
mials to an action of permutations on formal power series. There are
several ways one could do this, but we will use one that is particularly
simple: if π ∈ Sn and we have variables x1 , x2 , . . . , then we will define
π(xj ) = xπ(j) as usual for 1 ≤ j ≤ n, and we will set π(xj ) = xj for
j > n. Now for any formal power series f = f (x1 , x2 , . . .) and any per-
mutation π ∈ Sn , we define π(f ) by π(f ) ∶= f (π(x1 ), π(x2 ), . . .). With
this definition, we have the following natural analogue of Proposition
1.2.
Proposition 1.18. Suppose X = {xj }∞ j=1 is a set of variables, f and
g are formal power series in X, c is a constant, and π, σ ∈ Sn are
permutations. Then
(i) π(cf ) = cπ(f );

(ii) π(f + g) = π(f ) + π(g);
(iii) π(f g) = π(f )π(g);
(iv) (πσ)(f ) = π(σ(f )).
Proof. This is similar to the proof of Proposition 1.2. □
Now that we know how a permutation acts on a formal power

series, we are ready to discuss symmetric functions.
Definition 1.19. Suppose X = {xj }∞ j=1 is a set of variables and f is
a formal power series in X. We say f is a symmetric function in X
whenever, for all n ≥ 1 and all π ∈ Sn , we have π(f ) = f . We write
Λ(X) to denote the set of all symmetric functions in X, and we write
Λk (X) to denote the set of all symmetric functions in X that are
homogeneous of degree k. Often X is clear from context, in which
case we say f is a symmetric function, we write Λ to denote the set
of all symmetric functions, and we write Λk to denote the set of all
symmetric functions which are homogeneous of degree k.
When we had finitely many variables, we constructed the mono-

mial symmetric functions by adding all of the distinct images of a
given monomial. We can do the same thing when we have infinitely
many variables, but this time we get a formal power series, which is
not a polynomial in general.
Definition 1.20. Suppose λ is a partition and X = {xj }∞ j=1 is a set
of variables. Then the monomial symmetric function in X, which we
write as mλ (X), is the sum of all monomials Y for which there exists
n ≥ l(λ) and a permutation π ∈ Sn such that π(Y ) = xλ1 1 xλ2 2 ⋅ ⋅ ⋅ xλnn .
When the set X of variables is clear from context, we often omit

it, writing mλ instead of mλ (X). Similarly, if we write mλ and no
set of variables has been indicated, then we will assume our set of
variables is X = {xj }∞
j=1 .
Example 1.21. Compute m21 . Is m21 = m2 m1 ?
Solution. By definition m21 is the sum of the images of x21 x2 under

all permutations of x1 , x2 , . . ., so it is the sum of all monomials of the
form x2j xk , where j and k are distinct. Formally, we have

∞ ∞
m21 = ∑ ∑ x2j xk .
j=1 k=1
k≠j
The product m2 m1 = (x21 + x22 + ⋅ ⋅ ⋅)(x1 + x2 + ⋅ ⋅ ⋅) includes terms of

the form x3j , so it is not equal to m21 . However, we do have
m21 = (x21 + x22 + ⋅ ⋅ ⋅)(x1 + x2 + ⋅ ⋅ ⋅) − (x31 + x32 + ⋅ ⋅ ⋅),
so m21 = m2 m1 − m3 . □
As we might expect, the monomial symmetric functions form a

basis for Λk , as their polynomial counterparts did for Λk (Xn ).
Proposition 1.22. For all k ≥ 0, the set {mλ ∣ λ ⊢ k} of monomial

symmetric functions is a basis for Λk . In particular, dim Λk = p(k),
the number of partitions of k.
Proof. To show {mλ ∣ λ ⊢ k} is linearly independent, first note that

if we set xj = 0 in mλ for all j > n, then we obtain the monomial
symmetric polynomial mλ (Xn ). Now suppose we have constants aλ
for which ∑λ⊢k aλ mλ = 0. If we set xj = 0 for all j > k, then we find
∑λ⊢k aλ mλ (Xk ) = 0. But {mλ (Xk ) ∣ λ ⊢ k} is linearly independent
by Proposition 1.12, so aλ = 0 for all λ, as desired.
To show {mλ ∣λ ⊢ k} spans Λk , suppose f ∈ Λk . We argue by
induction on the number of terms in f of the form axa1 1 xa2 2 ⋅ ⋅ ⋅, where
a is a nonzero constant and a1 ≥ a2 ≥ ⋅ ⋅ ⋅. Note that for any such
term the sequence a1 , a2 , . . . is a partition of k, so there are at most
p(k) such terms. If f has exactly one term of the given form, then
all of its images under any permutation are also terms in f , and we
have f = amλ , where λ = a1 , a2 , . . . . Now suppose f has more than
one term of the given form. For any such term axa1 1 xa2 2 ⋅ ⋅ ⋅, all of the
images of this term must also be terms in f . Therefore, f − amµ ∈ Λk ,
where µ = a1 , a2 , . . . . Moreover, f − amµ has fewer terms of the given
form than f , so by induction f − amµ is a linear combination of the
elements of {mλ ∣ λ ⊢ k}. Now the result follows. □
1.4. Problems 19
1.4. Problems
1.1. Find all permutations π ∈ S4 for which π(f ) = f , where f (X4 ) =
x1 x22 x44 + x22 x43 x4 + x41 x22 x3 .
1.2. Find all permutations π ∈ S4 for which π(f ) = f , where f (X4 ) =
x1 x2 x23 x24 + x1 x22 x3 x24 + x21 x2 x23 x4 + x21 x22 x3 x4 .
1.3. Find a polynomial f in x1 , x2 , x3 which has π(f ) = sgn(π)f for
all π ∈ S3 , and which has x71 x23 as one of its terms.
1.4. For any set P of polynomials in x1 , . . . , xn , let Sn (P ) be the set
of permutations π ∈ Sn such that π(f ) = f for all f ∈ P . Prove
that for every P , the following hold.
(a) Sn (P ) is nonempty.
(b) Sn (P ) is closed under multiplication of permutations. That
is, if π, σ ∈ Sn (P ), then πσ ∈ Sn (P ).
(c) Sn (P ) is closed under taking inverses. That is, if π ∈
Sn (P ), then π −1 ∈ Sn (P ).
1.5. Prove or disprove: if P is a set of polynomials in x1 , . . . , xn ,
then for any permutation π ∈ Sn and any σ ∈ Sn (P ), we have
πσπ −1 ∈ Sn (P ).
1.6. For any set T ⊆ Sn , let Fix(T ) be the set of polynomials f (Xn )
such that τ (f ) = f for all τ ∈ T . Prove that for every T , the
following hold.
(a) Fix(T ) is a subspace of the vector space of polynomials in
x1 , . . . , xn with coefficients in Q.
(b) Fix(T ) is closed under multiplication. That is, if f ∈
Fix(T ) and g ∈ Fix(T ), then f g ∈ Fix(T ).
1.7. Prove or disprove: if T ⊆ Sn , then for any polynomial f ∈ Fix(T )
and any polynomial g(Xn ), we have f g ∈ Fix(T ).
1.8. Prove, or disprove and salvage: for any n ≥ 1 and any set T ⊆ Sn ,
we have T = Sn (Fix(T )).
1.9. Prove, or disprove and salvage: for any n ≥ 1 and any set P of
polynomials in x1 , . . . , xn , we have P = Fix(Sn (P )).
1.10. For any n ≥ 1, any k ≥ 0, and any T ⊆ Sn , let Fixk (T ) be the
set of polynomials in Fix(T ) which are homogeneous of degree
k. Show Fixk (T ) is a subspace of the space of all homogeneous

polynomials of degree k in x1 , . . . , xn with coefficients in Q.
1.11. For n ≥ 1, let Cn be the set containing just the permutation
(1, 2, . . . , n). Find and prove a formula for dim Fix2 (Cn ).
1.12. Show that the space Λ(Xn ) is infinite dimensional for all n ≥ 1.
1.13. Show that for all n, k ≥ 0, the space Λk (Xn ) is a subspace of
Λ(Xn ).
1.14. We showed in Proposition 1.12 that if n ≥ k, then Λk (Xn )
is finite-dimensional. Show Λk (Xn ) is also finite dimensional
when 0 ≤ k < n. More specifically, show dim Λk (Xn ) is the
number of partitions of k with at most n parts by finding a
basis whose elements are indexed by these partitions.
1.15. Show that for all n, k1 , k2 ≥ 0, if f ∈ Λk1 (Xn ) and g ∈ Λk2 (Xn ),
then f g ∈ Λk1 +k2 (Xn ).
1.16. Suppose n ≥ 1, k1 ≥ k2 ≥ 0, and k1 + k2 = n. Write the product
mk1 mk2 as a linear combination of {mλ ∣ λ ⊢ n}.
1.17. Write the product m4 m3 m2 m1 as a linear combination of
{mλ ∣ λ ⊢ 10}.
1.18. If we write the product m6 m5 m4 m3 m2 m1 as a linear combina-
tion of {mλ ∣ λ ⊢ 21}, what is the coefficient of m(12,9) ?
1.19. Suppose n ≥ 1 and k1 + k2 = n. Write the product m1k1 mk2 as
a linear combination of {mλ ∣ λ ⊢ n}.
1.20. If we write the product m1k1 +1 m1k2 +1 as a linear combination
of monomial symmetric functions, what is the coefficient of
m2,1k1 +k2 ?
1.21. Write ∏∞ ∞
j=1 (1 + xj ) and ∏j=1 (1 − xj ) in terms of the monomial
symmetric functions.
1.22. Write the sum
∑ mλ
λ
as an infinite product in as simple a form as possible. Here the

sum is over all partitions.
1.5. Notes 21
1.5. Notes
The background on formal power series we have developed here will
be enough for our work with symmetric functions. However, more
information is available in a variety of sources, including [Loe17, Ch.
7], [Niv69], and [Wil05, Ch. 2].
Chapter 2
The Elementary,
Complete Homogeneous,
and Power Sum
Symmetric Functions
We saw in Chapter 1 how the monomial symmetric functions arise

naturally when we group the terms of a symmetric function with
their images under the various permutations or, equivalently, when
we group terms according to their multisets of exponents. However,
several other families of symmetric functions appear in other contexts,
and these families give us other bases for Λk . In this chapter we
introduce the oldest of these families.
2.1. The Elementary Symmetric Functions

Suppose we have a polynomial f (z) whose leading coefficient is 1
(that is, f is monic), and whose roots are x1 , . . . , xn . Then one way
to factor f is as
f (z) = (z − x1 )(z − x2 ) ⋅ ⋅ ⋅ (z − xn ).
Since multiplication is commutative, we can permute these factors in

any way we like, and their product will still be f . As a result, f is
invariant under every permutation of x1 , . . . , xn . In particular, if we
23
24 2. The Elementary, Complete, and Power Sum Bases
Table 2.1. The coefficients of z k in Example 2.1
k coefficient of z k
0 x1 x2 x3 x4
1 −(x1 x2 x3 + x1 x2 x4 + x1 x3 x4 + x2 x3 x4 )
2 x1 x2 + x1 x3 + x1 x4 + x2 x3 + x2 x4 + x3 x4
3 −(x1 + x2 + x3 + x4 )
expand f in powers of z, then the coefficient of z k will be a symmetric

polynomial in x1 , . . . , xn .
Example 2.1. For 0 ≤ k ≤ 3, compute the coefficients of z k in the
polynomial (z − x1 )(z − x2 )(z − x3 )(z − x4 ).
Solution. When we expand the product and collect like powers of z,

we find
f (z) = x1 x2 x3 x4 − (x1 x2 x3 + x1 x2 x4 + x1 x3 x4 + x2 x3 x4 )z
+ (x1 x2 + x1 x3 + x1 x4 + x2 x3 + x2 x4 + x3 x4 )z 2
− (x1 + x2 + x3 + x4 )z 3 + z 4 .
Therefore, the coefficients of z k are as in Table 2.1. □
Notice in Example 2.1 that the coefficient of z k is a homogeneous

symmetric polynomial of degree 4 − k. Equivalently, the coefficient
of z 4−k is a homogeneous symmetric polynomial of degree k. In fact,
the coefficient of z 4−k is the sum of all products of exactly k distinct
xj ’s.
In general, each term in the expansion of (z − x1 ) ⋅ ⋅ ⋅ (z − xn ) is
a product of terms, one from each factor z − xj . Therefore, for any
k with 0 ≤ k ≤ n, the coefficient of z n−k is the sum of all products of
exactly k distinct xj ’s, up to a sign. This is the elementary symmetric
polynomial of degree k in x1 , . . . , xn .
Definition 2.2. For all positive integers n and k, the elementary
symmetric polynomial ek (Xn ) is given by
k
(2.1) ek (Xn ) = ∑ ∏ xj m = ∑ ∏ xj ,
1≤j1 <⋅ ⋅ ⋅<jk ≤n m=1 J⊆[n] j∈J
∣J∣=k
2.1. The Elementary Symmetric Functions 25
while the elementary symmetric function ek is given by
e k = ∑ ∏ xj .
J⊆P j∈J
∣J∣=k
(Here P is the set of positive integers.) By convention e0 (Xn ) = 1 and

e0 = 1, and if n < k, then ek (Xn ) = 0.
We note that (−1)k ek (Xn ) is the coefficient of z n−k in the poly-

nomial (z − x1 ) ⋅ ⋅ ⋅ (z − xn ).
For each k ≥ 0 the space Λk includes ek , but it contains none
of the other elementary symmetric functions ej . We would like the
elementary symmetric functions to form a basis for Λk , but apparently
we will need a lot more of them. In particular, we need to define a
symmetric function eλ for every partition λ ⊢ k. Since products of
symmetric functions are also symmetric functions, we have a natural
way to do this.
Definition 2.3. Suppose n ≥ 1 and λ is a partition. Then the ele-

mentary symmetric polynomial indexed by λ, written eλ (Xn ), is given
by
l(λ)
(2.2) eλ (Xn ) = ∏ eλj (Xn ).
j=1
Similarly, the elementary symmetric function indexed by λ, written

eλ , is given by
l(λ)
eλ = ∏ eλj .
j=1
We note that if n < λ1 , then eλ (Xn ) = 0.

We see that if λ ⊢ k, then eλ ∈ Λk , so Λk contains exactly p(k)
elementary symmetric functions for all k ≥ 0. If they are distinct, then
we have exactly the right number of elementary symmetric functions
to form a basis for this space. To see these symmetric functions
actually do form a basis for Λk , we look at how they are written as
linear combinations of the monomial symmetric functions. Our first
example is especially simple, but it gives us a fact we will find useful
later on.
Example 2.4. For all n ≥ k ≥ 1, write ek (Xn ) as a linear combination

of monomial symmetric polynomials, and then write ek as a linear
combination of monomial symmetric functions.
Solution. Since the monomial symmetric polynomials span Λk (Xn ),

we know there are constants ak,µ with
ek = ∑ ak,µ mµ .
µ⊢k
If ak,µ ≠ 0 for a given partition µ = µ1 , . . . , µl , then the coefficient

of xµ1 1 ⋅ ⋅ ⋅ xµl l is also ak,µ , since mµ (Xn ) is the only monomial sym-
metric polynomial in which this term appears. The only term in
ek (Xn ) whose exponents form a partition is x1 ⋅ ⋅ ⋅ xk , and this term
has coefficient 1, so ek (Xn ) = m1k (Xn ). Similarly, ek = m1k . □
In our next example we explore what happens to our linear com-

binations when we add more variables.
Example 2.5. Write e21 (X3 ), e21 (X4 ), and e21 (X5 ) as linear com-
binations of monomial symmetric polynomials, and then write e21 as
a linear combination of monomial symmetric functions.
Solution. By the definition of eλ (X3 ) we have

e21 (X3 ) = (x1 x2 + x1 x3 + x2 x3 )(x1 + x2 + x3 )
= x21 x2 + x21 x3 + x1 x2 x3 + x1 x22
+ x1 x2 x3 + x22 x3 + x1 x2 x3 + x1 x23 + x2 x23
= 3m111 (X3 ) + m21 (X3 ).
To compute e21 (X4 ), we first note that
e21 (X4 ) = (x1 x2 + x1 x3 + x1 x4 + x2 x3 + x2 x4 + x3 x4 )(x1 + x2 + x3 + x4 ).
Instead of computing every term in this product, we observe that for
any partition µ ⊢ 3 the coefficient of mµ (X4 ) in our expansion will
be equal to the coefficient of xµ1 1 xµ2 2 xµ3 3 , since mµ (X4 ) is the only
monomial symmetric polynomial in x1 , x2 , x3 , x4 in which xµ1 1 xµ2 2 xµ3 3
appears. We computed these coefficients when we computed e21 (X3 ),
so we have
e21 (X4 ) = 3m111 (X4 ) + m21 (X4 ).
Similarly, we also have

e21 (X5 ) = 3m111 (X5 ) + m21 (X5 ).
In fact, additional variables will not change our analysis, so we even
have
e21 = 3m111 + m21 . □
In the course of our solutions to Examples 2.4 and 2.5, we made

a useful observation: for any partition µ, the polynomial mµ (Xn )
l(µ) µ
is the only monomial symmetric polynomial in which ∏j=1 xj j ap-
pears. Therefore, in any linear combination of monomial symmet-
ric functions, the coefficient of mµ (Xn ) is equal to the coefficient of
l(µ) µj
∏j=1 xj . This observation allows us to learn about how to write the
symmetric function eλ as a linear combination of the monomial sym-
metric functions by looking at how to write the symmetric polynomial
eλ (Xn ) as a linear combination of the monomial symmetric polyno-
mials. More specifically, for any partitions λ and µ with ∣λ∣ = ∣µ∣ and
for any n ≥ 1, let Mλ,µ,n (e, m) be the rational number defined by
(2.3) eλ (Xn ) = ∑ Mλ,µ,n (e, m)mµ (Xn ).
µ⊢∣λ∣
Similarly, for any partitions λ and µ with ∣λ∣ = ∣µ∣, let Mλ,µ (e, m) be
the rational number defined by
(2.4) eλ = ∑ Mλ,µ (e, m)mµ .
µ⊢∣λ∣
With this notation, we have the following result.
Proposition 2.6. For any partitions λ and µ with ∣λ∣ = ∣µ∣, if n ≥

∣λ∣, then Mλ,µ,n (e, m) = Mλ,µ (e, m). In particular, if n ≥ ∣λ∣, then
Mλ,µ,n (e, m) is independent of n.
Proof. For convenience set k = l(µ) and l = l(λ).

As we saw in our solutions to Examples 2.4 and 2.5, the coef-
ficient Mλ,µ (e, m) is the coefficient of xµ1 1 ⋅ ⋅ ⋅ xµk k in eλ . If n ≥ ∣λ∣,
then this coefficient is determined by those terms in eλ1 , . . . , eλl in-
volving only x1 , . . . , xk . Since k ≤ ∣λ∣ ≤ n, the variables x1 , . . . , xk are
among x1 , . . . , xn , and Mλ,µ (e, m) is determined by those terms in
eλ1 , . . . , eλl involving only x1 , . . . , xn . This is also how Mλ,µ,n (e, m)

is determined, so Mλ,µ (e, m) = Mλ,µ,n (e, m). □
Building on the theme of Section 1.3, Proposition 2.6 is telling

us that if we have enough variables, then the algebraic relationships
among the elementary symmetric polynomials and the monomial sym-
metric polynomials do not depend on exactly how many variables we
have. With this in mind, we will pay particular attention to the co-
efficients Mλ,µ (e, m). In our last example we start to reveal some of
the combinatorics of these coefficients.
Example 2.7. Write e421 as a linear combination of monomial sym-

metric functions.
Solution. By Proposition 2.6 we can assume we have just the variables

x1 , x2 , x3 , x4 , x5 , x6 , and x7 . But even with this reduction it is still
impractical to write out all (74)(72)(71) = 5145 terms in e421 (X7 ) and
then group them into monomial symmetric polynomials. Fortunately,
we know that for any µ ⊢ 7, the coefficient of mµ in e421 is equal to
µ
the coefficient of ∏7j=1 xj j . And we also know each term in e421 is a
product of one monomial from each of e4 , e2 , and e1 .
2 3 5 7
Figure 2.1. The filling corresponding to x2 x3 x5 x7
To make our work easier, we represent each term of ek with a

filling of a 1 × k tile with distinct positive integers, in increasing order
from left to right, corresponding to the subscripts of the factors in
that term. For example, the term x2 x3 x5 x7 corresponds to the filling
of a 1 × 4 tile in Figure 2.1. By stacking and left-justifying these
fillings, we can represent each term in e421 as a filling of the Ferrers
diagram of (4, 2, 1) with positive integers, in which the entries in each
row are strictly increasing from left to right. For example, the filling
in Figure 2.2 corresponds to the product of the term x2 x3 x5 x7 from
e4 , the term x1 x4 from e2 , and the term x2 from e1 ; this product is
the term x1 x22 x3 x4 x5 x7 .
2
1 4
2 3 5 7
Figure 2.2. The filling corresponding to the product of

x2 x3 x5 x7 , x1 x4 , and x2
We have now translated our algebraic problem into the combina-

torial problem of counting the fillings of the Ferrers diagram of (4, 2, 1)
with certain properties. But before we actually do this counting, we
note some conditions under which we will have no such fillings. First,
each row of our filling has at most one 1, so we can have at most three
1’s in total. Therefore, if µ1 > 3, then the coefficient of mµ in e421 is
0. Similarly, if a filling has three 1’s, then only two rows have space
for a 2. Therefore, if µ1 = 3 and µ2 > 2, then the coefficient of mµ in
e421 is also 0. Continuing in this way, we see the coefficient of mµ in
e421 is nonzero only if µ is (3, 2, 12 ), (3, 14 ), (23 , 1), (22 , 13 ), (2, 15 ), or
(17 ).
Now that we have ruled out a bunch of partitions for which the
coefficient of mµ must be 0, we turn our attention to finding the
coefficients of the remaining mµ ’s, by counting fillings of the Ferrers
diagram of (4, 2, 1). For example, when µ = (22 , 13 ), we need to count
fillings of the Ferrers diagram of (4, 2, 1) with exactly two 1’s, two 2’s,
one 3, one 4, and one 5, in which the entries in each row are strictly
increasing from left to right. In such a filling there are three ways to
place the 1’s, which we see in Figure 2.3. Reading from left to right,
we call these cases A, B, and C.
In case A there are three ways to place the two 2’s, which we see
in Figure 2.4. Reading from left to right, we call these cases A1, A2,
1 1
1 1
1 1
Figure 2.3. The three ways to place the two 1’s in the Ferrers
diagram of (4, 2, 1)
2 2
1 1 2 1 2
1 2 1 2 1
Figure 2.4. The three ways to place the 2’s in case A
and A3. In cases A1 and A2 we can choose one of 3, 4, and 5 to

occupy the box that is the only empty box in its row, and the other
two entries must go in the bottom row in increasing order. In case
A3, we must place 3, 4, and 5 in the bottom row in increasing order.
Therefore, there are three fillings in cases A1 and A2, and one in case
A4, for a total of seven fillings in case A.
In cases B and C there is only one way to place the 2’s, since
they must go in different rows. In case B there are three ways to
place 3, 4, and 5, and there is one way to place 3, 4, and 5 in case
C. Therefore, there are three fillings in case B and one in case C, for
a total of eleven in cases A, B, and C. This means the coefficient of
m22111 in e421 is 11.
Using a similar analysis for the other partitions µ, we find the
coefficients in Table 2.2. □
Our solution to Example 2.7 includes a couple of observations that

also hold in the general case. First, for a given λ, we can characterize
a whole family of partitions µ with ∣µ∣ = ∣λ∣ for which Mλ,µ (e, m) = 0.
To describe this family, it is convenient to introduce an ordering on
the set of partitions.
Table 2.2. The coefficients of mµ in e421
µ coefficient of mµ
(3, 2, 12 ) 1
(3, 14 ) 4
(23 , 1) 3
(22 , 13 ) 11
(2, 15 ) 35
(17 ) 105
Definition 2.8. Suppose λ and µ are partitions. We say λ is greater

than µ in lexicographic order, and we write λ >lex µ, whenever there
is a positive integer m such that λj = µj for j < m and λm > µm . Here
we take λj = 0 if j > l(λ) and we take µj = 0 if j > l(µ).
As we see in our next example, the lexicographic ordering is es-

sentially the natural alphabetical ordering for partitions.
Example 2.9. Write the partitions of 6 in lexicographic order, from
largest to smallest.
Solution. To make a partition large in lexicographic order, we need

to make the early parts as large as possible. Thus (6) will be the
largest partition, and (5, 1) will come next. Continuing in this way,
we have
(6) >lex (5, 1) >lex (4, 2) >lex (4, 12 ) >lex (32 ) >lex (3, 2, 1)
>lex (3, 13 ) >lex (23 ) >lex (22 , 12 ) >lex (2, 14 ) >lex (16 ). □
Our solution to Example 2.9 suggests the lexicographic ordering

is a linear ordering of the set of partitions of a given n. In fact, in
Problems 2.5 and 2.6 we will ask you to show >lex is a linear order of
the set of all partitions.
Our solution to Example 2.7, on the other hand, suggests using
>lex to compare µ with the conjugate partition λ′ to find a collection of
partitions µ with Mλ,µ (e, m) = 0. In particular, we have the following
result.
Proposition 2.10. Suppose λ, µ are partitions with ∣λ∣ = ∣µ∣. Then
(i) if µ >lex λ′ then Mλ,µ (e, m) = 0;
(ii) Mλ,λ′ (e, m) = 1.
Proof. (i) By (2.2) we have

(2.5) eλ = eλ1 ⋅ ⋅ ⋅ eλl(λ) ,
µ
and we note that Mλ,µ (e, m) is the coefficient of xµ1 1 ⋅ ⋅ ⋅ xl(µ)
l(µ)
in this
′
product. If µ >lex λ , then by definition there exists m ≥ 1 such that
µm > λ′m and µj = λ′j for 1 ≤ j < m. Each factor eλj can contribute at
most one factor x1 to our term, so µ1 = λ′1 implies each factor eλj in
(2.5) contributes exactly one factor x1 . (This corresponds to filling

the first column of the Ferrers diagram of λ with 1’s.) Similarly, only
those eλj with λj ≥ 2 can contribute a factor x2 , so each such eλj
must contribute exactly one factor x2 . (This corresponds to filling
the second column of the Ferrers diagram of λ with 2’s.) Proceeding
in this way, we see that only those eλj with λj ≥ m can contribute a
factor xm to our term, so the exponent on xm is at most λ′m . Since
µl(µ)
µm > λ′m , the term xµ1 1 ⋅ ⋅ ⋅ xl(µ) does not appear in our product, and
the result follows.
(ii) Arguing as in the proof of (i), we see the only way to produce
λ′ λ′ ′
the term x1 1 ⋅ ⋅ ⋅ xl(λ ′ ) is to choose the term x1 ⋅ ⋅ ⋅ xλj from the factor
l(λ )
eλj in (2.5) for all j. Now the result follows. □
The converse of Proposition 2.10(i) turns out to be false: there

are partitions λ and µ with ∣λ∣ = ∣µ∣ and µ >/ lex λ′ which still have
Mλ,µ (e, m) = 0. Indeed, in Problem 2.8 we will invite you to find two
such partitions. Remarkably, there is a weaker ordering on partitions
for which the analogue of Proposition 2.10(i) and its converse both
hold. You will get to explore this ordering in Problems 2.9, 2.10, 2.11,
2.12, and 2.13.
Although we will get even more information from our solution to
Example 2.7, Proposition 2.10 is already enough to show the elemen-
tary symmetric functions of degree k form a basis for Λk .
Corollary 2.11. The set {eλ ∣ λ ⊢ k} of elementary symmetric func-

tions is a basis for Λk .
Proof. Let A be the p(k) × p(k) matrix whose rows and columns
are indexed by the partitions of k, in lexicographic order from small-
est to largest, and whose entries are given by Aλµ = Mλ′ ,µ (e, m).
By Proposition 2.10, A is a lower triangular matrix whose diagonal
entries are all equal to 1, so det A = 1 and A is invertible. Since
eλ′ = ∑µ⊢k Aλµ mµ , each monomial symmetric function mµ is a lin-
ear combination of elementary symmetric functions, and {eλ ∣ λ ⊢ k}
spans Λk by Proposition 1.12. But dim Λk = p(k) = ∣{eλ ∣ λ ⊢ k}∣, so
{eλ ∣ λ ⊢ k} must also be linearly independent. Therefore {eλ ∣ λ ⊢ k}
is a basis, which is what we wanted to prove. □
(12 ) (2)
(1)
(2) 1 0
(1) [ 1 ] [ ]
(12 ) 2 1
(13 ) (2, 1) (3)

(3) ⎡⎢ 1 0 0 ⎤⎥
⎢ ⎥
(2, 1) ⎢⎢ 3 1 0 ⎥⎥
(13 ) ⎢⎣ 6 3 1 ⎥⎦
(14 ) (2, 12 ) (22 ) (3, 1) (4)

(4) ⎡ 1 0 ⎤⎥
⎢ 0 0 0
⎢ ⎥
(3, 1) ⎢⎢ 4 1 0 0 0 ⎥⎥
(2 ) ⎢⎢ 6
2
2 1 0 0 ⎥⎥
⎢ ⎥
(2, 12 ) ⎢ 12 6 2 1 0 ⎥
⎢ ⎥
(14 ) ⎢⎣ 24 12 6 4 1 ⎥⎦
(15 ) (2, 13 ) (22 , 1) (3, 12 ) (3, 2) (4, 1) (5)

(5) ⎡ 1 0 ⎤⎥
⎢ 0 0 0 0 0
⎢ ⎥
(4, 1) ⎢⎢ 5 1 0 0 0 0 0 ⎥⎥
(3, 2) ⎢⎢ 10 3 1 0 0 0 0 ⎥⎥
⎢ ⎥
(3, 12 ) ⎢⎢ 20 7 2 1 0 0 0 ⎥⎥
(22 , 1) ⎢⎢ 30 12 5 2 1 0 0 ⎥⎥
⎢ ⎥
(2, 13 ) ⎢ 60 27 12 7 3 1 0 ⎥
⎢ ⎥
(15 ) ⎢⎣ 120 60 30 20 10 5 1 ⎥⎦
Figure 2.5. The coefficients Mλ′ ,µ (e, m) for ∣λ∣ = ∣µ∣ ≤ 5
In Figure 2.5 we have the matrices Aλµ = Mλ′ ,µ (e, m) from the
proof of Corollary 2.11 for ∣λ∣ = ∣µ∣ ≤ 5.
Our solution to Example 2.7 also suggests a combinatorial in-
terpretation of Mλ,µ (e, m) involving fillings of the Ferrers diagram
of λ. As we show next, this interpretation holds in general, and we
can rephrase it in terms of matrices of 0’s and 1’s to get another
description of Mλ,µ (e, m).
Proposition 2.12. The following hold for all partitions λ, µ ⊢ k.

(i) Mλ,µ (e, m) is the number of fillings of the Ferrers diagram

of λ with positive integers for which the entries in each row
are strictly increasing from left to right, and each integer j
appears exactly µj times.
(ii) Mλ,µ (e, m) is the number of k × k matrices in which every
entry is 0 or 1, the sum of the entries in row m is µm for
all m, and the sum of the entries in column j is λj for all
j.
(iii) Mλ,µ (e, m) is the number of ways to place k balls, consisting
of µm identical balls of type m for each m, into l(λ) urns,
so that the jth urn contains exactly λj balls, no two of which
have the same type.
Proof. (i) As above, we first note that Mλ,µ (e, m) is the coefficient
l(µ) l(λ)
of the term ∏m=1 xµmm in eλ = ∏j=1 eλj . With this in mind, suppose
l(λ) l(µ)
that for each j we have a term tj from eλj and ∏j=1 tj = ∏m=1 xµmm .
Then we can construct a filling of the Ferrers diagram of λ of the given
type by placing, for 1 ≤ j ≤ l(λ), the subscripts of the variables which
appear in tj in increasing order across the jth row of the diagram.
We can also invert this process: if we have a filling of the given type,
then for each j with 1 ≤ j ≤ l(λ) we can reconstruct tj as the product
∏k xk , which is over all k which appear in the jth row of the filling.
l(µ)
Therefore, we have a bijection between terms of the form ∏m=1 xµmm
in the product eλ1 ⋅ ⋅ ⋅ eλl(λ) and our fillings of the Ferrers diagram of
λ. Now the result follows.
(ii) Given a filling of the Ferrers diagram of λ as in part (i), place
a 1 in the mth entry of the jth column of a k × k matrix A whenever
row j of the given filling contains an m, and let the remaining entries
of A all be 0. By construction the sum of the entries in the jth column
of A will be λj . And since m appears exactly µm times in our given
filling, the sum of the entries in the mth row of A will be µm . We
can also invert this construction: if we have a k × k matrix A of the
given type, then for each j with 1 ≤ j ≤ l(λ) the entries in the jth
row of the associated filling will be the numbers of the rows in which
the jth column of A contains a 1. Now we have a bijection between
fillings of the Ferrers diagram of λ of the type given in part (i) and
matrices of the type given in part (ii), and the result follows.
(iii) Given a filling of the Ferrers diagram of λ as in part (i),
place a ball of type m in urn j whenever m appears in row j. By
construction urn j will contain λj balls for each j. And since m
appears exactly µm times in our given filling, we will use exactly µm
balls of type m. We can also invert this construction: if we have an
urn filling of the given type, then for each j with 1 ≤ j ≤ l(λ) the
entries in the jth row of the associated filling will be the numbers on
the balls in urn j. Now we have a bijection between fillings of the
Ferrers diagram of λ of the type given in part (i) and urn fillings of
the type given in part (iii), and the result follows. □
1 5
2 3
1 2 4
Figure 2.6. The filling F in Example 2.13
Example 2.13. Let λ = (3, 2, 2), and let F be the filling of the Ferrers
diagram for λ given in Figure 2.6. Find the corresponding 7×7 matrix
and way of placing seven balls in three urns described in Proposition
2.12.
Solution. The first row of F contains 1, 2, and 4, so the first column of

the associated matrix has a 1 in its first, second, and fourth positions,
and 0’s elsewhere. After constructing the other columns in the same
way, we find the associated matrix is
⎛1 0 1 0 0 0 0⎞
⎜1 1 0 0 0 0 0⎟
⎜ ⎟
⎜0 0⎟
⎜ 1 0 0 0 0 ⎟
⎜ ⎟
⎜1 0 0 0 0 0 0⎟ .
⎜ ⎟
⎜0 0 1 0 0 0 0⎟
⎜ ⎟
⎜ ⎟
⎜0 0 0 0 0 0 0⎟
⎝0 0 0 0 0 0 0⎠
4
2 3 5
1 2 1
urn 1 urn 2 urn 3
Figure 2.7. The filling of urns in the solution to Example 2.13
Similarly, since the first row of F contains 1, 2, and 4, the first urn
in our placement of balls in urns will contain balls of types 1, 2, and
4. Continuing in the same way, we find the associated placement of
balls in urns is as in Figure 2.7. □
The matrices with entries Mλ,µ (e, m) in Figure 2.5 all have a nice
symmetry: if you reflect any of them over the diagonal from the lower
left corner to the upper right corner, then the matrix is unchanged.
We can use one of our interpretations of Mλ,µ (e, m) in Proposition
2.12 to give an easy proof that this holds in general.
Corollary 2.14. For all partitions λ, µ ⊢ k, we have

Mλ,µ (e, m) = Mµ,λ (e, m).
Proof. For any partitions λ, µ ⊢ k, let Bλ,µ be the set of k×k matrices
in which every entry is 0 or 1, the sum of the entries in row m is µm
for all m, and the sum of the entries in column j is λj for all j. By
Proposition 2.12(ii) we have ∣Bλ,µ ∣ = Mλ,µ (e, m). The result follows
from the fact that the transpose map is a bijection between Bλ,µ and
Bµ,λ . □
Corollary 2.11 implies every element of Λ is a linear combination

of elementary symmetric functions eλ for various partitions λ, and
this expression is unique. But if we also allow ourselves to multiply
symmetric functions, then we can construct every element of Λ using
only the elementary symmetric functions {en }∞ n=1 . Our next result
amounts to showing this expression, too, is unique, but we couch it
in slightly different (though traditional) terminology.
Definition 2.15. Suppose aj ∈ Λ for j ≥ 1. We say the set {aj }∞ j=1

is algebraically independent whenever there is no nonzero polynomial
p(y1 , . . . , yn ) such that p(a1 , . . . , an ) = 0.
There is much we could say about which sets are algebraically in-
dependent and which are not. For example, it is easy to construct a
set which is not algebraically independent by starting with a nonzero
symmetric function and then including some polynomial function of
that symmetric function. For instance, any set including both e2 and
e32 − 6e2 is not algebraically independent. But as our next example
shows, there are more complicated ways for a set to fail to be alge-
braically independent.
Example 2.16. Show that the symmetric functions
m1 , m11 + m2 , m21 + m3 , m31 + m4 , m41 + m5 , . . .
are not algebraically independent.
Solution. With some patience we can check that

m31 = e111 = 6m111 + 3m21 + m3
and
(m11 + m2 )m1 = 3m111 + 2m21 + m3 .
Therefore, if p(y1 , y2 , y3 ) = y3 − 2y2 y1 + y13 , then
p(m1 , m11 + m2 , m21 + m3 ) = 0. □
We now show what we suggested earlier: the elementary sym-

metric functions {en }∞
n=1 are a canonical example of an algebraically
independent set.
Theorem 2.17. The set {en }∞
n=1 is algebraically independent.
Proof. Suppose f (y1 , . . . , yn ) is a polynomial with f (e1 , . . . , en ) = 0;

we show f (y1 , . . . , yn ) = 0. Note that each monomial in f (e1 , . . . , en )
is an elementary symmetric function eλ for some partition λ, so
f (e1 , . . . , en ) is a linear combination of elementary symmetric func-
tions. By Corollary 2.11 the elementary symmetric functions are lin-
early independent, so all of the coefficients in f (y1 , . . . , yn ) are equal
to 0. This implies f (y1 , . . . , yn ) = 0, as we claimed. □
Theorem 2.17 essentially says the elementary symmetric functions

{en }∞n=1 form an “algebraic basis” for Λ. That is, every element of
Λ can be uniquely written as a polynomial in {en }∞ n=1 , so we can
think of Λ as the set of all polynomials in the variables {en }∞ n=1 .
We sometimes take advantage of the point of view by substituting
values for the “variables” {en }∞
n=1 , rather than for the usual variables
x1 , x2 , . . ..
We conclude our discussion of the elementary symmetric func-
tions by returning to our starting point. We first encountered the
elementary symmetric functions by considering the expressions we
get when we write the coefficients of a generic polynomial in z in
terms of its roots x1 , . . . , xn . These coefficients are (specializations
of) the elementary symmetric functions e0 , e1 , . . ., which we used to
build all of the other elementary symmetric functions. Since e0 , e1 , . . .
is a sequence, we can’t resist studying its generating function. In fact,
our construction of the ej ’s guarantees this generating function will
have a nice product formula.
Proposition 2.18. The ordinary generating function for the sequence
{en }∞
n=0 of elementary symmetric functions is
∞ ∞
∑ en t = ∏(1 + xj t).
n
(2.6)
n=0 j=1
We often write E(t) to denote this generating function.
Proof. We can build each elementary symmetric function en uniquely

by adding the terms which result from deciding, for each j, whether
to include xj as a factor or not. This matches our computation of the
product on the right side of (2.6): we construct each term by deciding,
for each factor 1 + xj t, whether to use 1 or xj t as a factor. □
2.2. The Complete Homogeneous Symmetric

Functions
The elementary symmetric polynomial ek (Xn ) is a sum over all sub-
sets of [n] of size k, with no repeated elements allowed. In general,
if a problem is interesting when repetition is not allowed, then the
analogous problem in which repetition is allowed is also likely to be
2.2. The Complete Homogeneous Symmetric Functions 39
interesting. This suggests we should also consider a sum over all sub-
sets of [n] in which repeated elements are allowed. To distinguish this
sum from the corresponding sum over all subsets of [n], we will write
[[n]] to denote the multiset {1∞ , 2∞ , . . . , n∞ } with infinitely many
copies of each element of [n], and we will write J ⊆ [[n]] to mean J
is a submultiset of [[n]]. Similarly, we will write [P] to denote the
multiset consisting of infinitely many copies of each positive integer.
Definition 2.19. For all positive integers n and k, the complete

homogeneous polynomial hk (Xn ) is given by
k
hk (Xn ) = ∑ ∏ xjm = ∑ ∏ xj ,
1≤j1 ≤⋅ ⋅ ⋅≤jk ≤n m=1 J⊆[[n]] j∈J
∣J∣=k
while the complete homogeneous symmetric function hk is given by

k
hk = ∑ ∏ xj m = ∑ ∏ xj .
j1 ≤⋅ ⋅ ⋅≤jk m=1 J⊆[P] j∈J
∣J∣=k
By convention, h0 (Xn ) = 1 and h0 = 1. For any partition λ, the

complete homogeneous symmetric polynomial indexed by λ, written
hλ (Xn ), is given by
l(λ)
(2.7) hλ (Xn ) = ∏ hλj (Xn ),
j=1
and the complete homogeneous symmetric function indexed by λ, writ-

ten hλ , is given by
l(λ)
(2.8) hλ = ∏ hλj .
j=1
Since ∣{hλ ∣ λ ⊢ k}∣ = p(k) = dim Λk , if they are distinct, then we

have exactly the right number of complete homogeneous symmetric
functions of each degree to form a basis for Λk . Our proof that the
elementary symmetric functions form a basis for Λk inspires us to con-
sider how the complete homogeneous symmetric functions are written
as linear combinations of the monomial symmetric functions. As hap-
pened for the elementary symmetric functions, one case is especially
simple.
Example 2.20. For all k ≥ 1, write hk as a linear combination of

monomial symmetric functions.
λ
Solution. For every λ ⊢ k, one of the terms in hk is xλ1 1 ⋅ ⋅ ⋅ xl(λ)
l(λ)
, and
this term has coefficient 1. Therefore,
(2.9) hk = ∑ mλ . □
λ⊢k
As was the case for the elementary symmetric polynomials, as

long as we have enough variables, the algebraic relationships between
the complete homogeneous symmetric polynomials and the monomial
symmetric polynomials do not depend on exactly how many variables
we have. More specifically, for any partitions λ and µ with ∣λ∣ = ∣µ∣
and any n ≥ 1, let Mλ,µ,n (h, m) be the rational numbers defined by
hλ (Xn ) = ∑ Mλ,µ,n (h, m)mµ (Xn ).

µ⊢∣λ∣
Similarly, for any partitions λ and µ with ∣λ∣ = ∣µ∣, let Mλ,µ (h, m) be
the rational numbers defined by
(2.10) hλ = ∑ Mλ,µ (h, m)mµ .

µ⊢∣λ∣
Then we have the following analogue of Proposition 2.6.
Proposition 2.21. For any partitions λ and µ with ∣λ∣ = ∣µ∣, if n ≥

∣λ∣ then Mλ,µ,n (h, m) = Mλ,µ (h, m). In particular, if n ≥ ∣λ∣, then
Mλ,µ,n (h, m) is independent of n.
We can also write hλ as a linear combination of the monomial

symmetric functions by hand when ∣λ∣ is small.
Example 2.22. For each λ ⊢ 3, write hλ as a linear combination of

the monomial symmetric functions.
Solution. In view of Proposition 2.21, we can compute with hλ (X3 ).

When λ = (3), we have hλ = h3 = m3 + m21 + m111 by (2.9). When

λ = (2, 1), we have
h21 (X3 ) = (x21 + x22 + x23 + x1 x2 + x1 x3 + x2 x3 )(x1 + x2 + x3 )
= x31 + x32 + x33 + 2x21 x2 + 2x21 x3 + 2x1 x22 + 2x22 x3
+ 2x1 x23 + 2x2 x23 + 3x1 x2 x3 ,
so h21 = m3 + 2m21 + 3m111 . Similarly, when λ = (13 ), we have
h111 (X3 ) = (x1 + x2 + x3 )3
= x31 + x32 + x33 + 3x21 x2 + 3x21 x3 + 3x1 x22 + 3x22 x3
+ 3x1 x23 + 3x2 x23 + 6x1 x2 x3
so h111 = m3 + 3m21 + 6m111 . □
In Example 2.20 we found Mλ,µ (h, m) for all λ and µ with ∣λ∣ =
∣µ∣ = 3. In Figure 2.8 we summarize the results of Example 2.20
in a matrix, along with the corresponding matrices for ∣λ∣ = ∣µ∣ ≤
5. Problem 2.17 has combinatorial interpretations of these numbers
analogous to the interpretations of Mλ,µ (e, m) in Proposition 2.12.
As the matrices in Figure 2.8 suggest, the matrix we get when
we express the complete homogeneous symmetric functions in terms
of the monomial symmetric functions is not triangular in general.
(But you should take a look at its determinant! See Problem 2.23.)
As a result, it is not as easy to show the complete homogeneous
symmetric functions form a basis for Λk as it was to do this for the
elementary symmetric functions. But we can do it by using a simple
relationship between the ordinary generating function for {en }∞
n=0 and
the ordinary generating function for {hn }∞
n=0 .

{hn }∞
n=0 of complete homogeneous symmetric functions is
∞ ∞
1
∑ hn t = ∏
n
(2.11) .
n=0 j=1 1 − xj t
We often write H(t) to denote this generating function.
Proof. We can build each complete homogeneous symmetric function

hn uniquely by adding the terms which result from deciding, for each
(12 ) (2)
(1)
(12 ) 2 1
(1) [ 1 ] [ ]
(2) 1 1
(13 ) (2, 1) (3)

(1 ) ⎡⎢ 6
3
3 1 ⎤⎥
⎢ ⎥
(2, 1) ⎢⎢ 3 2 1 ⎥⎥
(3) ⎢⎣ 1 1 1 ⎥⎦
(14 ) (2, 12 ) (22 ) (3, 1) (4)

(14 ) ⎡⎢ 24 12 6 4 1 ⎤⎥
⎢ ⎥
(2, 12 ) ⎢⎢ 12 7 4 3 1 ⎥⎥
(2 ) ⎢⎢ 6
2
4 3 2 1 ⎥⎥
⎢ ⎥
(3, 1) ⎢ 4 3 2 2 1 ⎥
⎢ ⎥
(4) ⎢ 1 1 1 1 1 ⎥⎦
⎣
(15 ) (2, 13 ) (22 , 1) (3, 12 ) (3, 2) (4, 1) (5)

(1 ) ⎡⎢ 120
5
60 30 20 10 5 1 ⎤⎥
3 ⎢ ⎥
(2, 1 ) ⎢⎢ 60 33 18 13 7 4 1 ⎥⎥
(22 , 1) ⎢⎢ 30 18 11 8 5 3 1 ⎥⎥
⎢ ⎥
(3, 12 ) ⎢⎢ 20 13 8 7 4 3 1 ⎥⎥
(3, 2) ⎢⎢ 10 7 5 4 3 2 1 ⎥⎥
⎢ ⎥
(4, 1) ⎢ 5 4 3 3 2 2 1 ⎥
⎢ ⎥
(5) ⎢ 1 1 1 1 1 1 1 ⎥⎦
⎣
Figure 2.8. The coefficients Mλ,µ (h, m) for ∣λ∣ = ∣µ∣ ≤ 5
j, how many factors of xj to include. This matches our computation

of the product on the right side of (2.11): we construct each term by
1
deciding, for each factor 1−x jt
= 1 + xj t + (xj t)2 + ⋅ ⋅ ⋅, which power of
xj t to use as a factor. □
Comparing equation (2.11) with equation (2.6), we see E(−t) and

H(t) are multiplicative inverses. This gives us an elegant relationship
between the elementary symmetric functions and the complete homo-
geneous symmetric functions.
Proposition 2.24. For all n ≥ 1, we have

n
∑ (−1) ej hn−j = 0.
j
(2.12)
j=0
Proof. By equations (2.6) and (2.11) we have E(−t)H(t) = 1. When

n ≥ 1, the coefficient of tn on the right side of the identity is 0, and
on the left side it is ∑nj=0 (−1)j ej hn−j , so (2.12) must hold. □
This proof is pleasantly short, but it is not especially combina-

torial. With a little additional machinery, we can give a natural
combinatorial proof of equation (2.12).
Combinatorial Proof of Proposition 2.24. For any set or multi-

set J of positive integers, define xJ by xJ ∶= ∏j∈J xj . Now note that
ej = ∑J⊆P xJ and hj = ∑J⊆[P] xJ , where the sum for ej is over all
subsets of the positive integers of size j and the sum for hj is over
all submultisets of the positive integers of size j. In addition, for any
set J, define sgn(J) by sgn(J) ∶= (−1)∣J∣ . With this notation, (2.12)
becomes
∑ sgn(J1 )x 1 x 2 = 0,
J J
(J1 ,J2 )
where the sum is over all ordered pairs (J1 , J2 ) in which J1 is a subset
of P of size j and J2 is a submultiset of [P] of size n − j. If we set
sgn(J1 , J2 ) = sgn(J1 ) for each ordered pair (J1 , J2 ), then we just need
to give an involution I on these ordered pairs such that I has no fixed
points and sgn(I(J1 , J2 )) = − sgn(J1 , J2 ).
To construct I, suppose (J1 , J2 ) is given. Among all of the ele-
ments of J1 ∪ J2 , let k be the smallest. If k ∈ J1 , then set I(J1 , J2 ) ∶=
(J1 − {k}, J2 ∪ {k}), and if k ∈/ J1 , then set I(J1 , J2 ) ∶= (J1 ∪ {k},
J2 − {k}). In words, move one copy of k from J2 to J1 if you can, and
otherwise move k from J1 to J2 .
Moving the smallest element of J1 ∪ J2 between J1 and J2 does
not change the fact that it is the smallest element in J1 ∪ J2 , and our
construction guarantees J1 is always a set (rather than a multiset),
so we see I(I(J1 , J2 )) = (J1 , J2 ). Furthermore, since I changes the
size of J1 by one, it can have no fixed points, and sgn(I(J1 , J2 )) =
− sgn(J1 , J2 ), as desired. □
Proposition 2.24 gives us a new way to show {hλ ∣ λ ⊢ k} spans

Λk .
Proposition 2.25. For all k ≥ 0, the set {hλ ∣ λ ⊢ k} is a basis for

Λk .
Proof. When k = 0, we have Λ0 = Span{1}. Since h0 = e0 = 1, the

result is clear in this case.
When k = 1 we have Λ1 = Span{x1 + x2 + ⋅ ⋅ ⋅}. Since h1 = e1 =
x1 + x2 + ⋅ ⋅ ⋅, the result is also clear in this case.
Now suppose k ≥ 2; we argue by induction on k. Since
∣{hλ ∣ λ ⊢ k}∣=p(k)=dim Λk and Span{hλ ∣ λ ⊢ k}⊆Λk , it is sufficient
to show Λk ⊆ Span{hλ ∣ λ ⊢ k}. By Corollary 2.11, we have Λk =
Span{eλ ∣ λ ⊢ k}, so it is even enough to show eµ ∈ Span{hλ ∣ λ ⊢ k}
for all µ ⊢ k. To do this, we consider two cases.
If µ ≠ k, then eµ is a product eµ1 eµ2 ⋅ ⋅ ⋅ eµl(µ) in which, for each j,
we have µj < k and eµj ∈ Λµj . By induction, eµj ∈ Span{hλ ∣ λ ⊢ µj }
for all j, so eµ ∈ Span{hλ ∣ λ ⊢ k}.
If µ = k, then by (2.12) we have
k−1
ek = ∑ (−1)k+j+1 ej hk−j .
j=0
By induction, ej ∈ Span{hλ ∣ λ ⊢ j} for 0 ≤ j ≤ k − 1, so ek ∈

Span{hλ ∣ λ ⊢ k}, which is what we wanted to prove. □
Proposition 2.25 gives us an analogue of Theorem 2.17 for the

complete homogeneous symmetric functions.
Corollary 2.26. The set {hn }∞

Proof. This is similar to the proof of Theorem 2.17. □
2.3. The Power Sum Symmetric Functions

The elementary symmetric functions ek = m1k are at one extreme of
the set of monomial symmetric functions in the sense that (1k ) is the
partition of k with the largest number of parts. With this in mind,
2.3. The Power Sum Symmetric Functions 45
it seems natural to look at the monomial symmetric functions mµ in

which µ has the smallest number of parts, namely mk .
Definition 2.27. For all n, k ≥ 1 we set
n
pk (Xn ) = ∑ xkj
j=1
and
∞
pk = ∑ xkj .
j=1
For any partition λ, the power sum symmetric polynomial pλ (Xn )
indexed by λ is
l(λ)
(2.13) pλ (Xn ) = ∏ pλj (Xn ),
j=1
and the power sum symmetric function pλ indexed by λ is

l(λ)
(2.14) pλ = ∏ pλj .
j=1
Note that neither p0 (Xn ) nor p0 is defined.
Our work with the elementary and complete homogeneous sym-

metric functions suggests the next natural step is to explore how the
power sum symmetric functions are written as linear combinations
of the monomial symmetric functions. To start, for any nonempty
partitions λ and µ with ∣λ∣ = ∣µ∣ and any n ≥ 1, let Mλ,µ,n (p, m) be
the rational numbers defined by
pλ (Xn ) = ∑ Mλ,µ,n (p, m)mµ (Xn ).
µ⊢∣λ∣
Similarly, for any nonempty partitions λ and µ with ∣λ∣ = ∣µ∣, let
Mλ,µ (p, m) be the rational numbers defined by
(2.15) pλ = ∑ Mλ,µ (p, m)mµ .
µ⊢∣λ∣
Then we have the following analogue of Propositions 2.6 and 2.21.

Proposition 2.28. For any nonempty partitions λ and µ with ∣λ∣ =
∣µ∣, if n ≥ ∣λ∣, then Mλ,µ,n (p, m) = Mλ,µ (p, m). In particular, if n ≥
∣λ∣, then Mλ,µ,n (p, m) is independent of n.
In our first example we confirm we have in fact singled out mk .
Example 2.29. For each k ≥ 1, write pk as a linear combination of

Solution. Since pk has only terms of the form xkj , and each term has
coefficient 1, we see pk = mk . □
Using Proposition 2.28 to reduce our work to computations with

polynomials, we find the coefficients Mλ,µ (p, m) for ∣λ∣ = ∣µ∣ ≤ 5 in
Figure 2.9. These data suggest that when we write the power sum
symmetric functions as linear combinations of the monomial symmet-
ric functions and arrange our partitions in increasing lexicographic
order, the resulting matrices are upper triangular with nonzero en-
tries on their diagonals. We prove this next, and we use this result
to show {pλ ∣ λ ⊢ k} is a basis for Λk .
Proposition 2.30. For all k ≥ 1 and all nonempty partitions λ, µ

with ∣λ∣ = ∣µ∣, the following hold.
(i) If λ >lex µ, then Mλ,µ (p, m) = 0.
(ii) Mλ,λ (p, m) > 0.
Proof. (i) By definition we have
pλ = pλ1 ⋅ ⋅ ⋅ pλl(λ) ,
µ
and we note that Mλ,µ (p, m) is the coefficient of xµ1 1 ⋅ ⋅ ⋅ xl(µ)
l(µ)
in this
product. Each term in the expansion of this product is the product of
λ
one term from each pλj , and each such term has the form xmj for some
m. If all of these m are distinct, then λ = µ. If two of these m are
equal, then µ is obtained from λ by merging parts and rearranging
the resulting numbers into weakly decreasing order, in which case
µ >lex λ, and the result follows.
(ii) Note that in the expansion of the product
pλ = pλ1 ⋅ ⋅ ⋅ pλl(λ) ,
2.3. The Power Sum Symmetric Functions 47
(12 ) (2)
(1)
(12 ) 2 1
(1) [ 1 ] [ ]
(2) 0 1
(13 ) (2, 1) (3)

(1 ) ⎡⎢ 6
3
3 1 ⎤⎥
⎢ ⎥
(2, 1) ⎢⎢ 0 1 1 ⎥⎥
(3) ⎢⎣ 0 0 1 ⎥⎦
(14 ) (2, 12 ) (22 ) (3, 1) (4)

(14 ) ⎡⎢ 24 12 6 4 1 ⎤⎥
⎢ ⎥
(2, 12 ) ⎢⎢ 0 2 2 2 1 ⎥⎥
(2 ) ⎢⎢ 0
2
0 2 0 1 ⎥⎥
⎢ ⎥
(3, 1) ⎢ 0 0 0 1 1 ⎥
⎢ ⎥
(4) ⎢ 0 0 0 0 1 ⎥⎦
⎣
(15 ) (2, 13 ) (22 , 1) (3, 12 ) (3, 2) (4, 1) (5)

(1 ) ⎡⎢ 120
5
60 30 20 10 5 1 ⎤⎥
3 ⎢ ⎥
(2, 1 ) ⎢⎢ 0 6 6 6 4 3 1 ⎥⎥
(22 , 1) ⎢⎢ 0 0 2 0 2 1 1 ⎥⎥
⎢ ⎥
(3, 12 ) ⎢⎢ 0 0 0 2 1 2 1 ⎥⎥
(3, 2) ⎢⎢ 0 0 0 0 1 0 1 ⎥⎥
⎢ ⎥
(4, 1) ⎢ 0 0 0 0 0 1 1 ⎥
⎢ ⎥
(5) ⎢ 0 0 0 0 0 0 1 ⎥⎦
⎣
Figure 2.9. The coefficients Mλ,µ (p, m) for ∣λ∣ = ∣µ∣ ≤ 5
λ
we can obtain the term xλ1 1 ⋅ ⋅ ⋅ xl(λ)
l(λ)
by choosing, for each j, the term
λ
xj j from the factor pλj . Since the coefficients in pλj are all positive for
λ
all j, the coefficient of xλ1 1 ⋅ ⋅ ⋅ xl(λ)
l(λ)
is at least 1, so Mλ,λ (p, m) > 0. □
Corollary 2.31. For all k ≥ 1, the set {pλ ∣ λ ⊢ k} is a basis for Λk .
Proof. This is similar to the proof of Corollary 2.11. □
Corollary 2.32. The set {pn }∞

Proof. This is similar to the proof of Theorem 2.17. □
Earlier we found the ordinary generating functions for {en }∞ n=0

and {hn }∞
n=0 , which we used to show that the complete homogeneous
symmetric functions form a basis for Λk . Rather than finding the
ordinary generating function for {pn }∞n=1 , it turns out to be easier
∞
and more useful to find the ordinary generating function for { pnn }n=1 .
∞
Our expression for this generating function for { pnn }n=1 involves a
logarithm of a formal power series. This construction raises technical
issues, but addressing these issues would take us too far afield. So we
will proceed under the assumption that logarithms of formal power
series behave in analogy with logarithms of polynomials. The inter-
ested reader can find more details about this in a variety of sources,
including [Loe17, Ch. 7], [Niv69], and [Wil05, Ch. 2].

{ pnn }∞
n=1 of scaled power sum symmetric functions is
∞ ∞
pn n 1
(2.16) ∑ t = log(∏ ).
n=1 n j=1 1 − xj t
We often write P (t) to denote this generating function.
Proof. Since log ( 1−x

1
) = ∑∞ 1 n
n=1 n x , we have
∞
pn n
P (t) = ∑ t
n=1 n
∞ ∞
1
=∑∑ (xj t)n
n=1 j=1 n
∞
1
= ∑ log ( )
j=1 1 − xj t
⎛∞ 1 ⎞
= log ∏ ,
⎝j=1 1 − xj t ⎠

2.4. Problems 49
2.4. Problems
2.1. Compute Mλ,µ (e, m) for λ = (22 , 14 ) and µ = (32 , 12 ).
2.2. Find and prove a formula for Mλ,1n (e, m), where λ ⊢ n.
2.3. Find and prove a formula for Mλ,n (e, m), where λ ⊢ n.
2.4. Find and prove a formula for Mλ,µ (e, m), where λ ⊢ n and
µ = (n − 1, 1).
2.5. Show ≥lex (which means “>lex or =”) is a partial ordering on the
set of partitions.
2.6. Show that if λ and µ are distinct partitions, then λ >lex µ or
µ >lex λ. That is, show the lexicographic ordering is a total
ordering on the set of partitions.
2.7. Prove or disprove: if λ and µ are partitions with ∣λ∣ = ∣µ∣ and
λ >lex µ, then µ′ >lex λ′ .
2.8. Show that the converse of Proposition 2.10(i) is false. In partic-
ular, find partitions λ and µ with ∣λ∣ = ∣µ∣ for which Mλ,µ (e, m) =
0 even though λ′ >lex µ.
2.9. Suppose λ and µ are partitions. We say λ is greater than or
equal to µ in the dominance order, and we write λ ⊵ µ or
µ ⊴ λ, whenever ∑nj=1 λj ≥ ∑nk=1 µk for all n ≥ 0. Show that
Mλ,µ (e, m) ≠ 0 if and only if ∣λ∣ = ∣µ∣ and λ′ ⊵ µ.
2.10. Show ⊴ is a partial ordering on the set of partitions of n for all
n ≥ 0.
2.11. For each n ≥ 6, find partitions λ ⊢ n and µ ⊢ n for which neither
λ ⊴ µ nor µ ⊴ λ.
2.12. Prove or disprove: if λ and µ are partitions of n and λ ⊵ µ, then
λ ≥lex µ.
2.13. Prove or disprove: if λ and µ are partitions with ∣λ∣ = ∣µ∣ and
λ ⊵ µ, then λ′ ⊴ µ′ .
2.14. Suppose {fn }∞ n=1 is an algebraically independent set of symmet-
ric functions, each fn is homogeneous, and deg fn ≤ deg fn+1 for
all n ≥ 0. Furthermore, suppose every symmetric function can
be written as a polynomial in {fn }∞ n=1 . Show deg fn = n for all
n ≥ 0.
2.15. Write the formal power series

∞
∏(1 + xj + xj )
2
j=1
in terms of the elementary symmetric functions.

2.16. Show that for all k ≥ 1 and all partitions λ, µ ⊢ k, the quantity
Mλ,µ (h, m) defined in (2.10) is the number of fillings of the
Ferrers diagram of λ with positive integers for which the entries
in each row are weakly increasing from left to right, and each
integer j appears exactly µj times.
2.17. Show that for all k ≥ 1 and all partitions λ, µ ⊢ k, the following
hold.
(a) Mλ,µ (h, m) is the number of k × k matrices in which every
entry is a nonnegative integer, the sum of the entries in
row j is µj for all j, and the sum of the entries in column
j is λj for all j.
(b) Mλ,µ (h, m) is the number of ways to place k balls, consist-
ing of µm identical balls of type m for each m, into l(λ)
urns, so the jth urn contains exactly λj balls.
We will use this problem in our proof of Proposition 7.10.
2.18. Compute Mλ,µ (h, m) for λ = (22 , 14 ) and µ = (32 , 12 ).
2.19. Find and prove a formula for Mλ,1n (h, m), where λ ⊢ n.
2.20. Find and prove a formula for Mλ,n (h, m), where λ ⊢ n.
2.21. Find and prove a formula for Mλ,µ (h, m), where λ ⊢ n and
µ = (n − 1, 1).
2.22. Prove or disprove: for every λ, µ ⊢ k, we have Mλ,µ (h, m) =
Mµ,λ (h, m).
2.23. For each n ≥ 1, let An be the p(n)×p(n) matrix whose rows and
columns are indexed by partitions in lexicographic order from
smallest to largest, and for which the entry in row λ and column
µ is Mλ,µ (h, m). The matrices An for 1 ≤ n ≤ 5 are shown in
Figure 2.8. Prove or disprove: det(An ) = 1 for all n ≥ 1.
2.24. Show
n
−1
hk (Xn ) = ∑ xn−1−k
j ∏(xj − xl ) .
j=1 l≠j
2.4. Problems 51
2.25. Show that for all n ≥ 0, we have

en = det (h1−j+k )1≤j,k≤n .
Here hn = 0 for n < 0.
hn = det (e1−j+k )1≤j,k≤n .
Here en = 0 for n < 0.
2.27. For all k ≥ 1 and all partitions λ, µ ⊢ k, let Mλ,µ (p, m) be
defined by
pλ = ∑ Mλ,µ (p, m)mµ .
µ⊢k
Show that for all k ≥ 1 and all partitions λ, µ ⊢ k, the quantity
Mλ,µ (p, m) is the number of fillings of the Ferrers diagram of
λ with positive integers for which the entries in each row are
constant, and each integer j appears exactly µj times. We will
use this result in our proof of Proposition 7.17.
2.28. Show that for all k ≥ 1 and all partitions λ, µ ⊢ k, the following
hold.
(a) Mλ,µ (p, m) is the number of n × n matrices in which the
sum of the entries in row m is µm for all m, and for all j
exactly one entry of column j is λj and all other entries in
column j are 0.
(b) Mλ,µ (p, m) is the number of ways to place k balls, consist-
ing of µm identical balls of type m for each m, into l(λ)
urns, so that for all j the jth urn contains exactly λj balls,
all of which have the same type.
2.29. Compute Mλ,µ (p, m) for λ = (22 , 14 ) and µ = (32 , 12 ).
2.30. Find and prove a formula for Mλ,1n (p, m), where λ ⊢ n.
2.31. Find and prove a formula for Mλ,n (p, m), where λ ⊢ n.
2.32. Find and prove a formula for Mλ,µ (p, m), where λ ⊢ n and
µ = (n − 1, 1).
2.33. Find and prove a formula for Mλ,λ (p, m).
2.34. Prove or disprove: for every λ, µ ⊢ k, we have Mλ,µ (p, m) =
Mµ,λ (p, m).
Chapter 3
Interlude: Evaluations
of Symmetric Functions
Our interest in symmetric functions is based primarily on the rela-

tionships among them and how we can describe these relationships
combinatorially. Nevertheless, many apparently disparate combina-
torial quantities are, in fact, evaluations of symmetric functions at
carefully chosen points. This means, in particular, that a variety of
combinatorial identities and dualities are, in fact, just special cases of
symmetric function identities. In other words, symmetric functions
provide a common explanation for several families of combinatorial
facts. In this chapter we develop a handful of symmetric function
identities, which we then use to tie together binomial coefficients,
Stirling numbers, and q-binomial coefficients.
3.1. Symmetric Function Identities

Many combinatorial quantities we can recognize as evaluations of
symmetric functions, such as the binomial coefficients or the Stir-
ling numbers, satisfy recurrence relations similar to Pascal’s identity.
These recurrence relations come directly from the following simple
recurrence relations for the elementary and complete homogeneous
symmetric polynomials.
53
54 3. Evaluations of Symmetric Functions
Proposition 3.1. For all k ≥ 0 and all n ≥ 1, we have

(3.1) ek (Xn ) = ek (Xn−1 ) + xn ek−1 (Xn−1 )
and
(3.2) hk (Xn ) = hk (Xn−1 ) + xn hk−1 (Xn ).
Proof. To prove equation (3.1), note that we have two kinds of terms
in ek (Xn ): those with xn as a factor and those without xn as a factor.
Those without xn as a factor form ek (Xn−1 ), and when we take out
the common factor xn from those terms with xn as a factor, we obtain
ek−1 (Xn−1 ).
The proof of equation (3.2) is Problem 3.1. □
Equations (3.1) and (3.2) give us recurrence relations for various

specializations of the elementary and complete homogeneous sym-
metric functions, but these specializations are at the root of many
identities because the symmetric functions themselves satisfy a va-
riety of identities. We have already seen one of these identities in
(2.12), which says
n
∑ (−1) ej hn−j = 0
j
(3.3)
j=0
for n ≥ 1. Our next two identities are similar to this, but they give
us expressions for pk which are similar to the sum on the left side of
equation (3.3).
Proposition 3.2. For all k ≥ 1, we have

k
(3.4) pk = ∑ (−1)j−1 jej hk−j
j=1
and
k
(3.5) pk = ∑ (−1)k+j jek−j hj .
j=1
Proof. To prove equation (3.4), first note that when we expand the
products on the right in terms of x1 , x2 , . . ., the resulting terms corre-
spond to a choice of one term from ej , followed by a choice of one of
the variables xs in that term, followed by a choice of one term from
3.1. Symmetric Function Identities 55
1 3 4∗ 7 2 2 4
Figure 3.1. A filling indexing a term on the right side of

equation (3.4)
hk−j . We can keep track of these terms by starting with a 1 × j tile

and a 1 × (k − j) tile, filling the 1 × j tile with distinct positive integers
in increasing order, marking one of these positive integers, and then
filling the 1 × (k − j) tile with positive integers in weakly increasing
order. For example, when k = 7 and j = 3, the filled tiles in Figure 3.1
(with the 4 marked with a star in the left tile) correspond to choosing
the term x1 x3 x4 x7 from ej , marking the x4 , and then choosing the
term x22 x4 from hk−j .
We now describe a function f from the set of these marked, filled
pairs of tiles to itself. If a marked, filled pair T has j = 1 and all of
the entries in both tiles are equal, then we set f (T ) = T . Otherwise,
find the smallest number r which appears in either tile and which is
not the marked number in the first tile. If r does not appear in the
left tile (corresponding with the term from ej ), then remove one copy
of r from the right tile and insert it into the left tile. If r does appear
in the left tile, then remove it from the left tile and insert it in the
right tile.
When we apply f to the filling in Figure 3.1, we find r = 1, and
we obtain the filling in Figure 3.2.
Our function f has several important properties. First, notice
that if f (T ) ≠ T and we apply f to f (T ), then we will choose the
same number r as we did to compute f (T ), and we will put it back
in its original position. Therefore, f (f (T )) = T for all marked, filled
pairs of tiles. Second, notice that if f (T ) ≠ T , then the lengths of
the left tiles of T and f (T ) differ by 1, but their associated products
of terms are the same. Therefore, if f (T ) ≠ T , then T and f (T )
3 4∗ 7 1 2 2 4
Figure 3.2. The image of the filling in Figure 3.1 under f

correspond with terms in the right side of (3.4) which are negatives
of one another. In short, f is a sign-reversing involution.
Because f is a sign-reversing involution, only terms on the right
side of (3.4) corresponding with marked, filled pairs T of tiles with
f (T ) = T remain after cancellation. From our definition of f , we see
these are exactly the terms in pk .
The proof of equation (3.5) is Problems 3.3 and 3.4. □
We also have a pair of identities connecting the power sum sym-

metric functions with the elementary symmetric functions, and the
power sum symmetric functions with the complete homogeneous sym-
metric functions.
Proposition 3.3. For all k ≥ 1, we have

k
(3.6) khk = ∑ hk−j pj
j=1
and
k
(3.7) kek = ∑ (−1)j−1 ek−j pj .
j=1
Proof. To prove equation (3.6), first note that the terms in khk are
indexed by fillings of a 1 × k tile with positive integers in weakly
increasing order, in which exactly one entry has been marked. For
example, the filling in Figure 3.3 corresponds to the fourth copy of
x1 x43 x4 x26 x7 in h9 .
On the other hand, the terms on the right side of (3.6) are indexed
by pairs of tiles, one 1 × (k − j) and one 1 × j, in which the first tile is
filled with positive integers in weakly increasing order, and the second
tile is filled with j copies of one positive integer. For example, the
filling in Figure 3.4 corresponds to the product of x1 x23 x4 x26 x7 and x23
in the term with j = 2 on the right side of (3.6).
1 3 3 3∗ 3 4 6 6 7
Figure 3.3. A filling corresponding to a term on the left side

of equation (3.6)
3.2. Binomial Coefficients 57
1 3 3 4 6 6 7 3 3
Figure 3.4. A filling corresponding to a term on the right

side of equation (3.6)
To transform a marked filling of the first type into a pair of fillings

of the second type, suppose the marked number is r. Remove the
marked r and all copies of r to its right, and use them to make a tile
filled with r’s.
To transform a pair of fillings of the second type into a marked
filling of the first type, first mark the first entry of the second tile.
Then, if all of the entries in this tile are r, insert the entries in the
second tile immediately after the rightmost r in the first tile.
Note that the fillings in Figures 3.3 and 3.4 correspond with each
other under these maps.
Since these maps are inverses of one another, we have a bijection
between the terms on the left side of (3.6) and the right side, and the
result follows.
The proof of equation (3.7) is Problems 3.6 and 3.8. □
3.2. Binomial Coefficients

The easiest quantities to get as evaluations of symmetric functions
are the binomial coefficients. We could use equations (3.1) and (3.2)
along with the fact that (nk) is completely determined by (n0 ) = (nn) = 1
and Pascal’s relation (which says (nk) = (n−1
k
) + (n−1
k−1
) for 1 ≤ k ≤ n − 1)
to prove our next result by induction. But it is just as easy, and
arguably more illuminating, to give a somewhat more combinatorial
proof.
n
(3.8) ek (1, . . . , 1) = ( )
´¹¹ ¹ ¹ ¹ ¹¸ ¹ ¹ ¹ ¹ ¹¶ k
n
and
n+k−1
(3.9) hk (1, . . . , 1) = ( ).
´¹¹ ¹ ¹ ¹ ¹¸ ¹ ¹ ¹ ¹ ¹¶ k
n
Proof. To prove equation (3.9), first note that hk (1, . . . , 1) is the

´¹¹ ¹ ¹ ¹ ¹¸ ¹ ¹ ¹ ¹ ¹¶
n
number of terms in hk (Xn ), which in turn is the number of ways of
choosing k elements of [n] with repetition allowed. Some people call
this a “stars and bars problem”, while others call it a “flowershop
problem”. Whatever your terminology, we can represent each term
in hk (Xn ) as a sequence of k stars and n−1 dividers, or bars, in which
the number of stars (which could be 0) in the jth string of consecutive
stars from the left is the exponent on xj . For example, when n = 4
and k = 6, the string ∗∣∣ ∗ ∗ ∗ ∣ ∗ ∗ corresponds to the term x1 x33 x24 . We
can construct each string uniquely by choosing k positions out of the
n + k − 1 positions for the stars, so the number of terms is (n+k−1 k
).
The proof of equation (3.8) is similar and is Problem 3.12. □
We can now combine equations (3.8) and (3.9) with the identities
in the previous section to get a variety of binomial coefficient identi-
ties, essentially for free. As we will see, some of these identities will
be equivalent to familiar facts.
Proposition 3.5. For all m, n ≥ 1, we have

m
j n m+n−j−1
(3.10) ∑ (−1) ( )( ) = 0.
j=0 j m−j
Proof. In (3.3), replace n with m, evaluate the result with xr = 1 for

1 ≤ r ≤ n and xr = 0 for r > n, and use equations (3.8) and (3.9) to
eliminate ej and hm−j . □
It is worth noting that we can rearrange (3.10) to find

m
n m+n−j−1
∑ (−1) ( ) = 0,
j
j=0 j j − 1, n − j, m − j
(a+b+c)!
where (a+b+c
a,b,c
)= a!b!c!
is the usual multinomial coefficient.
Proposition 3.6. For all n, k ≥ 1, we have

k
n n+j−1
∑ (−1) )( ) = n.
k+j
(3.11) j(
j=1 k−j j
3.2. Binomial Coefficients 59
Proof. This is similar to the proof of equation (3.10), using (3.5) and
the fact that pk (1, . . . , 1) = n. □
´¹¹ ¹ ¹ ¹ ¹¸ ¹ ¹ ¹ ¹ ¹¶
n
It is worth noting that we can rearrange equation (3.11) to find

k
n+j−1
∑ (−1) ( ) = 1.
j+k
j=1 k − j, n − k + j, j − 1
The identities we have seen so far may not look familiar, but our
next one is much more common; some people call it the “hockey stick
identity” because the entries of Pascal’s triangle that it involves are
arranged in the shape of a hockey stick.
Proposition 3.7. For all n, k ≥ 1, we have
n+k−1 k−1
n+j−1
(3.12) ( )= ∑( ).
k−1 j=0 j
Proof. In (3.6) set xr = 1 for r ≤ n and xr = 0 for r > n and then use
(3.9) and the fact that pj (1, . . . , 1) = n to find
´¹¹ ¹ ¹ ¹ ¹¸ ¹ ¹ ¹ ¹ ¹¶
n
n+k−1 k
n+k−j−1
k( ) = ∑ n( ).
k j=1 k−j
Now divide both sides by n, use the fact that nk (n+k−1

k
) = (n+k−1
k−1
), and
replace j with k − j in the sum to obtain (3.12). □
Our last identities are two forms of the binomial theorem, one of
the first deep facts we learn about the binomial coefficients.
n
n
(3.13) (1 + t)n = ∑ ( )tk .
k=0 k
Proof. Set x1 = x2 = ⋅ ⋅ ⋅ = xn = 1 and xj = 0 for j > n in (2.6) and use

(3.8) to simplify the result. □
∞
n+k−1 k
(3.14) (1 − t)−n = ∑ ( )t .
k=0 k
Proof. This is similar to the proof of Proposition 3.8, using equations

(2.11) and (3.9). □
3.3. Stirling Numbers of the First and Second

Kinds
Given a finite set, the binomial coefficients tell us how many ways
there are to choose a subset of a given size. In contrast, the Stirling
numbers are concerned with how many ways there are to divide that
set into a given number of subsets, and possibly put some structure
on these subsets. More specifically, we define the Stirling numbers of
the first and second kinds in terms of permutations and set partitions,
as follows.
Definition 3.10. For all n ≥ 1 and all k ≥ 1, the Stirling number
of the first kind [nk] is the number of permutations of [n] with ex-
actly k cycles. (See Appendix C for more information on cycles in
permutations.)
Some authors prefer to discuss the signed Stirling numbers of the

first kind, which are given by (−1)n−k [nk].
To define the Stirling numbers of the second kind, we first need
to recall the notion of a set partition.
Definition 3.11. For any set A, a set partition of A is a set
{B1 , . . . , Bk } of nonempty subsets of A such that A = ⋃kj=1 Bj , and
if j ≠ l, then Bj ∩ Bl = ∅. The sets Bj are called the blocks of the
partition.
We note that changing the order of the blocks in a set partition

does not change the partition.
Definition 3.12. For all n ≥ 1 and all k ≥ 1, the Stirling number of
the second kind {nk} is the number of set partitions of [n] with exactly
k blocks.
The Stirling numbers of both kinds have much in common with

the binomial coefficients. For example, it is useful to arrange them
in triangles as in Figures 3.5 and 3.6, as we do with the binomial
3.3. Stirling Numbers of the First and Second Kinds 61
1
1 1
2 3 1
6 11 6 1
24 50 35 10 1
Figure 3.5. The top of the Stirling triangle of the first kind
1
1 1
1 3 1
1 7 6 1
1 15 25 10 1
Figure 3.6. The top of the Stirling triangle of the second kind
coefficients when we form Pascal’s triangle. And like the binomial

coefficients, in both Stirling triangles each entry is a certain combi-
nation of the entries above and to the left and above and to the right.
In particular, the Stirling triangles are determined by the following
initial conditions and recurrence relations.
Proposition 3.13. For all n ≥ 1, we have [n0 ] = 0 and [nn] = 1, and

for all n ≥ 2 and k with 1 ≤ k ≤ n − 1, we have
n n−1 n−1
(3.15) [ ]=[ ] + (n − 1)[ ].
k k−1 k
Proof. There is exactly one way for a permutation of [n] to have n

cycles, which is for each element to be its own cycle, so [nn] = 1. At
the other extreme, every permutation of a positive number of objects
must have at least one cycle, so [n0 ] = 0.
To prove equation (3.15), first note that in a permutation of [n],
the n is either alone in its cycle, or it is in a cycle of length 2 or more.
We will count the permutations in each of these two sets.
Each permutation of the first type with exactly k cycles consists
of the cycle (n) and an arbitrary permutation of [n − 1] with exactly
k − 1 cycles. Therefore there are [n−1
k−1
] of these permutations.
The permutations of the second type with exactly k cycles are

more complicated. If we remove the n from one of these permutations,
then we get a permutation of [n − 1] with exactly k cycles. But
the n can be in different places, so we can get each of these shorter
permutations from several different longer permutations. In fact, we
can arrange the longer permutation so each of its cycles begins with
its minimal element. When we do this, the n will not be the first entry
in its cycle, so the n can follow any of the n − 1 entries. Therefore, we
can construct each permutation of [n] with exactly k cycles uniquely
by choosing a permutation of [n − 1] wth k cycles, choosing one of
the n − 1 entries of this permutation, and placing the n immediately
after and in the same cycle as our chosen entry. As a result, there are
(n − 1)[n−1
k
] permutations of [n] with exactly k cycles in which the n
is in a cycle of length 2 or more.
Combining our results, we find that equation (3.15) holds. □
Proposition 3.14. For all n ≥ 1, we have {n1 } = {nn} = 1, and for all
n ≥ 2 and all k with 2 ≤ k ≤ n − 1, we have
n n−1 n−1
(3.16) { }={ } + k{ }.
k k−1 k
Proof. This is Problem 3.15. □
By combining Propositions 3.13 and 3.14 with equations (3.1)

and (3.2), we can see how the Stirling numbers of the first and second
kinds are evaluations of the elementary and complete homogeneous
symmetric functions, respectively.
Proposition 3.15. For all n ≥ 1 and all k with 0 ≤ k ≤ n, we have
n+1
(3.17) ek (1, 2, . . . , n) = [ ]
n+1−k
and
n+k
(3.18) hk (1, 2, . . . , n) = { }.
n
Proof. To prove (3.17), first note that e0 (1, 2, . . . , n) = 1 = [n−0+1

n+1
],
so the result holds for k = 0. Similarly, en (1, 2, . . . , n) = n! = n−n+1],
[ n+1
so the result holds for k = n.

3.3. Stirling Numbers of the First and Second Kinds 63
Now suppose n ≥ 2 and the result holds for n − 1 and all k with
0 ≤ k ≤ n. Then for any k with 1 ≤ k ≤ n, we can use (3.1), induction
on n, and (3.15) to find
ek (1, 2, . . . , n) = ek (1, 2, . . . , n − 1) + nek−1 (1, 2, . . . , n − 1)

n n
=[ ] + n[ ]
n−k n−k+1
n+1
=[ ],
n+1−k
which is what we wanted to prove.
The proof of equation (3.18) is Problem 3.16. □
As we did with the binomial coefficients, we can combine (3.17)

and (3.18) with our symmetric function identities to obtain a variety
of identities involving the Stirling numbers of each kind.
Proposition 3.16. For all m, n ≥ 1, we have

m
n+1 n+m−j
∑ (−1) [ ]{ } = 0.
j
(3.19)
j=0 n+1−j n
Proof. In equation (3.3), replace n with m, evaluate the result with

xr = r for 1 ≤ r ≤ n and xr = 0 for r > n, and use (3.17) and (3.18) to
eliminate ej and hm−j . □
Our symmetric function identities also enable us to express the

sum of the kth powers of the first n positive integers as an alternating
sum of products of Stirling numbers.
Proposition 3.17. For all k, n ≥ 1, we have

k
n+1 n+k−j
(3.20) 1k + 2k + ⋅ ⋅ ⋅ + nk = ∑ (−1)j−1 [ ]{ }.
j=1 n + 1 − j n
Proof. Evaluate equation (3.4) with xr = r for 1 ≤ r ≤ n and xr = 0

for r > n, and use (3.17) and (3.18) to eliminate ej and hk−j . Now
the result follows from the fact that, by definition, pk (1, 2, . . . , n) =
1k + 2k + ⋅ ⋅ ⋅ + nk . □
As we will see in Problem 3.17, in some ways equation (3.20) does

not help us find a formula for 1k + ⋅ ⋅ ⋅ + nk nearly as much as it helps
us use such a formula to find a formula for [nk] when k is near n.
The binomial theorem tells us the generating function for the
nth row of Pascal’s triangle factors nicely. We can use our symmetric
polynomial identities to show that the same thing happens for the
Stirling numbers of the first kind.
n
n k n−1
(3.21) ∑ [ ]t = ∏ (t + j).
k=1 j j=0
Proof. Set xj = j for 1 ≤ j ≤ n − 1 and xj = 0 for j ≥ n in equation

(2.6), and use (3.17) to find
n−1 n−1
n
∑[ ]tj = ∏ (1 + jt).
j=0 n − j j=1
Now reindex the sum on the left with k = n − j, replace t with 1/t,
and multiply both sides by tn to obtain (3.21). □
We can use a similar technique to obtain an identity involving

the Stirling numbers of the second kind.
∞
n+j j n 1
(3.22) ∑{ }t = ∏ .
j=0 n j=1 1 − jt
Proof. Set xj = j for 1 ≤ j ≤ n and xj = 0 for j > n in (2.11), and use

(3.18) to simplify the result. □
3.4. q-Binomial Coefficients

We have seen how several interesting families of numbers appear as
evaluations of symmetric polynomials, and how we can use symmetric
polynomial identities to recover, or maybe even discover, identities
involving these numbers. In this section we describe how to do the
same thing with a family of polynomial analogues of the binomial
coefficients. The story starts with inversions, which give us a simple
3.4. q-Binomial Coefficients 65
way to measure the extent to which a sequence of integers is not in

increasing order, and which we recall from Appendix C.
Definition 3.20. Suppose n ≥ 1. For any sequence π = a1 , . . . , an of

integers, an inversion is an ordered pair (j, k) with 1 ≤ j, k ≤ n, such
that j < k and aj > ak . We write inv(π) to denote the number of
inversions in π, and we sometimes call inv(π) the inversion number
of π.
Example 3.21. Find all of the inversions in the word π = 7192686,

and compute inv(π).
Solution. The inversions are the pairs of positions of the entries which
are in decreasing order. For π = 7192686 these pairs of positions are
(1, 2), (1, 4), (1, 5), (1, 7), (3, 4), (3, 5), (3, 6), (3, 7), and (6, 7), so
inv(π) = 9. □
It is rewarding to study the distribution of inv on many different

sets. For example, in Problems 3.18 and 3.19 we will glimpse some
of the interesting properties of the generating function for Sn with
respect to inv. However, the sets we are interested in are sets of
sequences of 0’s and 1’s.
Definition 3.22. For any nonnegative integers k and n with k ≤ n,

let Bn,k be the set of sequences of length n consisting of k 1’s and
n − k 0’s. The q-binomial coefficient (nk)q is the ordinary generating
function for Bn,k with respect to inv. That is,
n
( ) = ∑ q inv(π) .
k q π∈Bn,k
We can construct each sequence in Bn,k uniquely by choosing k

positions out of a total of n for the 1’s, so ∣Bn,k ∣ = (nk). As a result,
if we set q = 1 in the q-binomial coefficient (nk)q , we find (nk)1 = (nk).
So the q-binomial coefficients are polynomial analogues (commonly
called q-analogues) of the usual binomial coefficients.
We often arrange the binomial coefficients in Pascal’s triangle,
and we arrange the q-binomial coefficients in an analogous q-Pascal
triangle in Figure 3.7.
1
1 1
1 1+q 1
2 2
1 1+q+q 1+q+q 1
2 3 2 3 4 2 3
1 1+q+q +q 1 + q + 2q + q + q 1+q+q +q 1
Figure 3.7. The top of the q-Pascal triangle of q-binomial coefficients
Pascal’s triangle reminds us of Pascal’s relation, so we might ask

whether the q-binomial coefficients satisfy some sort of q-Pascal re-
lation. It is clear from the data in Figure 3.7 that the entries in the
q-Pascal triangle are not just the sum of the two entries above them,
as in Pascal’s triangle. But we can hope (nk)q is the sum of the entry
above and to its right, which is (n−1
k q
) , and something else. To see
whether this might be the case, we look at the differences (nk)q −(n−1
k q
) ,
which we have arranged in a triangle in Figure 3.8. Many of these
polynomials have powers of q as factors. When we pull these factors
out, and omit the 0’s from the left edge of the triangle, we obtain the
triangle in Figure 3.9. It appears from these results that
n n−1 n−1
( ) −( ) = q n−k ( ) .
k q k q k−1 q
We can prove this by splitting Bn,k into those sequences which begin
with 0 and those which begin with 1.
1
0 1
0 q 1
2 2
0 q q+q 1
3 2 3 4 2 3
0 q q +q +q q+q +q 1
Figure 3.8. The differences (n) − (n−1

k q
)
k q
1
q(1) 1
2 q(1 + q)
q (1) 1
3 2 2 2
q (1) q (1 + q + q ) q(1 + q + q ) 1
Figure 3.9. The differences (n) − (n−1

k q
) , rewritten
k q
Proposition 3.23. For all n ≥ 0, we have (n0 )q = (nn)q = 1, and for

all 1 ≤ k ≤ n − 1, we have
n n−1 n−1
(3.23) ( ) =( ) + q n−k ( ) .
k q k q k−1 q
Proof. Since Bn,0 = {0n } and Bn,n = {1n }, and 0n and 1n have no
inversions, we have (n0 )q = 1 and (nn)q = 1.
Now suppose n ≥ 1 and 1 ≤ k ≤ n − 1. Let Bn,k 0
be the set of
π ∈ Bn,k whose leftmost entry is 0, and let Bn,k be the set of π ∈ Bn,k
1
whose leftmost entry is 1. Note that each π ∈ Bn,k is in exactly one

0 1
of Bn,k and Bn,k .
If π ∈ Bn,k
0
, then removing the leftmost entry of π leaves a se-
0
quence in Bn−1,k , and this map is a bijection from Bn,k to Bn−1,k .
Furthermore, if π ∈ Bn,k , then the leftmost entry of π is not part of
0
any inversions in π. As a result, our removal map does not change

the inversion number. Therefore, the ordinary generating function
0
for Bn,k with respect to inv is (n−1
k q
) .
If π ∈ Bn,k
1
, then removing the leftmost entry of π leaves a se-
1
quence in Bn−1,k−1 , and this map is a bijection from Bn,k to Bn−1,k−1 .
However, if π ∈ Bn,k , then the leftmost entry of π forms an inversion
1
with every 0 in π. As a result, removing this entry reduces the in-

version number by the number of 0’s in π, which is n − k. Since this
happens for every π ∈ Bn,k
1 1
, the ordinary generating function for Bn,k
with respect to inv is q (k−1)q .
n−k n−1
0 1
Since Bn,k is the disjoint union of Bn,k and Bn,k , we now see that
equation (3.23) holds. □
There is no reason to favor the entry above and to the right of

(nk)q in the analysis that led us to Proposition 3.23. In fact, looking
instead at (nk)q − (n−1 ) leads to a different q-Pascal identity, which
k−1 q
we ask you to find and prove in Problem 3.21.
In addition to satisfying a natural recurrence relation, the bi-
nomial coefficients also have a simple formula in terms of factorials:
(nk) = k!(n−k)!
n!
. It turns out the q-binomial coefficients have an analo-
gous formula, involving a q-analogue of the factorial. To see what
this q-analogue of the factorial should be, we look at the entries
(n1 )q of the q-Pascal triangle in Figure 3.7. We know (n1 ) = n, so
we can hope (n1 )q is an appropriate q-analogue of n. It appears that
(n1 )q = 1 + q + ⋅ ⋅ ⋅ + q n−1 , so we make the following definition.
Definition 3.24. For any positive integer n, we define the q-integer

[n]q by setting [n]q = 1 + q + ⋅ ⋅ ⋅ + q n−1 , and we define the q-factorial
[n]q ! by setting [n]q ! = [n]q [n−1]q ⋅ ⋅ ⋅ [2]q [1]q . By convention, [0]q ! =
1.
1−q n
We note that [n]q = 1−q
.
As we might hope, we can replace the factorials in our formula
for (nk) with q-factorial to obtain a formula for (nk)q .
Proposition 3.25. For all integers n and k with 0 ≤ k ≤ n, we have

n [n]q !
(3.24) ( ) = .
k q [k]q ![n − k]q !
Now that we have the q-Pascal relation for the q-binomial coef-
ficients, we can compare it with equations (3.1) and (3.2) to try to
find a specialization of x1 , . . . , xn which will give us the q-binomial
coefficients. Equation (3.1) turns out to be more complicated than
we would like, but if we make the guess (in analogy with (3.9)) that
we will have hk (Xn ) = (n+k−1
k
)q , then (3.2) tells us
n+k−1 n+k−2 n+k−2

( ) =( ) + xn ( ) .
k q k q k−1 q
On the other hand, (3.23) says

n+k−1 n+k−2 n+k−2
( ) =( ) + q n−1 ( ) ,
k q k q k−1 q
so we can guess xn = q n−1 , and in general xj = q j−1 . In particular, we
have the following result.

n+k−1
(3.25) hk (1, q, q 2 , . . . , q n−1 ) = ( ) .
k q
Proof. We argue by induction on n + k.

If n + k = 1, then n = 1 and k = 0. In this case we have h0 (1) = 1
and (1+k−1
0
)q = (00)q = 1, so (3.25) holds in this case.
Now suppose k ≥ 0 and n ≥ 1 are given, and (3.25) holds for all
N ≥ 1 and all K ≥ 0 with N + K < n + k. By (3.2), our induction
hypothesis, and (3.23) we have
hk (1, q, q 2 , . . . , q n−1 ) = hk (1, q, q 2 , . . . , q n−2 )
+ q n−1 hk−1 (1, q, q 2 , . . . , q n−1 )
n+k−2 n+k−2
=( ) + q n−1 ( )
k q k−1 q
n+k−1
=( ) ,
k q
Now that we know we want xj = q j−1 , we should also evaluate

ek (1, q, q 2 , . . . , q n−1 ). We could hope ek (1, q, q 2 , . . . , q n−1 ) = (nk)q , but
this turns out to be too optimistic. For example, e3 (1, q, q 2 , q 3 ) =
q 3 +q 4 +q 5 +q 6 = q 3 (43)q has an extra power of q which does not appear
in (43)q . In fact, ek (1, q, q 2 , . . . , q n−1 ) is always a sum of powers of q,
and the smallest power of q which can appear is 1 ⋅ q ⋅ q 2 ⋅ ⋅ ⋅ q k = q (2) .
k
So the best we can hope for is the following result.
ek (1, q, q 2 , . . . , q n−1 ) = q (2) ( ) .

k n
(3.26)
k q
We can now combine our evaluations of ek (1, q, q 2 , . . . , q n−1 ) and

hk (1, q, q 2 , . . . , q n−1 ) with the identities in Section 3.1 to find several
q-binomial coefficient identities. For example, we have the following
q-analogue of the hockey stick identity.
Proposition 3.28. For all n ≥ 1 and all k ≥ 0, we have
n+k−1 k
n + k − j − 1 [nj]q
(3.27) k( ) = ∑( ) .
k q j=1 k−j q [j]q
Proof. Evaluate (3.6) with xr = q r−1 for 1 ≤ r ≤ n and xr = 0 for

r > n. Use (3.25) to eliminate hk and hk−j , and use the fact that
pj (1, q, q 2 , . . . , q n−1 ) = 1 + q j + q 2j + ⋅ ⋅ ⋅ + q j(n−1)

1 − q jn 1 − q
= ⋅
1 − q 1 − qj
[jn]q
=
[j]q
to eliminate pj . □
Equation (3.27) is a fine q-analogue of the hockey stick identity,

in the sense that if we set q = 1 and do a little algebra, we really
can recover the original hockey stick identity. But there is another,
simpler, q-analogue of the hockey stick identity, which we ask you to
prove in Problem 3.24.
Many other binomial coefficient identities have q-analogues, in-
cluding the binomial theorem. Happily, evaluating symmetric poly-
nomials gives us two versions of the q-binomial theorem.

n
(1 + t)(1 + qt) ⋅ ⋅ ⋅ (1 + q n−1 t) = ∑ q (2) ( ) tj .
j n
(3.28)
j=0 k q
Proof. Set xj = q j−1 for 1 ≤ j ≤ n and xj = 0 for j > n in (2.6), and

use (3.26) to simplify the result. □
3.5. Problems 71

∞
1 n+j−1 j
(3.29) = ∑( ) t .
(1 − t)(1 − qt) ⋅ ⋅ ⋅ (1 − q t) j=0
n−1 j q
Proof. This is similar to the proof of (3.28), using (2.11) and (3.25).
□
3.5. Problems
3.1. Prove equation (3.2). That is, prove that for all k ≥ 0 and all
n ≥ 1, we have
hk (Xn ) = hk (Xn−1 ) + xn hk−1 (Xn ).
3.2. Find the images of each of the pairs of tiles in Figure 3.10 under
the sign-reversing involution we used to prove (3.4).
2 3∗ 5 1 1 3 4 6
2 3∗ 5 2 2 3 4 5
2∗ 3 5 3 3 3 3 3
Figure 3.10. The pairs of tiles for Problem 3.2
3.3. Use a sign-reversing involution to prove (3.5), which says that

for k ≥ 1, we have
n
pk = ∑ (−1)k+j jek−j hj .
j=1
3.4. Use equation (3.4) to prove (3.5) directly.

3.5. Use generating functions to prove (3.4), as in our first proof of
(2.12).
3.6. Use a sign-reversing involution to prove (3.7), which says

k
kek = ∑ (−1)j−1 ek−j pj .
j=1

(2.12).
(2.12).
3.9. Prove that for all k ≥ 1, we have
kek = ∑ (−1)k1 k2 hk1 ek2 ek3 .
k1 +k2 +k3 =k
0≤k1 ,k2 ,k3 ≤k
3.10. Prove that for all k ≥ 1, we have

khk = ∑ (−1)k1 +k2 k3 ek1 ek2 hk3 hk4 hk5 .
k1 +k2 +k3 +k4 +k5 =k
0≤k1 ,k2 ,k3 ,k4 ,k5 ≤k
3.11. Prove that for all k ≥ 0 and all n ≥ 1, we have

k
n−j
ek (x1 + t, . . . , xn + t) = ∑ ( )ej (Xn )tk−j
j=0 k−j
and
k
n−1+k
hk (x1 + t, . . . , xn + t) = ∑ ( )hj (Xn )tk−j .
j=0 k − j
3.12. Prove (3.8), which says that for n ≥ 1 and k ≥ 0, we have

n
ek (1, . . . , 1) = ( ).
´¹¹ ¹ ¹ ¹ ¹¸ ¹ ¹ ¹ ¹ ¹¶ k
n
3.13. Use symmetric functions to show that for all n, k ≥ 1, we have

k−1
n n−1
∑ (−1) ( )=( ).
k−j−1
j=0 j k−1
3.14. Use Problem 3.11 to prove the binomial theorem.
3.15. Prove Proposition 3.14.
3.16. Prove equation (3.18).
3.6. Notes 73
3.17. (a) Prove that for all n ≥ 1, we have

n+1
1+2+⋅⋅⋅+n=( ),
2
2n + 1 n + 1
1 2 + 2 2 + ⋅ ⋅ ⋅ + n2 = ( ),
3 2
and
n+1 2
1 3 + 2 3 + ⋅ ⋅ ⋅ + n3 = ( ) .
2
(b) Use the results in part (a) to find formulas for [n−1
n
], [n−2
n
],
and [n−3].
n

∑ q
inv(π)
= [n]q !.
π∈Sn
3.19. In any sequence a1 , . . . , an of positive integers, a descent is a

number j with aj > aj+1 . The major index of the sequence,
written maj(a1 , . . . , an ), is the sum of the descents. Show that
for all n ≥ 1, we have
∑ q
maj(π)
= ∑ q inv(π) .
π∈Sn π∈Sn
3.20. Give a combinatorial interpretation of (nk)q when q = −1.

3.21. Conjecture a formula for (nk)q − (n−1 ) , and give a second q-
k−1 q
Pascal relation by proving your conjecture.
3.24. Prove the following q-analogue of the hockey stick identity:
n+k−1 k−1
n+j−1
( ) = ∑ qj ( ) .
k − 1 q j=0 j q
3.6. Notes
Some authors prefer to use s(n, k) to denote the signed Stirling num-
bers of the first kind, c(n, k) to denote ∣s(n, k)∣, and S(n, k) to denote
the Stirling number of the second kind. For our purposes, Knuth’s no-
tation [Knu92], which is also used by Benjamin and Quinn [BQ03],
is more appropriate, because it emphasizes the analogy between the

Stirling numbers and the binomial coefficients.
There are many proofs of the q-binomial theorem, some more
combinatorial than ours. For example, Bressoud gives a proof using
generating functions for (integer) partitions in his book on alternating
sign matrices [Bre99].
Chapter 4
Schur Polynomials and

Schur Functions
So far we have four different bases for the space Λk of homogeneous

symmetric functions of degree k: the monomial symmetric functions,
the elementary symmetric functions, the complete homogeneous sym-
metric functions, and the power sum symmetric functions. Each of
these bases has its advantages and disadvantages, but it turns out
there is a fifth basis which is more elegant than any basis we have
found so far. In this chapter we construct this basis in two different
ways—one combinatorial and the other using a ratio of determinants.
4.1. Schur Functions and Semistandard

Tableaux
In Chapter 2.16 and 2.17 we described analogous combinatorial in-
terpretations of the coefficients we obtain when we write the com-
plete homogeneous symmetric functions as linear combinations of the
monomial symmetric functions. As we show next, we can use these
combinatorial interpretations to give new definitions of both the el-
ementary and the complete homogeneous symmetric functions. To
state these results, we adopt the convention that for any filling T of
a Ferrers diagram with positive integers, we write xT to denote the
75
76 4. Schur Functions
1 4
2 2
3 6 1 6
Figure 4.1. A filling of the Ferrers diagram for (4, 22 ) with

associated monomial x21 x22 x3 x4 x26
monomial given by xT = ∏j∈T xj . For example, the monomial associ-

ated with the filling of the Ferrers diagram for (4, 22 ) in Figure 4.1 is
x21 x22 x3 x4 x26 . For any partition λ and any set A of positive integers,
we also write StrictRow(λ, A) to denote the set of fillings of the Fer-
rers diagram of λ with entries in A in which the entries in each row
are strictly increasing from left to right.
Proposition 4.1. For any partition λ and any integer n ≥ 1, we have
(4.1) eλ (Xn ) = ∑ xT
T ∈StrictRow(λ,[n])
and
(4.2) eλ = ∑ xT .
T ∈StrictRow(λ,P)
Proof. By equation (2.2) we have

l(λ)
eλ (Xn ) = ∏ eλj (Xn ).
j=1
The terms in the expansion of the product on the right are the prod-
l(λ)
ucts ∏j=1 tj , where tj is a term in eλj (Xn ). For each such product we
can construct a filling of the Ferrers diagram of λ of the given type by
placing, for 1 ≤ j ≤ l(λ), the subscripts of the variables which appear
in tj in increasing order in the jth row of the diagram. We can also
invert this construction: if we have a filling of the given type, then for
each j with 1 ≤ j ≤ l(λ) we can reconstruct tj as the product ∏k xk ,
which is over all k which appear in the jth row of the filling. There-
fore, we have a bijection between terms in eλ (Xn ) and our fillings of
the Ferrers diagram of λ. Moreover, by construction the image of a
filling T under this map is xT , and equation (4.1) follows.
The proof of (4.2) is similar to the proof of (4.1). □
4.1. Schur Functions and Semistandard Tableaux 77
4 4
1 4 6 1 2 4
1 2 4 1 4 6
Figure 4.2. The fillings of the Ferrers diagram of (32 , 1) for

the term x21 x2 x34 x6 in e331
Example 4.2. Find all fillings of the Ferrers diagram of (32 , 1) as-
sociated with the term x21 x2 x34 x6 in e331 .
Solution. Since e331 = e3 e3 e1 , we need to write x21 x2 x34 x6 as a product

of three factors, of degrees 3, 3, and 1, respectively. No subscripts are
repeated within a factor, so each factor must include x4 . Since the
last factor has degree 1, it cannot also include x1 , so each of the first
two factors must include x1 . Now there are two ways to complete
the factorization: x21 x2 x34 x6 = (x1 x2 x4 )(x1 x4 x6 )x4 and x21 x2 x34 x6 =
(x1 x4 x6 )(x1 x2 x4 )x4 . The associated fillings are as in Figure 4.2.
□
To state the corresponding result for the complete homogeneous

symmetric functions, we need some additional notation. For any par-
tition λ and any set A of positive integers, we write WeakRow(λ, A)
to denote the set of fillings of the Ferrers diagram of λ with entries
in A in which the entries in each row are weakly increasing from left
to right.
Proposition 4.3. For any partition λ and any integer n ≥ 1, we have
(4.3) hλ (Xn ) = ∑ xT
T ∈WeakRow(λ,[n])
and
(4.4) hλ = ∑ xT .
T ∈WeakRow(λ,P)
Our characterizations of eλ and hλ in equations (4.2) and (4.4)

both involve requirements on the rows of fillings of the Ferrers diagram
of λ, but include no restrictions on the columns of the filling. If we
move one of these requirements from the rows to the columns, then we
can define a new type of filling of a Ferrers diagram which interpolates
between these two old types of fillings.
Definition 4.4. For any partition λ, a semistandard tableau of shape
λ is a filling of the Ferrers diagram of λ in which the entries in the
columns are strictly increasing from bottom to top and the entries
in the rows are weakly increasing from left to right. If T is a se-
mistandard tableau, then we write sh(T ) to denote the shape of T .
We write SST(λ; n) to denote the set of all semistandard tableaux of
shape λ with entries in [n], and we write SST(λ) to denote the set
of all semistandard tableaux of shape λ with entries in P.
When n and ∣λ∣ are small, we can list all of the semistandard
tableaux in SST(λ; n) by hand.
Example 4.5. Find all semistandard tableaux of shape (22 , 1) with
entries in [3].
Solution. Each column contains a strictly increasing sequence of ele-

ments of {1, 2, 3}, so the first column must contain 1, 2, and 3 from
bottom to top. We can place any strictly increasing sequence of ele-
ments of {1, 2, 3} in the second column without violating the condition
that the rows be nondecreasing, so we have the semistandard tableaux
in Figure 4.3. □
3 3 3
2 2 2 3 2 3
1 1 1 1 1 2
Figure 4.3. The semistandard tableaux of shape (22 , 1) with

entries in [3]
We will sometimes want to discuss fillings of a Ferrers diagram

which have some of the properties of a semistandard tableau, but
possibly not all of them. To simplify matters when this happens, we
say a filling of a Ferrers diagram is column-strict whenever the entries
in each of its columns are increasing from bottom to top. Similarly,
we say a filling of a Ferrers diagram is row-nondecreasing whenever

the entries in each of its rows are weakly increasing from left to right.
Inspired by Propositions 4.1 and 4.3, we can now use our se-
mistandard tableaux to construct some new polynomials and formal
power series, which we hope will be symmetric polynomials and sym-
metric functions, respectively.
Definition 4.6. For any partition λ and any integer n ≥ 1, we write

sλ (Xn ) to denote the polynomial
(4.5) sλ (Xn ) = ∑ xT .
T ∈SST(λ;n)
Similarly, we write sλ to denote the formal power series
sλ = ∑ xT .
T ∈SST(λ)
We call sλ (Xn ) the Schur polynomial in x1 , . . . , xn indexed by λ and

we call sλ the Schur function indexed by λ.
Propositions 4.1 and 4.3 give us two families of examples of Schur

polynomials and Schur functions.
Example 4.7. Compute s1k (Xn ) and s1k .
Solution. The polynomial s1k (Xn ) is a sum over all fillings of a single
column with distinct integers in [n], in which the entries are strictly
increasing from bottom to top. If n < k, then there are no such
fillings, so our sum has no terms, and s1k (Xn ) = 0 in this case. If
n ≥ k, then there is exactly one such filling for each subset of [n]
of size k, and the monomial corresponding to a subset J is ∏j∈J xj .
Therefore, s1k (Xn ) = ek (Xn ) in this case. Similarly, s1k = ek . □
Example 4.8. Compute sk (Xn ) and sk .
Solution. The polynomial sk (Xn ) is a sum over all fillings of a single

row with integers in [n], in which the entries are weakly increasing
from left to right. There is exactly one such filling for each submultiset
of [n] of size k, and the monomial corresponding to such a submultiset
J is ∏j∈J xj . Therefore, sk (Xn ) = hk (Xn ). Similarly, sk = hk . □
In Examples 4.7 and 4.8 we saw that the Schur polynomials

(resp., functions) interpolate between the elementary symmetric poly-
nomials (resp., functions) and the complete homogeneous symmetric
polynomials (resp., functions). In our next few examples we look at
some of the Schur polynomials and Schur functions between these
extremes.
Example 4.9. Compute s21 (x1 ), s21 (X2 ), and s21 (X3 ). Write those
which are symmetric polynomials as linear combinations of the mono-
mial symmetric polynomials.
Solution. The polynomial s21 (x1 ) is a sum over all semistandard

tableaux of shape (2, 1) with entries in [1]. The leftmost column
of any such tableau must contain two distinct integers, since its en-
tries are strictly increasing from bottom to top. But we have only
two integers available, so there are no semistandard tableaux of the
sort we want. This means our sum has no terms, so by convention
s21 (x1 ) = 0.
To compute s21 (X2 ), we write down (in Figure 4.4) the two se-
mistandard tableaux of shape (2, 1) with entries in [2]. When we
write down the corresponding terms we find s21 (X2 ) = x21 x2 + x1 x22 =
m21 (X2 ).
2 2
1 1 1 2
Figure 4.4. The two semistandard tableaux of shape (2, 1)

with entries in [2]
Finally, we use the semistandard tableaux of shape (2, 1) with

entries in [3], which are shown in Figure 4.5, to find s21 (X3 ) =
m21 (X3 ) + 2m111 (X3 ). □
Example 4.10. Compute s221 (X2 ), s221 (X3 ), and s221 (X4 ). Write
those which are symmetric polynomials as linear combinations of the
monomial symmetric polynomials.
Solution. The polynomial s221 (X2 ) is a sum over all semistandard

tableaux of shape (22 , 1) with entries in [2]. The leftmost column
2 2 2
1 1 1 2 1 3
3 3 3
1 1 1 2 1 3
3 3
2 2 2 3
Figure 4.5. The eight semistandard tableaux of shape (2, 1)

with entries in [3]
of any such tableau must contain three distinct integers, since its
entries are strictly increasing from bottom to top. But we have only
two integers available, so there are no semistandard tableaux of the
sort we want. This means our sum has no terms, so by convention
s221 (X2 ) = 0.
To compute s221 (X3 ), we notice it is a sum over the set of semin-
standard tableaux in Example 4.5. Therefore, we have s221 (X3 ) =
x21 x22 x3 + x21 x2 x23 + x1 x22 x23 , which is m221 (X3 ).
To compute s221 (X4 ), we write down (in Figure 4.6) all 20 se-
mistandard tableaux of shape (22 , 1) with entries in [4]. Collecting
terms, we find
s221 (X4 ) = m221 (X4 ) + 2m2111 (X4 ).
We can also write down all 75 semistandard tableaux of shape

(22 , 1) with entries in [5] and collect terms to find
s221 (X5 ) = m221 (X5 ) + 2m2111 (X5 ) + 5m11111 (X5 ). □
All of the Schur polynomials we’ve seen so far are symmetric

polynomials, and all of the Schur functions we’ve seen are symmetric
functions. To prove the Schur polynomials are symmetric polynomi-
als and the Schur functions are symmetric functions, it will be helpful
to keep track of the number of entries of each type in a given semi-
standard tableau.
3 3 3 3 3 3
2 2 2 3 2 4 2 3 2 4 2 4
1 1 1 1 1 1 1 2 1 2 1 3
4 4 4 4 4 4
2 2 2 3 2 4 2 3 2 4 2 4
1 1 1 1 1 1 1 2 1 2 1 3
4 4 4 4 4
3 3 3 4 3 3 3 4 3 4
1 1 1 1 1 2 1 2 1 3
4 4 4
3 3 3 4 3 4
2 2 2 2 2 3
Figure 4.6. The 20 semistandard tableaux of shape (22 , 1)

with entries in [4]
Definition 4.11. For any semistandard tableau T , the content of T

is the sequence {µj }∞
j=1 , where µj is the number of j’s in T for each
j. When µj = 0 for j > n, then we sometimes abbreviate {µj }∞ j=1 as
{µj }nj=1 .
Our proof that the Schur polynomials and Schur functions are
symmetric uses a collection of combinatorial maps showing these poly-
nomials and formal power series are invariant under certain permuta-
tions. We combine these with an algebraic argument that it is enough
to check the result for only these permutations. We start with the
maps, which are called the Bender–Knuth involutions.
For each j ≥ 1, the Bender–Knuth involution βj is a function
from the set of semistandard tableaux with shape λ and content
{µl }∞
l=1 to the set of semistandard tableaux with shape λ and con-
tent µ1 , . . . , µj−1 , µj+1 , µj , µj+2 , . . . . To describe βj , suppose T is a
semistandard tableau with shape λ and content {µl }∞ l=1 , and consider
the columns of T . We only care about j’s and j + 1’s, so for us there
are only four types of columns: those containing both a j and a j + 1,
those containing a j but not a j + 1, those containing a j + 1 but not a
j, and those containing neither a j nor a j +1. We call a j (resp., j +1)
in T paired whenever there is a j + 1 (resp., j) in its column, and free
otherwise.
Now consider a row R of T . Like every row of T , the row R
has a number (possibly 0) of j’s, followed immediately by a number
(also possibly 0) of j + 1’s. Immediately above the j’s in R are some
(possibly no) j + 1’s, and then some (also possibly no) larger entries.
Therefore, in R we have a number of paired j’s, followed by a number
of free j’s. Similarly, immediately below the j + 1’s in R, we have
a number (possibly 0) of entries less than j, followed by a number
(possibly 0) of j’s. Therefore, in T we have a number of free j + 1’s,
followed by a number of paired j + 1’s.
To construct βj (T ), we start with the bottom row of T and move
to the top, doing the same thing to every row. Namely, if a row has
a free j’s followed by b free j + 1’s, then we replace that sequence of
free j’s and free j + 1’s with a sequence of b free j’s followed by a free
j + 1’s.
5
4 4 5 5 7 7 7
3 3 3 3 3 3 4 4 4 4
2 2 2 2 2 2 2 2 3 3 3 4 4 4 5 7
1 1 1 1 1 1 1 1 1 2 2 2 2 3 4 6
Figure 4.7. The tableau T for Example 4.12
Example 4.12. Find the image of the tableau T in Figure 4.7 under
the Bender–Knuth involution β3 .
Solution. In Figure 4.8 we have written the free 3’s and 4’s in T in
bold, and larger than the other entries. In the bottom row we have
5
4 4 5 5 7 7 7
3 3 3 3 3 3 4 4 4 4
2 2 2 2 2 2 2 2 3 3 3 4 4 4 5 7
1 1 1 1 1 1 1 1 1 2 2 2 2 3 4 6
Figure 4.8. The tableau T for Example 4.12 with the free
3’s and 4’s in bold
no free 3’s and one free 4, so we replace the free 4 with a (free) 3.
In the second row from the bottom we have one free 3 and two free
4’s, so we replace them with two (free) 3’s and one (free) 4. And in
the third row from the bottom we have four free 3’s and two free 4’s,
which we replace with two (free) 3’s and four (free) 4’s. When we’re
done, we have the tableau in Figure 4.9. □
5
4 4 5 5 7 7 7
3 3 3 3 4 4 4 4 4 4
2 2 2 2 2 2 2 2 3 3 3 3 4 4 5 7
1 1 1 1 1 1 1 1 1 2 2 2 2 3 3 6
Figure 4.9. The tableau β3 (T ) for Example 4.12
The Bender–Knuth involutions have several properties which will

be important to us, so we develop these next.
Lemma 4.13. If T is a semistandard tableau with shape λ and con-
tent {µl }∞
l=1 and j ≥ 1, then β(T ) is a semistandard tableau with shape
λ and content µ1 , . . . , µj−1 , µj+1 , µj , µj+2 , . . . .
Proof. By construction, T and βj (T ) have the same shape, and the

entries in each row of βj (T ) are weakly increasing from left to right.
To see βj (T ) is column-strict, first consider a box which contains a j
in T and a j + 1 in βj (T ). Since the j changed to a j + 1, it must have
been free in T , so the entry immediately above it in both T and βj (T )
is greater than j + 1. Therefore the entries in that column of βj (T )

are strictly increasing from bottom to top. Similarly, only free j + 1’s
in T can change to j’s in βj (T ), so the entries in a column in which
a j + 1 changed to a j are also strictly increasing from bottom to top.
All of the other columns of βj (T ) have the same entries as they do
in T , so βj (T ) is column-strict. Therefore, βj (T ) is a semistandard
tableau of shape λ.
To find the content of βj (T ), we first notice that if a box in T
does not contain a j or a j + 1, then it contains the same number in
T as it does in βj (T ). Therefore if l ≠ j and l ≠ j + 1, then βj (T )
has µl l’s. By construction, the number of j’s in βj (T ) is the number
of paired j’s in T plus the number of free j + 1’s in T . But the
number of paired j’s in T is equal to the number of paired j + 1’s
in T , since every column of T contains both a paired j and a paired
j + 1, or neither a paired j nor a paired j + 1. Therefore the number
of j’s in βj (T ) is equal to the number of paired j + 1’s in T plus the
number of free j + 1’s in T , which is µj+1 . Similarly (or using the
fact that T and βj (T ) have the same total number of entries), the
number of j + 1’s in βj (T ) is µj . Therefore, the content of βj (T ) is
µ1 , . . . , µj−1 , µj+1 , µj , µj+2 , . . . . □
Lemma 4.14. If T is semistandard tableau, then for any j ≥ 1 we

have βj (βj (T )) = T . In particular, βj is a bijection between the set of
semistandard tableaux with shape λ and content {µl }∞ l=1 and the set
of semistandard tableaux with shape λ and content µ1 , . . . , µj−1 , µj+1 ,
µj , µj+2 , . . . .
Proof. First note that applying the action of βj to a single row does
not change which j’s and j +1’s are free in any row. (As an aside, this
means we could apply the action of βj to the rows in any order, and
we would get the same tableau at the end.) Second, because βj only
changes j’s (resp., j + 1’s) with no j + 1 (resp., j) in their columns, if
a row in T has a free j’s and b free j + 1’s, then in βj (T ) it has b free
j’s and a free j + 1’s. Therefore, each row returns to its original state
if we apply βj again, so βj (βj (T )) = T . In particular, βj = βj−1 , so βj
is a bijection. □
The Bender–Knuth involutions have additional interesting prop-

erties, which we explore in Problems 4.6, 4.7, 4.8, and 4.9. For now,
though, we explain how to use the Bender–Knuth involutions to prove
the Schur polynomials and Schur functions are symmetric.
Proposition 4.15. Suppose λ ⊢ k and n ≥ 1. Then we have sλ (Xn ) ∈

Λk (Xn ) and sλ ∈ Λk .
Proof. We first show sλ (Xn ) ∈ Λk (Xn ).

If n < l(λ), then there are no semistandard tableaux of shape λ
with entries in [n], so the sum sλ (Xn ) has no terms and sλ (Xn ) =
0 ∈ Λk (Xn ).
Now suppose n ≥ l(λ). By construction, every term in sλ (Xn ) is
homogeneous of degree k = ∣λ∣, so we just need to show π(sλ (Xn )) =
sλ (Xn ) for all π ∈ Sn . To start, suppose π = (j, j + 1) for some j
with 1 ≤ j ≤ n − 1, and choose a monomial xµ1 1 ⋅ ⋅ ⋅ xµnn with total de-
gree µ1 + ⋅ ⋅ ⋅ + µn = ∣λ∣. (Notice these are the only terms that could
appear in sλ (Xn ) or π(sλ (Xn )).) By definition the coefficient of
xµ1 1 ⋅ ⋅ ⋅ xµnn in sλ (Xn ) is the number of semistandard tableaux with
shape λ and content µ = µ1 , . . . , µn . On the other hand, the coefficient
µ µj
of xµ1 1 ⋅ ⋅ ⋅ xµnn in π(sλ (Xn )) is the coefficient of xµ1 1 ⋅ ⋅ ⋅ xj j+1 xj+1 ⋅ ⋅ ⋅ xµnn
in sλ (Xn ), which is the number of semistandard tableaux with shape
λ and content µ1 , . . . , µj−1 , µj+1 , µj , µj+2 , . . . , µn . By Lemma 4.14
these coefficients are equal. Therefore, every term xµ1 1 ⋅ ⋅ ⋅ xµnn has
the same coefficient in both sλ (Xn ) and π(sλ (Xn )), so we must have
sλ (Xn ) = π(sλ (Xn )).
Now suppose π is an arbitrary permutation in Sn . By Proposition
C.5 there are positive integers j1 , . . . , jl such that
π = (j1 , j1 + 1)(j2 , j2 + 1) ⋅ ⋅ ⋅ (jl , jl + 1).
When we apply the result of the previous paragraph l times, we find
π(sλ (Xn )) = (j1 , j1 + 1)(j2 , j2 + 1) ⋅ ⋅ ⋅ (jl , jl + 1)(sλ (Xn ))

= sλ (Xn ),
so sλ (Xn ) is invariant under π, which is what we wanted to prove.

A similar argument shows sλ ∈ Λk . □
Now that we know sλ (Xn ) is always a symmetric polynomial and

sλ is always a symmetric function, our solution to Example 4.12 raises
an interesting question: Are the algebraic relationships between the
Schur polynomials and the monomial symmetric polynomials eventu-
ally independent of the number of variables? After all, increasing the
number of variables in that example added terms to our answer. To
make this question more precise, we make the following definition.
Definition 4.16. For any partitions λ and µ, we write Kλ,µ to denote

the number of semistandard tableaux of shape λ and content µ, and
we call the numbers Kλ,µ the Kostka numbers.
Note that for any partition λ we have

(4.6) sλ = ∑ Kλ,µ mµ .
µ⊢∣λ∣
As a result, if we followed our previous convention, then we would

write Mλ,µ (s, m) instead of Kλ,µ . However, Kλ,µ is the traditional
notation for these numbers, so we will stick with it. Our next result
answers our question about the algebraic relationships between the
Schur polynomials and the monomial symmetric polynomials.
Proposition 4.17. For any partitions λ and µ with ∣λ∣ = ∣µ∣ and any
n ≥ 1, let Kλ,µ,n be the number of semistandard tableaux with shape
λ and content µ, with entries in [n]. If n ≥ ∣λ∣, then Kλ,µ,n = Kλ,µ .
In particular, if n ≥ ∣λ∣, then Kλ,µ,n is independent of n.
Proof. Since µ is a partition and n ≥ ∣λ∣ = ∣µ∣, we must have µj = 0 for

j ≥ n. Therefore, any semistandard tableau with shape λ and content
µ contains no entry larger than n. This means Kλ,µ,n and Kλ,µ are
counting the same objects, so Kλ,µ,n = Kλ,µ . □
We would now like to show the Schur functions of degree k are a

basis for Λk . As we did for the elementary and power sum symmetric
functions, we do this by examining the coefficients we obtain when we
write sλ as a linear combination of monomial symmetric functions.
That is, we look at matrices of Kostka numbers, as in Figure 4.10.
As we might have hoped, when we arrange our partitions in in-
creasing lexicographic order, our matrices of Kostka numbers appear
(12 ) (2)
(1)
(12 ) 1 0
(1) [ 1 ] [ ]
(2) 1 1
(13 ) (2, 1) (3)

(1 ) ⎡⎢ 1
3
0 0 ⎤⎥
⎢ ⎥
(2, 1) ⎢⎢ 2 1 0 ⎥⎥
(3) ⎢⎣ 1 1 1 ⎥⎦
(14 ) (2, 12 ) (22 ) (3, 1) (4)

(14 ) ⎡⎢ 1 0 0 0 0 ⎤⎥
⎢ ⎥
(2, 12 ) ⎢⎢ 3 1 0 0 0 ⎥⎥
(2 ) ⎢⎢ 2
2
1 1 0 0 ⎥⎥
⎢ ⎥
(3, 1) ⎢ 3 2 1 1 0 ⎥
⎢ ⎥
(4) ⎢ 1 1 1 1 1 ⎥⎦
⎣
(15 ) (2, 13 ) (22 , 1) (3, 12 ) (3, 2) (4, 1) (5)

(1 ) ⎡⎢ 1
5
0 0 0 0 0 0 ⎤⎥
3 ⎢ ⎥
(2, 1 ) ⎢⎢ 4 1 0 0 0 0 0 ⎥⎥
(22 , 1) ⎢⎢ 5 2 1 0 0 0 0 ⎥⎥
⎢ ⎥
(3, 12 ) ⎢⎢ 6 3 1 1 0 0 0 ⎥⎥
(3, 2) ⎢⎢ 5 3 2 1 1 0 0 ⎥⎥
⎢ ⎥
(4, 1) ⎢ 4 3 2 2 1 1 0 ⎥
⎢ ⎥
(5) ⎢ 1 1 1 1 1 1 1 ⎥⎦
⎣
Figure 4.10. The matrices Kλ,µ for ∣λ∣ = ∣µ∣ ≤ 5
to be lower triangular with 1’s on their diagonals. In our next result

we prove this pattern continues.
Proposition 4.18. For all k ≥ 0, the following hold for all partitions
λ, µ ⊢ k.
(i) If µ >lex λ, then Kλ,µ = 0.

(ii) Kλ,λ = 1.
4.2. Schur Polynomials as Ratios of Determinants 89
Proof. (i) If µ >lex λ, then by definition there exists m such that

µm > λm and µj = λj for 1 ≤ j ≤ m − 1. No semistandard tableau can
have a 1 in its second row or higher, a 2 in its third row or higher, or a
j in its j + 1th row or higher, so a filling T of λ with content µ can be
a semistandard tableau only if it has j’s in every entry of the jth row
for 1 ≤ j ≤ m − 1. Since µm > λm , by the pigeonhole principle some
column of T must contain two m’s, so T cannot be a semistandard
tableau.
(ii) As in the proof of (i), a semistandard tableau of shape λ
and content λ must have j’s in every entry of its jth row for all j.
Since there is only one semistandard tableau like this, we must have
Kλ,λ = 1. □
As we will see in Problem 4.16, the converse of Proposition 4.18(i)

is false. Before we get there, though, we note that the Schur functions
of degree k really are a basis for Λk .
Proposition 4.19. For all k ≥ 0, the set {sλ ∣ λ ⊢ k} is a basis for
Λk .
4.2. Schur Polynomials as Ratios of

Determinants
So far we have used combinatorial methods to produce almost all of
our families of symmetric functions. Now we describe an algebraic
method for producing symmetric polynomials, which we use to give
an alternative definition of the Schur polynomials. We begin with
polynomials which are not symmetric, but which still interact with
permutations in a natural way.
Definition 4.20. Suppose n ≥ 1. We say a polynomial f (Xn ) is
alternating in x1 , . . . , xn whenever π(f ) = sgn(π)f for all π ∈ Sn .
Inspired by our construction of the monomial symmetric poly-

nomials, we can produce alternating polynomials by starting with a
monomial and adding all of its signed images under the elements of
the relevant permutation group.
Example 4.21. Find the alternating polynomial f (X3 ) with the

fewest terms which contains the monomial x31 x22 x3 .
Solution. Since f includes x31 x22 x3 and is alternating, it must also

include sgn(π)π(x31 x22 x3 ) = −x21 x32 x3 , where π = (12). Similarly, f
must include −x31 x2 x23 , −x1 x22 x33 , x1 x32 x23 , and x21 x2 x33 . Since
f (X3 ) = x31 x22 x3 − x21 x32 x3 − x31 x2 x23 − x1 x22 x33 + x1 x32 x23 + x21 x2 x33
is alternating, this is the desired polynomial. □
In contrast with the situation for symmetric polynomials, this

technique does not always produce a new alternating polynomial.
Example 4.22. Show that no alternating polynomial has the mono-
mial x1 x2 as a term.
Solution. If x1 x2 is a term in an alternating polynomial and π = (12),

then sgn(π)π(x1 x2 ) = −x1 x2 is also a term in that polynomial, which
is a contradiction. □
Example 4.22 is one instance of a much more general phenome-

non.
Proposition 4.23. If µ1 , . . . , µn are nonnegative integers such that
µj = µl for some j ≠ l, and f (Xn ) is an alternating polynomial in
Xn , then the coefficient of xµ1 1 xµ2 2 ⋅ ⋅ ⋅ xµnn in f (Xn ) is 0.
Proof. If axµ1 1 xµ2 2 ⋅ ⋅ ⋅ xµnn is a term in f (Xn ), then

sgn((jl))(jl)(axµ1 1 xµ2 2 ⋅ ⋅ ⋅ xµnn ) = −axµ1 1 xµ2 2 ⋅ ⋅ ⋅ xµnn
is also a term in f (Xn ). But if µj = µl , then these terms are equal,
so a = 0. □
Proposition 4.23 tells us we can only construct nonzero alternat-

ing polynomials from monomials if we start with a monomial whose
exponents are distinct. This includes zero exponents: there is no
alternating polynomial in x1 , x2 , x3 , x4 with x31 x4 as a term. How-
ever, as we show next, each monomial with distinct exponents does
in fact produce an alternating polynomial. In contrast with some
of the monomial symmetric polynomials, this alternating polynomial
has exactly one term for each permutation. This, in combination with
the signs associated with each term, allows us to write this alternating
polynomial as a determinant.
Proposition 4.24. If µ is a sequence with µ1 > µ2 > ⋅ ⋅ ⋅ > µn ≥ 0 and
(4.7) aµ (Xn ) = ∑ sgn(π)xµπ(1)
1
xµπ(2)
2
⋅ ⋅ ⋅ xµπ(n)
n
,
π∈Sn
then aµ (Xn ) is an alternating polynomial in x1 , . . . , xn . Moreover,

aµ (Xn ) is homogeneous of degree µ1 + ⋅ ⋅ ⋅ + µn , it has n! terms, and
µ
(4.8) aµ (Xn ) = det (xl j )1≤j,l≤n .
Proof. To show aµ (Xn ) is an alternating polynomial in Xn , suppose

σ ∈ Sn . Then by Proposition 1.2(i),(ii),
σ(aµ ) = ∑ sgn(π)xµσπ(1)
1
⋅ ⋅ ⋅ xµσπ(n)
n
.
π∈Sn
If τ = σπ, then π = σ −1 τ , and as π ranges over Sn , so does τ . There-

fore,
σ(aµ ) = ∑ sgn(σ −1 τ )xµτ (1)
1
⋅ ⋅ ⋅ sµτ (n)
n
.
τ ∈Sn
By Problem C.10, we have sgn(σ −1 τ ) = sgn(σ −1 ) sgn(τ ). Moreover,

we saw in Problem C.9 that inv(σ) = inv(σ −1 ), so sgn(σ) = sgn(σ −1 ).
Thus,
σ(aµ ) = ∑ sgn(σ) sgn(τ )xµτ (1)
1
⋅ ⋅ ⋅ xµτ (n)
n
,
τ ∈Sn
so σ(aµ ) = sgn(σ)aµ , and aµ (Xn ) is an alternating polynomial in Xn .
By construction, each term in aµ (Xn ) has total degree µ1 +⋅ ⋅ ⋅+µn ,
so aµ (Xn ) is homogeneous of degree µ1 + ⋅ ⋅ ⋅ + µn . Since µ1 , . . . , µn
are distinct, the terms of aµ (Xn ) are also distinct. Therefore, there
are n! of them. Finally, equation (4.8) follows immediately from (4.7)
and Proposition C.7. □
Our alternating polynomials aµ (Xn ) are indexed by sequences

µ1 > µ2 > ⋅ ⋅ ⋅ > µn ≥ 0, but it will be more convenient to index
them with partitions. To see how to do this, first consider how large
each µj must be, given that µ1 > µ2 > ⋅ ⋅ ⋅ > µn ≥ 0. For instance,
µn−1 > µn ≥ 0, so µn−1 ≥ 1. But now µn−2 > µn−1 ≥ 1, so µn−2 ≥ 2.
Arguing by induction, we find µn−j ≥ j for 0 ≤ j ≤ n−1. In other words,
our “smallest” indexing sequence is the sequence n − 1, n − 2, . . . , 2, 1,

which we denote by δn . With this in mind, we define a sequence λ
by setting λj = µj − δn (j) for 1 ≤ j ≤ n, and we sometimes abbreviate
λ = µ − δn or µ = λ + δn . By construction, λ is a partition with at most
n parts, and the map taking µ to λ is a bijection, since the inverse
of subtracting δn is adding δn . As a result, we can view aµ (Xn ) as
aλ+δn (Xn ), where λ is a partition with at most n parts.
Separating δn from λ in an indexing sequence µ allows us to see
more clearly the relationship between a generic aλ+δn (Xn ) and our
minimal alternating polynomial aδn (Xn ). Specifically, we can use
the fact that aλ+δn (Xn ) can be written as a determinant to factor
aδn (Xn ) completely and to show aδn (Xn ) always divides aλ+δn (Xn ).

(4.9) aδn (Xn ) = ∏ (xj − xl ).
1≤j<l≤n
Furthermore, for all n ≥ 1 and every partition λ with at most n parts,

there is a symmetric polynomial g(Xn ) with
(4.10) aλ+δn (Xn ) = g(Xn )aδn (Xn ).
Proof. We prove equations (4.9) and (4.10) together, but we start

by considering (4.10).
By (4.8) we have
λ +δn (j)
aλ+δn (Xn ) = det (xl j ) .
1≤j,l≤n
λ +δn (j)
If 1 ≤ k1 < k2 ≤ n and we set xk1 = xk2 in the matrix (xl j ) ,
1≤j,l≤n
then columns k1 and k2 are equal, so the determinant of this new ma-
λ +δ (j)
trix is 0. Therefore, if we view det (xl j n ) as a polynomial
1≤j,l≤n
in xk1 , then it has xk1 − xk2 as a factor. This is true for any k1 and
k2 with 1 ≤ k1 < k2 ≤ n, so
(4.11) aλ+δn (Xn ) = gλ (Xn ) ∏ (xj − xl )
1≤j<l≤n
for some polynomial gλ (Xn ).

If λ is the empty partition, then aλ+δn (Xn ) and ∏1≤j<l≤n (xj − xl )
both have total degree (n2 ), so in this case gλ (Xn ) is a constant. To
see which constant it is, consider the coefficient of xn−1 1 x2

n−2
⋅ ⋅ ⋅ xn−1
on both sides of (4.11). On the left this coefficient is 1, and on the
right it is gλ (Xn ), so in this case gλ (Xn ) = 1. Therefore, (4.9) holds,
and (4.11) can be rewritten as
aλ+δn (Xn ) = gλ (Xn )aδn (Xn ).
To complete the proof of (4.10) we just need to show gλ (Xn )

is a symmetric polynomial. To do this, suppose σ ∈ Sn . Then by
Proposition 1.2(iii) we have
σ(aλ+δn (Xn )) = σ (gλ (Xn )aδn (Xn ))
= σ(gλ (Xn ))σ(aδn (Xn ))
= sgn(σ)σ(gλ (Xn ))aδn (Xn ),
since aδn (Xn ) is alternating. But aλ+δn (Xn ) is also alternating, so
σ(aλ+δn (Xn )) = sgn(σ)aλ+δn (Xn )
= sgn(σ)gλ (Xn )aδn (Xn ).
When we combine these two computations, we find
sgn(σ)σ(gλ (Xn ))aδn (Xn ) = sgn(σ)gλ (Xn )aδn (Xn ).
And since sgn(σ)aδn (Xn ) ≠ 0, we must have σ(gλ (Xn )) = gλ (Xn ),
which means gλ (Xn ) is a symmetric polynomial. □
Our proof of Proposition 4.25 uses a fact about polynomials which

is worth mentioning. Specifically, we assume that if a polynomial in
Xn is divisible by all factors of the form xj − xl , where j < l, then it
is divisible by their product ∏1≤j<l≤n (xj − xl ). This might not hold
in other situations: 12 is divisible by both 6 and 4, but not by their
product. Our assumption turns out to be valid because our factors
xj − xl are irreducible, and polynomials in Xn with coefficients in C
factor uniquely into products of irreducibles.
Equation (4.10) gives us a new family of symmetric polynomials
a n (Xn )
in x1 , . . . , xn : for each partition λ with at most n parts, aλ+δ
δn (Xn )
is
a symmetric polynomial in x1 , . . . , xn . Moreover, since every term in
aλ+δn (Xn ) has degree ∣λ∣ + (n2 ) and every term in aδn (Xn ) has degree
(n2 ), our new symmetric polynomial associated with λ is homogeneous
of degree ∣λ∣. For small λ we can express these symmetric polynomials

in terms of our previous bases.
a (X )
Example 4.26. Suppose λ = (1) and n ≥ 1. Express aλ+δ n n
δn (Xn )
in
terms of the monomial symmetric polynomials, the elementary sym-
metric polynomials, the complete homogeneous symmetric polynomi-
als, the power sum symmetric polynomials, and the Schur polynomi-
als.
Solution. When n = 1, we have

aλ+δn (Xn ) det(x1 )
= = x1 ,
aδn (Xn ) det(1)
so our ratio is equal to m1 (x1 ) = e1 (x1 ) = h1 (x1 ) = p1 (x1 ) = s1 (x1 ).
When n = 2, we have
x2 x22
det ( 1 )
aλ+δn (Xn ) 1 1
=
aδn (Xn ) x1 − x2
= x1 + x2 ,
so our ratio is equal to m1 (X2 ) = e1 (X2 ) = h1 (X2 ) = p1 (X2 ) = s1 (X2 ).
When n = 3, we have
3
⎛x1 x32 x33 ⎞
det ⎜x1 x2 x3 ⎟
aλ+δn (Xn ) ⎝1 1 1⎠
=
aδn (Xn ) (x1 − x2 )(x1 − x3 )(x2 − x3 )
= x1 + x2 + x3 ,
so our ratio is equal to m1 (X3 ) = e1 (X3 ) = h1 (X3 ) = p1 (X3 ) = s1 (X3 ).
Our results so far suggest our ratio is equal to m1 (Xn ) = e1 (Xn ) =
h1 (Xn ) = p1 (Xn ) = s1 (Xn ) for all n. To prove this, first note that
for all n ≥ 2 we can set xn = 0 and expand the resulting determinants
along their last columns to find
aλ+δn (x1 , . . . , xn−1 , 0) = x1 x2 ⋅ ⋅ ⋅ xn−1 aλ+δn−1 (x1 , . . . , xn−1 )
and
aδn (x1 , . . . , xn−1 , 0) = x1 x2 ⋅ ⋅ ⋅ xn−1 aδn−1 (x1 , . . . , xn−1 ).
Therefore, by induction on n we have

aλ+δn (x1 , . . . , xn−1 , 0)
= x1 + ⋅ ⋅ ⋅ + xn−1 .
aδn (x1 , . . . , xn−1 , 0)
Since our desired ratio is a symmetric polynomial in x1 , . . . , xn , it
must be equal to x1 + ⋅ ⋅ ⋅ + xn , which is m1 (Xn ) = e1 (Xn ) = h1 (Xn ) =
p1 (Xn ) = s1 (Xn ). □
a (X )
Example 4.27. Suppose λ = (2) and n = 3. Express aλ+δ n n
δn (Xn )
in
terms of the monomial symmetric polynomials, the elementary sym-
metric polynomials, the complete homogeneous symmetric polynomi-
als, the power sum symmetric polynomials, and the Schur polynomi-
als.
Solution. We have
4
⎛x1 x42 x43 ⎞
det ⎜x1 x2 x3 ⎟
aλ+δn (Xn ) ⎝1 1 1⎠
=
aδn (Xn ) (x1 − x2 )(x1 − x3 )(x2 − x3 )
= x21 + x22 + x23 + x1 x2 + x1 x3 + x2 x3 ,
so our ratio is equal to m2 (X3 ) + m11 (X3 ) = e11 (X3 ) − e2 (X3 ) =
h2 (X3 ) = 21 p2 (X3 ) + 12 p11 (X3 ) = s2 (X3 ). □
n a (X )
Examples 4.26 and 4.27 suggest we can express aλ+δ n
δn (Xn )
most
simply in terms of Schur polynomials or complete homogeneous sym-
metric polynomials. Computing more examples sheds more light on
the situation: if λ = (2, 1), then we find
aλ+δn (Xn )
= s21 (Xn ) = h21 (Xn ) − h3 (Xn );
aδn (Xn )
if λ = (22 ), then we find
aλ+δn (Xn )
= s22 (Xn ) = h22 (Xn ) − h31 (Xn );
aδn (Xn )
and if λ = (2, 12 ), then we find
aλ+δn (Xn )
= s211 (Xn ) = h4 (Xn ) − h31 (Xn ) − h22 (Xn ) + h211 (Xn ).
aδn (Xn )
These suggest the following result.
Proposition 4.28. Suppose n ≥ 1 and λ is a partition with at most

n parts. Then
aλ+δn (Xn )
(4.12) sλ (Xn ) = .
aδn (Xn )
To give a combinatorial proof of (4.12), we need to interpret each

side as a generating function for some set of combinatorial objects.
This is difficult (though not impossible) for the ratio on the right side,
so instead of proving (4.12) directly, we prove the equivalent result
that
(4.13) sλ (Xn )aδn (Xn ) = aλ+δn (Xn ).
Our combinatorial proof of (4.13) will have several steps. First,

we will interpret the right side of (4.13) as a generating function for
Sn with respect to a certain weight function. Second, we will interpret
the left side of (4.13) as a generating function for ordered pairs (π, T ),
where π ∈ Sn and T is a certain kind of tableau, with respect to a
weight function similar to the weight function for the right side.
On both sides of our new equation some terms will be positive
and some will be negative, but in general the left side will have more
terms than the right. So our third step will be to use an involution
to cancel the extra terms on the left. And our fourth and final step
will be to give a bijection matching the remaining terms on the left
with the terms on the right.
To interpret the right side of (4.13) as a generating function for
Sn , we use the fact that aλ+δn (Xn ) is a determinant, and therefore a
sum over permutations. In particular, we have the following.
Definition 4.29. Suppose n ≥ 1. For any partition λ with at most n

parts and any π ∈ Sn , the Schur λ-weight of π, written schurwtλ (π),
is the monomial
n
n−j+λj
(4.14) schurwtλ (π) = ∏ xπ(j) .
j=1
Example 4.30. Find schurwtλ (π) for π = 561432 and λ = (4, 32 , 1).
Solution. Here n = 6, so we have

6
6−j+λj
schurwtλ (π) = ∏ xπ(j)
j=1
= x95 x76 x61 x34 x3 . □
Proposition 4.31. Suppose n ≥ 1. For any partition λ with at most

n parts, we have
(4.15) aλ+δn (Xn ) = ∑ sgn(π) schurwtλ (π).

π∈Sn
Proof. When we set µ = λ + δn in (4.7) and use (4.14) to simplify the

result, we find
1 +n−1 λ2 +n−2
aλ+δn (Xn ) = ∑ sgn(π)xλπ(1) xπ(2) ⋅ ⋅ ⋅ xλπ(n)
n
π∈Sn
= ∑ sgn(π) schurwtλ (π),

π∈Sn
We can use a similar approach to interpret the left side of (4.12)

as a sum over ordered pairs (π, T ), where π ∈ Sn and T ∈ SST(λ; n).
However, it will be useful to adopt an approach to T which ties it
more closely to π. To do this, we start with the following definition.
Definition 4.32. Suppose n ≥ 1. For any π ∈ Sn , let <π be the

ordering π(1) <π π(2) <π ⋅ ⋅ ⋅ <π π(n). Then for any partition λ
with at most n parts and any π ∈ Sn , a semistandard π-tableau of
shape λ is a filling of a Ferrers diagram of λ with 1, 2, . . . , n which is
column-strict and row-nondecreasing with respect to <π . We write
SSTπ (λ; n) to denote the set of all semistandard π-tableaux of shape
λ with entries in [n], and we write SSTπ (λ) to denote the set of all
semistandard π-tableaux of shape λ with entries in P.
Our first example of a collection of semistandard π-tableaux is

an analogue of Example 4.5.
Example 4.33. Find all π-tableaux of shape (22 , 1) with entries in

[3], where π = 312.
2 2 2
1 1 1 2 1 2
3 3 3 3 3 1
Figure 4.11. The semistandard 312-tableaux of shape (22 , 1)

with entries in [3]
Solution. By definition we have 3 <312 1 <312 2, so the tableaux we

want are those in Figure 4.11. □
Notice the tableaux in Example 4.33 are exactly the tableaux we

obtain from those in Example 4.5 by applying the permutation 312
to every entry. In general, for any π ∈ Sn , we have j < l if and only
if π(j) <π π(l) by our definition of <π . Therefore, if we write π(T )
to denote the tableau we obtain from a tableau T by applying π to
every entry of T , then T is a semistandard tableau if and only if π(T )
is a semistandard π-tableau.
We might expect that for a given π ∈ Sn , the generating function
for our π-tableaux will be a new symmetric polynomial. In other
words, for a given n ≥ 1, a given π ∈ Sn , and a given partition λ,
what can we say about ∑T ∈SSTπ (λ;n) xT ? When we try a few small
examples, we see it is just another way of writing the Schur polynomial
sλ (Xn ).

n parts and any π ∈ Sn , we have
(4.16) sλ (Xn ) = ∑ xT .
T ∈SSTπ (λ;n)
Proof. Since sλ (Xn ) is a symmetric polynomial in x1 , . . . , xn , when

we apply π to both sides of (4.5), we find
sλ (Xn ) = ∑ π (xT ) .
T ∈SST(λ;n)
But π(xT ) = xπ(T ) by the definitions of xT and π(T ), so

sλ (Xn ) = ∑ xπ(T ) .
T ∈SST(λ;n)
Now we can reindex the sum on the right, replacing T with π −1 (T )

and using the fact that we have π −1 (T ) ∈ SST(λ; n) if and only if
T ∈ SSTπ (λ; n) to find
sλ (Xn ) = ∑ xT ,
T ∈SSTπ (λ;n)
We can now interpret the left side of (4.13) as a generating func-

tion for ordered pairs (π, T ), where π ∈ Sn and T ∈ SSTπ (λ; n). In
particular, we have the following.

n parts we have
(4.17) sλ (Xn )aδn (Xn ) = ∑ ∑ sgn(π) schurwt∅ (π)xT .
π∈Sn T ∈SSTπ (λ;n)
Proof. If we set µ = δn in (4.7), then we find

n
sλ (Xn )aδn (Xn ) = ∑ sgn(π)sλ (Xn ) ∏ xn−j
π(j)
π∈Sn j=1
= ∑ sgn(π) schurwt∅ (π)sλ (Xn ),

π∈Sn
since schurwt∅ (π) = ∏nj=1 xn−j

π(j)
. Now we can use (4.16) to rewrite
sλ (Xn ) as a sum over π-tableaux, which leads to (4.17). □
We now have both sides of (4.13) written as generating functions,

and we’re ready to cancel some terms on the left. To start, for any
n ≥ 1 and any partition λ with at most n parts, let PT (n, λ) be the
set of ordered pairs (π, T ) in which π ∈ Sn and T ∈ SSTπ (λ; n). We
define a function from PT (n, λ) to itself. This function is similar to
a Bender–Knuth involution, so we write κ to denote it. We will also
abuse notation a bit by writing (κ(π), κ(T )) to denote the image of
(π, T ) under κ. Our construction of κ starts with the reading word
of a tableau, which we define next.
Definition 4.36. Suppose λ is a partition. For any filling T of the

Ferrers diagram of λ with positive integers, the reading word of T ,
written word(T ), is the word we get by reading the entries of T from
1 4
2 7
3 1 5
Figure 4.12. A filling of the Ferrers diagram for (3, 22 ) with

reading word 1427315
left to right in each row, starting with the top row and ending with
the bottom row.
As an example of a reading word, the reading word for the filling

in Figure 4.12 is 1427315.
The reading words for the tableaux which are fixed by our invo-
lution will have a particular structure, which is built on the order of
their entries. In particular, if T is a semistandard π-tableau for some
permutation π and T is fixed by our involution, then every tail seg-
ment of word(T ) will have at least as many π(1)’s as π(2)’s, at least
as many π(2)’s as π(3)’s, and in general at least as many π(j)’s as
π(j + 1)’s. To define our involution on tableaux whose reading words
do not have this property, we single out the rightmost entry for which
the tail segment starting at that entry violates this condition.
Definition 4.37. Suppose n ≥ 1, π ∈ Sn , and T is a semistandard

π-tableau with reading word k1 k2 ⋅ ⋅ ⋅. For each entry kj in this read-
ing word, let rj be the number of entries to the right of kj (including
kj itself) which are equal to kj . Similarly, for each entry kj = π(lj ),
let tj be the number of entries to the right of kj which are equal to
π(lj + 1). The π-climber of T or of word(T ) is the rightmost entry
kj such that tj > rj . If no such entry exists, then we say word(T )
is a Littlewood–Richardson π-word and T is a Littlewood–Richardson
π-tableau. When π is the identity permutation we sometimes say
word(T ) is a Littlewood–Richardson word and T is a Littlewood–
Richardson tableau.
Example 4.38. Suppose π = 4132. Determine which of the fillings

in Figure 4.13 are semistandard π-tableaux, which are Littlewood–
Richardson π-tableaux, and which entries are π-climbers.
2 2 3 2 3 3 2 2
1 3 2 1 3 3 1 1 1 1 1 1
4 1 1 4 4 4 4 4 4 4 4 4
Figure 4.13. Fillings of the Ferrers diagram of (32 , 2) for

Example 4.38
Solution. To start, notice each filling is both column-strict and row-

nondecreasing with respect to <π , so each filling is a π-tableau.
The reading word for the leftmost tableau is 22132411. At the
rightmost entry of this word we already have more π(2)’s than π(1)’s,
so the filling is not a Littlewood–Richardson π-tableau. Moreover,
the π-climber is the rightmost entry in the reading word, which is the
rightmost entry in the bottom row of the filling.
The reading word for the second filling from the left is 32133444.
The rightmost three entries of this word are π(1), but the fourth
entry from the right is π(3), so at this entry we have more π(3)’s
than π(2)’s, and the filling is not a Littlewood–Richardson π-tableau.
Moreover, the π-climber is the fourth entry from the right end of the
reading word, which is the rightmost entry in the second row from
the bottom of the filling.
The reading word for the third filling from the left is 33111444.
This is a Littlewood–Richardson π-word, so the filling is a Littlewood–
Richardson π-tableau.
The reading word for the rightmost filling is 22111444. The right-
most three entries of this word are π(1), and the next three entries
to the left are π(2), but the next entry to the left is π(4). So at the
seventh entry from the right we have more π(4)’s than π(3)’s, and
the filling is not a Littlewood–Richardson π-tableau. Moreover, the
π-climber is the seventh entry from the right end of the reading word,
which is the rightmost entry in the top row of the filling. □
Example 4.39. Suppose π = 35124. In Figure 4.14 we have four

incomplete fillings of the Ferrers diagram for (4, 3, 2). Determine
which ones can be completed to form a Littlewood–Richardson π-
tableau, and do so in all possible ways.
1
5
2
5
Figure 4.14. Partial fillings of the Ferrers diagram for

(4, 3, 2) for Example 4.39
Solution. In the upper left partial filling, the rightmost entry in the
bottom row must be greater than or equal to 5 = π(2) with re-
spect to <π , so it cannot be π(1). Therefore, the reading word for
the filling will not start with π(1), so it will not be a Littlewood–
Richardson π-word, and this partial filling cannot be completed to
form a Littlewood–Richardson π-tableau.
In order to complete the upper right filling to form a Littlewood–
Richardson π-tableau, we must put a 3 = π(1) in the rightmost box
in the bottom row. And since the entries in that row must be weakly
increasing with respect to <π from left to right, all of the entries in
this row must be 3. Now the next entry in the reading word for the
filling will be 1 = π(3), so this reading word cannot be a Littlewood–
Richardson π-word, and we cannot complete the filling to form a
Littlewood–Richardson π-tableau.
As with the upper right filling, in order to complete the lower left
filling to form a Littlewood–Richardson π-tableau, we must put a 3
in every entry in the bottom row. And for a similar reason, we must
put a 5 = π(2) in every entry in the second row from the bottom. And
again for a similar reason, we must put a 1 = π(3) as the rightmost
entry in the third row from the bottom. But 2 >π 1, so we cannot
complete the filling to form a Littlewood–Richardson π-tableau.
Using the same reasoning as in the last filling, we see we can

complete the lower right filling to form a Littlewood–Richardson π-
tableau in exactly one way: by putting a 3 in every entry in the
bottom row, a 5 in every entry of the second row, and a 1 in every
entry of the top row. □
Our work in Examples 4.38 and 4.39 suggests several useful facts
about π-climbers and Littlewood–Richardson π-tableaux.
Proposition 4.40. Suppose n ≥ 1, π ∈ Sn , and λ is a partition with
l(λ) ≥ n.
(i) There is a unique Littlewood–Richardson π-tableau of shape
λ, which is the filling in which every entry in row j is π(j)
for all j.
(ii) If T is a semistandard π-tableau of shape λ but T is not a
Littlewood–Richardson π-tableau, then its π-climber is the
rightmost entry in its row, and it is the leftmost entry in
word(T ) which is in row j but is not equal to π(j).
Proof. Suppose T is the filling in which every box in the jth row
from the bottom contains π(j), for all j. By construction, T is a
semistandard π-tableau. And because T has partition shape, word(T )
is a Littlewood–Richardson π-word and T is a Littlewood–Richardson
π-tableau.
Now suppose T is a semistandard π-tableau and there is a j such
that at least one box in the jth row from the bottom of T does not
contain π(j). Consider the rightmost entry k = π(l) of word(T ) with
this property. We claim this entry is the rightmost entry in its row
in T , and it is the π-climber in T .
To see our entry is the rightmost entry in its row, suppose it is
in the jth row from the bottom of T . Since T is column-strict with
respect to <π , this entry must be greater than π(j) with respect to <π .
But if it is not the rightmost entry in its row, then by construction
the rightmost entry in its row is π(j) <π k, which contradicts the fact
that T is row-nondecreasing with respect to <π .
To see our entry is the π-climber in T , note that the entries to
its right in the reading word for T , reading from right to left, are λ1
π(1)’s, followed by λ2 π(2)’s, up through λj−1 π(j − 1)’s. Since λ is

a partition, none of these entries can be the π-climber. On the other
hand, our entry π(l) is greater than π(j) with respect to >π , so when
we reach it we have more π(l)’s than π(l − 1)’s, so our entry is the
π-climber.
Because T has a π-climber, it is not a Littlewood–Richardson
π-tableau, so (i) holds. And our choice of entry shows (ii) holds. □
Now that we understand reading words and π-climbers, we are

ready to define κ.
If T is a Littlewood–Richardson π-tableau, then we set κ(π, T ) =
(π, T ), so suppose T is not a Littlewood–Richardson π-tableau. Then
T has a π-climber k = π(j), and by Proposition 4.40(ii) we know j > 1.
So we set l = π(j − 1), and we construct κ(π) by swapping k and l in
π. In symbols, we have κ(π) = (kl)π = π(j − 1, j).
Next we need to define κ(T ), but before we do, we make note of
a couple of useful facts about the relationship between π and κ(π).
Lemma 4.41. Suppose n ≥ 1, π ∈ Sn , and T is a semistandard π-

tableau which is not a Littlewood–Richardson π-tableau. Then the
following hold.
(i) sgn(π) = − sgn(κ(π)).

(ii) xk schurwt∅ (π) = xl schurwt∅ (κ(π)).
Here k = π(j) is the π-climber in T and l = π(j − 1).
Proof. (i) Since k and l are adjacent in π, swapping them to produce

κ(π) can only affect an inversion involving both of them. If k and l are
in increasing order in π, then they are in decreasing order in κ(π),
in which case κ(π) has one more inversion than π, and the results
follow. And if k and l are in decreasing order in π, then they are in
increasing order in κ(π), in which case κ(π) has one fewer inversion
than π, and the result follows.
(ii) Using the definitions of schurwt∅ (π) and schurwt∅ (κ(π)), we

find
n
xk schurwt∅ (π) = xk ∏ xn−m
π(m)
m=1
n
= xk xn−j
k xl
n−j+1 n−m
∏ xπ(m)
m=1
m≠j−1,j
n
= xl xn−j+1
k xn−j
l
n−m
∏ xπ(m)
m=1
m≠j−1,j
= xl schurwt∅ (κ(π)),
As we think about how to build κ(T ) from T , Lemma 4.41 tells

us we need to keep in mind that schurwt∅ (π) ≠ schurwt∅ (κ(π)), and
l <π k but k <κ(π) l. This second observation means T probably isn’t
even a κ(π)-tableau, because every column containing both a k and
an l has them in the wrong order.
To construct κ(T ) from T , first consider the row R containing the
π-climber k = π(j), and suppose R is the mth row from the bottom
of T . By Proposition 4.40(ii), the π-climber is the rightmost entry
in R. As in our construction of the Bender–Knuth involutions, we
call an entry k (resp., l) in T free whenever the entry immediately
below (resp., above) it in the same column is not l (resp., k), and it is
paired otherwise. In addition, we call the π-climber itself paired, even
though Proposition 4.40(ii) tells us the entry below it is π(m−1) <π l.
Because the entries in each row are weakly increasing with respect
to <π , in R and every row above R we first see a (possibly empty)
sequence of consecutive paired l’s, followed by a (possibly empty)
sequence of consecutive free l’s. We then see a (possibly empty) se-
quence of consecutive free k’s, followed by a (possibly empty) sequence
of consecutive paired k’s. We now change each paired l (resp., k) to a
k (resp., l). If a given row has a free l’s followed by b free k’s, then we
replace that sequence of l’s and k’s with a sequence of b k’s followed
by a l’s. In Figure 4.15 we see how this works on R, and in Figure
4.16 we see how it works on a row above R. We define κ(T ) to be
the resulting filling.
R
k
` ` k k
| {z }| {z }
a b
image of R
`
k k ` `
| {z }| {z }
b a
Figure 4.15. The effect of κ on the row R
row above R
k
` ` k k
| {z }| {z }
a b `
image of
row above R
`
k k ` `
| {z }| {z }
b a k
Figure 4.16. The effect of κ on a row above R
Example 4.42. Suppose π = 4132. For each of the semistandard

π-tableaux T in Figure 4.17, compute κ(π, T ) and κ(κ(π, T )).
2 2 3 2 3 3 2 2
1 3 2 1 3 3 1 1 1 1 1 1
4 1 1 4 4 4 4 4 4 4 4 4
Figure 4.17. The semistandard 4132-tableaux for Example 4.42

2 2 1 2 3 3 2 3
4 3 2 3 3 1 1 1 1 1 1 1
1 1 4 4 4 4 4 4 4 4 4 4
Figure 4.18. κ(T ) for the fillings T in Example 4.42 and

Figure 4.17
Solution. We first note that the semistandard 4132-tableaux in this

example are exactly the semistandard 4132-tableaux we saw in Ex-
ample 4.38.
As we saw in our solution to Example 4.38, the π-climber in the
leftmost filling is the 1 at the right end of the bottom row. Since
π(2) = 1, we have k = 1, j = 2, and l = 4. Therefore, κ(π) = 1432,
and κ(T ) is the leftmost semistandard π-tableau in Figure 4.18. In
this tableau the π-climber is again the rightmost entry of the bottom
row, which is now a 4. This means k = 4, j = 2, and l = 1. Therefore,
κ(κ(π)) = 4132, and when we compute κ(κ(T )) we find it is T .
In the second semistandard π-tableau from the left in Figure 4.17,
the π-climber is the 3 at the right end of the second row from the
bottom, so k = 3, j = 3, and l = 1. Therefore, κ(π) = 4312 and κ(T )
is the middle tableau in Figure 4.18. In this tableau the π-climber is
again the rightmost entry of the second row from the bottom, which
is now a 1. This means k = 1, j = 2, and l = 3. Therefore κ(κ(π)) = π,
and when we compute κ(κ(T )) we find it is again T .
As we saw in our solution to Example 4.38, the third tableau
from the left in Figure 4.17 is a Littlewood–Richardson π-tableau, so
in this case κ(κ(π)) = κ(π) = π and κ(κ(T )) = κ(T ) = T .
Finally, the π-climber in the rightmost tableau in Figure 4.17 is
the rightmost 2 in the top row, so k = 2, j = 4, l = 3, and κ(π) = 4123.
In addition, κ(T ) is the rightmost tableau in Figure 4.18. In this
tableau the π-climber is again the rightmost entry of the top row,
which is now a 3. This means k = 3, j = 4, and l = 2. Therefore,
κ(κ(π)) = π, and when we compute κ(κ(T )) we find once again that
it is T . □
Our solution to Example 4.42 suggests κ(T ) has the following

important properties.
tableau. Then the following hold.
(i) κ(T ) is a semistandard κ(π)-tableau.
(ii) If T is not a semistandard π-tableau, then xl xT = xk xκ(T ) ,
where k = π(j) is the π-climber in T and l = π(j − 1).
Proof. (i) We know T is column-strict and row-nondecreasing with

respect to <π , and we need to show κ(T ) is column-strict and row-
nondecreasing with respect to <κ(π) . We also know l and k are adja-
cent in π, with l immediately to the left of k, and κ(π) is π with l
and k swapped.
To show κ(T ) is column-strict with respect to <κ(π) , suppose C
is a column in T and κ(C) is the corresponding column in κ(T ),
and refer to Figure 4.19. If C contains neither k nor l, then it is
identical to κ(C). Furthermore, the entries of C appear in the same
order in π and κ(π), so the fact that the entries of C = κ(C) are
strictly increasing from bottom to top with respect to <π means they
are also strictly increasing from bottom to top with respect to <κ(π) .
Similarly, if C contains k (resp., l) but not l (resp., k), then κ(C) is
the same as C, except the entry k (resp., l) in C could become either
k or l in κ(C). But k and l are adjacent in both π and κ(π), and π
and κ(π) differ in no other entries. So the fact that the entries of C
are strictly increasing from bottom to top with respect to <π means
the entries of κ(C) are strictly increasing from bottom to top with
respect to <κ(π) . Finally, if C contains both k and l, then κ(C) also
contains both k and l, but in the other order. Since this matches
the transformation from π to κ(π), the fact that the entries of C are
strictly increasing from bottom to top with respect to <π once again
means the entries of κ(C) are strictly increasing from bottom to top
with respect to <κ(π) . Therefore, κ(T ) is column-strict with respect
to <κ(π) .
The proof that κ(T ) is row-nondecreasing with respect to <κ(π)
is similar, using the typical rows illustrated in Figures 4.20 and 4.21.
l l l l
a a a a
r r r r
g g g g
e e e e
k ` k k
or or
` `
` k
s s s s
m m m m
a a a a
l l l l
l l l l
C κ(C) C κ(C)
Figure 4.19. The action of κ on typical columns
small paired ` free ` free k paired k large

| {z }| {z }
a b
Figure 4.20. A typical row of T in the proof of Lemma 4.43(i)
small paired k free k free ` paired ` large

| {z }| {z }
b a
Figure 4.21. The image of a typical row of T in the proof of

Lemma 4.43(i)
(ii) The only entries which change in the transition from T to

κ(T ) are k’s and l’s, and they change to k’s and l’s, so
l(T ) k(T )
xT xl xk
= ,
xκ(T ) l(κ(T )) k(κ(T ))
xl xk
where k(T ) and l(T ) are the number of k’s and l’s in T , respectively.
Furthermore, in each row other than the row of the π-climber, T and
κ(T ) have the same number of k’s and the same number of l’s. And
in the row of the π-climber, one k in T (the π-climber itself) becomes
an l in κ(T ), so k(T ) − 1 = k(κ(T )) and l(T ) + 1 = l(κ(T )). This
means
xT xk
κ(T )
= ,
x xl

tableau. Then κ(κ(π, T )) = (π, T ).
Proof. If T is a Littlewood–Richardson π-tableau, then by definition

κ(π, T ) = (π, T ), which means κ(κ(π, T )) = (π, T ). So we can assume
T is not a Littlewood–Richardson π-tableau.
Using our previous notation, let the π-climber of T be k = π(j),
set l = π(j − 1), and suppose the π-climber is the rightmost entry
of the mth row from the bottom of T . By construction, the first
m − 1 entries of π and κ(π) are the same, and the entries in the
bottom m − 1 rows of T and κ(T ) are also the same. On the other
hand, the rightmost entry of the mth row from the bottom of κ(T )
is l = κ(π)(j) >κ(π) κ(π)(m), so this entry is the κ(π)-climber in
κ(T ). This means we obtain κ(κ(π)) from κ(π) by swapping k and
l, so κ(κ(π)) = π. This also means we obtain κ(κ(T )) from κ(T ) by
swapping paired k’s and l’s, and swapping adjacent blocks of free k’s
and l’s, exactly undoing the work we did to construct κ(T ) from T .
Therefore, κ(κ(T )) = T , which is what we wanted to prove. □
With κ and its relevant properties in hand, we are ready to prove

Proposition 4.28.
4.3. Problems 111
Proof of Proposition 4.28. By (4.17) we have

(4.18) sλ (Xn )aδn (Xn ) = ∑ ∑ sgn(π) schurwt∅ (π)xT .
π∈Sn T ∈SSTπ (λ;n)
Lemmas 4.41(i) and (ii) and 4.43(i) and (ii) tell us that if T is not
a Littlewood–Richardson π-tableau, k = π(j) is the π-climber in T ,
and l = π(j − 1), then
xl xk
sgn(π) schurwt∅ (π)xT = − sgn(κ(π)) schurwt∅ (κ(π)) xκ(T )
xk xl
= − sgn(κ(π)) schurwt∅ (κ(π))xκ(T ) .
Since κ is an involution by Lemma 4.44, our last computation tells
us any term on the right side of (4.18) which is not indexed by a
Littlewood–Richardson π-tableau cancels with another such term.
Therefore, (4.18) is equivalent to
(4.19) sλ (Xn )aδn (Xn ) = ∑ ∑ sgn(π) schurwt∅ (π)xT ,
π∈Sn T
where the inner sum on the right is over all Littlewood–Richardson

π-tableau T of shape λ. By Proposition 4.40(i) there is a unique such
Littlewood–Richardson π-tableau, so (4.19) becomes
n
sλ (Xn )aδn (Xn ) = ∑ sgn(π) schurwt∅ (π) ∏ xλπ(m)
m
.
π∈Sn m=1
By the definition of schurwtλ (π) we have

sλ (Xn )aδn (Xn ) = ∑ sgn(π) schurwtλ (π),
π∈Sn
and now Proposition 4.28 follows from (4.15). □
4.3. Problems
4.1. Find the image of the semistandard tableau T in Figure 4.22
under the Bender–Knuth involution β1 .
5 5 6 7
3 4 4 5 5 5 6
2 2 3 3 4 4 4 4 5 5 5
1 1 2 2 2 2 3 3 3 3 3 4
Figure 4.22. The semistandard tableau T for Problems 4.1,

4.2, 4.3, and 4.4

4.5. Each semistandard tableau of shape (2) corresponds in a nat-
ural way with a point (j, l) in the plane, where j ≤ l. Connect
two of these points whenever there is a Bender–Knuth involu-
tion mapping one of the corresponding tableaux to the other.
What does the resulting graph look like?
4.6. Show that if T is a semistandard tableau with content µ and
µ1 = µ2 , then β1 (T ) = T .
4.7. Show with examples that for any j ≥ 2 there is a semistandard
tableau T with βj (T ) ≠ T even though T has the same number
of j’s as it has j + 1’s.
4.8. The Bender–Knuth involution βj behaves like the permutation
σj = (j, j + 1) in the sense that it switches µj and µj+1 in the
content of a tableau, and it has βj = βj−1 . We saw in Problem
C.1 that σj σj+1 σj = σj+1 σj σj+1 . Find all j ≥ 1 with
βj βj+1 βj = βj+1 βj βj+1 .
4.9. (a) Show that if l > j + 1, then (j, j + 1)(l, l + 1) =

(l, l + 1)(j, j + 1).
(b) Find all j and l with l > j + 1 such that βj βl = βl βj .
4.10. Find an example of a set of semistandard tableaux which con-
tains at least two tableaux, which does not contain all semi-
standard tableaux of any shape, which is closed under all of the
Bender–Knuth involutions, and on which we have βj βj+1 βj =
βj+1 βj βj+1 for all j ≥ 1. Alternatively, prove no such set exists.
4.3. Problems 113
4.11. Suppose T is a set of semistandard tableaux which is closed

under all of the Bender–Knuth involutions. Let G(T ) be the
graph whose vertices are the tableaux in T and in which two
vertices are connected by an edge whenever there is a Bender–
Knuth involution mapping one to the other. Prove or disprove:
G(T ) is bipartite for all T .
4.12. Write the Schur function s22 (of degree 4) as a linear combina-
tion of the monomial symmetric functions, the elementary sym-
metric functions, the complete homogeneous symmetric func-
tions, and the power sum symmetric functions.
4.13. Suppose q is an indeterminate and n is a positive integer. Write
∑µ⊢n q l(µ)−1 mµ as a linear combination of Schur functions.
4.14. How many semistandard tableaux (of any shape) have content
k, k?
4.15. How many semistandard tableaux (of any shape) have content
k, k, k?
4.16. Show that the converse of Proposition 4.18(i) is false by finding
partitions λ and µ such that λ >lex µ but Kλ,µ = 0.
4.17. Show that for partitions λ and µ with ∣λ∣ = ∣µ∣, we have Kλ,µ ≠ 0
if and only if λ ⊴ µ.
4.18. Suppose λ ⊢ n, µ ⊢ n, and λ ≠ n. In addition, suppose λ1 ≠ µ1
and λ′1 ≠ µ′1 . Show that Kλ,µ = 1 if and only if λ = ((m + 1)m )
for some m and µ = (mm+1 ).
4.19. Prove or disprove: for all λ, µ ⊢ k, we have Kλ,µ = Kµ,λ .
4.20. Show that for all n and k with n ≥ k ≥ 0, we have
n
sn−k,1k = ∑ (−1)j+k+1 ej hn−j
j=k+1
and
k
(4.20) sn−k,1k = ∑ (−1)j+k ej hn−j .
j=0

n−1
(4.21) pn = ∑ (−1)j sn−j,1j .
j=0
We will use this result in our proof of the Murnaghan–Nakayama

rule in Chapter 9.
4.22. The sums in Problem 4.20 are partial sums from equation (2.12).
Suppose we write the analogous partial sums in (3.6) as linear
combinations of Schur functions. What patterns do you observe
in the resulting coefficients?
4.23. For each partition λ with ∣λ∣ ≤ 3, write hλ as a linear combina-
tion of Schur functions. Make a conjecture about the coefficients
you obtain.
4.24. Find and prove a formula for Kλ,µ , where λ = (n2 ) and µ =
2
(1n ). (Here λ ≠ (n, n). Instead, λ has one part, which is equal
to n2 .)
4.25. For any positive integer n, a tournament on n vertices, or a
tournament for short, is a graph whose vertices are 1, 2, . . . , n,
such that for each i ≠ j, either there is a directed edge (that is,
an arrow) from i to j, or there is a directed edge from j to i,
but not both. In addition, tournaments have no edges connect-
ing a vertex to itself (that is, tournaments do not have loops).
(We can think of the vertices in a tournament as teams and the
edges as a summary of the results of a match between the cor-
responding teams. For instance, if there is an edge from 3 to 5,
then team 3 defeated team 5.) There is one tournament on one
vertex, there are two tournaments on two vertices, and there are
eight tournaments on three vertices. How many tournaments
are there on n vertices?
4.26. For each positive integer n, we can construct a weight function
wt on tournaments with n vertices, as follows. If we have an
edge from a vertex i to a vertex j, then the weight of that edge
is xi if j > i and −xi if i > j. Then the weight of a tournament
T , written wt(T ), is the product of the weights of its edges.
Find and prove a product formula for the generating function
∑ wt(T ),
T
where the sum is over all tournaments T on n vertices.

4.3. Problems 115
4.27. We say a tournament is transitive, or acyclic, whenever it has

the property that if there is an edge from i to j, and an edge
from j to k, then the edge between i and k is directed from i to k.
Show that if a tournament on n vertices is transitive, then there
is a unique permutation π ∈ Sn such that for each i, j, the edge
between i and j is directed from i to j whenever π(i) < π(j).
(We can think of π as a ranking function for the teams in the
tournament: the first place team is π(1), the second place team
is π(2), and in general the jth place team is π(j).)
4.28. Express the generating function
∑ wt(T )
T
as a determinant, where the sum is over all transitive tourna-

ments.
4.29. In a tournament T on n vertices, the outdegree of a vertex j is
the number of vertices i ≠ j for which the edge between i and
j is directed from j to i. Show that if a tournament T is not
transitive, then it has two vertices with the same outdegree.
4.30. For any nontransitive tournament T , we define a tournament
T ′ as follows. Consider all pairs i < j of vertices such that i
and j have the same outdegree. Among these, consider those
pairs in which i is as small as possible. Among these, choose the
pair which minimizes j. Let T ′ be the tournament we obtain by
swapping i and j. Show that the map T ↦ T ′ is a sign-reversing,
weight-preserving involution with no fixed points. Explain why
this proves equation (4.9).
4.31. Let n ≥ k, and let Altk (Xn ) denote the vector space of alternat-
ing polynomials in x1 , . . . , xn which are homogeneous of degree
k. Find dim Altk (Xn ).
4.32. Suppose π = 612534. Compute κ(π, T ), where T is the semi-
standard π-tableau in Figure 4.23.
4.33. Suppose π = 37521468. Compute κ(π, T ), where T is the semi-
standard π-tableau in Figure 4.24.
4.34. Suppose π = 41532. Compute κ(π, T ), where T is the semistan-
dard π-tableau in Figure 4.25.
3
5 5 3
1 1 1
6 6 6 6
Figure 4.23. The semistandard 612534-tableau for Problem 4.32
1 1 1 1 4 4 6 6 6 8 8 8 8 8 8
2 2 2 2 1 1 1 1 1 1 4 4 4 6 6
7 5 5 5 5 5 5 2 2 2 2 1 1 1 1
3 3 3 7 7 7 7 7 5 5 5 5 2 2 2
Figure 4.24. The semistandard 37521468-tableau for Prob-

lem 4.33
2
3 3 3 2
1 1 5 3 3 3 2 2 2
4 4 4 1 1 5 5 5 3 3
Figure 4.25. The semistandard 41532-tableau for Problem 4.34
4.35. Suppose k ≥ j are positive integers, and compute Kλ,µ for λ =

(k, 1n−k ) and µ = (j, 1n−j ).
4.36. Suppose λ is a partition with r parts, and n and t are positive
integers with n ≥ r and t ≥ λ1 . Let λ̃ be the partition with
λ˜j = t − λr−j+1 . Show (x1 ⋅ ⋅ ⋅ xn )t sλ (x−1 −1
1 , ⋅ ⋅ ⋅ , xn ) = sλ̃ (Xn ).
4.4. Notes
One of the central problems of current interest in the study of sym-
metric functions is to explain how to write a given symmetric function
as a linear combination of Schur functions. This is of particular in-
terest if the coefficients involved are nonnegative integers (when our
4.4. Notes 117
symmetric functions are over Q) or polynomials whose coefficients

are all nonnegative integers (when our symmetric functions are over
a field of rational functions, such as Q(q) or Q(q, t)). For example,
Billey, Rhoades, and Tewari [BRT19] have recently done this for the
Boolean product polynomials Bn,k (Xn ), which are defined by
Bn,k (Xn ) = ∏ ∑ xj .
A⊆[n] j∈A
∣A∣=k
Similarly, many people have studied the modified Macdonald poly-

nomials over the past three decades, and Haiman has used [Hai01]
algebraic techniques to show these polynomials are Schur positive.
But we currently have no combinatorial proof of this fact. You can
read more about the modified Macdonald polynomials in Haglund’s
book on the q, t-Catalan polynomials [Hag07]. And you can read
about some of the algebraic reasons we are interested in Schur posi-
tivity, along with an analysis of how common it is, in Patrias’s recent
article [Pat19].
By (4.8), equation (4.9) is telling us how to evaluate a certain
determinant; this determinant is sometimes called Vandermonde’s de-
terminant, or the Vandermonde determinant. Our proof of this result
is algebraic, but we outline a proof of Gessel [Ges79] using certain
directed graphs in Problems 4.25 through 4.30. There are many other
proofs of this result, including one of Benjamin and Dresden [BD07]
using arrays of playing cards.
Our proof of Proposition 4.28 is based on an argument given by
Loehr [Loe11, Sec. 11.12], which uses a combinatorial interpretation
of aλ+δn (Xn ) involving sequences of beads on a string. However,
there are several other proofs, some of which are largely algebraic,
and others of which are more combinatorial. For instance, Proctor
gives a proof [Pro89] in which he shows both sides of (4.12) satisfy the
same recurrence relation, while Carbonara, Remmel, and Kulikauskas
[CRK95] give a combinatorial proof that involves augmenting semi-
standard tableaux with extra rows to the left.
Chapter 5
Interlude: A Rogues’
Gallery of Symmetric
Functions
The symmetric functions we have seen so far—monomial, elementary,

complete homogeneous, power sum, and Schur functions—are the pil-
lars of the symmetric function community. They are old, they are re-
lated to each other in both obvious and nonobvious ways, they have
many nice properties, and everyone knows and respects them. But
there are newer symmetric functions, many with interesting combina-
torial connections of their own. Our goal in this chapter is to study
some of these newer symmetric functions. We will be most concerned
with those arising from modifications of our definition of a semistan-
dard tableau, but we will also see an example coming from proper
colorings of graphs.
5.1. Skew Schur Functions

When we described the elementary symmetric functions and the com-
plete homogeneous symmetric functions in terms of fillings of Ferrers
diagrams in Propositions 4.1 and 4.3, we used Ferrers diagrams pri-
marily because these symmetric functions are indexed by partitions.
However, none of the conditions we placed on these diagrams involved
119
120 5. A Rogues’ Gallery of Symmetric Functions
their columns, so we could have just as easily viewed them as baskets

of rows, all jumbled together. Our description of the Schur functions
in terms of semistandard tableaux implicitly involves a way of orga-
nizing these baskets of rows: we put them together to form Ferrers
diagrams of partitions. But we could assemble these rows in other
ways. One common way to do this involves removing a “smaller”
Ferrers diagram from a “larger” one, as we describe next.
Definition 5.1. For any partitions λ and µ, we say µ is contained

in λ, and we write µ ⊆ λ, whenever µj ≤ λj for all j. In this case
we write λ/µ to denote the set of boxes in the Ferrers diagram of λ
which are not contained in the Ferrers diagram of µ. We call this set
of boxes the skew shape λ/µ, and we write ∣λ/µ∣ = ∣λ∣ − ∣µ∣ to denote
the number of boxes in this skew shape.
Note that for any partition λ we have ∅ ⊆ λ and λ/∅ = λ.
Definition 5.2. For any partitions λ ⊇ µ, a semistandard skew tableau

of shape λ/µ is a filling of λ/µ in which the entries in the columns
are strictly increasing from bottom to top and the entries in the rows
are weakly increasing from left to right. If T is a semistandard skew
tableau, then we write sh(T ) to denote the shape of T . We write
SST(λ/µ; n) to denote the set of all semistandard skew tableaux of
shape λ/µ with entries in [n], and we write SST(λ/µ) to denote the
set of all semistandard skew tableaux of shape λ/µ with entries in P.
Now we can replace semistandard tableaux with semistandard

skew tableaux in our definition of Schur polynomials and Schur func-
tions to obtain new symmetric polynomials and symmetric functions.
Definition 5.3. For all partitions µ ⊆ λ and any integer n ≥ 1, we

write sλ/µ (Xn ) to denote the polynomial
sλ/µ (Xn ) = ∑ xT .
T ∈SST(λ/µ;n)
Similarly, we write sλ/µ to denote the formal power series
sλ/µ = ∑ xT .
T ∈SST(λ/µ)
5.1. Skew Schur Functions 121
By convention, if µ ⊆/ λ, then we set sλ/µ (Xn ) = sλ/µ = 0. We call

sλ/µ (Xn ) the skew Schur polynomial in x1 , . . . , xn indexed by λ/µ,
and we call sλ/µ the skew Schur function indexed by λ/µ.
As we might hope, the skew Schur polynomials are symmetric

polynomials, and the skew Schur functions are symmetric functions.
Proposition 5.4. Suppose λ ⊢ k, µ ⊢ l, and µ ⊆ λ. Then for all

n ≥ 1, we have sλ/µ (Xn ) ∈ Λk−l (Xn ) and sλ/µ ∈ Λk−l .
Proof. Both sλ/µ (Xn ) and sλ/µ are homogeneous of degree k − l by

construction. We leave it as Problem 5.4 to use the Bender–Knuth
involutions to show they are symmetric. □
As we might also hope, as long as we have enough variables, the

algebraic relationships among the skew Schur polynomials, or between
the skew Schur polynomials and our other symmetric polynomials, do
not depend on exactly how many variables we have. To explain this
in more detail, we introduce the skew Kostka numbers, which depend
on the idea of the content of a semistandard skew tableau.
Definition 5.5. For any semistandard skew tableau T , the content

of T is the sequence {νj }∞
j=1 , where νj is the number of j’s in T for
each j.
Definition 5.6. For any partitions λ, µ, and ν with µ ⊆ λ, we write

Kλ/µ,ν to denote the number of semistandard skew tableaux of shape
λ/µ and content ν, and we call the numbers Kλ/µ,ν the skew Kostka
numbers.
By our definition of the skew Schur functions, for any partitions

µ ⊆ λ, we have
(5.1) sλ/µ = ∑ Kλ/µ,ν mν .

ν⊢∣λ∣−∣µ∣
We can write a similar equation in which we expand sλ/µ (Xn ) in

terms of the monomial symmetric polynomials. But as we show next,
if n is large enough, then the coefficients in this second equation are
also the skew Kostka numbers.
Proposition 5.7. For any partitions µ ⊆ λ and ν with ∣ν∣ = ∣λ∣ − ∣µ∣
and any n ≥ 1, let Kλ/µ,ν,n be the number of semistandard tableaux
with shape λ/µ and content ν, with entries in [n]. If n ≥ ∣λ∣ − ∣µ∣,
then Kλ/µ,ν,n = Kλ/µ,ν . In particular, if n ≥ ∣λ∣ − ∣µ∣, then Kλ/µ,ν,n is
independent of n.
Proof. As in our proof of Proposition 4.17, we know ν is a partition

and n ≥ ∣λ∣ − ∣µ∣ = ∣ν∣, so we must have νj = 0 for j ≥ n. Therefore,
any semistandard tableau with shape λ/µ and content ν contains no
entry larger than n. This means Kλ/µ,ν,n and Kλ/µ,ν are counting the
same objects, so Kλ/µ,ν,n = Kλ/µ,ν . □
Now that we know that the skew Schur polynomials and the skew
Schur functions are symmetric and that we can investigate algebraic
relationships involving skew Schur functions by looking at analogous
relationships involving skew Schur polynomials, we look at some ex-
amples to see what else we can discover about the skew Schur sym-
metric functions.
Example 5.8. Find all of the skew Schur polynomials in Λ2 (X2 ).
Solution. At first this looks daunting: there are many pairs of parti-
tions µ ⊆ λ with ∣λ∣ − ∣µ∣ = 2. But many of these pairs give us exactly
the same monomials, and therefore exactly the same skew Schur poly-
nomials. For example, (3, 1)/(12 ), (7)/(5), and (7, 32 , 1)/(5, 32 , 1)
all lead to the original Schur polynomial s2 (X2 ) = x21 + x1 x2 + x22 .
Similarly, (32 , 1)/(22 , 1), (92 )/(82 ), and (42 , 2, 13 )/(32 , 2, 13 ) all lead
to the other original Schur polynomial s11 (X2 ) = x1 x2 . And fi-
nally, (4, 3, 1)/(3, 2, 1), (8, 3)/(7, 2), and (52 , 32 , 22 , 1)/(5, 4, 32 , 2, 12 )
all look somewhat different, but they give us exactly the same mono-
mials, and their associated skew Schur polynomials are all equal to
s21/1 (X2 ) = x21 + 2x1 x2 + x22 = s1 (X2 )s1 (X2 ) = s2 (X2 ) + s11 (X2 ). □
The fact that s21/1 (X2 ) = s1 (X2 )s1 (X2 ), and therefore s21/1 = s21 ,
suggests a more general result about sλ/µ when λ/µ has more than
one piece. To make this more precise, suppose µ ⊆ λ are partitions.
We say two 1 × 1 squares in λ/µ are adjacent whenever they share an
edge. We also say a set of 1 × 1 squares in λ/µ is connected when-
ever any two squares a and b in the set are connected by a path
Figure 5.1. A skew shape with three connected components
a = a1 , a2 , . . . , am = b of squares in the set in which aj is adja-

cent to aj+1 . Then the connected components of λ/µ are the max-
imal connected sets of 1 × 1 squares in λ/µ. For example, if λ =
(10, 94 , 8, 7, 4, 3, 12 ) and µ = (86 , 4, 1), then λ/µ has three connected
components. In Figure 5.1 we have λ/µ, with one connected com-
ponent unshaded, one shaded in light gray, and one shaded in dark
gray.
As our solution to Example 5.8 suggests and as we prove next,
when λ/µ has more than one connected component, we can write
sλ/µ as a product of the skew Schur functions associated with the
connected components of λ/µ.
Proposition 5.9. Suppose µ ⊆ λ are partitions and λ/µ has con-

nected components λ(1) /µ(1) , . . . , λ(m) /µ(m) , where µ(1) , . . . , µ(m) ,
λ(1) , . . . , λ(m) are partitions and µ(j) ⊆ λ(j) for all j with 1 ≤ j ≤ m.
Then
m
sλ/µ = ∏ sλ(j) /µ(j) .
j=1
Proof. By construction, if two squares in the diagram of λ/µ are in

different connected components, then they are not in the same row or
in the same column. As a result, their entries are independent. This

means we can construct each semistandard skew tableau of shape λ/µ
uniquely by choosing, for each j with 1 ≤ j ≤ m, a semistandard skew
tableau of shape λ(j) /µ(j) . Therefore, the terms in sλ/µ are exactly
the products of one term from each sλ(j) /µ(j) , which is what we wanted
to prove. □
Proposition 5.9 tells us some skew Schur functions are actually

products of Schur (or skew Schur) functions. Among other things,
this means there are coincidences among skew Schur functions. For
example, s5321/22 = s5431/(32 ) . But as we see next, these are not the
only coincidences among skew Schur functions.
Example 5.10. Write s22/1 and s222/1 in terms of Schur functions.
Solution. By Proposition 5.7 we can compute s22/1 by computing

s22/1 (X3 ), for which we need the semistandard skew tableaux in Fig-
ure 5.2. The terms for these tableaux tell us s22/1 = m21 +2m111 = s21 .
By Proposition 5.7 we can compute s222/1 by instead computing
s222/1 (X5 ), but writing down all of the semistandard skew tableaux
we need seems daunting. So we warm up by computing s222/1 (X4 ), for
which we need the semistandard skew tableaux in Figure 5.3. These
tableaux tell us s222/1 (X4 ) = m221 (X4 ) + 2m2111 (X4 ) = s221 (X4 ).
1 2 2 2
1 1
1 3 2 3 3 3
1 1 1
1 3 2 3 3 3
2 2 2
Figure 5.2. The eight semistandard skew tableaux of shape

(22 )/(1) with entries in [3]
2 3 3 3 3 3
1 2 1 2 2 2
1 1 1
2 4 3 4 3 4 4 4 4 4
1 2 1 2 2 2 1 2 2 2
1 1 1 1 1
2 4 3 4 3 4 4 4 4 4 4 4
1 3 1 3 2 3 1 3 2 3 3 3
1 1 1 1 1 1
2 4 3 4 3 4 4 4 4 4 4 4
1 3 1 3 2 3 1 3 2 3 3 3
2 2 2 2 2 2
Figure 5.3. The 20 semistandard skew tableaux of shape

(23 )/(1) with entries in [4]
Now that we are warmed up, we can write down the 75 semistan-
dard skew tableaux of shape (23 )/(1) with entries in [5] and then
collect terms to find
s222/1 = m221 + 2m2111 + 5m11111 = s221 . □
Our solution to Example 5.10 tells us some skew Schur functions

are actually Schur functions, in nontrivial ways. In fact, if we rotate
Figure 5.2 by 180 degrees, and replace each entry a with 4−a, then we
obtain Figure 4.5. Similarly, if we rotate Figure 5.3 by 180 degrees,
and replace each entry a with 5 − a, then we obtain Figure 4.6. This
suggests that if the rotation of the diagram for λ/µ by 180 degrees is
the Ferrers diagram of a partition shape ν, then sλ/µ = sν . This turns
out to be true, for exactly the reason our examples suggest.
Proposition 5.11. Suppose µ ⊆ λ and ν are partitions with ∣ν∣ =

∣λ∣ − ∣µ∣, and let ν r be the skew shape we obtain by rotating ν by 180
degrees. If λ/µ = ν r , then sλ/µ = sν .
Proof. Set n = ∣ν∣ = ∣λ∣ − ∣µ∣; we will show sλ/µ (Xn ) = sν (Xn ).
For any sequence α = α1 , . . . , αn of nonnegative integers with
n in sλ/µ (Xn ).
α1 +⋅ ⋅ ⋅+αn = n, let Aλ/µ,α be the coefficient of x1α1 ⋅ ⋅ ⋅ xαn
By definition, Aλ/µ,α is the number of semistandard skew tableaux

of shape λ/µ and content α. If we rotate any such tableau by 180
degrees and replace each entry a with n + 1 − a, then we obtain a
semistandard tableaux of shape ν and content αn , αn−1 , . . . , α1 . Fur-
thermore, this map is a bijection, so the coefficient of xα 1 ⋅ ⋅ ⋅ xn in
1 αn
sλ/µ (Xn ) is equal to the coefficient of x1 ⋅ ⋅ ⋅ xn in sν (Xn ). But

αn α1
sν (Xn ) is symmetric, so this coefficient is also equal to the coefficient

of x1α1 ⋅ ⋅ ⋅ xα
n in sν (Xn ). Therefore, the coefficients in sλ/µ (Xn ) and
n
sν (Xn ) are all the same, so these polynomials are equal.

To finish the proof, note that Proposition 4.17 tells us that if we
write sν as a linear combination of monomial symmetric functions, we
get the same coefficients as if we write sν (Xn ) as a linear combination
of monomial symmetric polynomials, while Proposition 5.7 tells us the
same thing about sλ/µ and sλ/µ (Xn ). This means sλ/µ (Xn ) = sν (Xn )
implies sλ/µ = sν . □
Proposition 5.11 raises a question: Are there any other skew Schur
functions which are also Schur functions (of partition shapes)? We
already have enough tools to show that under certain conditions the
answer is no.
Proposition 5.12. Suppose µ ⊆ λ and ν are partitions with ∣ν∣ =
∣λ∣ − ∣µ∣, and let ν r be the skew shape we obtain by rotating ν by 180
degrees. If λ′j − µ′j ≥ λ′j−1 − µ′j−1 for all j ≥ 2 and sλ/µ = sν , then
λ/µ = ν or λ/µ = ν r .
Proof. We prove the contrapositive. That is, we show that if λ/µ ≠ ν

and λ/µ ≠ ν r , then sλ/µ ≠ sν . We consider three cases.
If λ/µ is a partition but λ/µ ≠ ν, then sν and sλ/µ are two different
Schur functions. Since the Schur functions are linearly independent,
we must have sλ/µ ≠ sν .
If there is a partition α such that λ/µ = αr , but α ≠ ν, then by

the other direction of this result we have sλ/µ = sα . As in our first
case, sα and sν are two different Schur functions, so sλ/µ = sα ≠ sν .
Now suppose λ/µ is not a partition, and we do not get the Ferrers
diagram of a partition if we rotate λ/µ by 180 degrees. Let l be the
number of boxes in the largest column of λ/µ, and for each j with
1 ≤ j ≤ l, let αj be the number of columns of λ/µ with at least j
boxes. Note that since λ′j − µ′j ≥ λ′j−1 − µ′j−1 for all j ≥ 2, we know αj
is also the number of boxes in the jth column of λ/µ. In particular,
α1 ≥ α2 ≥ ⋅ ⋅ ⋅ ≥ αl ≥ 1, so α = α1 , . . . , αl is a partition of ∣λ∣ − ∣µ∣.
By construction, α is the lexicographically largest partition for which
l αj
∏j=1 xj has a nonzero coefficient in sλ/µ . On the other hand, ν is the
ν
lexicographically largest partition for which ∏lj=1 xj j has a nonzero
coefficient in sν . If α ≠ ν, then sλ/µ ≠ sν , so we may assume α = ν.
In particular, for each j the jth column of λ/µ has the same number
of blocks as the jth column of ν, so we can construct λ/µ from the
diagram of ν by shifting some columns up.
Now consider the coefficient of x1 x2 ⋅ ⋅ ⋅ x∣ν∣ in sν and sλ/µ . Every
semistandard tableau of shape ν and content 1∣ν∣ can be transformed
into a semistandard tableau of shape λ/µ and content 1∣ν∣ by shifting
some columns up, and this map is injective. On the other hand,
because λ/µ is not a partition, there is a semistandard tableau T of
shape λ/µ and content 1∣ν∣ in which the 1 does not appear in the
leftmost column. Every semistandard tableau of shape ν and content
1∣ν∣ has its 1 in the its leftmost column, so T corresponds to no such
tableau. Therefore the coefficient of x1 x2 ⋅ ⋅ ⋅ x∣ν∣ in sν is less than the
coefficient of x1 x2 ⋅ ⋅ ⋅ x∣ν∣ in sλ/µ , which means sν ≠ sλ/µ . □
Proposition 5.12 still holds when we drop the requirement that

λ′j − µ′j ≥ λ′j−1 − µ′j−1 for all j ≥ 2 (even though our proof fails in this
case), which means we know exactly when a skew Schur function is
a Schur function. But if we return to Proposition 5.11, and ignore
the Schur function, we see we can read this result in another way.
In particular, Proposition 5.11 is telling us that if either λ/µ or its
180 degree rotation (λ/µ)r is the Ferrers diagram of a partition, then
sλ/µ = s(λ/µ)r . In other words, Proposition 5.11 is pointing us to
a coincidence among skew Schur functions. Can we find more such

coincidences?
We can start to explore this question by looking at sλ/µ and
s(λ/µ)r when λ/µ and (λ/µ)r are not necessarily the Ferrers diagram
of a partition. With some technological assistance we find
s2211/1 = s2111 + s221
and
s2221/11 = s2111 + s221 ,
so s2211/1 = s2221/11 . Similarly, we have
s321/1 = s221 + s311 + s32
and
s332/21 = s221 + s311 + s32 ,
so s321/1 = s332/21 . And if we examine the semistandard skew tableaux
indexing the terms in these symmetric functions, along with our proof
of Proposition 5.11, we reach the following result.
Proposition 5.13. Suppose µ ⊆ λ are partitions and (λ/µ)r is the di-

agram we obtain by rotating λ/µ by 180 degrees. Then sλ/µ = s(λ/µ)r .
Proof. This is similar to the proof of Proposition 5.11, using a bi-

jection between the tableaux indexing the terms in sλ/µ (Xn ) and the
tableaux indexing the terms in s(λ/µ)r (Xn ), where n = ∣λ∣ − ∣µ∣. □
We might hope Proposition 5.13 covers all situations in which

two skew Schur functions are equal. However, there are coincidences
of skew Schur functions that Proposition 5.13 does not explain. For
example, with yet more technological assistance we find
s54431/332 = s32211 + s3222 + s33111 + 2s3321 + 2s42111
+ 3s4221 + 3s4311 + 2s432 + s441 + s51111
+ 2s5211 + s522 + s531
and
s54221/311 = s32211 + s3222 + s33111 + 2s3321 + 2s42111
+ 3s4221 + 3s4311 + 2s432 + s441 + s51111
+ 2s5211 + s522 + s531 ,
5.2. Stable Grothendieck Polynomials 129
Figure 5.4. The Ferrers diagrams for (5, 42 , 3, 1)/(32 , 2) and (5, 4, 22 , 1)/(3, 12 )
so s54431/332 = s54221/311 . This is in spite of the fact that, as we see in

Figure 5.4, the Ferrers diagrams for the skew shapes (5, 42 , 3, 1)/(32 , 2)
and (5, 4, 22 , 1)/(3, 12 ) are not 180 degree rotations of one another.
This example belongs to a larger, somewhat well-understood family of
skew Schur function coincidences, but it remains an open problem to
find and classify all examples of these skew Schur function equalities.
This continues to be an area of active research.
5.2. Stable Grothendieck Polynomials

In Section 5.1 we constructed the skew Schur functions by expanding
our definition of the Schur functions to include tableaux of nonpar-
tition shapes. In this section we return to partition shapes, but we
use different entries: instead of building our tableaux by putting a
positive integer in each box, we now build our tableaux by putting a
set of positive integers in each box.
Definition 5.14. For any partition λ, a set-valued semistandard

tableau of shape λ is a filling of the Ferrers diagram of λ with nonempty
finite sets of positive integers such that if we create a filling of the
diagram with positive integers by replacing each set with one of its
elements, then the result is a semistandard tableau. For each set-
valued semistandard tableau T we write xT to denote the product
∏j∈T xj , which is over all elements of all of the sets in T .
Example 5.15. Find all of the set-valued semistandard tableaux of

shape (2, 1) with entries in [3], and compute ∑T xT , where the sum
is over all of these set-valued semistandard tableaux.
Solution. Because the elements of distinct sets in a column must be

in strictly increasing order from bottom to top, there are five ways to
fill the first column: with {1} in the lower left box and one of {2},
{3}, or {2, 3} in the upper left box, with {1, 2} in the lower left box
and {3} in the upper left box, or with {2} in the lower left box and
{3} in the upper left box. For each of the first three possibilities we
can put any nonempty subset of {1, 2, 3} in the lower right box, for
the last two options we can put any nonempty subset of {2, 3} in the
lower right box. If we also omit the braces and commas when we
write each set, then we obtain the set-valued tableaux in Figure 5.5.
When we add the terms associated with these tableaux, we obtain
m21 (X3 ) + 2m111 (X3 ) + m22 (X3 ) + 3m211 (X3 )
+ 2m221 (X3 ) + m222 (X3 ). □
Although the generating function for the set-valued tableaux of

shape λ with entries in [n] turns out to be a symmetric polynomial in
x1 , . . . , xn , we typically introduce a sign to get symmetric polynomials
and symmetric functions which appear in the study of certain geo-
metric objects called Schubert varieties (more specifically, structure
sheaves of the Schubert varieties in a flag variety).

Gλ (Xn ) to denote the polynomial
(5.2) Gλ (Xn ) = ∑(−1)∣T ∣−∣λ∣ xT .
T
Here the sum is over all set-valued semistandard tableaux of shape λ

with entries in [n]. For any such tableaux T , we write ∣T ∣ to denote
the sum of the sizes of the sets in T . Similarly, we write Gλ to denote
the formal power series
Gλ = ∑(−1)∣T ∣−∣λ∣ xT ,
T
where the sum is over all set-valued tableaux of shape λ with entries in
P. We call Gλ (Xn ) the stable Grothendieck polynomial in x1 , . . . , xn
indexed by λ, and we call Gλ the stable Grothendieck function indexed
by λ.
Example 5.17. Find G21 (X3 ).

2 3 23
1 1 1 1 1 1
2 3 23 3 3
1 2 1 2 1 2 12 2 2 2
2 3 23 3 3
1 3 1 3 1 3 12 3 2 3
2 3 23
1 12 1 12 1 12
2 3 23
1 13 1 13 1 13
2 3 23 3 3
1 23 1 23 1 23 12 23 2 23
2 3 23
1 123 1 123 1 123
Figure 5.5. The set-valued semistandard tableaux of shape

(2, 1) with entries in [3]
Solution. In our solution to Example 5.15 we saw the set-valued se-

mistandard tableaux with entries in [3] are as in Figure 5.5. Using
the appropriate signs, we find
G21 (X3 ) = 2m111 (X3 ) − m22 (X3 ) − 3m211 (X3 )

+ 2m221 (X3 ) − m222 (X3 ). □
Example 5.18. For each positive integer n, find G1n .

Solution. Because no number can be repeated in a set-valued semi-

standard tableau of shape 1n , we know every term in G1n has the
form xa1 xa2 ⋅ ⋅ ⋅ xak for positive integers a1 < a2 < ⋅ ⋅ ⋅ < ak , and the
sign on each term of degree k is (−1)k−n . Disregarding this sign for the
moment, the coefficient of xa1 xa2 ⋅ ⋅ ⋅ xak in G1n is the number of se-
quences consisting of a1 , . . . , ak (in that order) and n−1 bars, in which
no two bars are adjacent, and the sequence does not begin or end with
a bar. Indeed, one bijection from the set of these sequences to the
set of set-valued semistandard tableaux T with xT = xa1 xa2 ⋅ ⋅ ⋅ xak
puts the numbers before the first bar in the bottom box, the numbers
between the first two bars in the next box, etc. These sequences, in
turn, are in bijection with sequences of k − n stars and n − 1 bars, so
there are (n−1
k−1
) of them. Therefore, every term in G1n of degree k has
coefficient (−1)k−n (n−1 k−1
), so
∞ ∞
k−1 k−1
G1n = ∑ (−1)k−n ( )m1k = ∑ (−1)k−n ( )s1k . □
k=n n − 1 k=n n −1
We have seen that for any partition λ there is a positive in-

teger N such that if n ≥ N and we write the Schur polynomial
sλ (Xn ) as a linear combination of the monomial symmetric poly-
nomials mµ (Xn ), then the coefficients are independent of n. This is
not true for Gλ (Xn ): for some λ, as n grows we get more and more
nonzero coefficients when we write Gλ (Xn ) as a linear combination of
the monomial symmetric polynomials mµ (Xn ). On the other hand,
a modified version is true: for any partitions λ and µ, there is a pos-
itive integer N such that if n ≥ N , then the coefficient of mµ (Xn ) in
Gλ (Xn ) is independent of n. In other words, every coefficient even-
tually stabilizes, but each one stabilizes later than the last, so there
is no point at which every coefficient has stabilized. Nevertheless,
we can still show Gλ (Xn ) and Gλ are both symmetric. To do this,
we adapt the Bender–Knuth involutions we used to show the Schur
functions are symmetric.
Example 5.19. Let T be the set-valued tableau in Figure 5.6. Com-
pute β1 (T ), β2 (T ), β3 (T ), and β4 (T ).
Solution. In Figure 5.7 we have T with the free 1’s and 2’s in large
bold. There are two free 1’s and four free 2’s in the first row, so we
345 5 6 6
2 23 345 5 5
1 1 1 12 2 2 2 34
Figure 5.6. The set-valued tableau T in Example 5.19
345 5 6 6
2 23 345 5 5
1 1 1 12 2 2 2 34
Figure 5.7. The set-valued tableau T in Example 5.19 with
the free 1’s and 2’s in bold
345 5 6 6
2 23 345 5 5
1 1 1 1 1 12 2 34
Figure 5.8. The set-valued tableau β1 (T ) in the solution to

Example 5.19
345 5 6 6
23
2 2 45 5 5
1 1 1 12 3 3 3 34
Figure 5.9. The set-valued tableau β2 (T ) in the solution to

Example 5.19
replace them with four free 1’s and two free 2’s, moving the set 12
two boxes to the right. We find β1 (T ) is the tableau in Figure 5.8.
Working in a similar way, we find β2 (T ) is the tableau in Figure
5.9, β3 (T ) is the tableau in Figure 5.10, and β4 (T ) is the tableau in
Figure 5.11. □
345 5 6 6
2 234 45 5 5
1 1 1 12 2 2 2 34
Figure 5.10. The set-valued tableau β3 (T ) in the solution

to Example 5.19
34 45 6 6
2 23 34 4 45
1 1 1 12 2 2 2 35
Figure 5.11. The set-valued tableau β4 (T ) in the solution

to Example 5.19
Proposition 5.20. For any partition λ and any positive integer n,

the polynomial Gλ (Xn ) and the formal power series Gλ are both in-
variant under every π ∈ Sn .
Proof. As in the proof of Proposition 4.15, it is sufficient to show

Gλ (Xn ) and Gλ are both invariant under (j, j + 1) for each j with
1 ≤ j ≤ n − 1. As in that proof, we can check that the Bender–
Knuth involution βj is a bijection between the set-valued semistan-
dard tableaux of content µ1 , . . . , µj , µj+1 , µj+2 , . . . and the set-valued
semistandard tableaux of content µ1 , . . . , µj+1 , µj , µj+2 , . . . . If T is a
set-valued semistandard tableau of shape λ, then ∣βj (T )∣ = ∣T ∣ and
µ µj+1 µj+2
βj (T ) and T have the same shape. Therefore xµ1 1 ⋅ ⋅ ⋅ xj j xj+1 xj+2 ⋅ ⋅ ⋅
µ µj µj+2
and xµ1 1 ⋅ ⋅ ⋅ xj j+1 xj+1 xj+2 ⋅ ⋅ ⋅ have the same coefficient, which is what
we wanted to prove. □
Every semistandard tableau is a set-valued semistandard tableau,

and if T is a semistandard tableau of shape λ, then ∣T ∣ = ∣λ∣. All
other set-valued semistandard tableaux contain more numbers than
boxes, so for any partition λ, the symmetric function Gλ is sλ plus a
symmetric function of degree greater than ∣λ∣. This allows us to turn
these relationships around, by writing the Schur functions in terms
of the stable Grothendieck functions.
Example 5.21. For each positive integer n, write s1n as a linear

combination of stable Grothendieck functions.
Solution. We start with n = 1. Our solution to Example 5.18 says
G1 = s1 − s12 + s13 − s14 + ⋅ ⋅ ⋅
and
G12 = s12 − 2s13 + 3s14 − ⋅ ⋅ ⋅ ,
so
G1 + G12 = s1 − s13 + 2s14 − ⋅ ⋅ ⋅ .
We can now use our solution to Example 5.18 to eliminate s13 , s14 , . . .
in turn, to find
∞
s1 = ∑ G1k .
k=1
We can use the same process to find
s12 = G12 + 2G13 + 3G14 + ⋅ ⋅ ⋅ ,
s13 = G13 + 3G14 + 6G15 + ⋅ ⋅ ⋅ ,

and
s14 = G14 + 4G15 + 10G16 + ⋅ ⋅ ⋅ ,
which leads us to conjecture that
k−1
s1n = ∑ ( )G1k .
k≥n n − 1
We can use our solution to Example 5.18 to prove this. In par-

ticular, we have
∞ ∞ ∞
k−1 k−1 l−1
∑( )G1k = ∑ ∑(−1)l−k ( )( )s1l
k=n n − 1 k=n l=k n−1 k−1
∞
l−1 l
l−n
= ∑ (−1)l ( ) ( ∑ (−1)k ( )) s1l .
l=n n − 1 k=n l−k
The inner sum on the right is 0 unless l = n, in which case it is (−1)n .

So the entire expression is s1n , which is what we wanted to prove. □
Our expression for s1n as a linear combination of Grothendieck

polynomials is particularly nice, but it turns out we can write any
Schur function as a linear combination of Grothendieck polynomials.
Even better, the coefficients in this expansion count certain skew
tableaux.
Definition 5.22. Suppose µ ⊆ λ are partitions. An elegant tableau of

shape λ/µ is a semistandard tableau of shape λ/µ in which the entries
in the jth row (of λ) are in [j − 1]. We write fλµ to denote the number
of elegant tableau of shape λ/µ. Note that if µ ⊆/ λ, then fλµ = 0.
Example 5.23. Find all elegant tableaux of shape λ/µ, where λ =

(3, 2, 1).
Solution. If µ1 < 3, then we have a box in the first row of λ/µ, but
no positive integer is small enough for such a box. So we must have
µ1 = 3. In Figure 5.12 we have all of the elegant tableaux of shape
(3, 2, 1)/µ. □
2 2
1
2
1 1
1 1
1
Figure 5.12. The elegant tableaux of shape (3, 2, 1)/µ
n
Example 5.24. Find f11k , where k ≥ n.
Solution. In any elegant tableau of shape (1k )/(1n ), the entries are a
subset of [k − 1] of size k − n. In any subset of [k − 1] of size k − n the
smallest entry is at most (k − 1) − (k − n) + 1 = n, the second smallest
entry is at most (k − 1) − (k − n) + 2 = n + 1, and in general the jth
smallest entry is at most n + j. On the other hand, in an elegant
5.3. Dual Stable Grothendieck Polynomials 137
tableau of shape (1k )/(1n ), the entry in the jth box from the bottom
(of the skew shape) can be at most n + j. So every subset of [k − 1] of
size k − n corresponds to an elegant tableau of shape (1k )/(1n ), and
n
f11k = (k−n
k−1
) = (n−1
k−1
). □
Comparing our solutions to Examples 5.21 and 5.24 suggests the

following result.
Proposition 5.25. For any partition µ, we have
sµ = ∑ fλµ Gλ .
λ⊇µ
Proving Proposition 5.25 would take us too far afield, but one of
its consequences is worth pointing out.
Corollary 5.26. Every symmetric function is a (possibly infinite)

linear combination of stable Grothendieck functions.
Proof. This is immediate from Proposition 5.20, since every sym-

metric function is a linear combination of Schur functions. □
5.3. Dual Stable Grothendieck Polynomials

We have now constructed new symmetric functions by modifying the
semistandard tableaux in the definition of a Schur function to be of
skew shape or to have sets as entries instead of single integers. In this
section we return to tableaux of partition shapes whose entries are
positive integers, but we change our inequalities: rather than insisting
the entries in a column be strictly increasing from bottom to top, we
allow them to be weakly increasing from bottom to top.
Definition 5.27. For any partition λ, a reverse plane partition of

shape λ is a filling of the Ferrers diagram of λ with positive integers in
which the entries in each row are weakly increasing from left to right
and the entries in each column are weakly increasing from bottom
to top. We say such a tableau is row-nondecreasing and column-
nondecreasing.
3 3 3 3 3 3
1 1 1 2 1 3 2 2 2 3 3 3
2 2 2 2 2
1 1 1 2 1 3 2 2 2 3
1 1 1
1 1 1 2 1 3
Figure 5.13. The reverse plane partitions of shape (2, 1) with

entries in [3]
For any n ≥ 1 and any partition λ, we write RPP(λ; n) to denote

the set of reverse plane partitions with entries in [n], and we write
RPP(λ) to denote the set of reverse plane partitions with entries
in P. Our hope is that the generating function for RPP(λ; n) is a
symmetric polynomial, and the generating function for RPP(λ) is a
symmetric function. We look at a small example to see whether this
might happen.
Example 5.28. Compute the generating function for the reverse
plane partitions of shape (2, 1) with entries in [3], and determine
whether this generating function is a symmetric polynomial.
Solution. In Figure 5.13 we have the reverse plane partitions of shape

(2, 1) with entries in [3]; we find their generating function is
x31 + x32 + x33 + 2x21 x2 + 2x21 x3 + 2x22 x3 + x1 x22 + x1 x23 + x2 x23 + 2x1 x2 x3 .
Since the coefficient of x21 x2 is 2 but the coefficient of x1 x22 is 1,
this generating function is not a symmetric polynomial. In addition,
allowing more integers as entries in our reverse plane partitions will
not change these coefficients, so the generating function for reverse
plane partitions with entries in the positive integers is not always a
symmetric function. □
Our solution to Example 5.28 tells us the generating function for

reverse plane partitions of a given shape with respect to our usual
weight function is not a symmetric function in general. But it turns

out that if we consider a slightly different weight function, then we
do get a symmetric function. In particular, for any reverse plane
colj (T )
partition T , let xircont(T ) be the product ∏∞
j=1 xj , where colj (T )
is the number of columns of T which contain at least one j. Note
that if T is a semistandard tableau, then xircont(T ) = xT .
Example 5.29. Compute the generating function for the reverse

plane partitions of shape (2, 1) with entries in [3] with respect to
ircont, and determine whether this generating function is a symmetric
polynomial.
Solution. We have already listed these reverse plane partitions in Fig-

ure 5.13, and when we compute the generating function with respect
to ircont we get
x21 + x21 x2 + x21 x3 +x1 x2 + x1 x22
+x1 x2 x3 + x1 x3 + x1 x2 x3 + x1 x23 + x22 + x22 x3
+x2 x3 + x2 x23 + x23
= m2 (X3 ) + m21 (X3 ) + m11 (X3 ) + 2m111 (X3 )
= s21 (X3 ) + s2 (X3 ).
As we see, this generating function is a symmetric polynomial. □
Inspired by our success in Example 5.29, we make a new defini-

tion.

gλ (Xn ) to denote the polynomial
(5.3) gλ (Xn ) = ∑ xircont(T ) .
T ∈RPP(λ;n)
Similarly, we write gλ to denote the formal power series

gλ = ∑ xircont(T ) .
T ∈RPP(λ)
We call gλ (Xn ) the dual stable Grothendieck polynomial in Xn in-

dexed by λ, and we call gλ the dual stable Grothendieck function
indexed by λ.
1 2 2 2
1 1 2 1
T β1 (T )
Figure 5.14. A reverse plane partition T and its image

β1 (T ), which is not a reverse plane partition
2 2 2 1
1 2 1 2
1 1 2 1
T β1 (T )
Figure 5.15. A reverse plane partition T and its image

β1 (T ), which is not a reverse plane partition
We would like to use the Bender–Knuth involutions to show that

gλ (Xn ) and gλ are invariant under (j, j+1) for each j with 1 ≤ j ≤ n−1,
but we quickly run into a problem: there are reverse plane partitions
T for which βj (T ) is not a reverse plane partition for certain j. For
instance, in Figure 5.14 we have a reverse plane partition T and its
image β1 (T ), which is not a reverse plane partition. It looks like we
can fix this by swapping columns in β1 (T ), but as we see in Figure
5.15, there are examples in which there is no way to swap entire
columns in β1 (T ) to get a reverse plane partition.
Although the Bender–Knuth involutions do not map reverse plane
partitions to reverse plane partitions, if we focus our attention on
the columns of a reverse plane partition we can construct similar
maps which serve our purpose. To describe these maps, suppose T
is a reverse plane partition and j ≥ 1. We define the map γj by
constructing γj (T ) as follows.
To start, we modify each column of T separately. If a column
contains no j’s or j + 1’s, then we do not change it. Similarly, if a
column contains both a j and a j + 1, then we do not change it. But
if a column contains a j but no j + 1 (resp., a j + 1 but no j), then we
change every j (resp., j + 1) in the column to a j + 1 (resp., j), and
j+1 j+1
.. j + 1 .. j + 1
. .
.. .. .. ..
. . . .
j+1j+1 j+1j+1
7!
j+1 j j j+1
.. .. .. ..
. . . .
.. ..
j+1 . j .
j j+1
Figure 5.16. One way of modifying columns to get closer to

a reverse plane partition
we do not change any other entry in the column. We write flipj (T )

to denote the resulting filling.
In general, flipj (T ) is not a reverse plane partition. To transform
it into one, we first find the leftmost pair of adjacent columns with a
j +1 immediately to the left of a j. Note that if both of these columns
contain both a j and a j + 1, then T was not a reverse plane partition.
Now we modify these two columns as follows.
● If the left column has only j + 1’s and the right column has
only j’s, then we replace each j with j + 1 in the left column
and we replace each j + 1 with j in the right column.
● If the left column has only j + 1’s and the right column has
both j + 1’s and j’s, then we find the highest j in the right
column, and we change every j + 1 in the left column that
is at or below the height of this j to a j. Then we change
every j in the right column to a j + 1. We see the general
shape of this transformation in Figure 5.16.
● If the left column has both j’s and j + 1’s and the right
column has only j’s, then we find the highest j in the left
column, and we change every j in the right column that is
above the height of this j to a j + 1. Then we change every
j + 1 in the left column to a j. We see the general shape of
this transformation in Figure 5.17.
j+1 j
.. j .. j+1
. .
.. .. .. ..
. . . .
j+1 j j j+1
7!
j j j j
.. .. .. ..
. . . .
.. ..
j . j .
j j
Figure 5.17. One way of modifying columns to get closer to

a reverse plane partition
If the resulting filling is not a plane partition, then we repeat this

process, again finding the leftmost pair of adjacent columns with a
j + 1 immediately to the left of a j, and modifying those columns as
above.
We might hope each pair of columns we need to modify is further
to the right than the last, but it is sometimes necessary to backtrack.
Nevertheless, suppose we assign values to the columns of a filling so
that a column with j’s but no j + 1’s is worth two points, a column
with some j’s and some j + 1’s is worth one point, and all other
columns are worth no points. Now let the total value of a filling be
∑k jvk , where vk is the value of the kth column. Then the value
of any filling is a nonnegative integer, and we can check that our
modifications always transform a filling to another filling of smaller
value. Therefore, even though we might sometimes have to backtrack,
this process must eventually end. At that point we have the reverse
plane partition γj (T ).
By construction the number of columns of T which contain a j
is equal to the number of columns of flipj (T ) which contain a j + 1
and vice versa, and the swapping of partial columns we use to obtain
γj (T ) does not change these numbers. So the number of columns
of T which contain a j is equal to the number of columns of γj (T )
which contain a j + 1, and vice versa. And we could use a few more
technical results to show γj is an involution for all j. Instead of

pursuing these results, we refer the interested reader to the paper of
Galashin, Grinberg, and Liu [GGL16] for details, and we turn our
attention to the question of how to write the dual stable Grothendieck
functions as linear combinations of Schur functions.
Example 5.31. For each n ≥ 0, write gn as a linear combination of
Schur functions.
Solution. Since a semistandard tableau T of shape (n) cannot have

repeated entries in a column, these tableaux are exactly the reverse
plane partitions of shape (n), and we have xT = xircont(T ) for all of
them. Therefore, gn = sn for all n ≥ 0. □
Example 5.32. For each n ≥ 0, write g1n as a linear combination of
Schur functions.
Solution. Since the Ferrers diagram for (1n ) has just one column, ev-
ery term in g1n has the form xj1 xj2 ⋅ ⋅ ⋅ xjk for j1 < j2 < ⋅ ⋅ ⋅ < jk . There-
fore, g1n is a linear combination of Schur functions s1k . The coefficient
of s1k in this linear combination is the coefficient of x1 x2 ⋅ ⋅ ⋅ xk , which
is the number of reverse plane partitions of shape 1n with at least one
j for 1 ≤ j ≤ k, and no other entries. This is the number of solutions
to a1 + ⋅ ⋅ ⋅ + ak = n with aj ≥ 1 for all j, where we can interpret aj as
the number of times j appears in our plane partition. And this is the
number of sequences of k − 1 bars and n − k stars, or (n−1 k−1
). Therefore,
n
n−1
g1n = ∑ ( )s1k . □
k=1 k−1
When we compare our result in Example 5.21 with our result in

Example 5.32, we see there seems to be a relationship between the
Grothendieck polynomials and the dual stable Grothendieck polyno-
mials. In particular, we might be able to use this apparent relation-
ship to turn Proposition 5.25 into the following result.
Proposition 5.33. For any partition λ, we have
gλ = ∑ fλµ sµ ,
µ⊆λ
where fλµ is the number of elegant tableaux of shape λ/µ.

We will prove this result at the end of Section 8.1, after we have
developed an important combinatorial tool involving semistandard
tableaux.
5.4. The Chromatic Symmetric Function

So far we have looked at symmetric functions arising from tableaux
in various ways, all of which are somehow analogues of the Schur
functions. Now we take a different tack, by describing a way to use
proper colorings of graphs to produce symmetric functions.
To get started, recall that a simple graph (or a graph, for short)
is a set V of vertices, together with a set E of edges, which are
subsets of V of size 2. We say vertices x and y are adjacent whenever
{x, y} is an edge. We usually visualize a graph by representing the
vertices as dots, and the edges as lines or curves connecting their two
elements. Common graphs include K4 (the complete graph on four
vertices, which has four vertices and all possible edges between them),
K3,3 (the complete bipartite graph on two sets of three vertices each,
which consists of a set of three vertices, a set of three other vertices,
and all of the edges connecting vertices in different sets), and the
Petersen graph, which is shown in Figure 5.18.
The chromatic symmetric function of a graph is a generalization
of the chromatic polynomial of a graph, which in turn arises naturally
when we consider proper colorings of the vertices of a graph. To
Figure 5.18. The Petersen graph

5.4. The Chromatic Symmetric Function 145
1 2 1 2
Figure 5.19. A proper coloring of P4 with two colors
explain what all of this means, we start with colorings. If n ≥ 1, then

a coloring of a graph G with n colors is an assignment of a color from
[n] to each vertex. That is, a coloring of G is a function from its
vertex set to [n]. A proper coloring is a coloring with the property
that if two vertices are adjacent, then they have different colors.
Example 5.34. Let P4 be the path with four vertices. Find a proper
coloring of P4 with two colors, a proper coloring with three colors,
and a proper coloring with four colors.
Solution. In Figure 5.19 we have a proper coloring of P4 with two

colors. There is one other proper coloring of P4 with two colors, which
we can obtain from this one by swapping the 1’s and 2’s. There are
several proper colorings of P4 with three colors, and even more with
four colors. However, we are not required to use all of the available
colors, so the coloring in Figure 5.19 is also a proper coloring of P4
with three colors and a proper coloring of P4 with four colors. □
In our solution to Example 5.34 we mentioned that there are

many proper colorings of P4 with four colors, but we can be more
specific about this. In particular, for any graph G there is a function
χG for which χG (n) is the number of proper colorings of G with n
colors. In fact, in Problem 5.11 you will show χG (n) is a polynomial
in n, so we call χG the chromatic polynomial of G. Computing χG
for a graph like the Petersen graph is probably hard (though the
recurrence relation in Problem 5.10 is helpful), but we can do it for
some simpler graphs.
Example 5.35. Let Pk be the path with k vertices, and compute

χPk (n).
Solution. We color the vertices of Pk one at a time, starting with an

endpoint and working along the path. When we do this, we first find
there are n colors for the endpoint. This leaves n − 1 colors for the
endpoint’s neighbor. This, in turn, leaves n − 1 colors for the next

vertex in the path and n−1 colors for the vertex after that. In general,
we have n − 1 colors available for each successive vertex after the first
endpoint, so χPk (n) = n(n − 1)k−1 . □
Our solution to Example 5.35 suggests that if we can remove all

of the vertices of a graph G by removing one vertex of degree 1 at
a time, then we can compute χG (n). We call connected graphs of
this sort trees, and we have various ways of characterizing them in
Problems 5.12 and 5.15. Not only can we compute χT for every tree
T , but the answer we get only depends on the number of vertices in
T.
Proposition 5.36. If T is a tree with k vertices, then for all n ≥ 1

we have χT (n) = n(n − 1)k−1 .
Proof. We argue by induction on k.

If k = 1, then T consists of a single vertex, and χT (n) = n, which
is what we wanted to prove in this case.
Now suppose k > 1, and the result holds for all trees with k − 1
vertices. By definition, T has a vertex of degree 1. If we remove one of
these vertices from T , then we obtain a tree U with k − 1 vertices. We
can construct each proper coloring of T by choosing a proper coloring
of U , and then choosing a color for the endpoint we removed from T .
By induction, U has n(n − 1)k−2 proper colorings, and there are n − 1
choices for the color of the endpoint, so χT (n) = n(n − 1)k−1 . □
Although we did not phrase it this way, by definition the chro-

matic polynomial of a graph, evaluated at n, is a sum over all proper
colorings of the graph with n colors, in which each term is 1. The
chromatic symmetric function of a graph G is also a sum over all
proper colorings of G, but this time the summand for a given color-
ing is, in a certain sense, the product of the colors of the vertices.
More specifically, we have the following definition.
Definition 5.37. Suppose G is a graph with vertex set V and c is a

proper coloring of G. Then we define the weight of c, written wtG (c),
by setting
wtG (c) = ∏ xc(v) .
v∈V
We define the chromatic symmetric function of G, written XG , by
setting
XG = ∑ wtG (c),
c
where the sum on the right is over all proper colorings of G, with any
number of colors.
Before computing XG for a few specific graphs we make two ob-

servations.
First, XG is a symmetric function for every graph G. To see
this, suppose G is a graph and π ∈ Sn for some positive integer n.
If c is a coloring of the vertices of G with n colors, then π permutes
the colors of c to produce a new coloring cπ of G. Furthermore,
wtG (cπ ) = π(wtG (c)), and c is a proper coloring if and only if cπ is,
since π is a bijection. Therefore, π(XG ) = XG .
Second, for any graph G the chromatic polynomial χG (n) is a
specialization of XG , since χG (n) = XG (1, . . . , 1, 0, 0, . . .).
´¹¹ ¹ ¹ ¹ ¹¸ ¹ ¹ ¹ ¹ ¹¶
n
Example 5.38. Find the chromatic symmetric function for the com-
plete graph Kn in terms of any convenient set of symmetric functions.
Solution. Since all pairs of vertices are adjacent in Kn , every proper

coloring will use exactly n colors. For each choice j1 < j2 < ⋅ ⋅ ⋅ <
jn of colors, we have n! ways to assign j1 , . . . , jn to the vertices of
Kn . Therefore, each choice j1 < j2 < ⋅ ⋅ ⋅ < jn of colors contributes
n!xj1 ⋅ ⋅ ⋅ xjn to XKn . This means XKn = n!en . □
Example 5.39. Find the chromatic symmetric function for the graph
Knc with n vertices and no edges in terms of any convenient set of
Solution. We can construct each proper coloring of Knc uniquely by

choosing a color for each vertex. Choosing a color for a vertex is the
same as choosing a term from x1 + x2 + x3 + ⋅ ⋅ ⋅, so we must have
XKnc = e1n = h1n = mn1 . □
Our solution to Example 5.39 suggests we can compute the chro-

matic symmetric function of a disconnected graph by multiplying the
chromatic symmetric functions of its components. In particular, we
have the following result.
Proposition 5.40. Suppose a graph G consists of two graphs G1 and
G2 with disjoint vertex sets, and there is no edge in G connecting a
vertex in G1 with a vertex in G2 . Then
XG = XG1 XG2 .
Proof. We can construct each proper coloring c of G uniquely by

choosing a proper coloring c1 of G1 and a proper coloring c2 of G2 .
Since wtG (c) = wtG1 (c1 ) wtG2 (c2 ), we must have XG = XG1 XG2 . □
In general, computing XG by hand is hard, but we can do it

in some small examples by considering how the colors partition the
vertices. To be more specific about this, first suppose G is a graph
with vertex set V . A stable partition V1 , V2 , V3 , . . . of G is an ordered
set partition of V in which any of the sets Vj may be empty, and
if two vertices are connected by an edge, then they are in different
blocks of the partition. Each proper coloring c of G induces a stable
partition in which Vj is the set of vertices of color j. Conversely, each
stable partition induces a unique proper coloring of G. As a result,
we can use the language of stable partitions to write XG in terms of
Proposition 5.41. For any graph G with d vertices and any partition
λ ⊢ d, let zλ (G) be the number of stable partitions of G with ∣Vj ∣ = λj
for all j. Then we have
(5.4) XG = ∑ zλ (G)mλ .
λ⊢d
λ
Proof. By definition, the coefficient of xλ1 1 ⋅ ⋅ ⋅ xλl(λ) l(λ)
(and therefore
the coefficient of mλ in XG ) is the number of proper colorings of
G in which λj vertices have color j for all j. From our discussion
above we know we can construct each such proper coloring uniquely
by choosing a stable partition V1 , V2 , . . . of G and assigning the colors
1, 2, . . . , l(λ) to V1 , . . . , Vl(λ) in order. Therefore, the coefficient of mλ
in XG is zλ (G), which is what we wanted to prove. □
1 2 3 2 3 1
1 3 2 3 1 2
2 1 3 3 2 1
Figure 5.20. The six stable set partitions of the path with
three vertices with λ = (13 )
1 2 1
Figure 5.21. The stable set partition of the path with three
vertices with λ = (2, 1)
Using Proposition 5.41, we can find XG for a variety of small

graphs.
Example 5.42. Find XG as a linear combination of monomial sym-

metric functions, where G is the path with three vertices.
Solution. If λ = (13 ), then zλ (G) = 6, as shown in Figure 5.20.

If λ = (2, 1), then zλ (G) = 1, as shown in Figure 5.21.
Finally, if λ = (3), then zλ (G) = 0. Putting it all together, we
find XG = 6m111 + m21 . □
In Tables 5.1 and 5.2 we have XG for a variety of small graphs

G.
We know the chromatic symmetric function for the complete
graphs and for the graphs with no edges, but we might like to know
the chromatic symmetric function for some other families of graphs.
One natural family to consider is the family of paths. No pattern is
apparent when we express XPn in terms of the monomial symmetric
functions for small n, and as we see from the data in Table 5.3, the
situation is not much better when we express XPn in terms of the
elementary symmetric functions. But if we dig a little deeper, we can
use the elementary symmetric functions to give a recursive formula
for XPn .
Table 5.1. The chromatic symmetric functions of several

graphs with at most four vertices
G XG
6m111 + m21
24m1111 + 6m211 + 2m22
24m1111 + 4m211 + 2m22
24m1111 + 2m211
24m1111 + 4m211
24m1111 + 6m211 + m31

n
(5.5) XPn = en + ∑ (k − 1)ek XPn−k .
k=2
Proof. When n = 0, we have XP0 = 1 = e0 , and when n = 1, we have

XP1 = x1 + x2 + ⋅ ⋅ ⋅ = e1 , so suppose n ≥ 2.
If we imagine traversing Pn from one end to the other, then we
can classify the proper colorings of Pn according to whether they have
an edge along which the colors decrease, and if so, roughly where the
first decrease occurs.
The proper colorings with no edge along which the colors decrease
are exactly those for which the colors are strictly increasing, so the
generating function for these colorings is en .
On the other hand, we can construct each proper coloring of
Pn which contains a decrease as follows. First, choose a sequence
c1 < c2 < ⋅ ⋅ ⋅ < ck of positive integers. Then mark any entry cj < ck of
this sequence. Finally, choose a proper coloring of Pn−k , and let the
Table 5.2. The chromatic symmetric functions of several

graphs with five vertices
G XG
120m11111 + 36m2111 + 12m221 + 2m311 + m32
120m11111 + 36m2111 + 10m221 + 4m311 + m32
120m11111 + 36m2111 + 12m221 + 8m311 + m41
120m11111 + 30m2111 + 10m221
120m11111 + 30m2111 + 8m221 + 2m311 + m32
120m11111 + 24m2111 + 6m221
Table 5.3. The chromatic symmetric functions of paths with

six or fewer vertices
n XPn
0 1
1 e1
2 2e2
3 3e3 + e21
4 4e4 + 2e31 + 2e22
5 5e5 + 3e41 + 7e32 + 3e221
6 6e6 + 4e51 + 10e42 + 6e33 + 4e321 + 2e222
color of the leftmost vertex in this coloring be a. Note that the gener-
ating function for the objects we have constructed is (k − 1)ek XPn−k .
To turn these objects into proper colorings of Pn , we consider two
cases. If a ≠ cj , then our coloring is c1 , c2 , . . . , cj−1 , cj+1 , . . . , ck , cj ,

followed by our proper coloring of Pn−k . If a = cj , then our color-
ing is c1 , c2 , . . . , cj−1 , cj , cj+1 , . . . , ck , followed by our proper coloring
of Pn−k .
To see that every proper coloring of Pn which contains a decrease
is constructed uniquely by this process, suppose we have a proper
coloring of Pn which begins with d1 , . . . , dl , a, where d1 < d2 < ⋅ ⋅ ⋅ < dl
and dl > a. If a = dj for some j < l, then we must have been in the
second case above, and we can only build our coloring by taking k = l
and cm = dm for 1 ≤ m ≤ k. On the other hand, if a is not among
d1 , . . . , dl , then we must have been in the first case above. Now we
can only build our coloring by taking k = l + 1 and letting our chosen
sequence be d1 , . . . , dl , a, rearranged to be in increasing order. □
As often happens, our recurrence relation leads to a nice expres-

sion for the associated generating function.
Corollary 5.44. We have
∞ ∞
∑j=0 ej tj
∑ XPn t =
n
(5.6) .
n=0 1 − ∑∞
j=2 (j − 1)ej t
j
Proof. When we multiply (5.5) by tn and sum the result over all
n ≥ 0, we find
∞ ∞ ∞ n
∑ XPn t = ∑ en t + ∑ ∑ (k − 1)ek XPn−k t t
n n k n−k
n=0 n=0 n=0 k=2
∞ ∞ ∞
= ∑ en tn + ( ∑ (j − 1)ej tj )( ∑ XPn tn ).
n=0 j=2 n=0
When we solve this equation for ∑∞ n

n=0 XPn t , we obtain (5.6). □
We saw in Proposition 5.36 that every tree with k vertices has

the same chromatic polynomial, so we might wonder which trees with
k vertices have the same chromatic symmetric function. The data in
Tables 5.1 and 5.2 are helpful here: we see that no two trees with
five or fewer vertices have the same chromatic symmetric function.
This leads us to a remarkable question first posed by Stanley, which
remains open today. To state it, we first need some terminology.
Suppose G1 is a graph with vertex set V1 and G2 is a graph with vertex
5.5. Problems 153
set V2 . An isomorphism from G1 to G2 is a bijection f ∶ V1 → V2 such

that x, y ∈ V1 are adjacent in G1 if and only if f (x) and f (y) are
adjacent in G2 . We say G1 is isomorphic to G2 whenever there is an
isomorphism from G1 to G2 .
Question 5.45 ([Sta95, page 170]). Suppose T1 and T2 are trees
with XT1 = XT2 . Must T1 and T2 be isomorphic?
5.5. Problems
5.1. Suppose k ≥ 4, λ = (k, 2), and µ = (1). For a given positive
integer n, how many semistandard skew tableaux of shape λ/µ
are there with entries in [n]?
5.2. Show that if partitions µ ≠ λ have µ ⊆ λ, then µ <lex λ and
µ ⊴ λ.
5.3. If λ = (k n ) for positive integers k and n, then how many parti-
tions µ have µ ⊆ λ?
5.4. Suppose µ ⊆ λ are partitions and n is a positive integer. Use
the Bender–Knuth involutions to show that sλ/µ (Xn ) and sλ/µ
are symmetric.
5.5. For each n ≥ 2, how many elegant tableaux are there
of shape λ/µ, where λ = (n, n − 1, n − 2, . . . , 2) and µ =
(n − 2, n − 3, n − 4, . . . , 1)?
5.6. Let T be the reverse plane partition in Figure 5.22. For each
j, compute γj (T ), where γj is the generalized Bender–Knuth
involution described in Section 5.3.
3 3
1 2 3 5 6
1 2 3 4 5
1 2 3 4 5 5 6 6
1 1 3 3 5 5 5 6
1 1 1 3 5 5 5 6
Figure 5.22. The reverse plane partition T for Problem 5.6

5.7. Find the chromatic polynomial of Kk , the complete graph on k

vertices.
5.8. Suppose G is a connected graph with k vertices. Show that if

χG (n) = n(n − 1)k−1 , then G is a tree.
5.9. Find the chromatic polynomial of Ck , the cycle with k vertices.
5.10. For any graph G and any edge e in G, let G/e be the graph
we obtain from G by removing e, and let G/e be the graph we
obtain from G by removing e and identifying the two vertices
it connects. Show
χG (n) = χG/e (n) − χG/e (n).
5.11. Show that if G is a graph with k vertices, then χG (n) is a

polynomial in n of degree k.
5.12. Suppose n ≥ 3 and G is a graph. A cycle of length n in G is a

sequence v1 , . . . , vn of distinct vertices such that vj is adjacent
to vj+1 for 1 ≤ j ≤ n − 1 and vn is adjacent to v1 . Show a
connected graph G is a tree if and only if it has no cycles.
5.13. A crab is a graph with exactly one cycle. If G is a crab with k

vertices whose cycle has length l, then find χG (n).
5.14. For each n ≥ 3, let P 2k be the graph with vertices 1, 2, . . . , k

in which vertices i and j are adjacent whenever 0 < ∣i − j∣ ≤ 2.
Compute χP 2k (n).
5.15. Suppose G is a connected graph with k vertices and l edges.

Show G is a tree if and only if k − l = 1.
5.16. For any graph G, an acyclic orientation of G is a choice of orien-

tation for each edge, such that the resulting directed graph has
no cycles in which all of the edges are oriented in the same
direction. Show that if G is a graph with n vertices, then
(−1)n χG (−1) is the number of acyclic orientations of G.
5.5. Problems 155
Figure 5.23. The graphs for Problem 5.17
5.17. Find the chromatic symmetric functions for the graphs in Figure
5.23.
5.18. Find the chromatic symmetric functions for the graphs in Figure
5.24.
Figure 5.24. The graphs for Problem 5.18
5.19. Show that if graphs G and H have XG = XH , then G and H

have the same number of vertices.
5.20. Show that if graphs G and H have XG = XH , then G and H
have the same number of edges.
5.21. For each n ≥ 2, let Ln be the graph we get by removing one
edge from the complete graph with n vertices. Find and prove a
formula for XLn in terms of the elementary symmetric functions
and in terms of the Schur functions.
5.22. Show
∞
1
∑ XPn = .
n=0 1 − p1 t + p2 t2 − p3 t3 + ⋅ ⋅ ⋅
5.23. For each n ≥ 3, let Cn be the cycle with n vertices. Find and
prove a formula for the generating function
∞
n
∑ XCn t
n=3
in terms of the elementary symmetric functions.

5.6. Notes
Propositions 5.11 and 5.13 appear in Stanley’s Enumerative Combi-
natorics, Volume 2 [Sta99, Exercise 7.56(a)], and van Willigenburg
[vW05] uses Littlewood–Richardson tableaux to prove the more gen-
eral version of Proposition 5.12. Billera, Thomas, and van Willi-
genburg [BTvW06] prove a general result explaining the fact that
s(5,42 ,3,1)/(32 ,2) = s(5,4,22 ,1)/(3,12 ) .
The stable Grothendieck polynomials Gλ were first introduced
by Fomin and Kirillov [FK96] in another context, but it was Buch
[Buc02] who first gave a combinatorial formula for Gλ in terms of
set-valued tableaux. The dual stable Grothendieck polynomials were
defined in terms of reverse plane partitions by Lam and Pylyavskyy
[LP07]. Macdonald has gathered a variety of other Schur function
variations [Mac92].
Reverse plane partitions are connected with some rich enumer-
ative combinatorics, as are several other types of plane partitions.
Many books have more information on this, including [Bre99] and
[Sta99, Sec. 7.20-7.22].
Cho and van Willigenburg [CvW16] have found numerous bases
for Λ generated by chromatic symmetric functions. In particular, they
showed that for any family {Hk }k≥1 of graphs in which Hk has exactly
k vertices for each k, the symmetric functions {XHk }k≥1 generate Λ.
Martin, Morin, and Wagner [MMW08] have made some progress on
Stanley’s tree question, but the general problem remains open. In
related work, Orellana and Scott [OS14] have constructed an infinite
family of pairs of unicyclic graphs with the same chromatic sym-
metric function. Finally, several authors have studied variations and
generalizations of Stanley’s chromatic symmetric function, including
Gebhard and Sagan [GS01], Williams [Wil07], Humpert [Hum11],
and Shareshian and Wachs [SW16].
Chapter 6
The Jacobi–Trudi
Identities and an
Involution on Λ
Now that we have several bases for the vector space of symmetric
functions, we would like to know how to express these bases in terms
of one another. We have already made some progress in this direc-
tion: we have combinatorial descriptions of the coefficients we obtain
when we express the elementary symmetric functions, the complete
homogeneous symmetric functions, the power sum symmetric func-
tions, and the Schur functions in terms of the monomial symmetric
functions. However, we do not know how to express (for example)
the power sum symmetric functions in terms of the Schur functions.
In this chapter we tackle two related questions along these lines. In
particular, we show how to use determinants to express the Schur
functions in terms of the complete homogeneous symmetric functions
and then in terms of the elementary symmetric functions. Our results
will lead us to a new symmetry on the space of symmetric functions.
6.1. The First Jacobi–Trudi Identity

Our first goal is to express the Schur functions in terms of the com-
plete homogeneous symmetric functions. We already know how to do
157
158 6. The Jacobi–Trudi Identities and an Involution on Λ
this in some simple cases; for example, the fact that sn = hn follows
immediately from the definitions of sn and hn . To see what happens
more generally, we consider a slightly larger case.
Example 6.1. Express the Schur function s22 as a linear combina-
tion of complete homogeneous symmetric functions. Consider using
a software package like SageMath for your computations.
Solution. If we decide to forgo computer assistance, we can use Prob-

lem 2.16 and the data in Figure 2.8 to find
h4 = m4 + m31 + m22 + m211 + m1111 ,
h31 = m4 + 2m31 + 2m22 + 3m211 + 4m1111 ,
h22 = m4 + 2m31 + 3m22 + 4m211 + 6m1111 ,
h211 = m4 + 3m31 + 4m22 + 7m211 + 12m1111 ,
and
h1111 = m4 + 4m31 + 6m22 + 12m211 + 24m1111 .
When we solve this system, we get
m4 = 4h4 − 4h31 − 2h22 + 4h211 − h1111 ,
m31 = −4h4 + 7h31 + 2h22 − 7h211 + 2h1111 ,
m22 = −2h4 + 2h31 + 3h22 − 4h211 + h1111 ,
m211 = 4h4 − 7h31 − 4h22 + 10h211 − 3h1111 ,
and
m1111 = −h4 + 2h31 + h22 − 3h211 + h1111 .
Since s22 = m22 + m211 + 2m1111 , we have s22 = h2 h2 − h3 h1 . □
When we study our answer in Example 6.1, we see we have

h h3
s22 = det ( 2 ). Using similar computations, or possibly tech-
h1 h2
h h4
nological assistance, we find s32 = h3 h2 − h4 h1 = det ( 3 ) and
h1 h2
h h4
s33 = h3 h3 − h4 h2 = det ( 3 ). In fact, if we take h0 = 1 and
h2 h3
⎛ h2 h3 h4 ⎞
hn = 0 for n < 0, then we also find s22 = det ⎜ h1 h2 h3 ⎟ and
⎝h−2 h−1 h0 ⎠
6.1. The First Jacobi–Trudi Identity 159
h h3 ⎛ h2 h3 h4 ⎞
s21 = det ( 2 ) = det ⎜ h0 h1 h2 ⎟. Moving to longer par-
h0 h1 ⎝h−2 h−1 h0 ⎠
⎛ h2 h3 h4 ⎞
titions, we can also compute s211 = det ⎜ h0 h1 h2 ⎟ and s333 =
⎝h−1 h0 h1 ⎠
⎛h3 h4 h5 ⎞
det ⎜h2 h3 h4 ⎟. Taken together, these computations suggest the
⎝h1 h2 h3 ⎠
following result.
Theorem 6.2 (The First Jacobi–Trudi Identity). For any partition
λ and any k ≥ l(λ), we have
(6.1) sλ = det (hλj +l−j )1≤j,l≤k .
Here we take hn = 0 for all n < 0.
One can use matrix computations to prove this result, but we

prefer to give a combinatorial proof. To do this, we first need a
combinatorial description of hn for each n. Fortunately, the definition
of hn is combinatorial: for all n we have
(6.2) hn = ∑ ∏ xj ,
J⊆[P] j∈J
∣J∣=n
where the sum on the right is over all multisets of n positive integers.
To visualize one of these multisets, we imagine a bar graph in which
each element j of the multiset corresponds to a 1 × j bar, each bar
sits on the x-axis, and the bars are in weakly increasing order from
left to right with no gaps between consecutive bars. In Figure 6.1 we
see the bar graph for the multiset {1, 1, 3, 4, 4, 4}.
Figure 6.1. The bar graph for {1, 1, 3, 4, 4, 4}

Figure 6.2. The boundary path of the bar graph for {1, 1, 3, 4, 4, 4}
When we work with these bar graphs, we will find it convenient

to consider only the northwest boundary of the graph. In Figure 6.2
we have drawn the northwest boundary of the bar graph for our mul-
tiset {1, 1, 3, 4, 4, 4} as a thick, black line. To make these boundaries
precise, for any integer a and any n ≥ 0, we recall that a lattice path
from (a, 1) to (a + n, ∞) is an infinite sequence of north (0, 1) and
east (1, 0) steps which contains exactly n east steps. We write Γa,n to
denote the set of all lattice paths from (a, 1) to (a + n, ∞). In other
words, Γa,n is the set of all lattice paths beginning at (a, 1) which
have exactly n east steps.
As the graph in Figure 6.2 suggests, the lattice paths in Γa,n are
in bijection with multisets of n positive integers, so we can view hn
as a generating function for these paths. To do this, we need an ap-
propriate weight function. With this in mind, the h-weight of an east
step from (b, k) to (b+1, k) is xk , and the h-weight wth (α) of a lattice
path α is the product of the h-weights of its east steps. For example,
in Figure 6.3 each east step in the boundary path of the bar graph for
{1, 1, 3, 4, 4, 4} is labelled with its h-weight. In particular, the lattice
path associated with {1, 1, 3, 4, 4, 4} is EEN N EN EEEN ∞ , and its
h-weight is x21 x3 x34 . By construction, and by (6.2), for any integers a
and n we have
(6.3) hn = ∑ wth (α).

α∈Γa,n
Notice that we are free to choose any integer a in (6.3); soon we will
make an especially useful choice for this parameter.
x4 x4 x4
x3
x1 x1
Figure 6.3. The h-weights on the boundary of the bar graph

for {1, 1, 3, 4, 4, 4}
Now that we have a suitable combinatorial interpretation of hn ,

we turn our attention to the determinant on the right side of (6.1).
Using Proposition C.7 to expand this determinant in terms of permu-
tations, we find
k
det (hλj +l−j )1≤j,l≤k = ∑ (−1)inv(π) ∏ hλm +π(m)−m
π∈Sk m=1
k
= ∑ (−1)inv(π) ∏ ∑ wth (α).
π∈Sk m=1 α∈Γa,λm +π(m)−m
When we expand the product on the right we obtain a sum of products

of weights, in which each product is indexed by a sequence of lattice
paths, one for each weight factor in the product. In symbols, we have
k
det (hλj +l−j )1≤j,l≤k = ∑ (−1)inv(π) ∑ ∏ wth (αm ).
α1 ,...,αk m=1
π∈Sk
αj ∈Γa ,λ +π(j)−j
j j
As we mentioned before, in the inner sum on the right we are free

to start each lattice path αj at any point (aj , 1) that we find con-
venient. In other words, we can move our entire sequence α1 , . . . , αk
of lattice paths, or various individual paths in the sequence, left or
right to get another sequence of paths with the same collection of
h-weights. Some collections of starting points will turn out to be
more helpful than others, and to make our proof work we will take
aj = −π(j) for all j with 1 ≤ j ≤ k. In Figure 6.4 we have chosen
π = 231 and λ = (4, 32 ), which means α1 (in blue) begins at (−2, 1)
and has 4 + 2 − 1 = 5 east steps, α2 (in red) begins at (−3, 1) and has
Figure 6.4. A sequence of lattice paths for π = 231 and λ = (4, 32 )
3 + 3 − 2 = 4 east steps, and α3 (in green) begins at (−1, 1) and has

3 + 1 − 3 = 1 east step.
As we might guess from Figure 6.4, if we are given a sequence
α1 , . . . , αk of lattice paths with αj ∈ Γaj ,λj +π(j)−j for 1 ≤ j ≤ k, then
we can recover the associated permutation π and partition λ.
To recover π, notice that αj begins at (−π(j), 1), moves λj +πj −j
steps to the east (along with infinitely many steps north), and ends
at (−π(j) + λj + π(j) − j, ∞) = (λj − j, ∞). We know λj+1 ≤ λj ,
which means λj+1 − (j + 1) < λj − j. This, in turn, tells us the ending
x-coordinates of α1 , . . . , αk are distinct, with α1 ending furthest to
right, α2 ending next to the right, etc. If we now read our paths from
top to bottom, we see that αj leads from the jth point from the right
at the top to the π(j)th point from the right at the bottom. So we
can recover π by numbering the starting points and ending arrows 1
to k from right to left, and reading our sequence of paths downward,
as a function.
Now recovering λ is even easier than recovering π. From our
analysis above we know the rightmost ending arrow is the ending
arrow for α1 , which is (λ1 − 1, ∞). So we can recover λ1 by adding 1
to the x-coordinate of the path ending furthest to the right. Similarly,
we can recover λj by adding j to the x-coordinate of the path ending
j units from the right.
Figure 6.5. The family of lattice paths in Example 6.3
Example 6.3. Find the permutation π, the partition λ, and the h-

weights corresponding to the family of lattice paths in Figure 6.5.
Solution. In Figure 6.6 we have our sequence of lattice paths, with

the starting points and ending arrows numbered to help us find π.
From our discussion above we know the red path is α1 , the yellow
path is α2 , the light blue path is α3 , the dark blue path is α4 , and
the green path is α5 . Looking at the starting points for these paths,
we see π(1) = 4, π(2) = 1, π(3) = 2, π(4) = 3, and π(5) = 5. That is,
π = 41235.
To find λ, first note that α1 ends at (3, ∞), so λ1 −1 = 3 and λ1 = 4.
Similarly, α2 ends at (2, ∞), so λ2 − 2 = 2 and λ2 = 4. Continuing in
this way, we find λ = (42 , 3, 2).
Finally, by looking at the heights of the east steps in each path,
we find we have the weights wth (α1 ) = x22 x4 x35 x7 , wth (α2 ) = x3 x26 ,
wth (α3 ) = x1 x6 , wth (α4 ) = x5 , and wth (α5 ) = 1. □
Figure 6.6. The family of lattice paths in Example 6.3 with

the domain and range of π
Because we can recover π from a given sequence of lattice paths,

we can now write our determinant det (hλj +l−j )1≤j,l≤k as a sum over
sequences of lattice paths. In particular, for any partition λ and any
k with k ≥ l(λ), let Hλ,k be the set of sequences α1 , . . . , αk of lattice
paths such that αj ends at (λj −j, ∞) for each j, and each path starts
at a different point among (−1, 1), . . . , (−k, 1). For any such sequence
β = (α1 , . . . , αk ), we set wth (β) = ∏km=1 wth (αm ). With this notation,
we have
(6.4) det (hλj +l−j )1≤j,l≤k = ∑ (−1)inv(π) wth (β).

β∈Hλ,k
As we have seen, specifying which path begins at which point is equiv-

alent to specifying π.
Our proof of Theorem 6.2 involves two major steps. First, since
the right side of (6.1) (which we have rewritten in (6.4)) has negative
terms and the left side does not, we need to explain how to cancel
these negative terms. Second, once we have canceled the negative
terms, as well as some of the positive terms, on the right side of (6.1),
we need to explain how the remaining terms on the right correspond

with the terms on the left.
To describe how to cancel the negative terms on the right side of
(6.1), we define a function tswp (which is short for “tail-swap”) from
Hλ,k to Hλ,k .
If β ∈ Hλ,k and no two paths in β intersect, then we set tswp(β) =
β.
Now suppose β ∈ Hλ,k and at least one pair of paths in β in-
tersect. As above, let the individual paths in β be α1 , . . . , αk . To
start computing tswp(β), we first find the largest y-coordinate of any
point at which two paths in β intersect. That is, we find the highest
horizontal line L on which two paths in β intersect. Among all of the
intersection points with this y-coordinate, we choose the one with the
largest x-coordinate. That is, we choose the rightmost intersection
point on L. Note that if more than two paths intersect at this point,
then at least two of them must leave in the same direction (north or
east). If two paths leave to the north, then we have an intersection
point on a higher line, and if two paths leave to the east, then we
have an intersection point farther to the right on L. Both of these
contradict our choice of point, so it must be that exactly two paths
intersect at our chosen point; let these paths be αm and αw , where
m < w.
Now we observe that αm consists of a “before” part and an “after”
−
part. More specifically, if αm is the part of αm below our intersection
+ − +
point and αm is the part of αm above this point, then αm = αm , αm .
− +
Similarly, if αw is the part of αw below our intersection point and αw
− +
is the part of αw above this point, then αw = αw , αw . To construct
− + − +
tswp(β), replace αm with αw , αm , replace αw with αm , αw , and leave
all of the other paths unchanged. Note that, geometrically, we are
swapping the “tails” of αm and αw .
Example 6.4. Compute tswp(β) and tswp(tswp(β)) for the se-

quence β of lattice paths in Figure 6.7.
Solution. The highest row in which two paths in our sequence inter-
sect is the row at height 7. This row contains just one intersection
point, so the rightmost intersection point is the point in this row where
Figure 6.7. The family of lattice paths in Example 6.4
the red and blue paths intersect. When we swap the tails of these
paths, we obtain the sequence of paths in Figure 6.8. Note that the
permutation for β is 23145 and the partition is (5, 42 , 3, 2), while the
permutation for tswp(β) is 23415 and the partition is still (5, 42 , 3, 2).
When we compute tswp(tswp(β)), we find exactly the same in-
tersection point as we did when we computed tswp(β). So we once
again swap the tails of the red and blue paths, to find tswp(tswp(β)) =
β. □
We would like to use tswp to cancel all of the negative terms on

the right side of (6.1), so we hope tswp is a sign-reversing, h-weight-
preserving involution. Our solution to Example 6.4 suggests this is
the case, and we prove it is next.
Lemma 6.5. Suppose k ≥ 1 and λ is a partition with l(λ) ≤ k. For

any β ∈ Hλ,k , the following hold.
(i) If β has a pair of intersecting lattice paths, π is the per-
mutation for β, and σ is the permutation for tswp(β), then
sgn(σ) = − sgn(π).
Figure 6.8. The family of lattice paths in the solution to

Example 6.4
(ii) wth (tswp(β)) = wth (β).

(iii) tswp(tswp(β)) = β.
Proof. (i) By construction, we obtain σ from π by swapping the

adjacent entries π(j) and π(j + 1). Therefore, except for π(j) and
π(j +1), every pair of entries is in the same order in π as it is in σ. On
the other hand, π(j) and π(j + 1) form an inversion in π if and only
if they do not form an inversion in σ, so inv(π) and inv(σ) differ by
1. This means sgn(σ) = − sgn(π), which is what we wanted to prove.
(ii) The monomial wth (tswp(β)) is determined by the heights of
the east steps in tswp(β), while the monomial wth (β) is determined
by the heights of the east steps in β. But tswp(β) and β have the
same set of east steps, so we must have wth (tswp(β)) = wth (β).
(iii) First note that if no two lattice paths in β intersect, then by
definition tswp(β) = β, so tswp(tswp(β)) = β.
Now suppose two lattice paths in β intersect. Since β and tswp(β)
have the same set of intersection points, the intersection point we use
to compute tswp(tswp(β)) is exactly the same intersection point we
use to compute tswp(β). Furthermore, if the lattice paths in β which

− + − +
intersect at this point decompose as αm , αm and αw , αw , then the
lattice paths in tswp(β) which intersect at this point decompose as
− + − + − + − +
αm , αw and αw , αm . Now these paths become αm , αm and αw , αw
in tswp(tswp(β)), and all other paths are again unchanged, so β =
tswp(tswp(β)). In other words, when we compute tswp(tswp(β))
from tswp(β), we are unswapping the tails we swapped to compute
tswp(β) from β. □
Now we can use tswp to cancel terms on the right side of (6.1).
Lemma 6.6. Suppose k ≥ 1 and λ is a partition with l(λ) ≤ k. Then
we have
∑ (−1) wth (β) = ∑ wth (β),
inv(π)
(6.5)
β∈Hλ,k β∈IIλ,k
where IIλ,k is the set of lattice path families in Hλ,k in which no two
lattice paths intersect. (The notation II is intended to remind us of
an H with its intersections removed.)
Proof. If β ∈ Hλ,k and its associated permutation π is not 12 ⋅ ⋅ ⋅ k,

then π must have an inversion (j, l). By construction, the lattice
paths beginning at (−π(j), 1) and (−π(l), 1) must cross, so β has a
pair of intersecting lattice paths. By Lemma 6.5, the term on the left
side of (6.5) associated with β cancels with the term associated with
tswp(β). This leaves us with the sum on the right side of (6.5). □
To finish our proof of Theorem 6.2, we need to describe how the

terms on the right side of (6.5) correspond with the terms in sλ .
In particular, we complete the proof of Theorem 6.2 by providing
a weight-preserving bijection between sets of nonintersecting lattice
paths and semistandard tableaux.
Lemma 6.7. For any partition λ and any k ≥ l(λ), there is a bijection
η ∶ IIλ,k Ð→ SST(λ) such that wth (β) = xη(β) for all β ∈ IIλ,k .
Proof. Suppose β ∈ Hλ,k has β = (α1 , . . . , αk ), where αm ∈ Γ−m,λm

for 1 ≤ m ≤ k. For each m, 1 ≤ m ≤ k, let the entries in the mth row
of η(β) be the heights (that is, the y-coordinates) of the east steps
of αm , read from left to right. By definition there are λm of these
heights, and by construction they are in nondecreasing order, so we

have a row-nondecreasing filling of the Ferrers diagram of λ.
To see the resulting tableau is column-strict, first note that if the
wth entry in row m is greater than or equal to the entry immediately
above it in η(β), then the wth east step of αm+1 is at the same height
as, or below, the wth east step of αm . But the wth east step of αm+1
is in the column immediately to the left of the column containing the
wth east step of αm , so in this case αm+1 and αm must intersect. This
contradicts our choice of β, so η(β) is a semistandard tableau.
It is routine to reconstruct β from η(β), so η is a bijection. The
fact that wth (β) = xη(β) is immediate from the construction of η(β).
□
Example 6.8. If β is the family of lattice paths in Figure 6.9, then
find the corresponding semistandard tableau η(β).
Solution. Reading from right to left, the first path in β has east steps
at heights 1, 2, 2, and 5, so these are the entries in the bottom row
of η(β). Similarly, the entries in the second row from the bottom are
2, 4, and 4. Continuing in this way, we find η(β) is the tableau in
Figure 6.10. □
1
−4 −3 −2 −1 0 1 2 3
Figure 6.9. The family β of lattice paths for Example 6.8

6 7
5 5 5
2 4 4
1 2 2 5
Figure 6.10. The semistandard tableau η(β) in the solution

to Example 6.8
4 4 6
2 3 5
1 1 3 4 4
Figure 6.11. The semistandard tableau η(β) in Example 6.9
Example 6.9. Find the family β of lattice paths consisting of four

lattice paths for which η(β) is the semistandard tableau in Figure
6.11.
Solution. Since the bottom row of η(β) has entries 1, 1, 3, 4, and 4,

the rightmost lattice path in β must have east steps at exactly these
heights. Similarly, the second lattice path from the right in β must
have east steps at heights 2, 3, and 5, and the third lattice path from
the right in β must have east steps at heights 4, 4, and 6. Since η(β)
does not have a fourth row, the leftmost lattice path in β has no east
steps. Putting all of this together, we see β is as in Figure 6.12. □
Now we can assemble our maps tswp and η to prove Theorem

6.2.
6.2. The Second Jacobi–Trudi Identity 171
1
−4 −3 −2 −1 0 1 2 3 4
Figure 6.12. The lattice path family β in the solution to

Example 6.9
Proof of Theorem 6.2. For any partition λ and any k ≥ l(λ), we

have
det (hλj +l−j )1≤j,l≤k = ∑ (−1)inv(π) wth (β) (by (6.4))

β∈Hλ,k
= ∑ wth (β) (by (6.5))

β∈IIλ,k
= ∑ xT (by Lemma 6.7)

T ∈SST(λ)
= sλ ,
6.2. The Second Jacobi–Trudi Identity

Now that we have expressed the Schur function sλ in terms of the
complete homogeneous symmetric functions hµ , it is natural to ask
how to express the Schur function sλ in terms of the elementary sym-
metric functions eµ . When we do this in a few small examples, we
find
e e3
s21 = e2 e1 − e3 = det ( 2 ),
e0 e1
e e3
s22 = e2 e2 − e3 e1 = det ( 2 ),
e1 e2
⎛e2 e3 e4 ⎞
s31 = e2 e1 e1 − e2 e2 − e3 e1 + e4 = det ⎜e0 e1 e2 ⎟ ,
⎝0 e0 e1 ⎠
and
⎛e2 e3 e4 ⎞
s32 = e2 e2 e1 − e3 e1 e1 − e3 e2 + e4 e1 = det ⎜e1 e2 e3 ⎟ .
⎝0 e0 e1 ⎠
If we look at the subscripts in the diagonals of the matrices whose de-
terminants we are getting for sλ , we recognize the conjugate partition
λ′ , and if we look across the rows we see the subscripts increase by 1
every time we move to the next column to the right. This suggests
the following result.
Theorem 6.10 (The Second Jacobi–Trudi Identity). For any parti-
tion λ and any k ≥ λ1 , we have
(6.6) sλ = det (eλ′j +l−j ) .
1≤j,l≤k
Here we take en = 0 for all n < 0.
To give a combinatorial proof of this result similar to our proof

of Theorem 6.2, we need to interpret en as a generating function for
certain lattice paths with respect to a certain weight function. To
obtain an expression like this for hn , we assigned h-weight xk to each
east step at height k in a given lattice path. That is, we let the h-
weight of an east step be xk , where k is one more than the number
of north steps preceding that east step. This works when we are
trying to build hk , whose terms can have repeated subscripts, but it
cannot work for ek , which has no terms with repeated subscripts. As
a result, if the weight of one east step is xj , and it is immediately
followed by another east step, then the weight of that next east step
should be xj+1 . In other words, the weight of an east step should
take into account both the east and north steps preceding it. With
x7 x8 x9
x5
x1 x2
Figure 6.13. The e-weights on the boundary of the bar graph

for {1, 1, 3, 4, 4, 4}
this in mind, we let the e-weight of an east step be xk , where k is

one more than the total number of steps (east or north) preceding
that east step. In Figure 6.13 we have the e-weights for the steps
on the boundary of the bar graph for {1, 1, 3, 4, 4, 4}, which are in
contrast to the h-weights for this path in Figure 6.3. We define the
e-weight wte (α) of a lattice path α to be the product of the weights
of its east steps, so the e-weight of the lattice path in Figure 6.13 is
x1 x2 x5 x7 x8 x9 .
When dealing with e-weights, we will find it useful to start our
lattice paths at various heights, so we write Γa,b,n to denote the set
of all lattice paths from (a, b) to (a + n, ∞). In particular, we have
en = ∑ wte (α)
α∈Γa,b,n
for all integers a, b, and n.

Now that we can express en in terms of lattice paths, we consider
the determinant on the right side of (6.6). Using Proposition C.7 to
expand this determinant in terms of permutations, we find
k
det (eλ′j +l−j ) = ∑ (−1)inv(π) ∏ ∑ wte (α).
1≤j,l≤k m=1 α∈Γa,b,λ′
π∈Sk
m +π(m)−m
When we expand the product on the right, we obtain a sum of prod-

ucts of weights, in which each product is indexed by a sequence of
lattice paths, one for each weight factor in the product. In symbols,
we have
k
det (eλ′j +l−j ) = ∑ (−1)inv(π) ∑ ∏ wte (αm ).
1≤j,l≤k α1 ,...,αk m=1
π∈Sk
αj ∈Γ
aj ,bj ,λ′ +π(j)−j
j
As in our work with the first Jacobi–Trudi identity, we are free to

start each lattice path αj at any point (aj , bj ) we find convenient. If
we start αm at (−π(m), 1) for each m, as we did for hk , then requiring
the paths not to intersect in our noncancelling terms will force the
paths at the left end to begin with a long sequence of north steps.
This, in turn, will eliminate sequences of lattice paths for terms like
x21 , which must appear in our generating functions. To compensate
for this, we lift the starting points for the paths further to the left,
starting each path αm at (−π(m), π(m)), so that am = −π(m) and
bm = π(m).
We were able to recover π and λ from a given set of lattice paths
in our work on the first Jacobi–Trudi identity, and we can do the
same thing here, in basically the same way.
Example 6.11. Find the permutation π, the partition λ, and the

e-weights corresponding to the lattice path family in Figure 6.14.
Solution. In Figure 6.15 we have labelled each path with its corre-
sponding αj and numbered their starting points and ending arrows
Figure 6.14. The lattice path family for Example 6.11

Figure 6.15. The lattice path family for Example 6.11 with
information for π included
from right to left. If we now read our paths from top to bottom, we
see α1 (in black) tells us π(1) = 1, α2 (in yellow) tells us π(2) = 2, α3
(in green) tells us π(3) = 4, α4 (in red) tells us π(4) = 5, and α5 (in
blue) tells us π(5) = 3. Therefore, π = 12453.
To find λ, we note that for each j, the number of east steps in αj
is λ′j + π(j) − j, so λ′ = (4, 34 ) and λ = (53 , 1).
Finally, the e-weights for α1 , α2 , α3 , α4 , and α5 are x3 x5 x6 x8 ,
x1 x2 x3 , x1 x2 x4 x5 , x1 x2 x3 x4 , and x3 , respectively. □
Because we can recover π from a given sequence of lattice paths,

we can now write our determinant det (eλ′j +l−j ) as a sum over
1≤j,l≤k
sequences of lattice paths. In particular, for any partition λ and
any k with k ≥ λ1 , let Eλ,k be the set of sequences α1 , . . . , αk of
lattice paths such that αj ends at (λ′j − j, ∞) and each path starts
at a different point among (−1, 1), . . . , (−k, k). For any such sequence
β = (α1 , . . . , αk ), we set wte (β) = ∏km=1 wte (αm ). With this notation,
we have
(6.7) det (eλ′j +l−j ) = ∑ (−1)inv(π) wte (β).
1≤j,l≤k
β∈Eλ,k
Here we have again taken advantage of the fact that specifying which
path begins at which point is equivalent to specifying π.
Now we are ready to cancel all of the negative terms on the right
side of (6.7). We might be tempted to invent a new involution to do
this, but our old map tswp will do just fine, once we know it does not
change the e-weight of a family of lattice paths.
Lemma 6.12. For any k ≥ 1, any partition λ with λ1 ≤ k, and any

β ∈ Eλ,k , we have wte (tswp(β)) = wte (β).
Proof. As we saw in the proof of Lemma 6.5(ii), the families β and

tswp(β) have the same sets of east steps. Now note that if an east
step has leftmost point (l, m) and is on a lattice path whose starting
point is (−π(j), π(j)), then its e-weight is x(l+π(j))+(m−π(j)) = xl+m .
In particular, the e-weight of an east step does not depend on which
lattice path it belongs in, it only depends on the position of the step in
the plane. Therefore, each east step has the same e-weight in tswp(β)
as it has in β, which means wte (tswp(β)) = wte (β). □
Lemma 6.13. For any partition λ and any k ≥ λ1 , we have

∑ (−1) wte (β) = ∑ wte (β).
inv(π)
(6.8)
β∈Eλ,k β∈Ξλ,k
Here Ξλ,k is the set of β ∈ Eλ,k in which no two paths intersect. (The
notation Ξ is intended to remind us of an E with its intersections
removed.)
Proof. This is similar to the proof of Lemma 6.6, using Lemmas

6.5(i) and (iii) and Lemma 6.12. □
We can now interpret the right side of (6.8) in terms of semistan-

dard tableaux to obtain sλ′ .
Lemma 6.14. For any partition λ and any k ≥ λ1 , there is a bijection

ϵ ∶ Ξλ,k Ð→ SST(λ′ ) such that wte (β) = xϵ(β) for all β ∈ Ξλ,k .
Proof. For each β ∈ Ξλ,k , where β = (α1 , . . . , αk ), let ϵ(β) be the

filling of the Ferrers diagram of λ′ obtained by placing the subscripts
of the weights of αm in column m for each m. By construction ϵ(β)
is column strict and satisfies wte (β) = xϵ(β) . Moreover, it is routine
to reconstruct β from ϵ(β). We leave it as Problem 6.12 to show ϵ(β)
is row-nondecreasing if and only if no two paths in β intersect. □
10
1
−4 −3 −2 −1 0 1 2 3 4 5
Figure 6.16. The sequence of lattice paths in Example 6.15
Example 6.15. Find the semistandard tableau corresponding to the

sequence of nonintersecting lattice paths in Figure 6.16 under the map
in Lemma 6.14.
Solution. As we see in Figure 6.16, the weights for α1 are x4 , x5 , x6 ,

x8 , x10 , and x11 , so the first column of our semistandard tableau con-
tains 1, 4, 5, 6, 8, 10, and 11. Similarly, the second column contains 4,
6, and 7, the third column contains 5, and the fourth column contains
6. This gives us the semistandard tableau in Figure 6.17. □
We can now assemble the pieces we have developed to prove The-

orem 6.10.
11
10
8
6 7
5 6
4 4 5 6
Figure 6.17. The semistandard tableau in the solution to

Example 6.15
Proof of Theorem 6.10. For any partition λ and any k ≥ λ1 , we

have
det (eλ′j +l−j ) = ∑ (−1)inv(π) wte (β) (by (6.7))
1≤j,l≤k
β∈Eλ,k
= ∑ wte (β) (by (6.8))

β∈Ξλ,k
= ∑ xT (by Lemma 6.14)

T ∈SST(λ′ )
= sλ′ ,
6.3. The Involution ω

Throughout our study of symmetric functions, the elementary sym-
metric functions and the complete homogeneous symmetric functions
have played similar roles: each result involving one of these bases has
a corresponding result involving the other. The Jacobi–Trudi identi-
ties are an especially striking example of this. In fact, we can get each
of these identities from the other by interchanging e’s and h’s, if we
are also willing to replace sλ with sλ′ . This suggests an explanation
for the dual roles the e’s and the h’s are playing: there is an isomor-
phism from Λ to Λ which maps eλ to hλ , hλ to eλ , and sλ to sλ′ for
every partition λ. Each of these three requirements determines a lin-
ear transformation from Λ to Λ, so we first show these three functions
are identical.
6.3. The Involution ω 179
Proposition 6.16. The following are equivalent for any linear trans-
formation ω ∶ Λ → Λ.
(i) ω(eλ ) = hλ for all partitions λ.
(ii) ω(hλ ) = eλ for all partitions λ.
(iii) ω(sλ ) = sλ′ for all partitions λ.
Moreover, there is a unique linear transformation ω ∶ Λ → Λ which
satisfies (i)–(iii) above, and ω is an involution.
Proof. Since the elementary symmetric functions form a basis for Λ,

there is a unique linear transformation ωe ∶ Λ → Λ which satisfies (i).
Similarly, there is a unique linear transformation ωh ∶ Λ → Λ which
satisfies (ii), and there is a unique linear transformation ωs ∶ Λ → Λ
which satisfies (iii). Now apply ωh to (6.1) and use the linearity of
ωh and (6.6) to find that for any partition λ we have
ωh (sλ ) = ωh (det (hλi +j−i )1≤i,j≤k )

= det (eλi +j−i )1≤i,j≤k
= sλ′ .
Therefore ωh satisfies (iii), so ωh = ωs . We can use a similar argument

to show ωe = ωs , so ωe = ωs = ωh is the unique map which satisfies
(i)–(iii). The fact that this map is an involution follows from (iii),
since λ′′ = λ for all partitions λ. □
The fact that a function constructed to behave well with respect

to the elementary symmetric functions also behaves well with respect
to the Schur functions and the complete homogeneous symmetric
functions is a pleasant surprise. Even more remarkably, this function
also behaves well with respect to multiplication. To see this, we first
consider products of elementary symmetric functions and products of
complete homogeneous symmetric functions.
Definition 6.17. Suppose λ is a partition with ak parts of size k for

all k, and µ is a partition with bk parts of size k for all k. Then we
write λ ∪ µ to denote the partition with ak + bk parts of size k for all
k.
Products of elementary symmetric functions and products of com-

plete homogeneous symmetric functions behave in the same way: for
all partitions λ and µ we have eλ eµ = eλ∪µ and hλ hµ = hλ∪µ . This
similarity allows us to show that ω respects multiplication.
Proposition 6.18. For any symmetric functions f and g we have

ω(f g) = ω(f )ω(g).
Proof. Suppose f = ∑λ aλ eλ and g = ∑µ bµ eµ . Then we have f g =

∑λ,µ aλ bµ eλ∪µ , so ω(f g) = ∑λ,µ aλ bµ hλ∪µ . On the other hand, we
also have
ω(f )ω(g) = ∑ aλ bµ ω(eλ )ω(eµ )

λ,µ
= ∑ aλ bµ hλ hµ
λ,µ
= ∑ aλ bµ hλ∪µ ,
λ,µ
and the result follows. □
Our definition of ω describes how ω acts on the elementary sym-

metric functions, the complete homogeneous symmetric functions and
the Schur functions, but it would also be nice to know how it acts
on the power sum symmetric functions and the monomial symmetric
functions.
Starting with the power sum symmetric functions, we could com-
pute ω(pλ ) for various small partitions λ and look for a pattern. How-
ever, by Proposition 6.18 we know ω(pλ ) = ω(pλ1 )ω(pλ2 ) ⋅ ⋅ ⋅ ω(pλl(λ) ),
so it is enough to compute ω(pn ) for small n. When we do this, we
find ω(p1 ) = p1 , ω(p2 ) = −p2 , ω(p3 ) = p3 , ω(p4 ) = −p4 , and ω(p5 ) = p5 .
This suggests the following result.
(6.9) ω(pn ) = (−1)n−1 pn .

6.3. The Involution ω 181
Proof. If we apply ω to (3.4) and use linearity and Proposition 6.18

to simplify the result, then we find
n
ω(pn ) = ∑ (−1)j−1 jω(ej hn−j )
j=1
n
= ∑ (−1)j−1 jω(ej )ω(hn−j )
j=1
n
= ∑ (−1)j−1 jhj en−j
j=1
= (−1)n−1 pn .
Here we have used (3.5) in the last step. □
We can now use Propositions 6.18 and 6.19 to find ω(pλ ) for any
partition λ.
Proposition 6.20. For any partition λ we have
(6.10) ω(pλ ) = (−1)∣λ∣−l(λ) pλ .
Proof. By the definition of pλ and Propositions 6.18 and 6.19 we

have
l(λ)
ω(pλ ) = ω( ∏ pλj )
j=1
l(λ)
= ∏ ω (pλj )
j=1
l(λ)
= ∏ (−1)λj −1 pλj
j=1
= (−1)∣λ∣−l(λ) pλ ,
We now know how ω acts on eλ , hλ , sλ , and pλ , so it is natural

to ask how it acts on the monomial symmetric functions. To start to
understand ω(mλ ), we might like to express it in terms of one of our
bases. But which one?
Suppose we decide to write ω(mλ ) in terms of the complete homo-

geneous symmetric functions. Equation (2.3) and the data in Figure
2.5 tell us
e3 = m111 ,
e21 = m21 + 3m111 ,
and
e111 = m3 + 3m21 + 6m111 .
If we solve this system for m3 , m21 , and m111 , then we find
m3 = 3e3 − 3e21 + e111 ,
m21 = −3e3 + e21 ,
and
m111 = e3 .
Now we can apply ω and use the fact that ω(eµ ) = hµ to find ω(mλ )
for each λ ⊢ 3. In short, writing ω(mλ ) in terms of the complete
homogeneous symmetric functions is equivalent to inverting matrices
such as those in Figure 2.5. Similarly, writing ω(mλ ) in terms of the
elementary symmetric functions is equivalent to inverting matrices
such as those in Figure 2.8, writing ω(mλ ) in terms of the power sum
symmetric functions is equivalent to inverting matrices such as those
in Figure 2.9, and writing ω(mλ ) in terms of the Schur functions is
equivalent to inverting matrices such as those in Figure 4.10.
Inverting all of these matrices can be interesting, but we will
look at the one case we cannot quite do this way, in which we try to
write ω(mλ ) in terms of the monomial symmetric functions. We have
computed ω(mλ ) for some small partitions λ in Table 6.1.
Table 6.1. Expansions of ω(mλ ) in terms of monomial sym-

metric functions
λ ω(mλ )
(2, 1) −2m3 − m21
(22 ) m4 + m22
(3, 1) 2m4 + m31
(3, 2) −2m5 − m32
(22 , 1) 3m5 + m41 + 2m32 + m221
(32 ) m6 + m33
6.4. Problems 183
The data in Table 6.1 can lead to some interesting conjectures,

but we have no general results about these expansions to share. (But
see Problem 6.23.) Indeed, this is a symptom of a larger phenomenon:
the symmetric functions ω(mλ ) have not received as much attention
as the other symmetric functions we have seen so far. In recognition of
this, they are sometimes known as the forgotten symmetric functions.
Definition 6.21. For any partition λ, we write fλ to denote the

symmetric function ω(mλ ), and we call fλ the forgotten symmetric
function.
Although we cannot say much about the forgotten symmetric

functions, we do have the following result.
Proposition 6.22. For all k ≥ 0, the set {fλ ∣ λ ⊢ k} is a basis for

Λk .
Proof. We leave it as Problem 6.24 to show that if g ∈ Λk , then

ω(g) ∈ Λk , and we note this implies {fλ ∣ λ ⊢ k} ⊂ Λk .
To see {fλ ∣ λ ⊢ k} spans Λk , suppose f ∈ Λk . Since the monomial
symmetric functions form a basis for Λk , there exist constants aλ for
which ω(f ) = ∑λ⊢k aλ mλ . If we now apply ω and use the fact that ω
is linear and an involution, we find f = ∑λ⊢k aλ fλ .
To see {fλ ∣ λ ⊢ k} is linearly independent, suppose we have
constants aλ with ∑λ⊢k aλ fλ = 0. If we apply ω and use the fact
that ω is linear and an involution, we find ∑λ⊢k aλ mλ = 0. Since
the monomial symmetric functions are linearly independent, we must
have aλ = 0 for all λ. □
6.4. Problems
6.1. Among all families of lattice paths with associated permutation
π = 5172436, what is the smallest total degree the associated
h-weight can have?
π, what is the smallest total degree the associated h-weight can
have, in terms of the length of π and the entries of π?

π = 5172436, what is the smallest total degree the associated
e-weight can have?
π, what is the smallest total degree the associated e-weight can
have, in terms of the length of π and the entries of π?
6.5. Find the permutation π, the partition λ, and the h-weights
corresponding to the family of lattice paths in Figure 6.18.
Figure 6.18. The family of lattice paths in Problem 6.5
6.6. Find the permutation π, the partition λ, and the e-weights cor-
responding to the family of lattices paths in Figure 6.19.
6.7. Give an example of a family of lattice paths with associated
permutation π = 35124, associated partition λ = (42 , 3, 1), and
h-weight x31 x3 x44 x26 x27 .
6.8. Give an example of a family of lattice paths with associated
permutation π = 35124, associated partition λ = (42 , 3, 1), and
e-weight x41 x23 x34 x5 x27 .
6.4. Problems 185
6.9. Find the sequence of lattice paths tswp(β) for the sequence of
lattice paths β in Figure 6.20.

6.10. Find the sequence of lattice paths tswp(β) for the sequence of
lattice paths β in Figure 6.21.
6.11. If η is the bijection in Lemma 6.7 and β is the family of lattice
paths in Figure 6.22, then find η(β).

6.4. Problems 187
6.12. Complete the proof of Lemma 6.14 by showing ϵ(β) is row-

nondecreasing if and only if no two paths in β intersect.
6.13. If ϵ is the bijection in Lemma 6.14 and β is the family of lattice
paths in Figure 6.23, then find ϵ(β).
6.14. Sketch a sequence of nonintersecting lattice paths which corre-

sponds to the semistandard tableau T in Figure 6.24 under the
bijection η in Lemma 6.7. Use as few lattice paths as possible.
5
4 4
2 3 4 5
1 1 2 4
Figure 6.24. The semistandard tableau in Problems 6.14 and 6.15
6.15. Sketch a sequence of nonintersecting lattice paths which corre-

sponds to the semistandard tableau T in Figure 6.24 under the
bijection ϵ in Lemma 6.14. Use as few lattice paths as possible.
6.16. Let k be a positive integer. For any partition λ with k ≥ l(λ)

and any β ∈ Hλ,k in which two paths intersect, define a se-
quence stpw(β) of lattice paths by modifying the definition of
tswp(β) as follows. (Here stpw is a permutation/variation of
tswp.) First find the largest x-coordinate at which two paths
in β intersect. Then, among all of the intersection points with
this x-coordinate, choose the one with the largest y-coordinate
and swap the tails of the two paths which intersect at this point
to obtain stpw(β).
(a) Find a sequence β for which tswp(β) ≠ stpw(β).
(b) Is stpw a sign-reversing involution on the set of sequences
β ∈ Hλ,k in which two paths intersect?
6.17. Let k be a positive integer. For any partition λ with k ≥ l(λ)
and any β = (α1 , . . . , αk ) ∈ Hλ,k in which two paths intersect,
define a sequence sptw(β) of lattice paths as follows. (Here
sptw is a permutation/variation of stpw.) First find the largest
j for which αj intersects another lattice path among α1 , . . . , αk .
Then find the largest l < j for which αl intersects αj . Among
all of the points where αl and αj intersect, choose the last one
and swap the tails of αl and αj to obtain sptw(β).
(a) Find a sequence β for which tswp(β) ≠ sptw(β).
(b) Is sptw a sign-reversing involution on the set of sequences
β ∈ Hλ,k in which two paths intersect?
6.18. Show that if µ ⊆ λ are partitions, then
sλ/µ = det (hλj −µk +k−j )
and
sλ/µ = det (eλ′j −µ′k +k−j ) .
6.19. Find ω(m22 ) in terms of the monomial symmetric functions,

the elementary symmetric functions, the complete homogeneous
symmetric functions, the power sum symmetric functions, and
the Schur functions.
6.20. Show that if µ ⊆ λ are partitions, then ω(sλ/µ ) = sλ′ /µ′ .
6.21. Show that for all n ≥ 1, we have fn = (−1)n−1 pn .
6.22. Show that for all n ≥ 0, we have f1n = hn .
6.5. Notes 189
6.23. Suppose λ ⊢ n. Show that if

fλ = ∑ aµ mµ ,
µ⊢n
then all of the nonzero coefficients aµ have the same sign.

6.24. Show that if g ∈ Λk , then ω(g) ∈ Λk .
6.25. When we write s421 as a linear combination of the forgotten
symmetric functions, what is the coefficient of f2221 ?
6.26. When we write s53 as a linear combination of the forgotten
symmetric functions, for which partitions λ does fλ appear with
nonzero coefficient?
6.27. For any g ∈ Λk , define gn ∈ Λnk by
gn (x1 , x2 , . . .) = g(xn1 , xn2 , . . .).
Show that for all g ∈ Λm , we have
ω(gn ) = (−1)m(n−1) (ω(g))n .
6.28. Let An be the n×n matrix whose jlth entry is the binomial coef-
ficient (j+l−2
j−1
). Use a lattice path argument to evaluate det(An ).
6.5. Notes
The proofs of the Jacobi–Trudi identities we give here also appear
elsewhere, including in Bressoud’s book [Bre99] on alternating sign
matrices and Sagan’s book [Sag01] on the representation theory of
the symmetric group. They grow out of work of Lindström [Lin73]
and Gessel and Viennot [GV85] relating determinants to lattice path
counting problems.
We observed at the end of this chapter that computing ω(mλ )
is related to computing inverses of matrices of Kostka numbers. To
this end, Eğecioğlu and Remmel [ER90] have given a combinatorial
interpretation of the entries of these matrices.
Chapter 7
The Hall Inner Product
We have several bases for the space Λk of homogeneous symmetric

functions of degree k, but we would like to have even more linear
algebraic structure on this space. For example, in the familiar vector
space Rn we can use the dot product to define lengths of vectors and
angles between vectors, giving Rn a substantial geometric structure.
In this chapter we introduce an inner product on Λk , which gives us
a geometric structure on this space.
7.1. Inner Products on Λk

Since we have several bases for Λk , we can construct several inner
products on this space. In addition, for any two of our bases, we can
construct a function ⟨ , ⟩ ∶ Λk × Λk → Q with respect to which our
bases are dual, but which may or may not be an inner product. As
we will see, however, some of these functions coincide. To define these
functions, we will use the function δ on ordered pairs of partitions,
which is defined by
⎧
⎪
⎪1 if λ = µ,
δλ,µ = ⎨
⎪0 if λ ≠ µ.
⎪
⎩
Example 7.1. Let ⟨ , ⟩e be the inner product on Λ3 with ⟨eλ , eµ ⟩e =

δλ,µ , let ⟨ , ⟩p be the inner product on Λ3 with ⟨pλ , pµ ⟩p = δλ,µ , and
let ⟨ , ⟩s be the inner product on Λ3 with ⟨sλ , sµ ⟩s = δλ,µ . For each
191
192 7. The Hall Inner Product
λ, µ ⊢ 3, compute ⟨hλ , mµ ⟩e , ⟨hλ , mµ ⟩p , and ⟨hλ , mµ ⟩s . Use your

results to compare these inner products with the function ⟨ , ⟩h,m with
⟨hλ , mµ ⟩h,m = δλ,µ .
Solution. We can use the data in Figures 2.5 and 2.8 and some algebra
to find h3 = e3 −2e21 +e111 and m3 = 3e3 −3e21 +e111 . Using properties
of inner products, we find
⟨h3 , m3 ⟩e = ⟨e3 − 2e21 + e111 , 3e3 − 3e21 + e111 ⟩e
= ⟨e3 , 3e3 − 3e21 + e111 ⟩e + ⟨−2e21 , 3e3 − 3e21 + e111 ⟩e
+ ⟨e111 , 3e3 − 3e21 + e111 ⟩e
= 3⟨e3 , e3 ⟩e − 3⟨e3 , e21 ⟩e + ⟨e3 , e111 ⟩e − 6⟨e21 , e3 ⟩e
+ 6⟨e21 , e21 ⟩e − 2⟨e21 , e111 ⟩e + 3⟨e111 , e3 ⟩e
− 3⟨e111 , e21 ⟩e + ⟨e111 , e111 ⟩e
= 10.
We can perform similar computations for the other eight inner prod-
ucts or we can use technology to speed our work, but either way we
eventually reach the data in Table 7.1. Similarly, we can use the data
in Figure 2.9 to get the data in Table 7.2. And we can use the data
in Figure 4.10 to get the data in Table 7.3.
It is apparent from our data that ⟨ , ⟩s = ⟨ , ⟩h,m , but there is also
a hidden pattern. To see it, suppose we rescale ⟨ , ⟩p to eliminate the
Table 7.1. The values of ⟨hλ , mµ ⟩e for λ, µ ⊢ 3
(3) (2, 1) (13 )

(3) ⎡⎢ 10 −5 3 ⎤⎥
⎢ ⎥
(2, 1) ⎢⎢ 4 −1 0 ⎥⎥
(13 ) ⎢⎣ 1 0 0 ⎥⎦
Table 7.2. The values of ⟨hλ , mµ ⟩p for λ, µ ⊢ 3
(3) (2, 1) (13 )

(3) ⎡⎢ 1/3 1/6 −1/9 ⎤⎥
⎢ ⎥
(2, 1) ⎢⎢ 0 1/2 −1/6 ⎥⎥
(13 ) ⎢⎣ 0 0 1/6 ⎥⎦
7.1. Inner Products on Λk 193
Table 7.3. The values of ⟨hλ , mµ ⟩s for λ, µ ⊢ 3
(3) (2, 1) (13 )

(3) ⎡⎢ 1 0 0 ⎤⎥
⎢ ⎥
(2, 1) ⎢⎢ 0 1 0 ⎥⎥
(1 ) ⎢⎣ 0
3
0 1 ⎥⎦
Table 7.4. The modified values of ⟨hλ , mµ ⟩p for λ, µ ⊢ 3
(3) (2, 1) (13 )

(3) ⎡⎢ 1 0 0 ⎤⎥
⎢ ⎥
(2, 1) ⎢⎢ 0 1 0 ⎥⎥
(13 ) ⎢⎣ 0 0 1 ⎥⎦
fractions in Table 7.2. In particular, suppose we set ⟨pλ , pµ ⟩p = 0 if

λ ≠ µ and ⟨p3 , p3 ⟩p = 3, ⟨p21 , p21 ⟩p = 2, and ⟨p13 , p13 ⟩p = 6. (Note that
these values are the denominators of the diagonal entries in Table
7.2.) For this modified inner product we get the data in Table 7.4,
which means it is also equal to ⟨ , ⟩s = ⟨ , ⟩h,m . □
There is a tool we can use to recognize when choosing different

pairs of bases to be dual results in the same inner product, which
involves a method of combining a pair of bases of Λk into a symmetric
function in two sets of variables. To describe this method, we first
introduce a new set of variables y1 , y2 , . . . . To distinguish symmetric
functions in these variables from symmetric functions in x1 , x2 , . . ., we
write f (X) (resp., f (Y )) to denote the symmetric function f in the
variables x1 , x2 , . . . (resp., y1 , y2 , . . .). We then consider a new space
of formal power series which are symmetric in both of these sets of
variables.
Definition 7.2. For all k ≥ 0, we write Λk (X, Y ) to denote the set
of formal power series in the variables x1 , x2 , . . . and y1 , y2 , . . ., which
are symmetric in x1 , x2 , . . ., symmetric in y1 , y2 , . . ., homogeneous of
degree k in x1 , x2 , . . ., and homogeneous of degree k in y1 , y2 , . . . .
When we used logarithms of formal power series in the generating

∞
function for { pnn }n=1 in Proposition 2.33 and its proof, we sidestepped
some technical issues. Similar technical issues arise when we do linear
algebra with formal power series in two infinite sets of variables. Here,
too, we will leave these issues aside, assuming the linear algebra works
out in analogy with the finite case.
As we might expect, we can combine a set of linearly independent
symmetric functions in x1 , x2 , . . . with a set of linearly independent
symmetric functions in y1 , y2 , . . . to obtain a linearly independent set
in Λk (X, Y ).
Proposition 7.3. Suppose {uj ∣ 1 ≤ j ≤ l} and {vm ∣ 1 ≤ m ≤ n} are
linearly independent sets in Λk . Then the set {uj (X)vm (Y ) ∣ 1 ≤ j ≤
l, 1 ≤ m ≤ n} is linearly independent in Λk (X, Y ).
Proof. Suppose for 1 ≤ j ≤ l and 1 ≤ m ≤ n we have constants Ajm

with
l n
∑ ∑ Ajm uj (X)vm (Y ) = 0.
j=1 m=1
The set of symmetric functions in y1 , y2 , . . . is a vector space over the
set of rational functions in x1 , x2 , . . . , in which {vm (Y ) ∣ 1 ≤ m ≤
n} is linearly independent, so for each m with 1 ≤ m ≤ n, we have
∑j=1 Ajm uj (X) = 0. Since {uj (X) ∣ 1 ≤ j ≤ l} is linearly independent,
l
we must now have Ajm = 0 for all j, m with 1 ≤ j ≤ l and 1 ≤ m ≤ n,

Suppose {uλ ∣ λ ⊢ k} and {vµ ∣ µ ⊢ k} are bases for Λk . Then we

combine these bases to obtain an element of Λk (X, Y ) by setting
Fk (u, v) = ∑ uλ (X)vλ (Y ).
λ⊢k
As we show next, Fk (u, v) is closely related to the inner product on

Λk for which {uλ ∣ λ ⊢ k} and {vµ ∣ µ ⊢ k} are dual bases.
Proposition 7.4. Suppose {uλ ∣ λ ⊢ k}, {vλ ∣ λ ⊢ k}, {u′λ ∣ λ ⊢ k},
and {vλ′ ∣ λ ⊢ k} are bases for Λk , that ⟨ , ⟩ ∶ Λk ×Λk → Q is the function
on Λk × Λk for which ⟨uλ , vµ ⟩ = δλµ , and that ⟪, ⟫ ∶ Λk × Λk → Q is the
function on Λk × Λk for which ⟪u′λ , vµ′ ⟫ = δλµ . Furthermore, let A and
B be the p(k) × p(k) matrices whose rows and columns are indexed by
partitions of k and whose entries are defined by
(7.1) uλ = ∑ Aλµ u′µ
µ⊢k
7.1. Inner Products on Λk 195
and
(7.2) vλ = ∑ Bλµ vµ′ .
µ⊢k
Then the following are equivalent.

(i) At = B −1 .
(ii) ⟨ , ⟩ = ⟪, ⟫.
(iii) Fk (u, v) = Fk (u′ , v ′ ).
Proof. (i) ⇔ (ii) Observe that for all λ, µ ⊢ k, we have
⟪uλ , vµ ⟫ = ⟨⟨ ∑ Aλα u′α , ∑ Bµβ vβ′ ⟩⟩ (by (7.1) and (7.2))

α⊢k β⊢k
= ∑ ∑ Aλα Bµβ ⟪u′α , vβ′ ⟫ (by Definition A.15(i)–(iii))

α⊢k β⊢k
= ∑ Aλα Bµα .
α⊢k
Therefore, ⟪uλ , vµ ⟫ = δλ,µ for all λ, µ ⊢ k if and only if AB t = I,

which occurs if and only if BAt = I, which occurs if and only if
At = B −1 . On the other hand, ⟨ , ⟩ is the unique function on Λk × Λk
with ⟨uλ , vµ ⟩ = δλ,µ for all λ, µ ⊢ k. Therefore ⟨ , ⟩ = ⟪, ⟫ occurs if
and only if ⟪uλ , vµ ⟫ = δλ,µ for all λ, µ ⊢ k, which occurs if and only if
At = B −1 .
(i) ⇔ (iii) Observe that we have
Fk (u, v) = ∑ uλ (X)vλ (Y )
λ⊢k
= ∑ ( ∑ Aλα u′α (X))( ∑ Bµβ vβ′ (Y ))

λ⊢k α⊢k β⊢k
= ∑ ∑ ( ∑ Aλα Bλβ )u′α (X)vβ′ (Y ).

α⊢k β⊢k λ⊢k
−1
Now if A = B , then we can simplify the innermost sum to get to
t
Fk (u′ , v ′ ), which means Fk (u, v) = Fk (u′ , v ′ ).

Conversely, if Fk (u, v) = Fk (u′ , v ′ ), then by the same computation
we must have
(7.3) Fk (u′ , v ′ ) = ∑ ∑ ( ∑ Aλα Bλβ ) u′α (X)vβ′ (Y ).

α⊢k β⊢k λ⊢k
But Proposition 7.3 tells us the terms u′α (X)vβ′ (Y ) are linearly inde-
pendent, so if we compare the definition of Fk (u′ , v ′ ) with the right
side of (7.3), we see we must have
∑ Aλα Bλβ = δα,β .
λ⊢k
But this says exactly that At = B −1 . □
7.2. The Hall Inner Product and Cauchy’s

Formula
In our solution to Example 7.1 we saw three bilinear forms agree
on Λ3 : the inner product with respect to which the Schur functions
form an orthonormal basis, the bilinear form with respect to which
the complete homogeneous symmetric functions and the monomial
symmetric functions are dual, and a modified version of the inner
product with respect to which the power sum symmetric functions
form an orthonormal basis. The simplest of these descriptions seems
to be the first one, so we make the following definition.
Definition 7.5. The Hall inner product ⟨ , ⟩ ∶ Λ × Λ → Q is the unique
inner product on Λ such that
⟨sλ , sµ ⟩ = δλ,µ
for all partitions λ and µ.
Long before we learned about symmetric functions, we used the

usual dot product to capture the geometric structure of Rn . Now we
use the Hall inner product to do the same for Λk . In particular, for any
g ∈ Λk we can think of ⟨g, g⟩ as the square of the length of g, and for
any g1 , g2 ∈ Λk we can think of ⟨g1 , g2 ⟩ as the product of the lengths of
g1 and g2 and the cosine of the angle between them. When we study
geometry, we are particularly interested in transformations such as
rotations and reflections, which preserve geometric structures such as
lengths and angles. In this context that means we are interested in
transformations of the space of symmetric functions which preserve
the Hall inner product. So far we have only seen one transformation
of the space of symmetric functions, namely the involution ω, so it is
natural to ask whether ω preserves ⟨ , ⟩.
7.2. The Hall Inner Product and Cauchy’s Formula 197
Proposition 7.6. For all symmetric functions g1 and g2 we have
(7.4) ⟨ω(g1 ), ω(g2 )⟩ = ⟨g1 , g2 ⟩.
Proof. Since the Schur functions are a basis for Λ, there are scalars
aλ and bµ such that g1 = ∑λ aλ sλ and g2 = ∑µ bµ sµ . Using the linear-
ity of ⟨ , ⟩, we find
⟨g1 , g2 ⟩ = ⟨ ∑ aλ sλ , ∑ bµ sµ ⟩
λ µ
= ∑ ∑ aλ bµ ⟨sλ , sµ ⟩
λ µ
= ∑ aλ bλ .
λ
Similarly, since ω is linear, we have
⟨ω(g1 ), ω(g2 )⟩ = ⟨ω(∑ aλ sλ ), ω(∑ bµ sµ )⟩

λ µ
= ∑ ∑ aλ bµ ⟨sλ′ , sµ′ ⟩
λ µ
= ∑ aλ bλ ,
λ
Our solution to Example 7.1 suggests the complete homogeneous

symmetric functions and the monomial symmetric functions are dual
with respect to ⟨ , ⟩. And if that is true, then Proposition 7.6 tells
us the elementary symmetric functions and the forgotten symmetric
functions are also dual with respect to ⟨ , ⟩. Finally, Proposition 7.4
tells us a first step toward proving these results is to examine the
formal sum that characterizes this inner product. This sum turns out
to be the generating function for a natural collection of combinatorial
objects, which we introduce next.
Definition 7.7. For every n ≥ 0, a generalized permutation of length

a a2 ⋅ ⋅ ⋅ an
n is a 2 × n array [ 1 ] of positive integers in which the
b1 b2 ⋅ ⋅ ⋅ bn
following hold.
(1) a1 ≤ a2 ≤ ⋅ ⋅ ⋅ ≤ an .
(2) If aj = aj+1 , then bj ≤ bj+1 .
a a2 ⋅ ⋅ ⋅ an
For any generalized permutation π = [ 1 ] of length n
b1 b2 ⋅ ⋅ ⋅ bn
we write topwt(π) to denote the top weight of π, which is given by
n
topwt(π) = ∏ xaj ,
j=1
and we write bottomwt(π) to denote the bottom weight of π, which

is given by
n
bottomwt(π) = ∏ ybj .
j=1
We also write Sn to denote the set of generalized permutations of

length n, and we write S to denote the set of all generalized permu-
tations.
If we view a generalized permutation of length n as a sequence of

n columns, then we can find the generating function for all generalized
permutations with respect to their top and bottom weights.
Proposition 7.8. We have

∞ ∞
1
∑ topwt(π) bottomwt(π) = ∏ ∏ .
π∈S j=1 k=1 1 − xj yk
Proof. We can construct each generalized permutation uniquely as

follows.
1
● Choose a number (possibly zero) of columns of the form [ ].
1
The generating function for these columns is 1+x1 y1 +x21 y12 +
⋅ ⋅ ⋅ = 1−x11 y1 .
2
1
⋅ ⋅ ⋅ = 1−x12 y1 .
7.2. The Hall Inner Product and Cauchy’s Formula 199
2
2
⋅ ⋅ ⋅ = 1−x12 y2 .
Continue in this way, choosing a number (possibly zero) of columns of
j
the form [ ] for all positive integers j and k. The generating function
k
for these columns is 1−x1j yk . Once we have chosen all of our columns,
there is a unique way to assemble them into a generalized permuta-
j
tion. Choosing l columns of the form [ ] corresponds to choosing the
k
term (xj yk )l from the factor 1−x1j yk in the product ∏∞ ∞ 1
j=1 ∏k=1 1−xj yk ,
Remarkably, our next result connects the sum we wish to study

with the generating function for generalized permutations.
Theorem 7.9 (Cauchy’s Formula). We have

∞ ∞
1
(7.5) ∑ sλ (X)sλ (Y ) = ∏ ∏ ,
λ j=1 k=1 1 − xj yk
where the sum on the left is over all partitions.
In the next chapter we will give a spectacular combinatorial proof

of Theorem 7.9, but for now we just consider some of its consequences.
One of these consequences is that the monomial symmetric functions
and the complete homogeneous symmetric functions really are dual
with respect to the Hall inner product, as we suspected.
Proposition 7.10. We have

∞ ∞
1
(7.6) ∑ mλ (X)hλ (Y ) = ∏ ∏ ,
λ j=1 k=1 1 − xj yk
Proof. By Proposition 7.8, the right side of (7.6) is the generat-

ing function for generalized permutations with respect to topwt and
bottomwt. Now observe that we can encode each generalized permu-
tation π as a P × P matrix in which the entry in row j and column k is
the number of dominoes with top entry k and bottom entry j. More-
over, if A is the matrix associated with π, then topwt(π) = ∏∞
λj
j=1 xj
and bottomwt(π) = ∏∞
µj
j=1 yj , where µj is the sum of the entries in
row j of A and λj is the sum of the entries in column j of A. There-
fore, by Problem 2.17(a), the coefficient of ∏∞
λj µj
j=1 xj yj on the right
side of (7.6) is Mλ,µ (h, m), which is defined in (2.10).
Meanwhile, on the left side of (7.6) we can use the definition of
Mλ,µ (h, m) in (2.10) to find
∑ mλ (X)hλ (Y ) = ∑ ∑ Mλ,µ (h, m)mλ (X)mµ (Y ).
λ λ µ
Therefore, Mλ,µ (h, m) is also the coefficient of ∏∞ j j λ µ

j=1 xj yj on the left
side of (7.6), and the result follows. □
Corollary 7.11. For all partitions λ and µ we have

(7.7) ⟨mλ , hµ ⟩ = δλ,µ .
Proof. Cauchy’s formula and Proposition 7.10 tell us

∑ mλ (X)hλ (Y ) = ∑ sλ (X)sλ (Y ).
λ λ
Now Proposition 7.4 tells us that if ⟨ , ⟩ is the Hall inner product and
⟪, ⟫ is the bilinear form with ⟪mλ , hµ ⟫ = δλ,µ , then ⟨ , ⟩ = ⟪, ⟫. In
particular, ⟨mλ , hµ ⟩ = ⟪mλ , hµ ⟫ = δλ,µ . □
Since the Hall inner product is invariant under the involution

ω, we immediately find the elementary symmetric functions and the
forgotten symmetric functions are also dual bases.
Corollary 7.12. For all partitions λ and µ we have
⟨fλ , eµ ⟩ = δλ,µ .
Proof. By Propositions 6.16 and 7.6, Definition 6.21, and Corollary

7.11 we have
⟨fλ , eµ ⟩ = ⟨ω(mλ ), ω(hµ )⟩
= ⟨mλ , hµ ⟩
= δλ,µ . □
7.3. Inner Products of Power Sums 201
7.3. The Hall Inner Product on the Power Sum

Symmetric Functions
In our solution to Example 7.1 we saw p3 , p21 , and p111 are mutually
orthogonal with respect to the Hall inner product, but they are not
unit vectors: ⟨p3 , p3 ⟩ = 3, ⟨p21 , p21 ⟩ = 2, and ⟨p111 , p111 ⟩ = 6. Similar
computations tell us p2 and p11 are orthogonal with respect the Hall
inner product, but ⟨p2 , p2 ⟩ = 2 and ⟨p11 , p11 ⟩ = 2. Along the same
lines, p4 , p31 , p22 , p211 , and p1111 are also mutually orthogonal with
respect to the Hall inner product, but ⟨p4 , p4 ⟩ = 4, ⟨p31 , p31 ⟩ = 3,
⟨p22 , p22 ⟩ = 8, ⟨p211 , p211 ⟩ = 4, and ⟨p1111 , p1111 ⟩ = 24. Remarkably,
we have seen the numbers ⟨pλ , pλ ⟩ before, in our solution to Problem
C.4.
Definition 7.13. For any partition λ, let zλ be defined by
λ1
zλ = ∏ j nj nj !,
j=1
where nj is the multiplicity of j in λ.
As we saw in Problem C.4, for any partition λ ⊢ n, the quantity

n!/zλ is the number of permutations in Sn with cycle type λ. With
this quantity in hand, we can now describe how the Hall inner product
acts on the power sum symmetric functions.
Proposition 7.14. For all partitions λ and µ we have
⟨pλ , pµ ⟩ = zλ δλ,µ .
Cauchy’s formula and Proposition 7.4 together tell us we can

prove Proposition 7.14 by proving the following result.
Proposition 7.15. For any partition λ, let zλ be as in Definition
7.13. Then we have
∞ ∞
1 1
(7.8) ∑ pλ (X)pλ (Y ) = ∏ ∏ ,
z
λ λ j=1 k=1 1 − xj yk
At first it looks like we have replaced one problem we do not know

how to solve with another problem we also do not know how to solve.
But if we stare at (7.8) long enough, a strategy begins to emerge.

On the right side of this equation we have an infinite product that
looks suspiciously similar to the generating function H(t) in (2.11).
So perhaps we can start with something like that generating function,
but in the set of all products of x’s and y’s, and convert the result
to power sum symmetric functions. In order to do that, however, we
will need to know how to write hn as a linear combination of power
sum symmetric functions. We look at some small examples, to see if
we can spot a pattern.
Example 7.16. For each n ≤ 5, write hn as a linear combination of
power sum symmetric functions.
Solution. Using the data in Figure 2.9 and the fact that hn =
∑λ⊢n mλ , or using technology, we find h0 = p0 , h1 = p1 ,
1 1
h2 = p2 + p11 ,
2 2
1 1 1
h3 = p3 + p21 + p111 ,
3 2 6
1 1 1 1 1
h4 = p4 + p31 + p22 + p211 + p1111 ,
4 3 8 4 24
and
1 1 1 1 1 1 1
h5 = p5 + p41 + p32 + p311 + p221 + p2111 + p11111 . □
5 4 6 6 8 12 120
When we compare the denominators in our solution to Exam-
ple 7.16 with our values for zλ (which we are expecting to appear
anyway), we eventually reach the following result.
1
(7.9) hn = ∑ pλ .
z
λ⊢n λ
The fact that n!/zλ is the number of permutations of cycle type

λ suggests we should prove (7.9) in the form
n!
(7.10) n! hn = ∑ pλ .
z
λ⊢n λ
The left side of (7.10) is the generating function for ordered pairs
(π, J), where π ∈ Sn and J is a multiset of n positive integers, and
the weight of a pair (π, J) is ∏j∈J xj . Meanwhile, in view of Problem

2.27, the right side of (7.10) is the generating function for the ordered
pairs (σ, T ), where σ ∈ Sn and T is a filling of the Ferrers diagram of
the cycle type of σ such that the entries in each row of T are constant.
Here the weight of a pair (σ, T ) is xT .
To prove (7.10), we will give a weight-preserving bijection be-
tween the ordered pairs (π, J) on the left side and the ordered pairs
(σ, T ) on the right. To do this, it will be helpful to first reinterpret
the ordered pairs (σ, T ). In particular, note that if we write σ in cycle
notation, then there is a unique ordering of the cycles for which longer
cycles appear before shorter cycles, and if two cycles have the same
length, then the cycle with the larger minimal entry appears first. In
fact, we can even obtain a unique ordering of the entries of σ, if we
insist that in each cycle the smallest entry in the cycle appears first.
Our ordering of the cycles of σ gives us a natural bijection between
the cycles in σ and the rows in T . With this in mind, we can dispense
with T altogether, if we label each cycle with the entry that appears
in the associated row of T . We record these labeled permutations in
tables, in which the labels appear above the cycles. For example, in
Figure 7.1 we have an ordered pair (σ, T ) and in Figure 7.2 we have
the associated labeling of the cycles of σ.
(2) 3
(5) 3
(6 14) 2 2
(4 8 11) 2 2 2
(1 5 12 3) 7 7 7 7
(7 10 13 9) 3 3 3 3
Figure 7.1. An ordered pair (σ, T ) consisting of a permuta-

tion σ and a filling T
We can now use our new interpretation of the right side of (7.10)
to prove Proposition 7.17.
Proof of Proposition 7.17. We have seen it is sufficient to give

a weight-preserving bijection between ordered pairs (π, J), where
3 7 2 2 3 3
(7 10 13 9) (1 5 12 3) (4 8 11) (6 14) (5) (2)
Figure 7.2. The labeling of the cycles of the permutation

σ = (7 10 13 9)(1 5 12 3)(4 8 11)(6 14)(5)(2) corresponding
to the ordered pair in Figure 7.1
π ∈ Sn and J is a multiset of n positive integers, and the set of

permutations σ ∈ Sn whose cycles are labeled with positive integers.
To describe our bijection in one direction, suppose σ ∈ Sn is a
permutation whose cycles are labeled with positive integers. For each
j, 1 ≤ j ≤ n, let g(j) be the label of the cycle of σ which contains j,
and let J be the multiset {g(j) ∣ 1 ≤ j ≤ n}. Within each cycle, order
the entries so the smallest entry appears first. Then order the cycles
in weakly increasing order of their labels, and order the cycles with
a given label in decreasing order of their minimal elements. Observe
that this gives a unique ordering of [n]. Now remove the parentheses
and view the result as a permutation π, written in one-line notation.
Note that by construction, the weight of the ordered pair (π, J) is
equal to the weight of the labeled permutation σ.
To describe the inverse of this map, suppose (π, J) has π ∈ Sn and
J is a multiset of n positive integers. Write π in one-line notation, and
write the elements of J in weakly increasing order above the entries
of π. Now read π from left to right, and insert a divider between π(j)
and π(j + 1) whenever the entry above π(j) is not equal to the entry
above π(j + 1), or π(j + 1) is smaller than every entry to its left with
the same associated element of J written above. Insert parentheses
so that the entries of π between dividers form cycles, and let σ be the
permutation we have just written in cycle notation. By construction,
all of the entries of each cycle have the same number above them; let
this number be the label of the cycle.
Note that in our second construction the elements of each cycle
are ordered so that their smallest elements appear first, the cycles of
σ are ordered so that their labels are increasing, and the cycles with
the same label are written in order of decreasing minimal elements.
Therefore, our constructions are inverses of one another, and we have
the weight-preserving bijection we wanted. □
Example 7.18. Find the permutation in S9 whose cycles are la-

beled with positive integers which corresponds to the ordered pair
(π, J) = (859126347, {1, 1, 1, 3, 3, 5, 5, 5, 5}) under the bijection given
in the proof of Proposition 7.17.
Solution. When we write the elements of J in nondecreasing order

above the entries of π, we obtain the array in Figure 7.3. We must
now insert dividers where the labels in the top row change, which
gives us the array in Figure 7.4. Next we insert dividers immediately
to the left of those numbers in the bottom row which are smaller than
every number to their left in the same block (that is, with the same
label above), which gives us the array in Figure 7.5. Now we interpret
the dividers in the bottom row as breaks between cycles, which gives
us the permutation with labeled cycles in Figure 7.6. □
Now that we have an expression for hn as a linear combination of

the power sum symmetric functions, we are ready to prove Proposition
7.15.
Proof of Proposition 7.15. Let XY denote the set of products of

the form xj yk , so XY is the set of terms in the product
(x1 + x2 + ⋅ ⋅ ⋅)(y1 + y2 + ⋅ ⋅ ⋅),
and let hn (XY ) denote the complete homogeneous symmetric func-
tion in these variables. Then by (2.11) we have
∞ ∞ ∞
1
∑ hn (XY )t = ∏ ∏
n
,
n=0 j=1 k=1 1 − x j yk t
1 1 1 3 3 5 5 5 5
8 5 9 1 2 6 3 4 7
Figure 7.3. The first step in the solution to Example 7.18
1 1 1 3 3 5 5 5 5
8 5 9 1 2 6 3 4 7
Figure 7.4. The second step in the solution to Example 7.18

1 1 1 3 3 5 5 5 5
8 5 9 1 2 6 3 4 7
Figure 7.5. The third step in the solution to Example 7.18
3 5 1 5 1
(12) (347) (59) (6) (8)
Figure 7.6. The permutation with labeled cycles in the so-

lution to Example 7.18
and if we set t = 1, then we get

∞ ∞ ∞
1
∑ hn (XY ) = ∏ ∏ .
n=0 j=1 k=1 1 − xj yk
We now use (7.9) to eliminate hn (XY ), obtaining

∞ ∞ ∞
1 1
(7.11) ∑ ∑ pλ (XY ) = ∏ ∏ ,
n=0 λ⊢n zλ j=1 k=1 1 − xj yk
where pλ (XY ) is the power sum symmetric function in the variables

in XY . Observe that the sum on the left is just a sum over all parti-
tions. In addition, for any n ≥ 0, we have pn (XY ) = ∑∞ ∞
j=1 ∑k=1 xj yk =
n n
pn (X)pn (Y ). Therefore, pλ (XY ) = pλ (X)pλ (Y ). If we use this to

rewrite the left side of (7.11), then we obtain (7.8). □
As we mentioned earlier, we can use Proposition 7.15 to help us

prove Proposition 7.14.
Proof of Proposition 7.14. This is similar to the proof of Corol-

lary 7.11, using Cauchy’s formula and Propositions 7.4 and 7.15. □
7.4. Problems
7.1. Evaluate
∑ fλ (X)eλ (Y ),
λ
where the sum is over all partitions.
7.2. Find and prove a formula for ⟨en , hn ⟩.
7.3. Find and prove a formula for ⟨h1n , h1n ⟩.
7.5. Notes 207
7.4. Without using technology, compute ⟨h13 , p733 ⟩.

7.5. Find and prove a formula for ⟨en , pn ⟩.
7.6. Give a combinatorial interpretation of the inner product ⟨hλ , hµ ⟩
for any partitions λ and µ.
7.7. Show that for any partition µ we have
hµ = ∑ Kν,µ sν .
ν⊢∣µ∣
Here Kν,µ is the Kostka number of Definition 4.16.

7.8. Given n ≥ 0, show we have
∑ Kλ,µ Kλ,ν = Mµ,ν (h, m)
λ⊢n
for all partitions µ, ν of n.
7.9. For all n ≥ 0, find and prove a formula for en as a linear combi-
nation of the power sum symmetric functions.
7.10. Suppose k ≥ 0 and u1k , . . . , uk are the symmetric functions
we obtain by applying Gram–Schmidt orthogonalization to
m1k , . . . , mk , where m1k , . . . , mk and u1k , . . . , uk are in increas-
ing lexicographic order of their indexing partitions. For each
partition λ ⊢ k, find uλ in terms of the Schur functions. (Here
you may assume we are using coefficients in R instead of Q, if
necessary.)
7.5. Notes
Although it is now named for Hall, the Hall inner product was first
introduced by Redfield [Red27].
Chapter 8
The Robinson–
Schensted–Knuth
Correspondence
In Chapter 7 we saw that the formal power series ∑ sλ (X)sλ (Y ) is

λ
closely related to the Hall inner product on Λ. After examining the
terms of this series for which ∣λ∣ ≤ 3, we conjectured that
∞ ∞
1
(8.1) ∑ sλ (X)sλ (Y ) = ∏ ∏ ,
λ j=1 k=1 1 − xj yk
where the sum on the left is over all partitions. This result is known
as Cauchy’s formula, and in this chapter we give it a combinatorial
proof.
The sum on the left side of (8.1) has a natural combinatorial
interpretation: it is the generating function for the ordered pairs
(P, Q) of semistandard tableaux of the same shape, with respect to
the weight xQ y P . The product on the right side of (8.1) also has
a combinatorial interpretation. To describe it, recall that a general-
a a2 ⋅ ⋅ ⋅ an
ized permutation of length n is a 2 × n array [ 1 ] of
b1 b2 ⋅ ⋅ ⋅ bn
positive integers such that a1 ≤ a2 ≤ ⋅ ⋅ ⋅ ≤ an , and if aj = aj+1 , then
bj ≤ bj+1 . We showed in Proposition 7.8 that the right side of (8.1) is
the generating function for generalized permutations π with respect
209
210 8. The Robinson–Schensted–Knuth Correspondence
to the weight topwt(π) bottomwt(π), where topwt(π) = ∏nj=1 xaj and

bottomwt(π) = ∏nj=1 ybj .
In view of our combinatorial interpretations of its two sides, equa-
tion (8.1) is equivalent to
∑ x y = ∑ topwt(π) bottomwt(π),
Q P
(8.2)
(P,Q) π∈S
where the sum on the left is over all ordered pairs (P, Q) of semi-
standard tableaux with sh(P ) = sh(Q). This means we can prove
Cauchy’s formula by giving a bijection R from the set of generalized
permutations to the set of ordered pairs of semistandard tableaux of
the same shape such that if R(π) = (P, Q), then topwt(π) = xQ and
bottomwt(π) = y P .
8.1. RSK Insertion: Constructing P (π)

Roughly speaking, we will construct the semistandard tableau P (π)
in R(π) = (P (π), Q(π)) by starting with an empty tableau (that is,
one with no boxes) and inserting the entries in the bottom row of π, in
order from left to right, one entry at a time. Some of these insertions
will push entries that were in the bottom row to higher rows, but at
the end of each insertion we will still have a semistandard tableau.
Before we give a general description of this insertion procedure, we
illustrate it with a few examples.
To start, suppose we are in the middle of constructing P (π) for
some generalized permutation π, we have reached the semistandard
tableau T in Figure 8.1, and the next entry we need to insert is 5. If
we can, we always insert our new entry in a new box at the end of the
bottom (first) row of T . Doing this gives us the filling in Figure 8.2.
Since this filling is a semistandard tableau, it is the next semistandard
5
2 4
1 3 3 5
Figure 8.1. The semistandard tableau T we get in the con-

struction of a certain P (π)
8.1. RSK Insertion: Constructing P (π) 211
5
2 4
1 3 3 5 5
Figure 8.2. The tableau we get by inserting 5 into the

tableau T in Figure 8.1
tableau in our construction of P (π).

At this point some new notation will be useful. If T is a semi-
standard tableau and c is a positive integer, then we write rc (T ) to
denote the semistandard tableau we obtain by inserting c into T . We
have just found that if T is the semistandard tableau in Figure 8.1,
then r5 (T ) is the semistandard tableau in Figure 8.2.
In our first example the filling we obtained by putting c into a
new box at the end of the first row of T was a semistandard tableau,
but this is not always the case. In particular, this only works if c
is greater than or equal to the rightmost entry of the first row. To
describe what we do when c is too small, suppose we want to insert
4 into the semistandard tableau T in Figure 8.1. Instead of adding
a new box to the first row, we will replace one of the entries in that
row with the 4. If we replace either of the first two entries, then the
resulting filling will not be a semistandard tableau. On the other
hand, if we replace either of the last two entries, then the resulting
filling will be a semistandard tableau. We always place a new entry
as far to the right as possible, so it will always replace the leftmost
entry greater than it. In this case, we replace the 5 with our 4, and
we say the 4 “bumps” the 5.
We now have the semistandard tableau in Figure 8.3, but we also
have the 5 we bumped out of the first row. We try to insert it in
5
2 4
1 3 3 4
Figure 8.3. The tableau T from Figure 8.1 after the 4 bumps
the 5. The 5 is waiting to be inserted into the second row.
5
2 4 5
1 3 3 4
Figure 8.4. The semistandard tableau r4 (T )
5
5 5 4 4
2 4 3 2 3 2 3
1 2 3 5 1 2 3 5 1 2 3 5
Figure 8.5. The bumps in the construction of the semistan-

dard tableau r2 (T )
the second row, just as we inserted the 4 in the first row. Our 5 is
greater than or equal to every entry in the second row, so we place it
in a new box at the end of the row. This gives us r4 (T ), which is the
semistandard tableau in Figure 8.4.
In our last example we ended up modifying two rows of T to get
r4 (T ). As you might imagine, we sometimes get much longer chains
of bumps. For example, suppose we want to find r2 (T ), where T is
still the semistandard tableau in Figure 8.1. In Figure 8.5 we have
the intermediate steps in this process: the 2 bumps the leftmost 3
out of the first row, that 3 bumps the 4 out of the second row, that 4
bumps the 5 out of the third row, and we put that 5 in a new box at
the end of the (previously empty) fourth row to obtain the tableau in
Figure 8.6.
5
4
2 3
1 2 3 5
Figure 8.6. The semistandard tableau r2 (T )

With our examples in mind, we can now describe how to construct

rc (T ) in general. We insert a number a into a row of T in one of two
ways. If a is greater than or equal to every entry in the row, then we
add a box to the end of the row and put a in the new box. Otherwise,
we find the leftmost entry b in the row which is greater than a, and
we replace b with a. Note that after we make this replacement, a is
the rightmost entry in the row which is less than b; this observation
will be useful when we invert our construction of P (π).
Now to construct rc (T ), we first insert c into the bottom row
of T . If we do this by adding a new box to the row, then we are
done. Otherwise, c has replaced some number c1 ; we insert c1 into
the second row. If we do this by adding a new box to the row, then
we are done. Otherwise, c1 has replaced some number c2 ; we insert
c2 into the third row. We continue this process until we have added
a new box to the end of some row of T .
Now that we know how to insert a number into a semistandard
tableau, we can apply the process repeatedly to construct a semistan-
dard tableau from any finite sequence of positive integers: the semi-
standard tableau associated with the sequence a1 , . . . , an is the filling
we obtain by starting with the empty tableau, then inserting a1 , then
inserting a2 , etc. That is, the semistandard tableau associated with
a1 , . . . , an is ran (ran−1 (⋅ ⋅ ⋅ ra2 (ra1 (∅)) ⋅ ⋅ ⋅)). We write P (a1 , . . . , an )
to denote this tableau.
Example 8.1. Compute P (24312).
4
2 2 4 2 3
P (2) P (24) P (243)
4 4
2 2 3
1 3 1 2
P (2431) P (24312)
Figure 8.7. The intermediate tableaux in the construction of P (24312)

Solution. In Figure 8.7 we have the tableaux we obtain from each

insertion on the way to P (24312). □
The best way to build your intuition for the insertion process the
first time you see it is to compute a variety of examples. Try the
insertions in the next example yourself before checking your answers.
Example 8.2. Let α1 = 12132434, α2 = 21421343, α3 = 22134143,
α4 = 34231241, and α5 = 41312243. For each j, which of the semi-
standard tableaux in Figure 8.8 (if any) is P (αj )?
4
3 3
2 2 4 2 2 2 3 4
1 1 3 3 4 1 1 3 1 1 2 3 4
T1 T2 T3
4
4 3 3
3 4 2 2
1 1 2 2 3 1 1 4
T4 T5
Figure 8.8. The semistandard tableaux for Example 8.2
Solution. When we apply the insertion algorithm to each sequence,

we find P (α1 ) = T3 , P (α2 ) = T1 , P (α4 ) = T5 , and P (α5 ) = T4 . P (α3 )
is not among the given tableaux. □
We can now define P (π) for any generalized permutation π.

a a2 ⋅ ⋅ ⋅ an
Definition 8.3. If π = [ 1 ] is a generalized permuta-
b1 b2 ⋅ ⋅ ⋅ bn
tion, then we write P (π) to denote the filling P (b1 , b2 , . . . , bn ).
We will soon show P (π) is a semistandard tableau for any gener-

alized permutation π, but to do so it is useful to pay attention to two
9
7
6 6 9
2 5 8
1 3 7 8
Figure 8.9. The semistandard tableau T for Example 8.4
sequences that arise when we insert a number c into a semistandard

tableau T . One is the sequence of numbers we insert in each row
along the way. For each j, we write cj−1 to denote the number we
insert in row j in our construction of rc (T ). We call the sequence
c0 , . . . , cn the bumping sequence associated with rc (T ), and we note
that c = c0 .
The other sequence we pay attention to is the sequence of boxes
into which we insert the cj ’s, including the new box we create at the
end of the insertion process. We call this sequence the bumping path
9 9
7 7
6 6 9 6 6 9 8
2 5 8 7 2 5 7
1 3 5 8 1 3 5 8
9 9
7 9 7 9
6 6 8 6 6 8
2 5 7 2 5 7
1 3 5 8 1 3 5 8
Figure 8.10. The steps in the construction of r5 (T ) in Ex-

ample 8.4
associated with rc (T ). To get a feel for how these sequences behave,

it is helpful to keep track of them for a variety of insertions.
Example 8.4. Find the bumping sequence and the bumping path
associated with r5 (T ), where T is given in Figure 8.9.
Solution. In Figure 8.10 we see the intermediate steps in the con-

struction of r5 (T ). Here the bumping sequence is 5, 7, 8, 9 and the
squares in the bumping path are shaded. □
Example 8.5. Find the bumping sequences and the bumping paths
associated with r4 (T ) and r4 (r4 (T )), where T is the semistandard
tableau in Figure 8.11. Here the bumping sequence is 4, 5, 6, 9 and
the squares in the bumping path are shaded.
6 6 9
3 4 6 7 7
1 2 4 4 5 5 6
Figure 8.11. The semistandard tableau T for Example 8.5
Solution. In Figure 8.12 we have the intermediate steps in the con-

struction of r4 (T ).
6 6 9 6 6 9 6
3 4 6 7 7 5 3 4 5 7 7
1 2 4 4 4 5 6 1 2 4 4 4 5 6
9
6 6 6
3 4 5 7 7
1 2 4 4 4 5 6
Figure 8.12. The steps in the construction of r4 (T ) in Ex-

ample 8.5
9 9
6 6 6 6 6 6 7
3 4 5 7 7 5 3 4 5 5 7
1 2 4 4 4 4 6 1 2 4 4 4 4 6
9
6 6 6 7
3 4 5 5 7
1 2 4 4 4 4 6
Figure 8.13. The steps in the construction of r4 (r4 (T )) in

Example 8.5
In Figure 8.13 we have the intermediate steps in the construction

of r4 (r4 (T )) from r4 (T ). Here the bumping sequence is 4, 5, 7 and
the squares in the bumping path are shaded dark gray. □
As our examples suggest, and as we show next, bumping se-

quences and bumping paths have several useful properties. First of
all, bumping sequences are strictly increasing.
Lemma 8.6. Suppose T is a semistandard tableau, c is a positive
integer, and c = c0 , c1 , . . . , cn is the bumping sequence associated with
rc (T ). Then c0 < c1 < ⋅ ⋅ ⋅ < cn .
Proof. By definition cj bumps cj+1 for each j ≤ n − 1. By construc-

tion, this means cj+1 is the leftmost entry in its row which is greater
than cj , so cj < cj+1 . □
Secondly, bumping paths always move up and weakly to the left.

Lemma 8.7. Suppose T is a semistandard tableau, c is a positive in-
teger, and s1 , . . . , sn is the bumping path associated with rc (T ). Then
for each j with 1 ≤ j ≤ n − 1, either sj+1 is immediately above sj or
sj+1 is above and to the left of sj .
Proof. In T the numbers in sj and sj+1 are cj and cj+1 , respectively.

By construction, cj+1 is the leftmost entry in its row which is greater
than cj . But T is column-strict, so if there is an entry immediately

above cj , then it is greater than cj . Therefore either that entry is
cj+1 , or some entry to its left is cj+1 . □
Finally, when we insert a number and then a weakly larger num-

ber, the first bumping path is to the left of the second bumping path.
Lemma 8.8. Suppose T is a semistandard tableau, c ≤ d, s1 , . . . , sn is

the bumping path associated with rc (T ), and t1 , . . . , tm is the bumping
path associated with rd (rc (T )). Then n ≥ m and sj is strictly to the
left of tj for all j with 1 ≤ j ≤ m.
Proof. We argue by induction on the number of rows in T .

When T has no rows the result is immediate from our insertion
algorithm, so suppose T has at least one row and the result holds for
all semistandard tableaux with fewer rows than T has, and for all c
and d with c ≤ d. Since our given c and d have c ≤ d, we know d must
be inserted to the right of c in the first row of rc (T ). Therefore, d1
was to the right of c1 in T , so c1 ≤ d1 .
Now we are inserting c1 ≤ d1 into the semistandard tableau we
obtain from T by removing its first row, and the result follows by
induction. □
We can now show our insertion algorithm always produces semi-

standard tableaux.
Proposition 8.9. For any sequence a1 , . . . , an of positive integers,

P (a1 , . . . , an ) is a semistandard tableau. In particular, for any gen-
eralized permutation π, the filling P (π) is a semistandard tableau.
Proof. The result is clear when n = 0 or n = 1, so suppose n ≥

2; we argue by induction on n. In fact, we show that after every
row insertion that occurs in the construction of P (a1 , . . . , an ) from
P (a1 , . . . , an−1 ), the intermediate filling is a semistandard tableau.
To do this, suppose we are in the middle of the insertion process, and
we are about to insert cj−1 into the jth row. By induction, we are
inserting cj−1 into a row in a semistandard tableau.
b b
cj cj−1
cj−2 cj−2
before after
inserting inserting
cj−1 cj−1
Figure 8.14. Inserting cj−1 immediately above cj−2
When we insert cj−1 , we only change the jth row, and since we
put cj−1 in place of the leftmost entry greater than it, the entries in
the resulting row are weakly increasing from left to right.
Similarly, when we insert cj−1 , we only change the entries in the
column into which we insert it. In fact, we only change one entry of
that column. We consider two cases.
In the first case, suppose we insert cj−1 in the box immediately
above cj−2 . In this case the relevant entries are as in Figure 8.14. We
know by induction that cj < b, and by Lemma 8.6 we have cj−1 < cj < b.
On the other hand, Lemma 8.6 also tells us cj−2 < cj−1 , so our new
filling is a semistandard tableau.
In the second case, suppose we insert cj−1 in a box above and to
the left of cj−2 . In this case the relevant entries are as in Figure 8.15.
As in the previous case, we have cj−1 < b. We also know cj−1 was in
b b
cj cj−1
a · · · cj−2 a · · · cj−2
before inserting cj−1 after inserting cj−1
Figure 8.15. Inserting cj−1 above and to the left of cj−2

the box now occupied by cj−2 , which is to the right of a. Since that
filling was semistandard by induction, we must have cj−1 ≥ a. But
if cj−1 = a, then cj−2 would have bumped the leftmost a, not cj−1 .
Therefore, cj−1 > a and our new filling is a semistandard tableau. □
We now know how to construct P (π) for any generalized permu-

tation π, but to complete our proof of Cauchy’s formula, we also need
to construct Q(π). We will do this in the next section. To conclude
this section, we illustrate the usefulness of the insertion process we
used to construct P (π) by using it in an outline of a bijective proof of
our expansion formula for the dual stable Grothendieck functions in
terms of the Schur functions. In particular, we will prove Proposition
5.33, which says that for any partition λ we have
(8.3) gλ = ∑ fλµ sµ ,
µ⊆λ
where fλµ is the number of elegant tableaux of shape λ/µ.

To prove equation (8.3), we give a bijection which decomposes a
reverse plane partition T of shape λ into a pair (S, U ), where S is
a semistandard tableau of shape µ, U is an elegant tableau of shape
λ/µ, and xircont(T ) = xS . This bijection works one row at a time, from
the top of the Ferrers diagram of T to the bottom, so we start with
those T with few or no rows.
If T has no rows, then S must also have no rows, so S and U are
both empty.
If T has exactly one row, then the associated elegant tableau
U must also have one row, and therefore no entries. Since we need
xircont(T ) = xS , we must have S = T .
Now suppose T has exactly two rows, T1 and T2 , so λ = (∣T1 ∣, ∣T2 ∣).
To construct (S, U ), we start with the intermediate pair (Sint , Uint )
we obtain by applying our bijection to the reverse plane partition
consisting of just T2 . To construct S from Sint , first add an empty
box to the top of each of the leftmost ∣T1 ∣ columns of Sint , to obtain
a partial filling of shape λ. Now let T̂1 be the row we obtain from
T1 by removing those entries in T1 which are equal to the entries
immediately above them in T2 , and use the RSK insertion algorithm
to insert the entries of T̂1 into Sint , from smallest to largest. If ∣T̂1 ∣ <
1 2 4 5
1 1 3 5 5
Figure 8.16. A reverse plane partition with two rows
2 4 1 1
1 2 4 5 1 2 4 5 1 1 3 5 5
Figure 8.17. Building S and U for the reverse plane partition

in Figure 8.16
∣T1 ∣, then our shape has empty boxes; put a bold 1 in each empty box.
By the same argument we used to prove Proposition 8.9 the boxes
with nonbold entries form a semistandard tableau; this tableau is S,
and the bold entries form U .
By construction, U has shape λ/µ, where µ = sh S. By Lemma
8.8, no two boxes of U are in the same column, so U is a semistandard
skew tableau. To see that U is an elegant tableau, first note that the
bottom row of Sint is T2 . Furthermore, we can check that each entry
of T1 that we insert bumps the entry immediately above it in T2 (if
there is such an entry) out of the bottom row of Sint . Therefore, the
bottom row of our new filling is exactly T1 . In particular, this row
contains no bold 1’s, so U is an elegant tableau. In Figure 8.17 we
see the construction of S (in black) and U (in bold) from T1 , where
T is the reverse plane partition in Figure 8.16.
Now suppose T has exactly n ≥ 3 rows, T1 , . . . , Tn , so λj = ∣Tj ∣
for each j with 1 ≤ j ≤ n. To construct (S, U ), we first construct the
intermediate pair (Sint , Uint ) that we obtain by applying our bijection
to the reverse plane partition consisting of T2 , T3 , . . . , Tn . We view
Sint and Uint as being joined into a filling of shape (λ2 , . . . , λn ), with
the entries of Uint in bold.
To start, add an empty box to the top of each of the leftmost λ1
columns of our filling, add one to each bold entry, and move each bold
entry up one box in its column. This will leave exactly one empty
box in each of the leftmost λ1 columns, between the bold and nonbold
entries. Now let T̂1 be the row we obtain from T1 by removing those
entries in T1 which are equal to the entries immediately above them

in T2 , and use the RSK insertion algorithm to insert the entries of
T̂1 into our filling, from smallest to largest. This will fill some of the
empty boxes, but by Lemma 8.8 we will only need the empty boxes
we have already added. Now place a bold 1 in each remaining empty
box. Let S be the filling consisting of the nonbold entries, and let U
be the filling consisting of the bold entries.
By induction and the same argument we used to prove Proposi-
tion 8.9, we know S is a semistandard tableau. By construction, the
shape of U is λ/ sh S. As in the previous case, the bottom row of S
is T1 , so U has no entries in its bottom row. Since Uint is an elegant
tableau by induction, and no two of the boxes we add to Uint to get
U are in the same column, U is also an elegant tableau. In Figure
8.19 we see the construction of S (in black) and U (in bold) from Sint
and Uint , where T is the reverse plane partition in Figure 8.18.
To show our map is a bijection, we just need to describe its in-
verse. That is, we need to explain how to reconstruct T from the
2 3 5
1 3 4 4 6
1 3 4 4 5 5 6
Figure 8.18. A reverse plane partition with three rows
2
2 5 1 2 5
1 3 4 4 6 1 3 4 4 6
1 1 2
2 5 6 1 1
1 3 4 4 5 5 6
Figure 8.19. Building S and U from Sint and Uint for the
reverse plane partition in Figure 8.18
8.2. Constructing Q(π) 223
associated pair (S, U ). We saw that the bottom row of S is the bot-
tom row of T , so we start by letting T1 be the bottom row of S. To
construct (Sint , Uint ) from (S, U ), we first continue to view S and U
as being joined into a filling of shape λ, and we remove the bold 1’s
from this filling. Now some columns contain an empty box and some
do not; apparently the highest nonbold entries in the columns with
no empty boxes are in the boxes that were added in the last step of
the construction of S from Sint . And by Lemma 8.8 these boxes were
added from upper left to lower right. We can reverse RSK insertion
on these boxes from lower right to upper left (see the next section for
more discussion of reversing RSK insertion), subtract one from each
bold entry and move it down one box in its column. Now the leftmost
λ1 columns will have an empty box at the top; when we remove these
boxes we obtain (Sint , Uint ). We have seen the second row of T is
the bottom row of Sint , so we let T2 be the bottom row of S, and we
repeat the process until we have removed all of the boxes from S and
U.
8.2. Constructing Q(π)

Now that we know how to construct the semistandard tableau P (π)
for any generalized permutation π, we can complete our bijection
(and our proof of Cauchy’s formula) by explaining how to construct
Q(π). To see how we might do this, it is useful to recall that we
are constructing a bijection from a set of generalized permutations to
a set of ordered pairs of semistandard tableaux of the same shape.
To invert this map, we will need to be able to recover π from its
image (P (π), Q(π)). Our hope, then, is that Q(π) will give us just
enough information to allow us to reverse the construction of P (π).
So we ask ourselves, if we already have P (π), how do we reverse its
construction, and what information do we need to actually carry the
process out?
The construction of P (π) is a sequence of steps in which we insert
a number into a row. If an insertion did not result in a number
being bumped out of the row, then we must have added a new box
to the end of the row for the newly inserted number. In this case,
we know exactly which number was inserted: it was the rightmost
number in the row. On the other hand, if an insertion did result in a

number being bumped, and we know what number was bumped, then
that number was the leftmost number greater than the number being
inserted. Therefore, as we noted earlier, the number inserted became
the rightmost number less than the number that was bumped.
Putting our observations together, we see that if we know which
box was added in the last step of the construction of P (π), then we
can put its entry back where it came from in the row below. Then we
can put the entry that bumped it back where it came from in the next
row below, etc. Eventually, we find out which entry bumped an entry
out of the first row; that entry must be the last number inserted. We
can see how this works in the next example.
Example 8.10. Suppose we obtained the semistandard tableau in

Figure 8.20 by inserting a number c into a semistandard tableau T .
For each box that could have been added as the new box at the end
of the insertion process, find c and T .
Solution. When we insert a number into a semistandard tableau, the

new box we eventually add is always at the right end of its row and
the top of its column. In the given tableau there are three boxes that
could be the last new box: the box containing 4 at the end of the first
row, the box containing 7 at the end of the second row, and the box
containing 6 at the end of the third row.
If the last box added was the box containing 4 at the end of the
first row, then no bumping occurred and we must have inserted c = 4
into the tableau in Figure 8.21.
If the last box added was the box containing 7 at the end of the
second row, then it must have been bumped by the rightmost 4 in the
6 6
4 5 5 7
3 3 4 4 4
Figure 8.20. The semistandard tableau for Example 8.10

6 6
4 5 5 7
3 3 4 4
Figure 8.21. If we insert c = 4 into this tableau, then we

obtain the tableau in Figure 8.20.
6 6
4 5 5
3 3 4 4 7
Figure 8.22. If we insert c = 4 into this tableau, then we

obtain the tableau in Figure 8.20.
first row, so we must have inserted c = 4 into the tableau in Figure

8.22.
Finally, if the last box added was the box containing 6 at the end
of the third row, then two bumps must have occurred, and we can
reverse the process as in Figure 8.23. □
In our solution to Example 8.10, we noted that the box added to

a semistandard tableau T in the construction of rc (T ) must be at the
6 6
6
4 5 5 7 4 5 6 7
5
3 3 4 4 4 3 3 4 4 4
6
4 5 6 7
3 3 4 4 5
c=4
Figure 8.23. The intermediate steps when we remove the 6
from the top row of the semistandard tableau in Figure 8.20
right end of its row and the top of its column. We call such a box an
outer corner of the shape of T .
As Example 8.10 and the discussion before it suggest, if we specify
an outer corner of a semistandard tableau, then we can remove it
using a reverse insertion process. We start this process by removing
the specified outer corner from T . Suppose this corner is in the kth
row, and it has entry ck−1 . Now we find the rightmost entry ck−2 in
the (k − 1)th row which is less than ck−1 , and we replace it with ck−1 .
We repeat this process until we have removed an entry c0 from the
first row of T . As we note next, this process always produces a new
semistandard tableau.
Proposition 8.11. Suppose T is a semistandard tableau and s is an

outer corner in T . Then the filling we obtain from T by using s in the
reverse insertion process described above is a semistandard tableau.
We now see that if we know in which order boxes were added

in the construction of P (π), then we can reconstruct π. This means
Q(π) needs to record the order in which the boxes were added in
the construction of P (π). On the other hand, if the top row of π is
a1 , . . . , an , then Q(π) must have a1 , . . . , an as its entries, to ensure
xQ(π) = xtopwt(π) . With this in mind, we construct Q(π) as follows.
a1 a2 ⋅ ⋅ ⋅ an
Definition 8.12. If π = [ ] is a generalized permu-
b1 b2 ⋅ ⋅ ⋅ bn
tation, then we write Q(π) to denote the filling of the Ferrers dia-
gram of sh(P (π)) that we obtain by placing aj in the box we add to
P (b1 , . . . , bj−1 ) to produce P (b1 , . . . , bj ).
Example 8.13. Find Q(π1 ), Q(π2 ), and Q(π3 ) for the general-
1 1 2 3 3 1 1 2 3 4
ized permutations π1 = [ ], π2 = [ ], and
2 4 3 1 2 2 4 3 1 2
1 2 3 4 5
π3 = [ ].
2 4 3 1 2
Solution. In each case we are computing P (24312) as in Example 8.1.

If we insert the entries of the top row of our generalized permutation
3 3 4
2 3 2 4 3 5
1 1 1 1 1 2
Q(π1 ) Q(π2 ) Q(π3 )
Figure 8.24. The semistandard tableaux Q(π1 ), Q(π2 ), and

Q(π3 ) in Example 8.13
into each new box as we add it, we find Q(π1 ), Q(π2 ), and Q(π3 ) are
as in Figure 8.24. □
Our construction of Q(π) guarantees it has the same shape as

P (π), but we also need to show Q(π) is always a semistandard
tableau.
Proposition 8.14. If π is a generalized permutation, then Q(π) is

a semistandard tableau.
Proof. The result is clear when the length of π is 0 or 1, so we assume

this length is at least 2, and we argue by induction.
For each j, when we add aj to our growing tableau, we add it
at the right end of a row and the top of a column. Since a1 ≤ ⋅ ⋅ ⋅ ≤
aj , when we add aj it is greater than or equal to every other entry
currently in the tableau, so in particular it is greater than or equal
to every other entry in its row and in its column. By induction, this
means the entries in every row of Q(π) are weakly increasing from
left to right.
To show the entries in every column of Q(π) are strictly increasing
from bottom to top, first note that by induction the only column in
which this could fail is the column containing aj . To see it does not
fail, suppose we have aj = aj−r for some positive integer r. Since π is a
generalized permutation, we must also have bj ≥ bj−r . Now by Lemma
8.8 the box we add when we insert bj is to the right (and possibly
below) the box we add when we insert bj−r , so aj and aj−r cannot be
in the same column. Therefore, no two entries in the same column
of Q(π) are equal, so the entries in every column must be strictly
increasing from bottom to top. This means Q(π) is a semistandard

tableau. □
Now that we know how to construct Q(π), we can define the RSK
correspondence.
Definition 8.15. The RSK correspondence is the function R from
the set of generalized permutations of length n to the set of ordered
pairs of semistandard tableaux with the same shape, and n boxes,
which is defined by R(π) = (P (π), Q(π)).
As our discussion above suggests, we will show R is a bijection

by constructing its inverse. However, at first glance it appears Q(π)
might not tell us the exact order in which we added boxes to P (π),
because Q(π) may have repeated entries. If Q(π) has three copies of
its largest entry, then which one is in the box added last? Fortunately,
as we see in our next example, Lemma 8.8 will tell us the answer.
Example 8.16. Find π if R(π) is the ordered pair of semistandard
tableaux in Figure 8.25.
5 5 4 5
3 4 2 4
2 2 5 1 3 4
P (π) Q(π)
Figure 8.25. The ordered pair R(π) for Example 8.16
Solution. The largest entry in Q(π) is 5, and it occurs just once, so it

must be in the last box added in the construction of P (π). When we
remove it and extract the corresponding entry from P (π), we find we
must have inserted 2 into the tableau in Figure 8.26 to obtain P (π).
5
Therefore, the rightmost column of π is [ ], and we need to continue
2
the process on the ordered pair in Figure 8.27.
The largest entry in our new tableau Q is 4, which appears in
three boxes. By the definition of a generalized permutation, the en-
tries inserted into P to produce these boxes must have been inserted
5
3 5
2 4 5
Figure 8.26. The first tableau we obtain from P (π) in our

solution to Example 8.16
5 4
3 5 2 4
2 4 5 1 3 4
P Q
Figure 8.27. The ordered pair of semistandard tableaux we

obtain after one step in our solution to Example 8.16
5 2
3 5 1 3
P Q
Figure 8.28. The ordered pair of semistandard tableaux we

obtain after removing the 4’s from Q(π) in our solution to
Example 8.16
in weakly increasing order, so by Lemma 8.8, we should remove the

boxes in Q containing 4 in order from lower right to upper left. When
4 4 4 5
we remove entries in this order, we find the tail of π is [ ],
2 4 5 2
and we reach the ordered pair in Figure 8.28.
Now that we have no more repeated numbers in Q, we can con-
1 2 3 4 4 4 5
tinue to find π = [ ]. □
5 3 5 2 4 5 2
We can now define a map E from the set of ordered pairs of

semistandard tableaux of the same shape with n boxes to the set of
2 × n arrays of positive integers. As in our example, E operates by
removing one box at a time from P and Q, and building E(P, Q)
from right to left. If at some stage j boxes remain in Q, then aj is
the largest of the remaining entries of Q. Among all of the boxes in
Q containing aj , we remove the one farthest to the right. We now use

the reverse insertion process to remove the entry in the corresponding
box of P . Then bj is the entry we eventually remove from the first
row of P in this reverse insertion process.
We believe E = R−1 , but to prove this, we need to know E has
the correct range. The key to the proof will be the following lemma.
Lemma 8.17. Suppose P is a semistandard tableau and ck is an

entry in an outer corner of P in the (k + 1)st row of P . Let P ′ be the
semistandard tableau we obtain by using the reverse insertion process
to remove the box containing ck from P , and let c0 be the number
in the first row of P removed in this process. Now suppose dm is an
entry in an outer corner of P ′ in the (m + 1)st row of P ′ , and m ≥ k,
so dm is to the left of ck in P . If d0 is the number in the first row
of P ′ we remove when we use the reverse insertion process to remove
the box containing dm from P ′ , then d0 ≤ c0 .
Proof. Let ck , ck−1 , . . . , c0 be the numbers in P which are moved to

a lower row in P when we remove the box containing ck from P .
Similarly, let dm , dm−1 , . . . , d0 be the numbers in P ′ which are moved
to a lower row in P ′ when we remove the box containing dm from P ′ .
Since ck is in an outer corner in P , and since it is gone when we
move dk+1 into ck ’s former row, dk+1 must displace an entry dk from
that row that was to the left of ck . Therefore, dk ≤ ck . Now when
we put dk into row k, we know ck is already there and dk will replace
an entry dk−1 to the left of ck . Since ck replaced ck−1 , in P we must
have had dk−1 in the same row as and to the left of ck−1 . Therefore,
dk−1 ≤ ck−1 . By induction, d0 ≤ c0 . □
Proposition 8.18. For any ordered pair (P, Q) of semistandard tab-

leaux of the same shape with n boxes, E(P, Q) is a generalized per-
mutation of length n.
Proof. Since the entries of Q are exactly the entries of the top row
of E(P, Q), we see the length of E(P, Q) is the number of boxes in
a a2 ⋅ ⋅ ⋅ an
Q. For convenience, set E(P, Q) = [ 1 ].
b1 b2 ⋅ ⋅ ⋅ bn
When j boxes remain in Q, we always choose aj to be the largest

remaining entry of Q, which means a1 ≤ ⋅ ⋅ ⋅ ≤ an . So we just need to
show that if aj = aj+1 , then bj ≤ bj+1 .
By our construction of E(P, Q), if aj = aj+1 , then aj appears to
the left of and weakly above aj+1 in Q. In addition, because aj and
aj+1 are the largest entries in their tableaux, aj is in an outer corner
of Q, and aj+1 is in an outer corner of the tableau we obtain from
Q by removing aj . Together, these observations mean that if ck is
the entry in P in the box corresponding to the box in Q containing
aj , and dm is the entry in P in the box corresponding to the box in
Q containing aj+1 , then ck and dm satisfy the hypotheses of Lemma
8.17. Therefore, if c0 is the number we remove from the first row of
P when we remove the box containing ck , and d0 is the number we
remove from the first row of the resulting tableau when we remove
the box containing dm , then d0 ≤ c0 . But d0 = bj and c0 = bj+1 , so we
have bj ≤ bj+1 , which is what we wanted to prove. □
Theorem 8.19. For every n ≥ 0, the map R is a bijection between

the set of generalized permutations of length n and the set of or-
dered pairs (P, Q) of semistandard tableaux with n boxes and the
same shape. Furthermore, for every generalized permutation π we
have xtopwt(π) y bottomwt(π) = xQ(π) y P (π) .
Proof. By construction P (π) and Q(π) have the same shape for
any generalized permutation π. Therefore, by Propositions 8.9 and
8.14, for each n ≥ 0 the map R is indeed a function from the set
of generalized permutations of length n to the set of ordered pairs
(P, Q) of semistandard tableaux with n boxes and the same shape.
On the other hand, by Proposition 8.18 we also know that for each
n ≥ 0 the map E is a function from the set of ordered pairs (P, Q) of
semistandard tableaux with n boxes and the same shape to the set of
generalized permutations of length n. Finally, by construction R and
E are inverse functions, so they are both bijections.
By construction, the entries of P (π) are exactly the entries in the
bottom row of π, and the entries of Q(π) are exactly the entries of
the top row of π, so the last claim also holds. □
8.3. Implementing RSK with Growth Diagrams

One of the attractions of the RSK correspondence is the extent to
which it connects properties of and relationships among generalized
permutations with properties of and relationships among the corre-
sponding ordered pairs of semistandard tableaux. Even better, we
can often illuminate these connections by describing the RSK corre-
spondence in new and surprising ways. To illustrate these ideas, we
look at a natural operation on generalized permutations.
Definition 8.20. For any generalized permutation π, let π −1 be the

generalized permutation we obtain from π by swapping the top and
bottom rows and by rearranging the columns of the resulting array
to form a generalized permutation.
1 1 2 3 4 4 4
Example 8.21. Find π −1 for π = [ ].
1 3 1 3 2 2 4
Solution. When we swap the top and bottom rows of π, we obtain

1 3 1 3 2 2 4
[ ]. And when we rearrange the columns of this
1 1 2 3 4 4 4
1 1 2 2 3 3 4
array, we find π −1 = [ ]. □
1 2 4 4 1 3 4
By construction we have (π −1 )−1 = π, and you can check that if

π is a permutation in the usual sense (that is, a bijection from [n]
to [n]), then π −1 is the inverse function for π. So we might ask, how
can we construct (P (π −1 ), Q(π −1 )) directly from (P (π), Q(π)), with-
out actually computing π −1 ? In this section we answer this question
by reformulating the RSK correspondence in terms of objects called
growth diagrams.
To start to describe how to construct a growth diagram, suppose
a a2 ⋅ ⋅ ⋅ an
we have a generalized permutation π = [ 1 ]. Let W
b1 b2 ⋅ ⋅ ⋅ bn
be the maximum of a1 , . . . , an , let H be the maximum of b1 , . . . , bn ,
and construct a rectangular grid of squares with H rows and W
columns. Number the columns 1, 2, . . . , W from left to right and the
rows 1, 2, . . . , H from bottom to top. For each j and k with 1 ≤ j ≤ W
8.3. Implementing RSK with Growth Diagrams 233
2 2
1 1
0 1
Figure 8.29. The first step in our growth diagram example
and 1 ≤ k ≤ H, enter the number of columns in π with top entry j

and bottom entry k in the box in column j and row k.
1 1 1 2 2 2 2
As a running example, let π = [ ]. Then
2 3 3 1 2 3 3
n = 7, W = 2, H = 3, and we have the initial diagram in Figure 8.29.
At each vertex of our growth diagram we will place a finite se-
quence of positive integers. For now we make no further assumptions
about these sequences, but we will eventually show they are all par-
titions. We start by placing the empty sequence at every vertex in
the leftmost column and at every vertex in the bottom row. In our
running example we get the diagram in Figure 8.30.
We now place a sequence of positive integers at each of the re-
maining vertices, working from lower left to upper right, using the
following growth rule. Suppose our partially filled diagram contains
the configuration in Figure 8.31, where λ, µ, and ν are sequences of
positive integers. Let m be the sequence of positive integers whose
jth term mj is the minimum of µj and νj , and let M be the sequence
whose jth term is the maximum of µj and νj . Note that if we identify
µ and ν with their Ferrers diagrams, then m = µ ∩ ν and M = µ ∪ ν.
To construct the sequence α that we will place in the upper right
corner of our square, first set α1 = M1 + k. Then, for j ≥ 2, set
αj = Mj + mj−1 − λj−1 . We will see shortly that mj ≥ λj for all j, so to
obtain αj we are starting with the maximum of µj and νj , and then

adding the amount by which the minimum of µj−1 and νj−1 exceeds
λj−1 . When we apply this rule repeatedly in our running example,
we get the diagram in Figure 8.32.
Drawing the Ferrers diagrams of the partitions at the vertices of
the upper right square in our running example sheds some additional
light on the growth rule. In Figure 8.33 we have this square, along
with m and M . In m the box with the dot is the box in m which is
not in λ; this box moves to the box with the dot in α. And in α the
shaded boxes are the other two boxes we added to M to get α.
To show how to use the growth diagram of π to construct P (π)
and Q(π), we first need to develop some of its properties. To start,
we can use a short chain of inequalities to show that if µ and ν are
partitions, then M and m are also partitions. For example, if j ≥ 2
and Mj = µj , then we have
Mj = µj ≤ µj−1 ≤ max(µj−1 , νj−1 ) = Mj−1 .
Similar inequalities hold if Mj = νj , so M1 ≥ M2 ≥ ⋅ ⋅ ⋅ and M is a

partition. And we can use a similar argument to show m is also a
partition.
Next we show λ ⊆ µ and λ ⊆ ν, which will also mean λ ⊆ m.
2 2
;
1 1
;
0 1
; ; ;
Figure 8.30. The second step in our growth diagram example

μ
k
λ ν
Figure 8.31. The local configuration to which we apply our

growth rule
; (3) (5; 2)
2 2
; (1) (2; 1)
1 1
; ; (1)
0 1
; ; ;
Figure 8.32. The complete growth diagram in our running example
Figure 8.33. The upper right square in the growth diagram

for our running example
Lemma 8.22. If λ, µ, and ν are arranged in our growth diagram

as in Figure 8.31, then we have λj ≤ µj and λj ≤ νj for all j, which
implies λj ≤ mj for all j.
Proof. The results hold when λ = µ = ν = ∅, so we can assume by

induction they hold for a given λ, µ, and ν. Then we need to show
that if we use our growth rule to construct α, then µj ≤ αj and νj ≤ αj
for all j.
By induction we know λj ≤ µj ≤ Mj for all j. Furthermore, α1 =
M1 +k, so α1 ≥ µ1 . In addition, if j ≥ 2, then by induction mj−1 ≥ λj−1 ,
so mj−1 − λj−1 ≥ 0, which means αj = Mj + mj−1 − λj−1 ≥ Mj ≥ µj .
A similar argument shows αj ≥ νj for all j. □
We will also need a somewhat more subtle technical result about

the relationship between λ and M .
Lemma 8.23. If λ, µ, and ν are arranged in our growth diagram as

in Figure 8.31, then Mj ≤ λj−1 for all j ≥ 2.
Proof. The result holds when λ = µ = ν = ∅, so we can assume by

induction it holds for a given λ, µ, and ν. Then we need to show that
if we use our growth rule to construct α, then αj ≤ µj−1 and αj ≤ νj−1
for all j ≥ 2.
To prove this, first note that λj−1 − Mj ≥ 0 for all j ≥ 2 by induc-
tion. On the other hand, by the definition of m we have νj−1 − mj−1 ≥
0. Therefore, λj−1 − Mj + νj−1 − mj−1 ≥ 0, so Mj + mj−1 − λj−1 ≤ νj−1 .
That is, αj ≤ νj−1 .
A similar argument shows αj ≤ µj−1 , and the result follows. □
In our construction of the growth diagram of π we assumed only

that the entries at the vertices were sequences of nonnegative integers.
Now we can use Lemma 8.23 to show they are actually partitions.
Proposition 8.24. For any generalized permutation π, the sequence

at each vertex of the growth diagram of π is a partition.
Proof. Since the empty sequence is a partition, we can argue by

induction, assuming λ, µ, and ν in our growth rule are partitions. To
show α is also a partition, first note that M1 −m1 ≥ 0 by the definitions
of m and M , and λ1 −M2 ≥ 0 by Lemma 8.23, so M1 −m1 +λ1 −M2 ≥ 0
and M1 − m1 + λ1 − M2 + k ≥ k. Rearranging this last inequality, we
find α1 − α2 ≥ k.
Similarly, for all j ≥ 3 we have Mj−1 − mj−1 ≥ 0 by the definitions

of m and M , we have mj−2 −λj−2 ≥ 0 by Lemma 8.22, and λj−1 −Mj ≥ 0
by Lemma 8.23. Therefore,
Mj−1 − mj−1 + mj−2 − λj−2 + λj−1 − Mj ≥ 0.
Rearranging this last inequality, we find αj−1 − αj ≥ 0, so αj−1 ≥ αj
for all j ≥ 3. Therefore, α is a partition. □
As we will show, for any generalized permutation π, the common

shape of P (π) and Q(π) is the partition in the upper right corner of
the growth diagram of π. To actually construct P (π) and Q(π), we
use two fillings of the partitions in the growth diagram of π.
To construct the P -filling of the growth diagram, first label the
vertical edges in each column 1, 2, . . . , H from bottom to top. By
Lemma 8.22, if ν is immediately below α in the growth diagram, then
ν ⊆ α. In α, fill all of the boxes which are also in ν with the same
entry as in ν, and fill all of the remaining boxes with the label of
the vertical edge from ν to α. In our running example we obtain
the fillings in Figure 8.34. Here we have removed the boxes from the
Ferrers diagrams, listing only their entries, to save space.
If you use the RSK correspondence to construct P (π) in our
running example, you will discover it is exactly the filling in the upper
right corner of the P -filling of the growth diagram for π. As we show
next, this is not a coincidence.
Proposition 8.25. For any generalized permutation π, the semistan-
dard tableau in the upper right corner of the P -filling of the growth
diagram of π is P (π).
Proof. To start, assign coordinates to the vertices of the growth di-

agram of π as if they were in the first quadrant of the plane, so the
lower left corner has coordinates (0, 0) and the upper right corner
has coordinates (W, H). For each vertex (a, b), let Ins(a, b) be the
semistandard tableau we obtain by using the RSK correspondence to
insert the entries of the bottom row of π which correspond to the
squares below and to the left of the vertex (a, b) in the growth dia-
gram. In addition, for each vertex (a, b), let P (a, b) be the semistan-
dard tableau at vertex (a, b) in the P -filling of the growth diagram
of π. We need to show Ins(W, H) = P (W, H). Since Ins(a, b) and

P (a, b) are both the empty tableau (and therefore are equal) when-
ever a = 0 or b = 0, we can assume by induction Ins(a, b) = P (a, b)
whenever (a, b) ≠ (W, H).
We first claim Ins(W, H) and P (W, H) agree on all of their entries
which are less than H.
To see this, first note that by the construction of our P -labeling,
in P (W, H) the entries which are less than H form exactly the semi-
standard tableau P (W, H − 1).
On the other hand, note that Ins(W, H) is the result of taking
the sequence we inserted to produce Ins(W, H − 1), including some
additional H’s at various places, and inserting the resulting sequence.
Since no H can move any smaller number in the insertion process, the
numbers less than H will form the same tableau in Ins(W, H) as they
do in Ins(W, H − 1). By induction this is P (W, H − 1), so Ins(W, H)
and P (W, H) agree on all of their entries which are less than H.
; 233 23
12333
3 2 3 2 3
; 2 2
12
2 1 2 1 2
; ; 1
1 0 1 1 1
; ; ;
Figure 8.34. The P -filling for our running example

To complete the proof, let µ be the shape of Ins(W − 1, H) =

P (W −1, H), let λ be the shape of Ins(W −1, H −1) = P (W −1, H −1),
let ν be the shape of Ins(W, H − 1) = P (W, H − 1), and let k be the
number in the upper right square of the growth diagram of π. Note
that we know exactly how many H’s are in each row of P (W, H):
there are M1 + k − ν1 H’s in the first row of P (W, H) and Mj + mj−1 −
λj−1 − νj in row j, for all j ≥ 2. Since H is the largest entry in both
P (W, H) and Ins(W, H), it can only appear in contiguous strings at
the end of a row. So we just need to show the number of H’s in the
first row of Ins(W, H) is M1 + k − ν1 , and the number in row j, for
j ≥ 2, is Mj + mj−1 − λj−1 − νj .
To count the H’s in each row of Ins(W, H), first note that we
obtain Ins(W, H) from Ins(W − 1, H) by inserting a nondecreasing
sequence of positive integers. By Lemma 8.8, this process bumps
each H in Ins(W, H) at most once. By construction, the jth row of
P (W − 1, H) contains exactly µj − λj H’s, and since P (W − 1, H) =
Ins(W − 1, H) by induction, this is also true of the jth row of
Ins(W − 1, H). Since Ins(W, H − 1) is the tableau made by the num-
bers less than H, a copy of H in Ins(W − 1, H) is bumped up to
the next row if and only if it is in a box in µ that also appears in
ν. Therefore, for any j ≥ 1, the number of H’s bumped from row
j to row j + 1 in the transition from Ins(W − 1, H) to Ins(W, H) is
min(νj − λj , µj − λj ) = mj − λj . We know the number of H’s in row
j of Ins(W, H) is the number of H’s in row j of Ins(W − 1, H), plus
the number which enter from row j − 1, minus the number that leave.
When j = 1 this is µ1 − λ1 − (m1 − λ1 ) + k = µ1 − m1 + k = M1 + k − ν1 ,
where we have used the fact that m1 + M1 = µ1 + ν1 in the last step.
And when j ≥ 2, this is
µj − λj + mj−1 − λj−1 − (mj − λj ) = µj + mj−1 − λj−1 − mj

= Mj + mj−1 − λj−1 − νj ,
where we have again used the fact that mj + Mj = µj + νj in the last

step.
We now see Ins(W, H) and P (W, H) have the same number of
H’s in each row, so they must be equal. □
Proposition 8.25 gives us a couple of interesting facts about P -

fillings for free.
Corollary 8.26. For any generalized permutation π, the P -fillings at
the vertices of the growth diagram of π are all semistandard tableaux.
Corollary 8.27. For any generalized permutation π, if α is a par-
tition immediately above the partition ν in the growth diagram of π,
then in each column of the Ferrers diagram of α there is at most one
box which is not in ν.
By looking carefully at the diagram in Figure 8.35 we can see

Corollary 8.27 is a restatement of Lemma 8.23.
αj
νj−1
| {z }
αj−1
Figure 8.35. A diagram connecting Corollary 8.27 with

Lemma 8.23
To build Q(π), we fill the growth diagram of π horizontally rather

than vertically. More specifically, to construct the Q-filling of the
growth diagram, first label the horizontal edges in each row 1, 2, . . . , W
from left to right. By Lemma 8.22, if µ is immediately to the left of
α in the growth diagram, then µ ⊆ α. In α, fill all of the boxes
which are also in µ with the same entry as in µ, and fill all of the
remaining boxes with the label of the horizontal edge from µ to α. In
our running example we obtain the fillings in Figure 8.36.
Proposition 8.28. For any generalized permutation π, the semistan-
dard tableau in the upper right corner of the Q-filling of the growth
diagram of π is Q(π).
Proof. Assign coordinates to the vertices of the growth diagram of π

as in the proof of Proposition 8.25. For each (a, b), let Q(a, b) be the
tableau at vertex (a, b) in the Q-filling of the growth diagram of π and
let Fill(a, b) be the tableau Q(π(a, b)), where π(a, b) is the generalized
permutation corresponding with the squares below and to the left of
; 111 22
1 2 11122
2 2
; 1 2
1 2 12
1 1
; ; 2
1 2
0 1
; ; ;
1 2
Figure 8.36. The Q-filling for our running example
(a, b) in the growth diagram of π. We need to show Q(W, H) =

Fill(W, H). Since Q(a, b) and Fill(a, b) are both the empty tableau
(and therefore are equal) when a = 0 or b = 0, we can assume by
induction that Q(a, b) = Fill(a, b) whenever (a, b) ≠ (W, H).
By construction, we obtain Q(W, H) from Q(W −1, H) by insert-
ing W in all of the boxes in Q(W, H) which are not in Q(W − 1, H).
On the other hand, we construct Fill(W, H) from Fill(W − 1, H) in
the same way: we insert W in all of the boxes in Fill(W, H) which
are not in Fill(W − 1, H). Since Fill(W − 1, H) = Q(W − 1, H) by
induction, we must have Fill(W, H) = Q(W, H). □
We can now answer the question we posed at the start of the

section: we obtain (P (π −1 ), Q(π −1 )) from (P (π), Q(π)) by swapping
P (π) and Q(π).
Theorem 8.29. If π is a generalized permutation, then P (π −1 ) =

Q(π) and Q(π −1 ) = P (π).
Proof. Observe that we obtain the growth diagram of π −1 by re-

flecting the growth diagram of π over a line from the lower left
corner to the upper right corner. Since this reflection switches the
P -labeling and the Q-labeling, we must have P (π −1 ) = Q(π) and
Q(π −1 ) = P (π). □
8.4. Problems
1 1 2 2 3 4 4
8.1. Compute P (π) and Q(π) for π = [ ].
1 5 2 2 1 1 4
8.2. Find the generalized permutation π for which P (π) and Q(π)
are as in Figure 8.37.
4 5
3 4 6 4 5 7
1 3 3 2 3 3
P (π) Q(π)
Figure 8.37. The tableaux P (π) and Q(π) for Problem 8.2
8.3. Suppose 1, 1, 3, 3, 3, 4, 4, 5 is the top row of a generalized permu-

tation π and P (π) is the semistandard tableau in Figure 8.38.
What are Q(π) and π?
5
3 5
2 4
1 2 2
Figure 8.38. The tableau P (π) for Problem 8.3
8.4. Suppose the diagram in Figure 8.39 is part of a growth diagram

for a generalized permutation. Find λ and k.
8.5. Suppose the diagram in Figure 8.40 is part of a growth diagram
for a generalized permutation. Find µ and k.
8.4. Problems 243
k
λ
Figure 8.39. Part of a growth diagram for Problem 8.4
Figure 8.40. Part of a growth diagram for Problem 8.5
8.6. Give a constructive bijection between generalized permutations

and P × P matrices of nonnegative integers with finite support.
8.7. For which generalized permutations π does P (π) have shape n?
8.8. For which generalized permutations π does P (π) have shape
1n ?
8.9. For which generalized permutations π does P (π) have a hook
shape?
8.10. For any sequence a1 , . . . , an and any j with 1 ≤ j ≤ n, let f (j)
be the length of the longest weakly increasing subsequence of
a1 , . . . , an whose last entry is aj . For example, if our sequence is
2, 3, 2, 3, 1, 5, then we have the values of f in Table 8.1. Now for
each k, let rk = aj , where j is the largest number with f (j) = k.
Table 8.1. The values of f in our example in Problem 8.10
j 1 2 3 4 5 6
f (j) 1 2 2 3 1 4
In our example, r1 = a5 = 1, r2 = a3 = 2, r3 = a4 = 3, and

r4 = a6 = 5. Show r1 , r2 , . . . are the entries in the bottom row
of P (a1 , . . . , an ), read from left to right, and the length of this
row is the length of the longest weakly increasing subsequence
of a1 , . . . , an .
8.11. Show that for any sequence a1 , . . . , an of positive integers, the
length of the longest strictly decreasing subsequence of a1 , . . . ,
an is the length of the leftmost column of the semistandard
tableau P (a1 , . . . , an ).
8.12. Suppose a1 , . . . , an is a sequence of distinct positive integers.
Compute P (a1 , . . . , an ) and P (an , an−1 , . . . , a1 ) for several such
sequences, and make a conjecture about the relationship be-
tween P (a1 , . . . , an ) and P (an , an−1 , . . . , a1 ) in general.
8.13. Show that if π ∈ Smn+1 , then π contains an increasing subse-
quence of length m + 1 or a decreasing subsequence of length
n + 1.
8.14. Prove that
∞ ∞
∑ sλ (X)sλ′ (Y ) = ∏ ∏ (1 + xk yj ).
λ j=1 k=1
(This will require a modification of the RSK algorithm.)

8.15. Prove that
1
∑ sλ (X) = ∞ .
λ ∏j=1 (1 − xj ) ∏k>j≥1 (1 − xj xk )
Suppose T is a semistandard tableau with n boxes. We say T is

a standard tableau whenever its entries are 1, 2, . . . , n, each of which
appears exactly once in T . For any partition λ, we write fλ to denote
the number of standard tableaux of shape λ.
8.16. Show
2
∑ (fλ ) = n!
λ⊢n
for all n ≥ 0.
8.5. Notes 245
8.17. Show that for all n ≥ 0, fnn is the Catalan number Cn .

8.18. How many standard tableaux with exactly n boxes have at most
two rows?
8.19. Show that the number of standard tableaux with exactly n
boxes and at most three rows is the Motzkin number Mn .
(The Motzkin numbers may be defined by M0 = 1 and Mn =
Mn−1 + ∑nk=2 Mk−2 Mn−k for n ≥ 1.)
8.20. For any partition λ, a saturated partition chain of type λ is a
sequence λ0 , λ1 , . . . , λn of partitions such that λ0 is the empty
partition, λn = λ, and for all j with 0 ≤ j ≤ n − 1, the partition
λj+1 is the partition λj with one additional box. Give a con-
structive bijection between the set of saturated partition chains
of type λ and the set of standard tableaux of shape λ.
8.21. For any partition λ, let Pλ be the poset whose elements are the
boxes in the Ferrers diagram of λ, and for which boxes s and t
have s ≥ t whenever t is weakly above and to the right of s. A
linear extension of Pλ is an ordering of the elements of Pλ such
that if s appears before t, then s ≥ t in Pλ . Give a constructive
bijection between the set of linear extensions of Pλ and the set
of standard tableaux of shape λ.
8.22. Show ∑λ⊢n fλ is the number of involutions in Sn for all n ≥ 0.
8.23. Show that the number of fixed points in an involution π ∈ Sn is
the number of columns of odd length in P (π).
8.5. Notes
The RSK algorithm was first introduced independently by Robinson
[dBR38] and Schensted [Sch61] in the context of permutations. It
was extended to generalized permutations and applied to Cauchy’s
formula (among other things) by Knuth [Knu70]. Since then, nu-
merous authors have given generalizations and analogues in various
settings; van Leeuwen has written an overview of some of these re-
sults [vL96]. The growth diagram approach to the RSK algorithm
was introduced by Fomin [Fom95].
Schensted’s invention of the RSK algorithm was motivated by his
interest in studying longest increasing subsequences in permutations.
Much more has been discovered since Schensted’s work. You can find
out about many of these developments in Romik’s book [Rom15].
There is a beautiful formula for the number of standard tableaux
of shape λ, which involves the hook lengths of λ (see Problem B.5
for the definition of a hook in the Ferrers diagram of a partition). In
particular, the number of standard tableaux of shape λ is
n!
,
∏(j,k)∈λ hj,k
where the product in the denominator is over all boxes in the dia-
gram of λ, and hj,k is the hook length of the box (j, k). Greene,
Nijenhuis, and Wilf [GNW79] have given an elegant probabilistic
proof of this formula, while Novelli, Pak, and Stoyanovskii [NPS97]
have given a more involved bijective proof. The hook-length formula
was discovered independently by Frame and Robinson (working to-
gether) and Thrall within hours of each other. For more details on
this story, and more information on bijective proofs of the formula,
see [Sag01, Sec. 3.10].
Chapter 9
Special Products
Involving Schur
Functions
So far we have found several bases for the vector space of symmetric
functions, we have discovered how to express some of these bases in
terms of others of them, and we have studied their relationship with a
useful inner product. In short, we have pursued a variety of questions
involving linear combinations of symmetric functions. In this chapter
we turn our attention to products of symmetric functions.
As a general goal, if λ and µ are partitions, and aλ and bµ are
elements of one of our named bases of Λ, then we would like to know
how to write aλ bµ as a linear combination of elements of each of
our named bases. In Definition 6.17 and the discussion following
it we described how to do this when aλ and bµ are both complete
homogeneous symmetric functions or are both elementary symmetric
functions. Next we consider products of Schur functions.
Our eventual goal is to give a combinatorial description of the
coefficients we obtain when we write a product sλ sµ of Schur functions
as a linear combination of Schur functions. In situations like this it is
typical to first consider the special cases λ = (1n ) and λ = (n). Next
we usually look at the case in which λ is a hook, which is a partition of
247
248 9. Special Schur Function Products
the form (k, 1n−k ), since these sometimes serve to interpolate between
the partitions (1n ) and (n). These are the cases we consider in this
chapter; they will lead us to combinatorial formulas for the coefficients
we obtain when we write en sµ , hn sµ , and pn sµ as linear combinations
of Schur functions.
9.1. The Pieri Rules

Our first goal is to describe how to write hn sµ as a linear combination
of Schur functions. As usual, we start with some examples in the hope
of formulating a conjecture about these products. Since s1 = e1 = h1 ,
s11 = h2 , and s2 = e2 , the smallest nontrivial example involves s21 .
Example 9.1. Write h2 s21 as a linear combination of Schur func-

tions.
Solution. There are several ways we could compute this product.

We could write h2 and s21 as linear combinations of the monomial
symmetric functions, compute the relevant products to get h2 s21 as
a linear combination of the monomial symmetric functions, and then
invert the matrix giving the Schur functions in terms of the monomial
symmetric functions to get h2 s21 as a linear combination of Schur
functions.
Alternatively, we could use a Jacobi–Trudi identity to write s21 as
a linear combination of the complete homogeneous symmetric func-
tions, multiply by h2 , and then invert the matrix giving the Schur
functions in terms of the complete homogeneous symmetric functions
to get h2 s21 as a linear combination of Schur functions.
Finally, we could use technology.
Whichever method we use, we find h2 s21 = s41 + s32 + s311 + s221 .
In Figure 9.1 we see the partitions λ for which sλ appears in the
Schur expansion of h2 s21 . The partition (2, 1) is contained in each
of these partitions, so in each partition we have shaded the boxes in
λ/(2, 1). □
9.1. The Pieri Rules 249
Figure 9.1. The partitions in the Schur expansion of h2 s21
Computing as in our solution to Example 9.1, we also find

h1 s21 = s31 + s22 + s211 ,
h3 s21 = s51 + s42 + s411 + s321 ,
h3 s22 = s52 + s421 + s322 ,
and
h4 s22 = s62 + s521 + s422 .
These results suggest hn sµ is always a simple sum of Schur functions,
but it is not so clear exactly which Schur functions actually appear.
To figure this out, it is useful to look at the Ferrers diagrams of the
partitions λ for which sλ appears in the Schur expansion of hn sµ . In
Figure 9.2 we have the Ferrers diagrams of the partitions λ for which
sλ appears in the Schur expansion of h1 s21 , in Figure 9.3 we have
the Ferrers diagrams of the partitions λ for which sλ appears in the
Schur expansion of h3 s21 , in Figure 9.4 we have the Ferrers diagrams
of the partitions λ for which sλ appears in the Schur expansion of
h3 s22 , and in Figure 9.5 we have the Ferrers diagrams of the partitions
λ for which sλ appears in the Schur expansion of h4 s22 . A close
examination of these results (and perhaps others) suggests that if sλ
is a term in the Schur expansion of hn sµ , then µ ⊆ λ and no two boxes
of λ/µ are in the same column. We will need names for skew shapes
of this type and for their conjugates, which will appear in the Schur
expansion of en sµ .

Definition 9.2. Suppose µ ⊆ λ are partitions. We say λ/µ is a

horizontal strip of length k whenever it contains exactly k boxes, no
two of which are in the same column. Similarly, we say λ/µ is a
vertical strip of length k whenever it contains exactly k boxes, no two
of which are in the same row.
We can now describe how to write hn sµ as a linear combination of

Schur functions. This result is known as the first Pieri rule; remark-
ably, the central ingredient in its proof is the RSK correspondence.
Theorem 9.3 (The First Pieri Rule). For any nonnegative integer
n and any partition µ, we have
(9.1) hn sµ = ∑ sλ ,
λ
where the sum on the right is over all partitions λ such that µ ⊆ λ
and λ/µ is a horizontal strip of length n.
Proof. By definition
hn sµ = ∑ xP ∏ xj ,
(P,J) j∈J
where the sum on the right is over all ordered pairs (P, J) in which
P is a semistandard tableau of shape µ and J is a multiset of n
positive integers. Therefore it is sufficient to give a bijection between
this set of ordered pairs and the set of semistandard tableaux P ′ for
which µ ⊆ sh(P ′ ), sh(P ′ )/µ is a horizontal strip of length n, and
′
xP ∏j∈J xj = xP .
To describe our bijection, suppose P is a semistandard tableau
of shape µ and J is a multiset of positive integers j1 ≤ j2 ≤ ⋅ ⋅ ⋅ ≤ jn .
Use the RSK insertion algorithm to insert j1 , j2 , . . . , jn into P , in that
order, to obtain a filling P ′ of shape λ.
By Proposition 8.9, we know P ′ is a semistandard tableau, and
′
by construction xP = xP ∏j∈J xj . We also know by construction that
µ ⊆ λ, and the fact that j1 ≤ ⋅ ⋅ ⋅ ≤ jn combined with Lemma 8.8
implies no two of the boxes added to P in the construction of P ′ will
be in the same column. In other words, λ/µ is a horizontal strip.
To describe the inverse of our bijection, suppose we are given
a semistandard tableau P ′ of shape λ, where µ ⊆ λ and λ/µ is a
horizontal strip of length n. Use the inverse of the RSK insertion
algorithm to remove the boxes in λ/µ from P ′ , starting with the
rightmost box in λ/µ and moving to the left. Lemma 8.17 tells us
that if the numbers we remove from P ′ in this way are jn , jn−1 , . . . , j1 ,
then j1 ≤ ⋅ ⋅ ⋅ ≤ jn . And by Proposition 8.11 the resulting filling P is a
semistandard tableau, which must be of shape µ. Since this process
reverses our first process step-by-step, the two maps must be inverse
bijections. □
Since hλ = hλ1 ⋅ ⋅ ⋅ hλl , where l = l(λ), we can use (9.1) repeatedly

to write the product hλ sµ in terms of Schur functions. When we do,
we find that the resulting coefficients are the skew Kostka numbers
we introduced in Definition 5.6.

(9.2) hλ sµ = ∑ Kν/µ,λ sν .
ν⊇µ
Proof. Throughout we set l = l(λ); we argue by induction on l.

When l = 1, Theorem 9.3 tells us
hλ1 sµ = ∑ sν ,
ν⊇µ
where the sum on the right is over all ν ⊇ µ for which ν/µ is a horizon-
tal strip of length λ1 . But a horizontal strip of length λ1 has exactly
one filling of content λ1 , so Kν/µ,λ1 = 1, and the result holds for l = 1.
Now suppose l > 1 and the result holds for all λ of length l − 1.
In addition, notice that for any ν ⊇ µ we can group fillings of ν/µ
of content λ according to the horizontal strip formed by those boxes
which contain l. This implies that for any ν ⊇ µ, we have
Kν/µ,λ = ∑ Kξ/µ,α ,
ν⊇ξ⊇µ
where α = λ1 , . . . , λl−1 and the sum on the right is over all ξ ⊇ µ for
which ν/ξ is a horizontal strip of length λl . With this in mind, by
induction we now have
hλ sµ = sµ hλ1 ⋅ ⋅ ⋅ hλl−1 hλl
= ∑ Kξ/µ,α sξ hλl
ξ⊇µ
= ∑ ∑ Kξ/µ,α sν
ξ⊇µ ν⊇ξ
= ∑ Kν/µ,λ sν ,
ν⊇µ
Example 9.5. Find the semistandard skew tableaux counted by the

coefficient of s433 in the Schur expansion of h331 s21 .
1 2 3 2 2 3
1 2 1 2
1 2 1 1
Figure 9.6. The semistandard skew tableaux of shape

(4, 32 )/(2, 1) and content (32 , 1) in Example 9.5
Solution. These tableaux are the semistandard skew tableaux of shape

(4, 32 )/(2, 1) and content (32 , 1). This shape has a column of height
3, which must contain 1, 2, and 3. It also has a column of height 2,
which must now contain 1 and 2. There are two ways to place the re-
maining 1 and 2 in the last two boxes, so we obtain the semistandard
skew tableaux in Figure 9.6. □
Among other things, Proposition 9.4 tells us that if we write the

complete homogeneous symmetric functions as linear combinations of
Schur functions, the resulting coefficients are the Kostka numbers.
Corollary 9.6. For all partitions λ ⊢ n, we have

hλ = ∑ Kν,λ sν .
ν⊢n
Proof. Set µ = ∅ in Proposition 9.4, and note that s∅ = 1 and

Kν/∅,λ = Kν,λ . □
It is useful to observe that we can also use the Hall inner product
to obtain Corollary 9.6. To do this, first note that since the Schur
functions form a basis for Λ, there are constants Mλ,µ (h, s) such that
hλ = ∑µ Mλ,µ (h, s)sµ . Since the Schur functions are an orthonormal
basis for Λ with respect to the Hall inner product, we have
⟨sν , hλ ⟩ = ⟨sν , ∑ Mλ,ζ (h, s)sζ ⟩

ζ
= ∑ Mλ,ζ (h, s)⟨sν , sζ ⟩

ζ
= Mλ,ν (h, s).

On the other hand, the complete homogeneous symmetric functions
and the monomial symmetric functions are dual bases with respect
to the Hall inner product. Therefore, we also have
⟨sν , hλ ⟩ = ⟨∑ Kν,ζ mν , hλ ⟩
ζ
= ∑ Kν,ζ ⟨mζ , hλ ⟩
ζ
= Kν,λ ,
and the result follows.
Our next goal is to write en sµ as a linear combination of Schur
functions. Fortunately, we can do this by applying ω to the first Pieri
rule. The resulting formula is known as the second Pieri rule.
Theorem 9.7 (The Second Pieri Rule). For any nonnegative integer
n and any partition µ, we have
(9.3) en sµ = ∑ sλ ,
λ
where the sum on the right is over all partitions λ ⊇ µ such that λ/µ
is a vertical strip of length n.
Proof. Apply ω to (9.1) and use Propositions 6.18 and 6.16(ii), (iii)
to simplify the result. □
Theorem 9.7 follows immediately from Theorem 9.3 when we ap-

ply ω, but having a combinatorial proof will be helpful when we con-
sider more general products of Schur functions. To give such a proof,
first observe that
en sµ = ∑ xP ∏ xj ,
(P,J) j∈J
where the sum on the right is over all ordered pairs (P, J) in which
P is a semistandard tableau of shape µ and J is a set of n positive
integers. To prove (9.3), we need to give a bijection between the
set of such ordered pairs and the set of semistandard tableaux P ′
for which µ ⊆ sh(P ′ ), sh(P ′ )/µ is a vertical strip of length n, and
′
xP ∏j∈J xj = xP . In the proof of Theorem 9.3 we used the RSK
insertion algorithm to insert the elements of J into P , from smallest
to largest. But Lemma 8.8 tells us that inserting the elements of J
in this order always produces a semistandard tableau P ′ for which
sh(P ′ )/µ is a horizontal strip, rather than the vertical strip we need.
Fortunately, we have a useful analogue of Lemma 8.8 telling us how

to produce vertical strips.
Lemma 9.8. Suppose T is a semistandard tableau, c > d, s1 , . . . , sn is

the bumping path associated with rc (T ), and t1 , . . . , tm is the bumping
path associated with rd (rc (T )). Then n < m and for all j with 1 ≤ j ≤
n, either sj = tj or sj is to the right of tj .
Proof. This is similar to the proof of Lemma 8.8. □
Lemma 9.8 tells us that if we insert the elements of J into P in

decreasing order, then each bumping path is strictly longer than the
last, so the boxes we add to P will form a vertical strip. This is how
we construct the bijection we want.
As we did for sµ hn , we can use the second Pieri rule to write sµ eλ
as a linear combination of Schur functions, in which the coefficients
are skew Kostka numbers.
(9.4) eλ sµ = ∑ Kν ′ /µ′ ,λ sν .
ν
Proof. Apply ω to (9.2) and use Propositions 6.18 and 6.16(ii), (iii)
to simplify the result. □
As a special case of Proposition 9.9, we can use the Kostka num-

bers to write the elementary symmetric functions as linear combina-
tions of Schur functions.
Corollary 9.10. For all partitions λ ⊢ n, we have
eλ = ∑ Kµ′ ,λ sµ .
µ⊢n
Proof. Set µ = ∅ in Proposition 9.9, and note that s∅ = 1 and

Kν ′ /∅,λ = Kν ′ ,λ . □
9.2. The Murnaghan–Nakayama Rule

Our final goal in this chapter is to write pn sµ as a linear combination
of Schur functions. We will then use this result to write pλ as a
linear combination of Schur functions, obtaining a result known as
the Murnaghan–Nakayama rule. As we did with the first Pieri rule,
we begin with some small examples.
Example 9.11. Write p2 s21 as a linear combination of Schur func-

tions.
Solution. As in Example 9.1, we can do this in a variety of ways.

To do it without technological help, we could start with the fact
that s21 = 31 (p111 − p3 ) to write p2 s21 in terms of the power sum
symmetric functions, and then write the result in terms of Schur
functions. Alternatively, we could write both p2 and s21 in terms
of the monomial symmetric functions, multiply, and then write the
result in terms of Schur functions. Whichever method we choose, we
find
p2 s21 = s41 − s2111 .
In Figure 9.7 we have the Ferrers diagrams of the partitions λ for
which sλ appears in this expansion. The partition (2, 1) is contained
in each of these partitions, so in each diagram we have shaded the
boxes in λ/(2, 1). □
Computing as in our solution to Example 9.11, we also find

p3 s21 = s51 − s33 − s222 + s21111 ,
p4 s22 = s62 − s53 + s22211 − s221111 ,
Figure 9.7. The partitions in the Schur expansion of p2 s21

9.2. The Murnaghan–Nakayama Rule 257
and
p5 s32 = s82 − s64 + s3331 − s322111 + s3211111 .
These results suggest pn sµ is always a simple sum of Schur functions,
some of which appear with coefficient −1. To determine which Schur
functions actually appear in the Schur expansion of pn sµ and with
what signs, it is helpful to look at Ferrers diagrams again. In Figure
9.8 we have the Ferrers diagrams of the partitions λ for which sλ
appears in the Schur expansion of p3 s21 , in Figure 9.9 we have the
Ferrers diagrams of the partitions which appear in the Schur expan-
sion of p4 s22 , and in Figure 9.10 we have the Ferrers diagrams of the
partitions which appear in the Schur expansion of p5 s32 .

For each Ferrers diagram associated with pn sµ , we have shaded

the boxes in the diagram which are not in µ. In some cases these
boxes form a vertical strip, in others they form a horizontal strip,
and in some they form a strip along the border which is neither a
vertical strip nor a horizontal strip. We can describe what seems to
be going on more precisely with the help of the next definition.
Definition 9.12. Suppose µ ⊆ λ are partitions. We say λ/µ is a
border strip whenever it contains no 2×2 square of boxes. We say two
boxes in a border strip are adjacent whenever they share an edge, and
we say a border strip is connected whenever every pair of its boxes
is connected by a sequence of adjacent boxes. If λ/µ is a connected
border strip, then the height of λ/µ, written ht (λ/µ), is the number
of (nonempty) rows in λ/µ.
It appears from our examples that sλ is a term in the Schur

expansion of pn sµ if and only if λ/µ is connected border strip with
n boxes. In addition, the coefficient of sλ in pn sµ appears to be
(−1)ht(λ/µ)−1 .
To start to show that our observations hold in general, we use
some identities proved earlier to rewrite pn sµ in terms of Schur func-
tions. First, we can use (4.21) to write pn as a sum of Schur functions,
obtaining
n−1
pn sµ = ∑ (−1)j sn−j,1j sµ .
j=0
We don’t yet know how to express the product sn−j,1j sµ as a linear
combination of Schur functions, but we do know how to express the
hook Schur function sn−j,1j in terms of elementary and complete ho-
mogeneous symmetric functions. In particular, we can use (4.20) to
eliminate sn−j,1j , obtaining
n−1 j
pn sµ = ∑ ∑ (−1)k hn−k ek sµ .
j=0 k=0
Now we can use the Pieri rules to find

n−1 j
pn sµ = ∑ ∑ (−1)k ∑ hn−k sµv
j=0 k=0 µv ⊇µ
n−1 j
= ∑ ∑ (−1)k ∑ ∑ sµvh ,
j=0 k=0 µv ⊇µ µvh ⊇µv
where the third sum is over all partitions µv ⊇ µ for which µv /µ is

a vertical strip of length k and the fourth sum is over all partitions
µvh ⊇ µv for which µvh /µv is a horizontal strip of length n − k.
In this last sum we construct each of the Schur functions sµvh by
choosing j with 0 ≤ j ≤ n − 1, then choosing k with 0 ≤ k ≤ j, then
choosing µv so µ ⊆ µv and µv /µ is a vertical strip of length k, and
then choosing µvh so µv ⊆ µvh and µvh /µv is a horizontal strip of
length n − k. So each term is indexed by a pair consisting of a number
j and a picture like the one in Figure 9.11, in which µ = (4, 3, 1, 1),
Figure 9.11. Part of a typical indexing pair in our sum for pn sµ

the boxes in µv /µ are shaded light gray, and the boxes in µvh /µv are
shaded dark gray. In each of these pairs we are also required to have
j ≥ ∣µv /µ∣.
We would like to reorder our sum so the partition µvh appears
first, because we expect to eventually reach a sum over partitions that
we obtain by adding n boxes to µ, and this is exactly how we obtain
µvh . When we do this, we find
n−1
∣µ /µ∣
(9.5) pn sµ = ∑ ∑ ∑ (−1) v sµvh ,
µvh ⊇µ µv ⊆µvh
j=∣µv /µ∣
µv ⊇µ
where the middle sum is over all partitions µv such that µ ⊆ µv ,

µv ⊆ µvh , µv /µ is a vertical strip, and µvh /µv is a horizontal strip.
We suspect many terms on the right side of (9.5) will cancel, but
to see how to organize this cancellation, we need to understand the
middle sum. That is, we need to understand how, if we are given a
partition µvh which contains µ and has n more boxes than µ, we can
separate the n new boxes into an inner vertical strip and an outer
horizontal strip. To see what is involved in this separation process,
try the following example.
Example 9.13. Let µ = (8, 7, 5, 1), let n = 13, let α = (10, 9, 8, 4, 2, 1),
and let β = (9, 9, 8, 4, 3, 1). Note that we obtain both α and β by
adding n boxes to µ. For both α and β, find all ways of separating
these new boxes into an inner vertical strip and an outer horizontal
strip.
Solution. In Figure 9.12 we have numbered the boxes in α/µ.
13
12 11
10 9 8
7 6 5
4 3
2 1
Figure 9.12. The diagram of α, with the boxes in α/µ numbered

13
12 11 10
9 8 7
6 5 4
3 2
1
Figure 9.13. The diagram of β, with the boxes in β/µ numbered
To start, note that if box 1 is in our vertical strip, then so is box

2. But boxes 1 and 2 are in the same row, so they cannot be in the
same vertical strip. Therefore, box 1 must be in our horizontal strip.
Similarly, boxes 3, 5, 6, 8, 9, and 11 must be in our horizontal strip.
On other hand, if box 2 is in our horizontal strip, then so is box
3. But boxes 2 and 3 are in the same column, so they cannot be in
the same horizontal strip. Therefore, box 2 must be in our vertical
strip. Similarly, boxes 4, 10, and 12 must also be in our vertical strip.
We have now assigned each box to a strip, except for boxes 7 and
13. We can check that each of these boxes can be assigned to either
strip, so we have four possible separations.
In Figure 9.13 we have numbered the boxes in β/µ.
Note that if box 8 is in our vertical strip, then so is box 9. But
boxes 8 and 9 are in the same row, so they cannot be in the same
vertical strip. Therefore, box 8 is in our horizontal strip.
On the other hand, if box 8 is in our horizontal strip, then so is
box 10. But boxes 8 and 10 are in the same column, so they cannot
be in the same horizontal strip. Therefore box 8 cannot be in either
strip, and there is no way to separate the boxes in β/µ into an inner
vertical strip and an outer horizontal strip. □
We can use the reasoning we saw in our solution to Example 9.13

to make some general observations about how to separate the boxes
in µvh /µ.
Lemma 9.14. Suppose µ ⊆ µvh are partitions and µvh /µ can be sepa-
rated into an inner vertical strip and an outer horizontal strip. Then
the following hold.
(i) Each box in µvh /µ which is not the top box in its column is
in the vertical strip.
(ii) Each box in µvh /µ which is not the leftmost box in its row
(in µvh /µ) is in the horizontal strip.
Proof. (i) If such a box were not in the vertical strip, then it would
be in the horizontal strip, as would every box above it in the same
column. Since there is at least one box above it, this would mean the
horizontal strip has at least two boxes in the same column, contra-
dicting the fact that it is a horizontal strip.
(ii) This is similar to the proof of (i). □
We can combine the two parts of Lemma 9.14 to show µvh /µ can
only be separated in certain situations.
Lemma 9.15. Suppose µ ⊆ µvh are partitions and µvh /µ can be sepa-
rated into an inner vertical strip and an outer horizontal strip. Then
µvh /µ can contain no 2 × 2 square.
Proof. If µvh /µ contains a 2 × 2 square, then Lemma 9.14 tells us its

lower right box must be in both the vertical strip and the horizontal
strip, contradicting the fact that these strips are disjoint. □
Lemma 9.15 tells us that if µvh /µ can be separated, then it is a

border strip, possibly with several connected components. In Figure
9.14 we have one such border strip, with three connected components.
Lemma 9.14 tells us that in each border strip some of the boxes can
only be in the inner vertical strip, while others can only be in the
outer horizontal strip. In Figure 9.15 we have labelled the boxes that
must be in the vertical strip with a v, and those that must be in the
horizontal strip with an h.
In our example, each connected component has exactly one box
that Lemma 9.14 does not assign to a strip. This unassigned box is
always the one in the upper left of the component; we call this box
the head of the component. To see that this holds in general, suppose
Figure 9.14. A ribbon with three connected components
v
v h h h
v
v
h h
v h
v h h
v
Figure 9.15. A ribbon with boxes labelled according to the

strip they must be in
we have a connected component of a border strip, and we start in

the head of the component. Since no box in the component is to the
head’s left, and no box in the component is above it, Lemma 9.14 does
not assign the head to a strip. Now traverse the component from box
to adjacent box, moving down and to the right. By Lemma 9.15 no
box in the component has boxes both directly below and directly to
the right, so each box has a unique next box in the component. If
we enter a box from the left, then Lemma 9.14 assigns that box to
the horizontal strip. If we enter a box from above, then Lemma 9.14
assigns that box to the vertical strip. We enter every box other than
the head of the component from exactly one of these directions, so
every box but the head is assigned to a unique strip. As we show
next, we are free to put the heads of the components in either strip.
Lemma 9.16. Suppose µ ⊆ µvh are partitions and µvh /µ can be sep-
arated into an inner vertical strip and an outer horizontal strip. Fur-
thermore, suppose we label the head of each connected component of
µvh /µ with v or h. Then there is a separation of µvh /µ in which the
heads labelled h are in the horizontal strip and the heads labelled v
are in the vertical strip.
Proof. First note that if two boxes in µvh /µ are in different connected
components, then they are not in the same row or column, so we can
assume without loss of generality that µvh /µ is connected.
Let s be the head of µvh /µ, label each box of µvh /µ which is not
at the top of its column with a v, and label each box which is not the
leftmost box in its row with an h. We make several observations.
First, no two boxes in the same column are labelled h, and no
two boxes in the same row are labelled v. Second, if a box labelled v
is in the same row (resp., column) as a box labelled h, then the box
labelled v is to the left (resp., below) the box labelled h. And finally,
every box in the same row as s is labelled h and every box in the
same column as s is labelled v.
Our last observation tells us that labelling s with h cannot give
us two boxes labelled h in the same column, and labelling s with v
cannot give us two boxes labelled v in the same row. This, combined
with our first observation, tells us that for either labelling of s, the
boxes labelled v will form a vertical strip and the boxes labelled h
will form a horizontal strip. And our second observation tells us the
resulting vertical strip will be inside the horizontal strip. □
Lemma 9.16 tells us we can rewrite (9.5) as

n−1
(9.6) pn sµ = ∑ ∑ ∑ (−1)v(l) sµvh ,
µvh ⊇µ l j=v(l)
where the middle sum is over all labellings l of the heads of the com-
ponents of µvh /µ with v and h, and v(l) is the total number of boxes
labelled v in µvh /µ. We now prove our main result by explaining how
to cancel most of the terms on the right side of (9.6).
Theorem 9.17. For all n ≥ 0 and any partition µ, we have
pn sµ = ∑(−1)ht(λ/µ)−1 sλ .
λ
Here the sum is over all λ ⊇ µ such that λ/µ is a connected border
strip with exactly n boxes, and ht (λ/µ) is the height of λ/µ.
Proof. We start with (9.6), and we give our proof in two stages.
By Lemma 9.16, we are free to switch the label of the head of the
topmost component of µvh /µ between v and h. Each switch changes
v(l) by one, reversing the sign of (−1)v(l) . So all terms for which we
can make this switch cancel. The terms for which we cannot make
this switch are exactly those in which j = v(l) and the head of the
topmost component of µvh /µ is labelled h: if we make the switch in
one of these terms, then we get a term with v(l) > j, which does not
appear in our sum. Therefore, we have
(9.7) pn sµ = ∑ ∑(−1)v(l) sµvh ,
µvh ⊇µ l
where the inner sum is over all labellings of the heads of the com-
ponents of µvh /µ with v and h in which the head of the topmost
component is labelled h, and v(l) is the total number of boxes la-
belled v in µvh /µ.
Equation (9.7) is close to the result we want, but it includes
terms associated with border strips which are not connected. To see
how these terms cancel, suppose µvh /µ is a border strip which is not
connected. By Lemma 9.16 we are free to switch the label on the
head of the second connected component from the top between v and
h. As before, this reverses the sign of the term but changes nothing
else, so all terms with a second connected component cancel.
Now we can take the sum on the right side of (9.7) to be over the
set of connected border strips with exactly n boxes. To obtain the
result we want to prove, we just need to note that in our labellings the
top row contains no v’s, and each row below the top contains exactly
one v, so v(l) = ht (µvh /µ) − 1. □
We can apply Theorem 9.17 repeatedly with µ = ∅ to write pλ as

a linear combination of Schur functions. The terms in this expansion
are indexed by the ways in which we can build λ using border strips,
which leads us to the idea of a border strip tableau of shape λ.
Definition 9.18. For any partitions λ ⊢ n and µ ⊢ n, a border strip

tableau of shape λ and type µ is a filling of (the Ferrers diagram of)
λ with positive integers in [n] such that if λ(j) is the set of boxes
which contain 1, 2, . . . , j, then for all j with 0 ≤ j ≤ l(µ), λ(j + 1)/λ(j)
is a connected border strip with exactly µj+1 boxes. For any border
strip tableau T , we set
l(µ)
sgn(T ) = ∏ (−1)ht(λ(j+1)/λ(j))−1 .
j=0
Example 9.19. Find all border strip tableaux of shape (3, 2, 2) with
type (3, 2, 1, 1).
Solution. There are three ways to place the innermost border strip
of length 3: as a horizontal strip, as a hook of shape (2, 1), and as
a vertical strip. In the first case there are two ways to place the
next border strip (which has length 2), each of which determines the
remaining border strips. In the second case there is no way to place
the next border strip. In the third case there are two ways to place
the next border strip, one of which determines the remaining border
strips, and the other of which has two choices for the placement of the
remaining two border strips. Figure 9.16 shows the resulting border
strip tableaux. □
Theorem 9.20. For any partitions λ ⊢ n and µ ⊢ n, let Mλ,µ (p, s)

be defined by
pλ = ∑ Mλ,µ (p, s)sµ .
µ⊢n
3 4 2 4 1 4
2 2 2 3 1 3
1 1 1 1 1 1 1 2 2
1 4 1 3
1 2 1 2
1 2 3 1 2 4
Figure 9.16. The border strip decompositions in Example 9.19
Then we have
Mλ,µ (p, s) = ∑ sgn(T ),

T
where the sum on the right is over all border strip tableaux of shape
µ with type λ.
Proof. If l(λ) = 1, then λ = n. In this case the claim says Mn,µ (p, s)
is the sum of sgn(T ), over all border strip tableaux T of shape µ with
type n. But there is a border strip tableau of shape µ and type n if
and only if µ is a hook n−j, 1j for some j with 0 ≤ j ≤ n−1. Moreover,
in this case there is exactly one border strip tableau of shape µ, which
has sign (−1)j . Therefore, the claim says
n−1
pn = ∑ (−1)j sn−j,1j ,
j=0
which we proved as (4.21).

Now suppose l(λ) ≥ 2; we argue by induction on l(λ). Set l = l(λ)
and α = λ1 , . . . , λl−1 . In addition, for any partitions µ and ν, let
BST (µ, ν) be the set of all border strip tableaux of shape µ with
type ν. Using our induction hypothesis and Theorem 9.17, we find
pλ = pα pλl
= ∑ ∑ sgn(Tα )sν pλl
ν⊢n−λl Tα ∈BST (ν,α)
= ∑ ∑ sgn(Tα ) ∑(−1)ht(µ/ν)−1 sµ ,
ν⊢n−λl Tα ∈BST (ν,α) µ
where the inner sum on the last line is over all partitions µ such that
ν ⊆ µ and µ/ν is a border strip of size λl . When we combine Tα
with µ/ν, we obtain a border strip tableau T of shape µ and type λ,
and by construction we have sgn(T ) = sgn(Tα )(−1)ht(µ/ν)−1 . Every
border strip tableau of shape µ and type λ is uniquely constructed in
this way, so we have
pλ = ∑ ∑ sgn(T )sµ ,
µ⊢n T ∈BST (µ,λ)
Now we can use the Hall inner product and Theorem 9.20 to
write the Schur functions as linear combinations of the power sum
Corollary 9.21. For any partitions λ ⊢ n and µ ⊢ n, let Mλ,µ (s, p)

be defined by
sλ = ∑ Mλ,µ (s, p)pµ .
µ⊢n
Then we have
Mλ,µ (s, p) = zµ−1 ∑ sgn(T ),
T
where the sum on the right is over all border strip tableaux of shape
λ and type µ.
Proof. By Proposition 7.14 we have
⟨sλ , pµ ⟩ = ⟨ ∑ Mλ,ν (s, p)pν , pµ ⟩

ν⊢n
= zµ Mλ,µ (s, p).
9.3. Problems 269
But by the definition of the Hall inner product we also have
⟨sλ , pµ ⟩ = ⟨sλ , ∑ Mµ,ν (p, s)sν ⟩

ν⊢n
= Mµ,λ (p, s).
Now the result follows from Theorem 9.20. □
For those readers who are familiar with representations of finite

groups, we note that Mλ,µ (p, s) is closely connected with the repre-
sentation theory of the symmetric group Sn . In particular, with some
additional work we can show Mλ,µ (p, s) is the value of the irreducible
character χµ of Sn associated with µ, evaluated on the conjugacy
class of permutations with cycle type λ.
9.3. Problems
9.1. How many border strips are contained in the partition (k n ),
where k and n are positive integers?
9.2. If we write h4 s221 as a linear combination of Schur functions,
then for which partitions λ does sλ have a nonzero coefficient?
9.3. If we write e4 s221 as a linear combination of Schur functions,
9.4. Suppose µ is a partition. For each n ≥ 0, let fh (n) be the sum of
the coefficients when we write hn sµ as a linear combination of
Schur functions. Similarly, for each n ≥ 0, let fe (n) be the sum
of the coefficients when we write en sµ as a linear combination of
Schur functions. Characterize those µ for which fh (n) = fe (n)
for all n ≥ 0.
9.5. If we write p4 s221 as a linear combination of Schur functions,
9.6. Find a skew shape λ/µ for which we have sλ/µ = e3 h4 h2 e5 h1 .
9.7. Show that for all partitions µ ⊢ n and ν ⊢ n, we have
∑ Mµ,λ (p, s)Mν,λ (p, s) = δµ,ν zµ .
λ⊢n
9.8. Show that for all partitions λ ⊢ n and µ ⊢ n, we have
−1
∑ zν Mν,λ (p, s)Mν,µ (p, s) = δλ,µ .
ν⊢n
9.9. Show that if λ = (m, m − 1, m − 2, . . . , 2, 1), then sλ is a polyno-

mial in the odd power sums p1 , p3 , . . . .
9.10. For any partition λ, define a border strip decomposition of λ to
be a partition of the squares of the Ferrers diagram of λ into
(nonempty) border strips. Show that the number of border
strip decompositions of λ is a product of Fibonacci numbers.
Chapter 10
The
Littlewood–Richardson
Rule
The symmetric function ideas we have discussed so far lead to nu-

merous new questions, but two in particular stand out. First, for any
partitions µ and ν, the product sµ sν is a symmetric function, so it
must be a linear combination of Schur functions. That is, there must
be constants cλµ,ν with
(10.1) sµ sν = ∑ cλµ,ν sλ ,
λ
where the sum on the right is over all partitions λ with ∣λ∣ = ∣µ∣ + ∣ν∣.
Similarly, for any partitions µ ⊆ λ, the skew Schur function sλ/µ is a
symmetric function, so it too must be a linear combination of Schur
functions. That is, there must be constants dλµ,ν with
(10.2) sλ/µ = ∑ dλµ,ν sν ,

ν
where the sum on the right is over all partitions ν with ∣λ∣ = ∣µ∣ + ∣ν∣.
The Pieri rules give us combinatorial descriptions of cλµ,ν when µ = (n)
or µ = (1n ). And in Chapter 5 we found conditions on λ and µ which
guarantee dλµ,ν = 0 except for a unique ν, for which dλµ,ν = 1. In this
271
272 10. The Littlewood–Richardson Rule
chapter we show cλµ,ν = dλµ,ν for all partitions µ, ν, and λ, and we

prove these numbers count a particular kind of tableau.
10.1. Products of Tableaux

To give a combinatorial description of the coefficients cλµ,ν in (10.1),
we might hope for a combinatorial model for the product sµ sν . More
specifically, by definition sµ and sν are generating functions for semi-
standard tableaux of a given shape, so their product is a generating
function for ordered pairs of these tableaux. If we can find a way to
combine the tableaux in such an ordered pair to produce an appro-
priate new tableau, then we might be able to describe the resulting
coefficients cλµ,ν .
To describe one way to combine two semistandard tableaux, recall
from Definition 4.36 that the reading word of a semistandard tableau
T , written word(T ), is the word we obtain by reading the entries of
each row of T from left to right, starting with the top (last) row and
moving down to the bottom (first) row.
Definition 10.1. For any semistandard tableaux T1 and T2 , the
product T1 ∗r T2 is the semistandard tableau we obtain by using the
RSK insertion algorithm to insert word(T2 ) into T1 .
Example 10.2. Compute all of the products T1 ∗r T2 in which T1
and T2 are semistandard tableaux of shape (2, 1) with entries in [3].
Solution. After 64 applications of the RSK insertion algorithm, we

obtain the results in Table 10.1. □
Our solution to Example 10.2 gives us some algebraic information,

showing ∗r is not commutative in general, but its combinatorics is
much more intriguing. By construction we have xT1 ∗r T2 = xT1 xT2 , so
the terms in the product s21 (X3 )s21 (X3 ) are exactly the weights of
the entries in Table 10.1. But something more is happening in our
example. At first glance, ∗r gives us no control over the shapes of the
tableaux in Table 10.1. More specifically, we can compute directly
that
s21 (X3 )s21 (X3 ) = s42 (X3 ) + s33 (X3 )
(10.3)
+ 2s321 (X3 ) + s222 (X3 ).
10.1. Products of Tableaux 273
Table 10.1. All products of semistandard tableaux of shape

(2, 1) with entries in [3]
2 3 2 3 2 3 3 3
∗r 11 11 12 12 13 13 22 23
2 22 23 22 23 22 23 23 23
11 1111 1111 1112 1112 1113 1113 1122 1123
3 3 33 3 33 3 33 33 33
11 2 1111 2 1112 2 1113 1122 1123
1111 1112 1113
2 222 223 22 223 22 22 23 23
12 111 111 1122 112 1123 1133 1222 1223
3 3 3 3 3 3 3 33 33
12 22 23 2 23 2 2 1222 1223
111 111 1122 112 1123 1133
2 3 233 3 233 3 23 233 23
13 22 111 22 112 22 1133 122 1233
111 113 113
3 3 333 3 333 3 33 333 33
13 23 111 23 112 23 1133 122 1233
111 112 113
3 3 3 3 3 3 3 33 33
22 22 22 2 23 2 2 2222 2223
112 113 1222 122 1223 1233
3 33 3 3 3 3 3 333 33
23 22 23 23 23 23 2 222 2233
11 113 122 123 123 1333
But we have no reason to think the tableaux in Table 10.1 will have the
shapes (4, 2), (3, 3), (3, 2, 1), and (2, 2, 2) appearing on the right side
of (10.3), even though their weights are the terms in the associated
Schur functions. So it is a pleasant surprise to see that when we group
the tableaux in Table 10.1 by their shapes, we get exactly the sets of
tableaux for which the Schur functions in (10.3) are the generating
functions. In other words, ∗r appears to be a combinatorial model not
just of multiplication of Schur functions, but actually of multiplication
of the individual terms in those Schur functions. We will soon see
these observations hold in general.
We think we have a combinatorial model for the multiplication
happening in (10.1), but we might also want a combinatorial model
for whatever is going on in (10.2). The terms on the left side of
(10.2) are indexed by semistandard tableaux of a specific skew shape,
while the terms on the right are indexed by semistandard tableaux of
various partition shapes. So we would like a combinatorial method of

converting a semistandard tableau of skew shape to a semistandard
tableau of partition shape with the same weight. If we view the boxes
in our given tableau as tiles which can slide either down or to the left,
then there is a natural way to make this conversion, called jeu de
taquin.
To describe jeu de taquin, suppose µ ⊆ λ are partitions and we
have a semistandard tableau T of shape λ/µ. To start, choose a blank
box in µ which is at the top of its column and the right end of its row.
(That is, choose an outer corner of µ.) There is a unique way to slide
a box into the blank space so that the resulting object is column strict
and row nondecreasing. Perform that slide, interchanging the blank
box with a box in the diagram. Repeat this process until the blank
box has no box above it or to its right. Now the number of boxes in
µ has decreased by one; choose an outer corner in this new partition
to be the new blank box, and repeat this sliding process. Continue
choosing outer corners and sliding until the shape is no longer skew.
Definition 10.3. For any semistandard skew tableau T , the rectifi-
cation of T , written rect(T ), is the filled partition shape obtained via
the sliding process described above.
The Ferrers diagrams of most partitions have several outer cor-

ners, so in principle the rectification of a given skew tableau could
depend on the particular outer corners chosen in the sliding process.
That is, rect(T ) might not be well defined. We will soon show rect(T )
is well defined, but for now we will settle for an example of the sliding
process in action.
Example 10.4. Compute the rectification of the semistandard skew
tableau T of shape (5, 3, 3, 3, 1)/(2, 1, 1) in Figure 10.1.
Solution. If we choose the top box in the left column of µ and perform
the sliding operation, we obtain the objects in Figure 10.2. Here we
have colored the blank box light gray at each step. In Figure 10.3 we
see each of the intermediate tableaux which occur in the computation
of rect(T ), for a particular sequence of choices of outer corner. In each
tableau, we have shaded the boxes in the sliding path that produced
that tableau. □
2
1 5 6
4 4
2 3
1 1 3
Figure 10.1. The semistandard skew tableau T in Example 10.4
2 2
1 5 6 5 6 2 5 6
4 4 1 4 4 1 4 4
2 3 2 3 2 3
1 1 3 1 1 3 1 1 3
Figure 10.2. The first sequence of slides in Example 10.4
2 5 6 2 5 6
1 4 4 1 4 4
2 3 2 3
1 1 3 1 1 3
5 6 5 6
2 4 4 4 4
1 2 3 2 2 3
1 1 3 1 1 1 3
Figure 10.3. All sequences of slides in Example 10.4

4
2 2
W
Figure 10.4. The semistandard tableau W in Example 10.5
4 2 2
2 4 2
2 2 4
S1 S2 S3
Figure 10.5. The possible semistandard skew tableaux in

Example 10.5
Example 10.5. Find all semistandard skew tableaux S of shape

(3, 2, 1)/(2, 1) for which rect(S) is the semistandard tableau W in
Figure 10.4.
Solution. Since the entries of S must be 2, 2, and 4, the only possi-

bilities are the semistandard skew tableaux S1 , S2 , and S3 in Figure
10.5. Since tiles never slide up in jeu de taquin, and the tile con-
taining 4 in S3 is in the first row, rect(S3 ) ≠ W . But we can check
rect(S1 ) = rect(S2 ) = W . □
Although we already have a way of computing products of se-

mistandard tableaux which seems to model (10.1), we can use jeu de
taquin to define a second product on the set of semistandard tableaux.
Definition 10.6. Suppose T1 and T2 are semistandard tableaux. Let
T be the semistandard skew tableau we obtain by identifying the lower
right corner of T1 with the upper left corner of T2 . (See Figure 10.6.)
We define T1 ∗j T2 by setting T1 ∗j T2 = rect(T ), and we call ∗j the
jeu de taquin product.
Example 10.7. Compute all of the products T1 ∗j T2 in which T1
and T2 are semistandard tableaux of shape (2, 1) with entries in [3].
Solution. After 64 applications of the jeu de taquin rectification al-

gorithm, we obtain the results in Table 10.2. □
T1
T2
Figure 10.6. Identifying the corners of tableaux T1 and T2

to compute T1 ∗j T2
Table 10.2. All jeu de taquin products of semistandard

tableaux of shape (2, 1) with entries in [3]
2 3 2 3 2 3 3 3
∗j 11 11 12 12 13 13 22 23
2 22 23 22 23 22 23 23 23
11 1111 1111 1112 1112 1113 1113 1122 1123
3 3 33 3 33 3 33 33 33
11 2 1111 2 1112 2 1113 1122 1123
1111 1112 1113
2 222 223 22 223 22 22 23 23
12 111 111 1122 112 1123 1133 1222 1223
3 3 3 3 3 3 3 33 33
12 22 23 2 23 2 2 1222 1223
111 111 1122 112 1123 1133
2 3 233 3 233 3 23 233 23
13 22 111 22 112 22 1133 122 1233
111 113 113
3 3 333 3 333 3 33 333 33
13 23 111 23 112 23 1133 122 1233
111 112 113
3 3 3 3 3 3 3 33 33
22 22 22 2 23 2 2 2222 2223
112 113 1222 122 1223 1233
3 33 3 3 3 3 3 333 33
23 22 23 23 23 23 2 222 2233
11 113 122 123 123 1333
Remarkably, ∗r and ∗j agree for all of the pairs of tableaux in

Examples 10.2 and 10.7! This might inspire you to compare ∗r and
∗j on a few larger examples, where you will see they also agree. We
will soon prove that for any semistandard tableaux T1 , and T2 , we
have T1 ∗r T2 = T1 ∗j T2 .
10.2. Knuth Equivalence

The product ∗r is a strange hybrid: it begins and ends with semistan-
dard tableaux, but it uses words in an essential way in the middle.
To study ∗r , we imagine defining it entirely in terms of words. That
is, suppose we have semistandard tableaux T1 and T2 , and we want
to compute T1 ∗r T2 . In addition, suppose we replace T1 and T2 with
their reading words word(T1 ) and word(T2 ). In the simplest of all
worlds, word(T1 ∗r T2 ) would be the concatenation of word(T1 ) and
word(T2 ). Of course, if there is any bumping at all involved in the
computation of T1 ∗r T2 , then this will not be the case. So we set our
sights a little lower: we try to define an equivalence relation on words
with the property that the bumping that happens in the course of the
RSK insertion algorithm does not change the equivalence class of the
associated reading word, even though it changes the word itself.
To see what our prospective equivalence relation should look like,
suppose word(T1 ) has just one entry. That is, suppose we have a
semistandard tableau T , into which we are row-inserting a constant
c. We begin by putting c at the right end of word(T ) and designating
this entry to be the traveler. As we proceed different entries will be
the traveler, but the traveler will always be the entry we are inserting
into the next row. We now view row-insertion as a sequence of steps,
each of which consists of either swapping the traveler with the entry
immediately to its left, or letting the traveler pass its title to the entry
immediately to its left. To see how to execute these steps, suppose we
are inserting c into a row whose entries are a1 , a2 , . . . , ak , and consider
what must occur to accomplish our row-insertion.
If c < a1 , then c must trade places with each of ak , ak−1 , . . . , a2 in
turn, until we reach a1 , c, a2 , . . . , ak . Then a1 becomes the traveler,
and c has bumped a1 .
10.2. Knuth Equivalence 279
If c < a2 but c ≥ a1 , then c must first trade places with each

of ak , ak−1 , . . . , a3 , to reach a1 , a2 , c, a3 , . . . , ak . Then a2 becomes the
traveler, a2 trades places with a1 , and c has bumped a2 .
In general, if c < aj but c ≥ aj−1 , then c must first trade places
with each of ak , ak−1 , . . . , aj+1 . Then aj becomes the traveler, it trades
places with each of aj−1 , aj−2 , . . . , a1 , and c has bumped aj .
We are using two types of steps in this process. In one type of
step we have consecutive entries x and z, and the entry y immediately
to their left satisfies x < y ≤ z. In this case, x and z can trade places.
These are the swaps we use to move c until it is just to the right of
the entry it will bump, and in these cases we have x = c, as well as
y = ai−1 and z = ai for some i ≥ j. In the other type of step we have
consecutive entries x and z, and the entry y immediately to their
right satisfies x ≤ y < z. In this case, x and z can also trade places.
These are the swaps we use to move aj (the entry bumped by c) until
it is to the left of all other entries in its row, and in these cases we
have z = aj , as well as x = ai and y = ai+1 for some i < j. These two
transformations on words are central to our equivalence relation, so
we give them names.
Definition 10.8. Two sequences of positive integers (that is, two

words) are related by an elementary Knuth transformation whenever
one of the following holds.
(1) There are entries x, y, and z such that x < y ≤ z, and there
are words w1 and w2 such that one sequence has the form
w1 yxzw2 ,
and the other sequence has the form
w1 yzxw2 .
That is, the two sequences are identical, except that the
consecutive entries x and z have been swapped.
(2) There are entries x, y, and z such that x ≤ y < z, and there
are words w1 and w2 such that one sequence has the form
w1 xzyw2 ,
and the other sequence has the form
w1 zxyw2 .
That is, the two sequences are identical, except that the
consecutive entries x and z have been swapped.
As we see in our next example, we can use the elementary Knuth

transformations to turn various sets of words, including the set of
permutations of a given length, into graphs.
Example 10.9. Let G be the graph whose vertices are the permu-
tations of [4], in which two vertices are connected whenever they are
related by an elementary Knuth transformation. Sketch G.
Solution. For any permutation π, we can apply an elementary Knuth

transformation to π whenever we have an index j such that either
π(j − 1) is between π(j) and π(j + 1) (numerically), or π(j + 1) is
between π(j) and π(j − 1). For instance, if π = 1234, then no such
indices exist, so no elementary Knuth transformation applies, and π
is not adjacent to any other permutation. In contrast, we can apply
an elementary Knuth transformation to 3142 in one way: the 3 is
immediately to the left of the adjacent entries 1 and 4. Therefore,
3142 is adjacent to 3412 in G. Similarly, we can apply an elementary
Knuth transformation to 1432 in one way: the 3 is immediately to the
right of the adjacent entries 1 and 4. Therefore, 1432 is adjacent to
4132 in G. Examining each permutation of [4] in this way, we obtain
the graph in Figure 10.7. Some of the vertices in this graph are in
boxes; we will explain the significance of these boxes shortly. □
By construction, row insertion can be accomplished with a se-

quence of elementary Knuth transformations, and we would like all
of the intermediate words in this process to be equivalent. This de-
termines our equivalence relation.
Definition 10.10. We say two words w1 and w2 are Knuth equiva-

lent, and we write w1 ∼K w2 , whenever there exist words v1 , . . . , vn
such that v1 = w1 , vn = w2 , and vj and vj+1 are related by an elemen-
tary Knuth transformation for all j with 1 ≤ j ≤ n − 1.
1234
2413 2143
1342 1324 3124
2134 2314 2341
1243 1423 4123
3421 3241 3214
4312 4132 1432
2431 4231 4213
3142 3412
4321
Figure 10.7. The graph G in Example 10.9
In Example 10.9, two permutations were Knuth equivalent ex-

actly when they were in the same connected component of G. In gen-
eral, our definition of Knuth equivalence guarantees that the Knuth
equivalence classes for a set of words are exactly the connected com-
ponents of the corresponding graph on those words.
We can now show Knuth equivalence has the property we orig-
inally sought: when we view ∗r as a product on words, the reading
word of a product of tableaux is Knuth equivalent to the concatena-
tion of the reading words of the tableaux themselves.
Theorem 10.11. For any semistandard tableau T and any word w =

c1 c2 ⋅ ⋅ ⋅ cn , we have word(T )w ∼K word(rcn (⋅ ⋅ ⋅ rc2 (rc1 (T )))).
Proof. If w has just one entry c1 , then, as we noted in the discus-

sion leading up to Definition 10.8, we can transform word(T )c1 into
word(rc1 (T )) with a sequence of elementary Knuth transformations.
Therefore the result holds in this case. Now suppose w has two or
more entries; we argue by induction on the length of w.
By induction we have
word(T )c1 ⋅ ⋅ ⋅ cn−1 ∼K word(rcn−1 (⋅ ⋅ ⋅ rc1 (T ))),
so there is a sequence of elementary Knuth transformations which
turns
word(T )c1 ⋅ ⋅ ⋅ cn−1 into word(rcn−1 (⋅ ⋅ ⋅ rc1 (T ))).
The same sequence of elementary Knuth transformations turns
word(T )c1 ⋅ ⋅ ⋅ cn−1 cn into word(rcn−1 (⋅ ⋅ ⋅ rc1 (T )))cn ,
so
word(T )c1 ⋅ ⋅ ⋅ cn−1 cn ∼K word(rcn−1 (⋅ ⋅ ⋅ rc1 (T )))cn .
But rcn−1 (⋅ ⋅ ⋅ rc1 (T )) is also a semistandard tableau, so by the first
part of this proof we have
word(rcn−1 (⋅ ⋅ ⋅ rc1 (T )))cn ∼K word(rcn (rcn−1 (⋅ ⋅ ⋅ rc1 (T )))).
Now the result follows. □
Corollary 10.12. For any semistandard tableaux T1 and T2 we have

word(T1 ) word(T2 ) ∼K word(T1 ∗r T2 ).
Proof. Set w = word(T2 ) in Theorem 10.11. □
Theorem 10.11 also tells us something interesting about the rela-

tionship between a word w and the reading word of the semistandard
tableaux P (w).
Corollary 10.13. For any word w we have word(P (w)) ∼K w.
Proof. Let T be the empty tableau in Theorem 10.11. □

a x b1 bk
c1 ··· cm d1 dk
Figure 10.8. The two relevant rows of the tableau T1 in the

proof of Theorem 10.14
a b1 bk
c1 ··· c m x d1 dk
Figure 10.9. The two relevant rows of the tableau T2 in the

proof of Theorem 10.14
Corollary 10.12 suggests we should also consider the relationship

between word(T1 ∗j T2 ) and word(T1 ) word(T2 ), which leads to the
following results.
Theorem 10.14. For any semistandard tableau T of skew shape,

word(T ) ∼K word(rect(T )).
Proof. By induction it is sufficient to show that if T1 is a filling we

obtain in the process of using using jeu de taquin slides to transform
T into rect(T ), and we obtain T2 from T1 by making a single slide,
then word(T1 ) and word(T2 ) are Knuth equivalent.
To start, note that if we obtain T2 from T1 by making a horizontal
slide, then word(T1 ) = word(T2 ), so these words are Knuth equivalent.
Now suppose we obtain T2 from T1 by making a vertical slide.
Then we can assume the two rows involved in the slide transformed
from the configuration in Figure 10.8 to the configuration in Figure
10.9. In particular, if x is the entry that slides, then there are words
w1 and w2 , and entries b1 ≤ b2 ≤ ⋅ ⋅ ⋅ ≤ bk and c1 ≤ c2 ≤ ⋅ ⋅ ⋅ ≤ cm , such
that
word(T1 ) = w1 xb1 ⋅ ⋅ ⋅ bk c1 ⋅ ⋅ ⋅ cm w2
and
word(T2 ) = w1 b1 ⋅ ⋅ ⋅ bk c1 ⋅ ⋅ ⋅ cm xw2 .
We describe how to use elementary Knuth transformations to trans-
form word(T1 ) into word(T2 ).
First observe that if k ≥ 1 (and we take x to be b0 ), then we have

c1 < bk−1 ≤ bk , so we can transform word(T1 ) into
w1 xb1 ⋅ ⋅ ⋅ bk−1 c1 bk c2 ⋅ ⋅ ⋅ cm w2 .
We can repeat this process until c1 is immediately to the right of x. In
the same way, we can move each of c2 , . . . , cm past each of b1 , . . . , bk ,
to obtain
w1 xc1 ⋅ ⋅ ⋅ cm b1 ⋅ ⋅ ⋅ bk w2 .
Now observe that c1 ≤ c2 < x, so we can transform our word into
w1 c1 xc2 ⋅ ⋅ ⋅ cm b1 ⋅ ⋅ ⋅ bk w2 .
Similarly, for all j ≤ m, we have cj−1 ≤ cj < x, so we can transform
our word into
w1 c1 ⋅ ⋅ ⋅ cm−1 xcm b1 ⋅ ⋅ ⋅ bk w2 .
If m ≥ 1, then w1 is nonempty, so suppose the rightmost entry of w1
is a and w1 = w1′ a. Then for all j ≤ m we have cj−1 ≤ cj < a, and as a
result, we can transform our word into
w1′ c1 ⋅ ⋅ ⋅ cm−1 axcm b1 ⋅ ⋅ ⋅ bk w2 .
Now we have cm < a ≤ x, so we can transform our word into
w1′ c1 ⋅ ⋅ ⋅ cm−1 acm xb1 ⋅ ⋅ ⋅ bk w2 .
Then we can return a to its original position, obtaining
w1 c1 ⋅ ⋅ ⋅ cm xb1 ⋅ ⋅ ⋅ bk w2 .
We have now moved x past c1 , . . . , cm , but we will need even more

alphabetical acrobatics to get x past b1 , . . . , bk . To begin this process,
note that w2 must have at least k entries, so there are numbers d1 ≤
d2 ≤ ⋅ ⋅ ⋅ ≤ dk and a word w2′ so that w2 = d1 ⋅ ⋅ ⋅ dk w2′ . Now for each
j ≤ k, we have d1 < bj−1 ≤ bj , so we can transform our word into
w1 c1 ⋅ ⋅ ⋅ cm xb1 d1 b2 ⋅ ⋅ ⋅ bk d2 ⋅ ⋅ ⋅ dk w2′ .
Similarly, we can move each of d2 , . . . , dk−1 to the left, obtaining
w1 c1 ⋅ ⋅ ⋅ cm xb1 d1 b2 d2 b3 d3 ⋅ ⋅ ⋅ bk dk w2′ .
Now we have x ≤ d1 < b1 , so we can transform our word into
w1 c1 ⋅ ⋅ ⋅ cm b1 xd1 b2 d2 b3 d3 ⋅ ⋅ ⋅ bk dk w2′ .
10.3. The Relationship Between P and word 285
This is an example of a general situation: whenever we have the string

xd1 d2 ⋅ ⋅ ⋅ dk bk+1 dk+1 in our word, if j ≤ k, then dj ≤ dj+1 < bk+1 , so
we can obtain xbk+1 d1 ⋅ ⋅ ⋅ dk dk+1 . By doing this repeatedly, we can
eventually transform our word into
w1 c1 ⋅ ⋅ ⋅ cm b1 ⋅ ⋅ ⋅ bk xd1 ⋅ ⋅ ⋅ dk w2′ .
Finally, we can reverse our initial steps to obtain
w1 b1 ⋅ ⋅ ⋅ bk c1 ⋅ ⋅ ⋅ cm xw2 ,
which is word(T2 ). □
Corollary 10.15. For any semistandard tableaux T1 and T2 we have

word(T1 ) word(T2 ) ∼K word(T1 ∗j T2 ).
Proof. The skew tableau we build from T1 and T2 by identifying

the lower right corner of T1 with the upper left corner of T2 has
reading word word(T1 ) word(T2 ), and the rectification of this tableau
is T1 ∗j T2 by definition, so the result follows from Theorem 10.14. □
10.3. The Relationship Between P and word

The results at the end of Section 10.2, and Corollary 10.13 in particu-
lar, suggest we should take a more systematic look at the relationship
between P and word. To do this, we fix throughout this section a
nonempty multiset A of positive integers. Let W (A) be the set of
permutations of A, and let T (A) be the set of semistandard tableaux
whose multiset of entries is A. In Figure 10.10 we have W (A) (on
the left) and T (A) (on the right) for A = {1, 2, 2, 3}.
As we have seen, we have a function P ∶ W (A) → T (A) given by
the RSK insertion algorithm, and we have a function word ∶ T (A) →
W (A) which computes reading words. In addition, we have the equiv-
alence relation ∼K on W (A). In Figure 10.11 we have the maps P
and word for A = {1, 2, 2, 3}. As Figure 10.11 suggests, and as we
show next, P is constant on the equivalence classes of ∼K .
Theorem 10.16. For all w1 , w2 ∈ W (A), if w1 ∼K w2 , then P (w1 ) =

P (w2 ).
1223 1 2 2 3
2132 2 3
2312 1 2
1232
3
1322
1 2 2
3122
2231
2
2213
1 2 3
2123
2321 3
3221 2
3212 1 2
Figure 10.10. The sets W (A) and T (A) for A = {1, 2, 2, 3}
Proof. By induction, we just need to show that if w1 and w2 are

related by an elementary Knuth transformation, then P (w1 ) = P (w2 ).
Suppose there are words u and v, and specific entries x < y ≤ z,
such that w1 = uyzxv and w2 = uyxzv. Since w1 and w2 are identical
to the left of x and to the right of z, it is sufficient to show that for
any semistandard tableau T , we have rx (rz (ry (T ))) = rz (rx (ry (T ))).
(In our case, T = P (u).) To do this, first note that by Lemma 8.8,
the bumping path of z in ry (T ) is strictly to the right of the bumping
path of y in T . On the other hand, arguing as in the proof of Lemma
8.8, we can show that the bumping path of x in ry (T ) is weakly
to the left of the bumping path of y in T . Therefore, the bumping
paths of z and x in ry (T ) have no common boxes, so we must have
rx (rz (ry (T ))) = rz (rx (ry (T ))).
P
1223 1 2 2 3
word
2132 2 3
2312 1 2
1232
3
1322
1 2 2
3122
2231
2
2213
1 2 3
2123
2321 3
3221 2
3212 1 2
Figure 10.11. The maps P and word, and the equivalence

classes for ∼K , for A = {1, 2, 2, 3}
Now suppose there are words u and v, and entries x ≤ y < z, such
that w1 = uxzyv and w2 = uzxyv. As before, it is sufficient to show
ry (rz (rx (T ))) = ry (rx (rz (T ))) for any semistandard tableau T . We
can check this directly when T is empty, so suppose T has at least
one row. We argue by induction on the number of rows in T .
The rest of the proof is an examination of several cases, most
of which we leave to the reader. We outline the cases involved and
handle two of them in detail.
Our first case division involves x and y: either x = y or x < y.
Suppose x = y, and the first row of T has entries a1 , . . . , aj+k+l ,
where
a1 ≤ ⋅ ⋅ ⋅ ≤ aj ≤ x = y < aj+1 ≤ ⋅ ⋅ ⋅ ≤ aj+k ≤ z < aj+k+1 ≤ ⋅ ⋅ ⋅ ≤ aj+k+l .

In this case we have several subcases, according to the values of k and

l. For example, suppose k ≥ 2 and l ≥ 1. When we insert x, then z,
and then y into the bottom row, our new bottom row is
a1 , . . . , aj , x, y, aj+3 , . . . , aj+k , z, aj+k+2 , . . . , aj+k+l ,
and we need to insert aj+1 , then aj+k+1 , and then aj+2 into the next
row. On the other hand, when we insert z, then x, and then y into
the bottom row, our new bottom row is again
a1 , . . . , aj , x, y, aj+3 , . . . , aj+k , z, aj+k+2 , . . . , aj+k+l ,
and we need to insert aj+k+1 , then aj+1 , and then aj+2 into the next
row. Since we know aj+1 ≤ aj+2 < aj+k+1 , by induction we will obtain
the same tableau in both cases.
We leave the remaining cases in which x = y to the reader.
Now suppose x < y, and the entries of the first row of T are
a1 , . . . , aj+k+l+n , where
a1 ≤ ⋅ ⋅ ⋅ ≤ aj ≤ x,
x < aj+1 ≤ ⋅ ⋅ ⋅ ≤ aj+k ≤ y,

y < aj+k+1 ≤ ⋅ ⋅ ⋅ ≤ aj+k+l ≤ z,
and
z < aj+k+l+1 ≤ ⋅ ⋅ ⋅ ≤ aj+k+l+n .
In this case we have several more subcases, according to the values of
k, l, and n. For example, suppose k = l = 0 and n ≥ 2. When we insert
x, then z, and then y into the bottom row, our new bottom row is
a1 , . . . , aj , x, y, aj+3 , . . . , aj+n ,
and we need to insert aj+1 , then aj+2 , and z into the next row. On
the other hand, when we insert z, then x, and then y into the bottom
row, our new bottom row is again
a1 , . . . , aj , x, y, aj+3 , . . . , aj+n ,
and this time we need to insert aj+1 , then z, and then aj+2 . Since we
know z < aj+1 ≤ aj+2 , by the other case in this proof we will obtain
the same semistandard tableau.
We leave the remaining cases in which x < y to the reader. □
Theorem 10.16 tells us the map P induces a function from the set
W (A)/ ∼K of equivalence classes of ∼K to T (A). Abusing notation,
we write P to denote this function. We also note that by composing
word with the function mapping a word in W (A) to its Knuth equiv-
alence class, we get a function from T (A) to W (A)/ ∼K . We abuse
notation here, too, writing word to denote this function.
Corollary 10.13 tells us the composition word ○P is the identity
on W (A)/ ∼K , and Figure 10.11 suggests P ○ word is the identity on
T (A). Next we show this holds in general.
Theorem 10.17. For all T ∈ T (A), we have P (word(T )) = T .
Proof. For each j, let rj be the jth row of T , so word(T )=rl rl−1 ⋅ ⋅ ⋅ r1 ,
where l is the number of rows in T . Since the entries of rl are weakly
increasing, P (rl ) is a copy of the top row of T . Because T is column-
strict, each entry of rl−1 is strictly less than the corresponding entry of
rl . And since T is weakly increasing across rows, each entry of rl−1 is
greater than or equal to the entry to its right in rl−1 . Therefore, each
entry of rl−1 bumps the corresponding entry of rl up to the second
row, and P (rl rl−1 ) is a copy of the top two rows of T . This occurs
for each row: for each j, P (rl ⋅ ⋅ ⋅ rl−j ) is a copy of the top j + 1 rows
of T . In particular, P (word(T )) = P (rl ⋅ ⋅ ⋅ r1 ) = T , which is what we
wanted to prove. □
Now we know exactly how the maps P and word are related.
Corollary 10.18. The functions P ∶ W (A)/ ∼K → T (A) and

word ∶ T (A) → W (A)/ ∼K are inverse bijections, and in particular
they are both one-to-one and onto.
Proof. This follows from Corollary 10.13 and Theorem 10.17. □
We can now show that rectification of skew tableaux is well de-

fined, which means the jeu de taquin product ∗j is also well defined.
Corollary 10.19. If T is a semistandard tableau of skew shape, then

rect(T ) does not depend on the outer corners we choose in its com-
putation.
Proof. Suppose T1 and T2 are semistandard tableaux of partition

shape that we obtain by rectifying T , possibly choosing different outer
corners along the way. By Theorem 10.14 we know word(T1 ) and
word(T2 ) are both Knuth equivalent to word(T ), and are therefore
Knuth equivalent to each other. Since word is one-to-one, we must
have T1 = T2 . □
Corollary 10.20. Suppose T1 and T2 are semistandard tableaux of

partition shape. Then T1 ∗j T2 does not depend on the outer corners
we choose in its computation.
Proof. This follows from Corollary 10.19, since T1 ∗j T2 is the recti-

fication of a specific semistandard tableau of skew shape. □
We can now verify what we observed in Examples 10.2 and 10.7:

the products ∗r and ∗j are the same.
Corollary 10.21. If T1 and T2 are semistandard tableaux, then

T1 ∗r T2 = T1 ∗j T2 .
Proof. In Corollary 10.12 we saw word(T1 ∗r T2 ) is Knuth equivalent

to word(T1 ) word(T2 ), and in Corollary 10.15 we saw word(T1 ∗j T2 )
is Knuth equivalent to word(T1 ) word(T2 ). Therefore, word(T1 ∗r T2 )
and word(T1 ∗j T2 ) are Knuth equivalent to each other. Since word
is one-to-one, we must have T1 ∗r T2 = T1 ∗j T2 . □
In view of Corollary 10.21, we just write ∗ to denote the common

product ∗r = ∗j . It will be useful to note how this product interacts
with P and the concatenation of words.
Corollary 10.22. Suppose π = αβ, where α = a1 ⋅ ⋅ ⋅ am and β =

b1 ⋅ ⋅ ⋅ bn . Then P (π) = P (α) ∗ P (β).
Proof. By Corollaries 10.12 and 10.13 we have

word(P (α) ∗ P (β)) ∼K word P (α) word P (β)
∼K αβ.
Now if we apply P and use Theorem 10.17, then we find P (α)∗P (β) =
P (αβ) = P (π), which is what we wanted to prove. □
10.4. The Littlewood–Richardson Rule 291
10.4. The Littlewood–Richardson Rule

We now have two descriptions of a product on semistandard tableaux
which appears to model the multiplication of terms in Schur functions.
That is, we know there are constants cλµ,ν with
(10.4) sµ sν = ∑ cλµ,ν sλ ,
λ⊢∣µ∣+∣ν∣
and it appears our product ∗ maps ordered pairs of semistandard

tableaux indexing terms on the left side of (10.4) to the semistandard
tableaux indexing terms on the right side of (10.4). To prove this,
we will study how we can factor terms on the right into products of
terms on the left. Here some new notation will be useful.
Definition 10.23. Suppose λ, µ, and ν are partitions and T is a
semistandard tableau of shape λ. We write T (µ, ν, T ) to denote the
set of ordered pairs (U, V ) for which U is a semistandard tableau of
shape µ, V is a semistandard tableau of shape ν, and U ∗ V = T .
If we can show ∣T (µ, ν, T )∣ depends only on µ, ν, and λ, and not on

our specific choice of T , then we will know ∗ models the multiplication
in (10.4), and we will have cλµ,ν = ∣T (µ, ν, T )∣ for every semistandard
tableau T of shape λ.
We also have a rectification algorithm, which appears to convert
terms in a skew Schur function to terms in a classical Schur function.
That is, we know there are constants dλµ,ν with
(10.5) sλ/µ = ∑ dλµ,ν sν ,
ν⊢∣λ∣−∣µ∣
and it appears rectification of skew tableaux maps the semistandard

skew tableaux indexing terms on the left side of (10.5) to the se-
mistandard tableaux indexing terms on the right side of (10.5). To
prove this, we will study how we can “unrectify” a given semistan-
dard tableau into a skew tableau of a given target shape, which will
correspond to transforming tableaux indexing terms on the right side
of (10.5) into tableaux indexing terms on the left side of (10.5). Here
we also have some useful notation.
Definition 10.24. Suppose λ, µ, and ν are partitions and W is a
semistandard tableau of shape ν. We write S(λ/µ, W ) to denote the
set of semistandard tableaux S of shape λ/µ with rect(S) = W . Note

that if µ ⊆/ λ, then S(λ/µ, W ) is empty.
If we can show ∣S(λ/µ, W )∣ depends on µ, ν, and λ but not on our

specific choice of W , then we will know rectification models the rela-
tionship between the terms on the left side of (10.5) and the terms on
the right, and we will have dλµ,ν = ∣S(λ/µ, W )∣ for every semistandard
tableaux W of shape ν.
If we can do everything we have described so far, then we will be
able to show cλµ,ν = dλµ,ν by giving a bijection between T (µ, ν, T ) and
S(λ/µ, W ). As it turns out, having such a bijection will give us the
other results we need almost for free, so we start there.
Stripped of its technical details, our bijection is fairly easy to
describe. If we have semistandard tableaux U and V with U ∗ V = T ,
then the associated semistandard skew tableau S ∈ S(λ/µ, W ) will
be the recording tableau we obtain from W when we insert word(V )
into U . In the other direction, if we have S ∈ S(λ/µ, W ), then we can
use S to remove the entries of V from T , leaving U .
To describe our maps in more detail, suppose µ, ν, and λ are par-
titions, T is a semistandard tableau of shape λ, and W is a semistan-
dard tableau of shape ν. Given (U, V ) ∈ T (µ, ν, T ), we can construct
s s2 ⋅ ⋅ ⋅ s∣ν∣
a corresponding S ∈ S(λ/µ, W ) as follows. Let σ = [ 1 ]
b1 b2 ⋅ ⋅ ⋅ b∣ν∣
be the generalized permutation with P (σ) = V and Q(σ) = W . Con-
struct rb∣ν∣ (rb∣ν∣−1 (⋅ ⋅ ⋅ rb1 (U ))), and let S be the semistandard skew
tableau we obtain by inserting s1 , . . . , s∣ν∣ into the new boxes of λ/µ
as they are added.
Definition 10.25. Suppose µ, ν, and λ are partitions, T is a se-
mistandard tableau of shape λ, and W is a semistandard tableau of
shape ν. For each (U, V ) ∈ T (µ, ν, T ), we write Ψµ,ν,λ,T,W (U, V ) to
denote the semistandard skew tableau S we constructed above. When
there is no possibility of confusion, we abbreviate Ψ = Ψµ,ν,λ,T,W .
Example 10.26. Suppose µ = ν = (2, 1), λ = (3, 2, 1), and T and
W are the semistandard tableaux in Figure 10.12. From Examples
10.2 and 10.7 we know that if U1 , U2 , V1 , and V2 are as in Figure
10.13, then T (µ, ν, T ) = {(U1 , V1 ), (U2 , V2 )}. And from Example 10.5
3
4
2 3
2 2
1 1 2
W
T
Figure 10.12. The semistandard tableaux T and W in Ex-
amples 10.26 and 10.28
3 2 3
1 3 1 2 1 2
U1 V1 U2 = V2
Figure 10.13. The semistandard tableaux U1 , U2 , V1 , and

V2 in Examples 10.26 and 10.28
4 2
2 4
2 2
S1 S2
Figure 10.14. The semistandard skew tableaux S1 and S2

in Examples 10.26 and 10.28
we know that if S1 and S2 are as in Figure 10.14, then S(λ/µ, W ) =

{S1 , S2 }. Compute Ψ(U1 , V1 ) and Ψ(U2 , V2 ).
Solution. To compute Ψ(U1 , V1 ), we first invert R to find that if σ1

is a generalized permutation with P (σ1 ) = V1 and Q(σ1 ) = W , then
2 2 4
σ1 = [ ]. When we insert 2, 2, and 1 into U1 , our intermedi-
2 2 1
ate tableaux are as in Figure 10.15. The corresponding intermediate
tableaux for 2, 2, and 4 are as in Figure 10.16, so Ψ(U1 , V1 ) = S1 .
Similarly, we find that if σ2 is a generalized permutation with
2 2 4
P (σ2 ) = V2 and Q(σ2 ) = W , then σ2 = [ ]. Using this, we find
1 3 2
Ψ(U2 , V2 ) = S2 . □
3
3 3 3 3 2 3
1 2 1 2 2 1 1 2
Figure 10.15. The intermediate tableaux we obtain when we

insert 2, 2, and 1 into U1
4
2 2 2
2 2
Figure 10.16. The intermediate steps in the construction of

Ψ(U1 , V1 ) in Example 10.26
We will soon show Ψµ,ν,λ,T,W is a function from T (µ, ν, T ) to

S(λ/µ, W ), at which point we will want to understand its inverse.
To describe this inverse, suppose µ, ν, and λ are partitions, T is a
semistandard tableau of shape λ, and W is a semistandard tableau of
shape ν. For each S ∈ S(λ/µ, W ), we can construct a corresponding
(U, V ) ∈ T (µ, ν, T ) as follows. Fill in the boxes of µ in S with en-
tries which are smaller than all entries of S to create a semistandard
u ⋅ ⋅ ⋅ u∣µ∣ t1 ⋅ ⋅ ⋅ t∣λ∣−∣µ∣
tableau U S of shape λ. Let π = [ 1 ] be
a1 ⋅ ⋅ ⋅ a∣µ∣ c1 ⋅ ⋅ ⋅ c∣λ∣−∣µ∣
the generalized permutation with P (π) = T and Q(π) = U S. Then
we set U = P (a1 ⋅ ⋅ ⋅ a∣µ∣ ) and V = P (c1 ⋅ ⋅ ⋅ c∣λ∣−∣µ∣ ).
In principle, our construction of U might depend on the par-
ticular way we fill µ to obtain U S. To see U is actually indepen-
dent of this choice, suppose we fill µ with entries smaller than all
entries of S in another way to obtain a semistandard tableau U ′ S
of shape λ. Then we know there is a generalized permutation π ′ =
v ⋅ ⋅ ⋅ v∣µ∣ y1 ⋅ ⋅ ⋅ y∣λ∣−∣µ∣
[ 1 ] with P (π ′ ) = T and Q(π ′ ) = U ′ S.
d1 ⋅ ⋅ ⋅ d∣µ∣ f1 ⋅ ⋅ ⋅ f∣λ∣−∣µ∣
Now consider what happens when we use U S and U ′ S to extract en-
tries from T . Because all of the entries of S are larger than all of the
entries of U and U ′ , we will first remove from T the boxes in λ/µ.
These are the same for both U S and U ′ S, so yj = tj and fj = cj for all
j. Once we have removed the boxes in λ/µ, we will be left with the
same tableau for both U S and U ′ S. For U S this will be P (a1 ⋅ ⋅ ⋅ a∣µ∣ )
and for U ′ S this will be P (d1 ⋅ ⋅ ⋅ d∣µ∣ ), so these tableau must be equal.
Therefore, U is independent of the filling we use to construct U S.
Definition 10.27. Suppose µ, ν, and λ are partitions, T is a se-

mistandard tableau of shape λ, and W is a semistandard tableau
of shape ν. For each S ∈ S(λ/µ, W ), we write Ωµ,ν,λ,T,W (S) to de-
note the ordered pair (U, V ) of semistandard tableaux we constructed
above. When there is no possibility of confusion, we abbreviate
Ω = Ωµ,ν,λ,T,W .
Example 10.28. Suppose µ = ν = (2, 1), λ = (3, 2, 1), and T and W

are the semistandard tableaux in Figure 10.12. Compute Ω(S1 ) and
Ω(S2 ), where S1 and S2 are as in Figure 10.14.
4
2 2
1 1 2
Figure 10.17. The semistandard tableau U S in our solution

to Example 10.28
Solution. To compute Ω(S1 ), we first construct U S as in Figure 10.17.

If π1 is a generalized permutation with P (π1 ) = T and Q(π1 ) = U S,
1 1 2 2 2 4
then π1 = [ ]. Now Ω(S1 ) = (P (331), P (221)) =
3 3 1 2 2 1
(U1 , V1 ), where U1 and V1 are as in Figure 10.13. Similarly, we find
Ω(S2 ) = (U2 , V2 ). □
Now that we have candidates for our bijections between T (µ, ν, T )

and S(λ/µ, W ), we need to show they have the correct ranges. This
will require two technical lemmas. The first involves the relationship
between a given word and the word we obtain by removing its k
smallest entries. In this context, we say an entry a in a given word is
smaller than an entry b in that word whenever a < b, or a = b and a
is to the left of b.
Lemma 10.29. Suppose w1 and w2 are Knuth equivalent words, and

w1′ and w2′ are the words we obtain by removing the k smallest entries
from w1 and the k smallest entries from w2 . Then w1′ and w2′ are
Knuth equivalent.
Proof. By induction it is enough to show the result holds when k = 1.

Suppose w1 and w2 are Knuth equivalent, let b be their smallest
entry, and let w1′ and w2′ be w1 and w2 with b removed. Since w1
and w2 are Knuth equivalent, we have a sequence of words w1 =
v1 , v2 , . . . , vn−1 , vn = w2 such that for each j, the words vj and vj+1 are
related by an elementary Knuth transformation. For each j, let vj′ be
vj with b removed. If vj and vj+1 are related by an elementary Knuth
transformation which does not involve b, then vj′ and vj+1 ′
are related
by the same elementary Knuth transformation. If vj and vj+1 are
related by an elementary Knuth transformation which does involve
b, then b must play the role of x. Therefore, vj′ = vj+1 ′
. Removing
some terms if necessary from v1 , . . . , vn , we see w1 and w2′ are Knuth
′ ′ ′
equivalent. □
a a2 ⋅ ⋅ ⋅ an
Lemma 10.30. Suppose π = [ 1 ] is a generalized per-
b1 b2 ⋅ ⋅ ⋅ bn
mutation and T is a semistandard tableau. Let
U = rbn (rbn−1 (⋅ ⋅ ⋅ rb1 (T ))),
and let S be the semistandard skew tableau we obtain by placing
a1 , . . . , an in the boxes we add to T in the construction of U so that for
each j we place aj in the jth box we add to T . Then rect(S) = Q(π).
Proof. Before we get to the details of the proof, we need to set some
additional notation.
First, for any generalized permutation α, we write bottom(α) to
denote the bottom row of α.
Second, let T − be the filling we obtain by subtracting the largest
entry of T from every entry of T . Note that T − is a semistan-
dard tableau, even though none of its entries are positive. Now
c c2 ⋅ ⋅ ⋅ cm
let σ = [ 1 ] be the generalized permutation with
d1 d2 ⋅ ⋅ ⋅ dm
R(σ) = (T, T − ), so P (σ) = T and Q(σ) = T − . Finally, let τ be
c ⋅ ⋅ ⋅ cm a1 ⋅ ⋅ ⋅ an
the concatenation of σ and π, so τ = [ 1 ].
d1 ⋅ ⋅ ⋅ dm b1 ⋅ ⋅ ⋅ bn
Observe that by construction Q(τ ) is T − with S attached, and we
have P (τ ) = rbn (rbn−1 (⋅ ⋅ ⋅ rb1 (T ))).
To start the proof, recall from Definition 8.20 that τ −1 is the gen-
eralized permutation we obtain by reordering the columns of the array
d ⋅ ⋅ ⋅ dm b1 ⋅ ⋅ ⋅ bn
[ 1 ]. This means it is sufficient to show
c1 ⋅ ⋅ ⋅ cm a1 ⋅ ⋅ ⋅ an
word(Q(π)) ∼K bottom(π −1 ) ∼K word(S) ∼K word(rect(S)),
since the result will then follow from Corollary 10.18.
To prove word(Q(π)) ∼K bottom(π −1 ), first recall from Theorem
8.29 that Q(π) = P (π −1 ). Combining this with Corollary 10.13 and
the fact that P (π −1 ) = P (bottom(π −1 )), we find
word(Q(π)) = word(P (π −1 )) ∼K bottom(π −1 ).
To prove bottom(π −1 ) ∼K word(S), first note that by Corollary

10.13 and Theorem 8.29 we have bottom(τ −1 ) ∼K word(Q(τ )). Now
note that Q(τ ) contains S by construction, the entries of word(Q(τ ))
are c1 , . . . , cm , a1 , . . . , an in some order, the entries of word(S) are
a1 , . . . , an in the same order as they appear in word(Q(τ )), and
cj < ak for all j and k. Therefore, word(S) is word(Q(τ )) with its
m smallest entries removed. Similarly, bottom(π −1 ) is bottom(τ −1 )
with its m smallest entries removed. By Lemma 10.29 we must have
bottom(π −1 ) ∼K word(S).
Finally, the fact that word(S) ∼K word(rect(S)) follows from
Theorem 10.14. □
We can now show Ψµ,ν,λ,T,W is a function from T (µ, ν, T ) to

S(λ/µ, W ) and Ωµ,ν,λ,T,W is a function from S(λ/µ, W ) to T (µ, ν, T ).
Proposition 10.31. Suppose µ, ν, and λ are partitions, T is a se-
shape ν. If (U, V ) ∈ T (µ, ν, T ), then Ψµ,ν,λ,T,W (U, V ) ∈ S(λ/µ, W ).
Proof. Recall that to construct Ψ(U, V ) we first find σ with P (σ) =

s s2 ⋅ ⋅ ⋅ s∣ν∣
V and Q(σ) = W . If σ = [ 1 ], then we construct
b1 b2 ⋅ ⋅ ⋅ b∣ν∣
rb∣ν∣ (⋅ ⋅ ⋅ rb1 (U )), and we let S = Ψ(U, V ) be the semistandard skew

tableau we obtain by inserting s1 , . . . , s∣ν∣ into the new boxes of λ/µ
as they are added.
Since U ∗ V = T has shape λ, by construction Ψ(U, V ) has shape
λ/µ. And Lemma 10.30 tells us rect(Ψ(U, V )) = W , so Ψ(U, V ) ∈
S(λ/µ, W ). □
Proposition 10.32. Suppose µ, ν, and λ are partitions, T is a se-

shape ν. If S ∈ S(λ/µ, W ), then Ωµ,ν,λ,T,W (S) ∈ T (µ, ν, T ).
Proof. Recall that to construct Ω(S) we first fill the boxes in µ

with entries which are smaller than all of the entries in S to form a
semistandard tableau U S of shape λ, and then we find π with P (π) =
u ⋅ ⋅ ⋅ u∣µ∣ t1 ⋅ ⋅ ⋅ t∣λ∣−∣µ∣
T and Q(π) = U S. If π = [ 1 ], then
a1 ⋅ ⋅ ⋅ a∣µ∣ c1 ⋅ ⋅ ⋅ c∣λ∣−∣µ∣
Ω(S) = (U, V ), where U = P (a1 , . . . , a∣µ∣ ) and V = P (c1 , . . . , c∣λ∣−∣µ∣ ).
By construction and Corollary 10.22, we have U ∗ V = P (π) = T .
And since every entry of U S in µ is less than every entry in S, the
entries in µ must be u1 , . . . , u∣µ∣ . In particular, the shape of U is µ.
t t ⋅ ⋅ ⋅ t∣λ∣−∣µ∣
To see the shape of V is ν, set τ = [ 1 2 ], so
c1 c2 ⋅ ⋅ ⋅ c∣λ∣−∣µ∣
P (τ ) = V . By Lemma 10.30 we have Q(τ ) = rect(S) = W . Therefore,
sh V = sh P (τ ) = sh Q(τ ) = sh W = ν. □
To show Ψµ,ν,λ,T,W and Ωµ,ν,λ,T,W are bijections, we show they

are inverses of one another.
Theorem 10.33. Suppose λ, µ, and ν are partitions, T is a semi-

standard tableau of shape λ, and W is a semistandard tableau of shape
ν. Then Ψµ,ν,λ,T,W and Ωµ,ν,λ,T,W are inverses of one another, and
are therefore bijections between T (µ, ν, T ) and S(λ/µ, W ).
s s2 ⋅ ⋅ ⋅ s∣ν∣
Proof. Suppose (U, V ) ∈ T (µ, ν, T ), let σ = [ 1 ] be
b1 b2 ⋅ ⋅ ⋅ b∣ν∣
the generalized permutation with P (σ) = V , and let Q(σ) = W . Insert
b1 , . . . , b∣ν∣ into U , and insert s1 , . . . , s∣ν∣ into the corresponding boxes
of the diagram of λ/µ to obtain S = Ψ(U, V ).
To construct Ω(Ψ(U, V )) = Ω(S), first let Uc be a semistandard

tableau of shape µ whose entries are all less than every entry of S.
Then let U S be the semistandard tableau of shape λ we get by com-
u ⋅ ⋅ ⋅ u∣µ∣ t1 ⋅ ⋅ ⋅ t∣λ∣−∣µ∣
bining Uc and S. Let π = [ 1 ] be the
a1 ⋅ ⋅ ⋅ a∣µ∣ c1 ⋅ ⋅ ⋅ c∣λ∣−∣µ∣
generalized permutation with P (π) = T and Q(π) = U S, and set
U1 = P (a1 ⋅ ⋅ ⋅ a∣µ∣ ) and V1 = P (c1 ⋅ ⋅ ⋅ c∣λ∣−∣µ∣ ). We claim U = U1 and
V = V1 .
v v2 ⋅ ⋅ ⋅ v∣µ∣
To prove our claim, let σ ′ = [ 1 ] be the generalized
d1 d2 ⋅ ⋅ ⋅ d∣µ∣
permutation with P (σ ) = U and Q(σ ) = Uc . Now if π ′ = σ ′ σ =
′ ′
v ⋅ ⋅ ⋅ v∣µ∣ s1 ⋅ ⋅ ⋅ s∣ν∣
[ 1 ], then P (π ′ ) = P (σ ′ ) ∗ P (σ) = U ∗ V = T
d1 ⋅ ⋅ ⋅ d∣µ∣ b1 ⋅ ⋅ ⋅ b∣ν∣
and Q(π ′ ) = U S. But we also have P (π) = T and Q(π) = U S, so
we must have π = π ′ . Therefore, U = P (σ ′ ) = P (a1 ⋅ ⋅ ⋅ a∣µ∣ ) = U1 and
V = P (σ) = P (c1 ⋅ ⋅ ⋅ c∣ν∣ ) = V1 .
Now suppose S ∈ S(λ/µ, W ), fill the boxes of µ in S with en-
tries which are smaller than all entries of S to create a semistandard
u ⋅ ⋅ ⋅ u∣µ∣ t1 ⋅ ⋅ ⋅ t∣λ∣−∣µ∣
tableau U S of shape λ, and let π = [ 1 ]
a1 ⋅ ⋅ ⋅ a∣µ∣ c1 ⋅ ⋅ ⋅ c∣λ∣−∣µ∣
be the generalized permutation with P (π) = T and Q(π) = U S. Set
U = P (a1 ⋅ ⋅ ⋅ a∣µ∣ ) and V = P (c1 ⋅ ⋅ ⋅ c∣λ∣−∣µ∣ ).
s s2 ⋅ ⋅ ⋅ s∣ν∣
To construct Ψ(Ω(S)) = Ψ(U, V ), find σ = [ 1 ]
b1 b2 ⋅ ⋅ ⋅ b∣ν∣
with P (σ) = V and Q(σ) = W . Construct rb∣ν∣ (⋅ ⋅ ⋅ rb1 (U )), and let
S1 = Ψ(Ω(S)) be the semistandard skew tableau of shape λ/µ that
we obtain by inserting s1 , . . . , s∣ν∣ into the new boxes of λ/µ as they
are added. We claim S = S1 .
t t ⋅ ⋅ ⋅ t∣λ∣−∣µ∣
To prove our claim, suppose τ = [ 1 2 ]. Then
c1 c2 ⋅ ⋅ ⋅ c∣λ∣−∣µ∣
P (τ ) = P (c1 ⋅ ⋅ ⋅ c∣λ∣−∣µ∣ ) = V = P (σ). In addition, by Lemma 10.30
and our construction, we have
Q(τ ) = rect(S) = W = rect(S1 ) = Q(σ).
Therefore, τ = σ.
u u2 ⋅ ⋅ ⋅ u∣µ∣
Now we can construct both S and S1 by using [ 1 ]
a1 a2 ⋅ ⋅ ⋅ a∣µ∣
to build U , then constructing rb∣ν∣ (⋅ ⋅ ⋅ rb1 (U )) and building a semi-
standard skew tableau of shape λ/µ by inserting s1 , . . . , s∣ν∣ into the
new boxes of λ/µ as they are added. Therefore, S = S1 .
We have now shown Ω○Ψ is the identity on T (µ, ν, T ) and Ψ○Ω is
the identity on S(λ/µ, W ), so Ψ and Ω are inverses of one another. □
As we promised earlier, we can use our bijections Ψ and Ω to

show that ∣T (µ, ν, T )∣ depends on the shape of T but not on T itself.
Corollary 10.34. If T1 and T2 are semistandard tableaux of the same

shape and µ and ν are partitions, then
∣T (µ, ν, T1 )∣ = ∣T (µ, ν, T2 )∣.
Proof. Fix a semistandard tableau W of shape ν. By Theorem 10.33

we know ∣T (µ, ν, T1 )∣ and ∣T (µ, ν, T2 )∣ are both equal to ∣S(λ/µ, W )∣,
so they are equal to each other. □
We can also show ∣S(λ/µ, W )∣ depends on the shape of W , but

not on W itself.
Corollary 10.35. If W1 and W2 are semistandard tableaux of the

same shape and λ and µ are partitions, then
∣S(λ/µ, W1 )∣ = ∣S(λ/µ, W2 )∣.
Last but not least, we can also connect our expressions for sµ sν
and sλ/µ as linear combinations of Schur functions.
Corollary 10.36. Suppose λ, µ, and ν are partitions. If cλµ,ν is

defined as in (10.4) and dλµ,ν is defined as in (10.5), then cλµ,ν = dλµ,ν .
Proof. Let T be a semistandard tableau of shape λ, and let W be a

semistandard tableau of shape ν. If cλµ,ν is defined as in (10.4), then
cλµ,ν = ∣T (µ, ν, T )∣. Similarly, if dλµ,ν is defined as in (10.5), then dλµ,ν =
∣S(λ/µ, W )∣. These two quantities are equal by Theorem 10.33. □
Now that we know cλµ,ν = dλµ,ν , we would like to have a com-

binatorial interpretation of this common quantity. We have cλµ,ν =
∣S(λ/µ, W )∣ for any semistandard tableau W of shape ν, so we might
try to understand this quantity by choosing W to have a particularly
simple form. The simplest possible form W can have is the one in
which every entry in the jth row of W is j, for all j. So our question
becomes, which semistandard skew tableaux of shape λ/µ rectify to
this W ?
To answer this question, we consider the reading words of our
tableaux. Theorem 10.14 tells us the reading words of our semistan-
dard skew tableaux will be Knuth equivalent to the reading word
of W . Now our question is, which words are Knuth equivalent to
n ⋅ ⋅ ⋅ n n − 1 ⋅ ⋅ ⋅ n − 1 ⋅ ⋅ ⋅ 1 ⋅ ⋅ ⋅ 1?
Example 10.37. Find all words which are Knuth equivalent to the
word 332211.
Solution. After some experimentation, we find these words are

332211, 323211, 332121, 323121, and 321321. □
After examining many more examples, we are led back to the

Littlewood–Richardson words we first saw in Definition 4.37. We
describe them more directly here.
Definition 10.38. We say a word a1 ⋅ ⋅ ⋅ an is a Littlewood–Richardson

word whenever every tail ak ⋅ ⋅ ⋅ an of the word has at least as many
copies of j as it has copies of j + 1 for every j.
Theorem 10.39. Suppose µ, ν, and λ are partitions, W is the se-

mistandard tableau of shape ν in which every entry in the jth row is
j for all j, and S is a semistandard skew tableau of shape λ/µ. Then
rect(S) = W if and only if S and W have the same number of j’s for
each j, and word(S) is a Littlewood–Richardson word.
Proof. (⇒) Suppose rect(S) = W . Since we obtain W by sliding the

numbers in S around, S and W have the same number of j’s for each
j. In addition, by Theorem 10.14,
word(S) ∼K n ⋅ ⋅ ⋅ n ⋅ ⋅ ⋅ 1 ⋅ ⋅ ⋅ 1.
Since W has partition shape, word(W ) is a Littlewood–Richardson

word. Therefore, by induction we just need to show that elementary
Knuth transformations change Littlewood–Richardson words into
Littlewood–Richardson words. To do this, we consider our two types
of elementary Knuth transformations separately.
Suppose x ≤ y < z, so the words w1 xzyw2 and w1 zxyw2 are
related by an elementary Knuth transformation. If w1 xzyw2 is a
Littlewood–Richardson word, then the only way w1 zxyw2 can fail
to be a Littlewood–Richardson word is if xyw2 is not a Littlewood–
Richardson word. But xzyw2 is a Littlewood–Richardson word, so
this would mean z = x − 1, which contradicts the fact that z > x.
On the other hand, if w1 zxyw2 is a Littlewood–Richardson word,
then the only way w1 xzyw2 could fail to be a Littlewood–Richardson
word would be if zyw2 were not a Littlewood–Richardson word. But
zxyw2 is a Littlewood–Richardson word, so this would mean z = x + 1
and y = x. It would also mean xyw2 has exactly the same number
of x’s as x + 1’s. But this implies yw2 has more x + 1’s than x’s,
contradicting the fact that zxyw2 is a Littlewood–Richardson word.
The case in which x < y ≤ z and the words w1 yxzw2 and w1 yzxw2
are related by an elementary Knuth transformation is similar.
(⇐) Suppose S and W have the same number of j’s for each j,
and word(S) is a Littlewood–Richardson word. For each j, let µj be
the number of j’s in S. By Problem 10.1, we know word(S) is Knuth
equivalent to
n ⋅ ⋅ ⋅ n ⋅ ⋅ ⋅ 1 ⋅ ⋅ ⋅ 1,
´¹¹ ¹ ¸¹¹ ¹ ¶ ²
µn µ1
which is word(W ). By Theorem 10.14, this means word(rect(S)) is
Knuth equivalent to word(W ). Since word is one-to-one, we must
have W = rect(S). □
We can now gather the main results of this section into a single
theorem. Our proof of this result is short, but it provides an outline
of the ideas we have developed in this section.
Theorem 10.40 (The Littlewood–Richardson Rule for Symmetric
Functions). For any partitions λ, µ, and ν, let cλµ,ν be the number of
semistandard skew tableaux of shape λ/µ and content ν whose reading
10.5. Problems 303
words are Littlewood–Richardson words. Then

sµ sν = ∑ cλµ,ν sλ
λ
and
sλ/µ = ∑ cλµ,ν sν .
ν
Proof. Since the Schur functions form a basis for the space of all
symmetric functions, for all partitions µ, ν, and λ there are constants
cλµ,ν and dλµ,ν with
sµ sν = ∑ cλµ,ν sλ
λ
and
sλ/µ = ∑ dλµ,ν sν .
ν
Let T be a semistandard tableau of shape λ, and let W be the semi-
standard tableau of shape ν in which every entry in the jth row is j.
By Corollary 10.34 we have cλµ,ν = ∣T (µ, ν, T )∣, by Corollary 10.35 we
have dλµ,ν = ∣S(λ/µ, W )∣, and by Theorem 10.33 we have cλµ,ν = dλµ,ν .
By Theorem 10.39 we know S(λ/µ, W ) is the set of semistandard
skew tableaux of shape λ/µ and content ν whose reading words are
Littlewood–Richardson words. □
10.5. Problems
10.1. Suppose a1 < a2 < ⋅ ⋅ ⋅ < am are integers. For any word w in
a1 , . . . , am , let µj (w) be the number of times aj appears in w,
and write µ(w) to denote the sequence µ1 (w), . . . , µm (w). We
say a word w = b1 ⋅ ⋅ ⋅ bn in a1 , . . . , am is a Littlewood–Richardson
word whenever µ(bj ⋅ ⋅ ⋅ bn ) is a partition for all j. Show that if
w is a Littlewood–Richardson word, then w is Knuth equivalent
to
am ⋅ ⋅ ⋅ am ⋅ ⋅ ⋅ a1 ⋅ ⋅ ⋅ a1 .
´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹¸ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹¶ ´¹¹ ¹ ¹ ¹ ¹ ¹ ¸¹ ¹ ¹ ¹ ¹ ¹ ¶
µm (w) µ1 (w)
We use this result in our proof of Theorem 10.39.
10.2. How many permutations in Sn are Knuth equivalent to no other
permutations?
10.3. Show two permutations π1 and π2 are Knuth equivalent if and

only if their reversals are Knuth equivalent. Is the same result
also true for words?
10.4. For any word w and any positive integer k, let L(w, k) be the
largest sum we can obtain by adding the lengths of k disjoint
nondecreasing subsequences of w. Prove that if T is a semistan-
dard tableau of shape λ, then L(word(T ), k) = λ1 + λ2 + ⋅ ⋅ ⋅ + λk .
10.5. Show that if two words w1 and w2 are Knuth equivalent, then
L(w1 , k) = L(w2 , k) for all k.
10.6. Show that for any word w, the sum of the lengths of the bottom
k rows of P (w) is L(w, k).
10.7. Suppose w1 and w2 are Knuth equivalent words, and w1′ and
w2′ are the words we obtain by removing the k largest entries
from w1 and the k largest entries from w2 . Show w1′ ∼K w2′ .
10.8. Show
sλ/n = ∑ sν ,
ν
where the sum on the right is over all partitions ν ⊆ λ for which
λ/ν is a horizontal strip of length n. Find and prove a similar
formula for sλ/1n .
10.9. Suppose µ and λ are partitions. Without using any of the results
in this chapter, show
⟨sλ , sµ f ⟩ = ⟨sλ/µ , f ⟩
for every symmetric function f . 2
10.10. Let (l1 , . . . , lr−1 ), (m1 , . . . , mr−1 ), and (n1 , . . . , nr−1 ) be sequen-
ces of nonnegative integers for some r ≥ 1. A BZ pattern of type
(r, l, m, n) is an assignment of integers to the non-corner points
of an (r + 1) × (r + 1) × (r + 1) triangular lattice such that the
following hold.
● The sum of the values at the points in the jth horizontal
row from the top is lj .
● The sum of the values at the points in the jth northwest
diagonal from the left is mj .
● The sum of the values at the points in the jth southwest
diagonal from the right is nj .
10.5. Problems 305
● All of the partial sums along rows and diagonals from left
to right, from southeast to northwest, and from northeast
to southwest are nonnegative.
In Figure 10.18 we see which rows and diagonals sum to which
values, and the directions in which the partial sums in the last
condition are computed. Show that the Littlewood–Richardson
coefficient cλµ,ν is the number of BZ patterns of type (r, l, m, n),
where r ≥ max(l(λ), l(µ), l(ν)), lj = λr−j − λr−j+1 for all j, mj =
µj − µj+1 for all j, and nj = νj − νj+1 for all j.
m3
m2 `1
m1 `2
`3
n3 n2 n1
Figure 10.18. The values of the sums and the directions of
the partial sums in the definition of a BZ pattern
10.11. Suppose l1 , . . . , lr is a sequence of 0’s and 1’s. If we interpret

each 0 as a unit step down and each 1 as a unit step to the right,
then l1 , . . . , lr traces out the northeast boundary of the Ferrers
diagram of a partition. We call this partition the partition asso-
ciated with l1 , . . . , lr . For example, the partition associated with
01011011 is 3, 1, as is the partition associated with 10110111.
Now suppose r ≥ 1 and (l1 , . . . , lr−1 ), (m1 , . . . , mr−1 ), and
(n1 , . . . , nr−1 ) are sequences of 0’s and 1’s, all with the same
number of 0’s and the same number of 1’s. A puzzle of type
(r, l, m, n) is a tiling of an equilateral triangle of side length r
0 0 1 1
1 0
0 1
0 1
0 1
0 0 1 1
0 1
1 1 0 0
0 1
Figure 10.19. The tiles allowed in puzzles in Problem 10.11
with copies of the edge-labeled tiles in Figure 10.19 (in the given
orientations: no rotations or reflections are allowed) such that
the following hold.
● If two tiles share an edge, then that edge has the same
label in both tiles.
● The labels on the edges from left to right along the north-
west boundary are m1 , . . . , mr .
● The labels on the edges from left to right along the north-
east boundary are n1 , . . . , nr .
● The labels on the edges from left to right along the south
boundary are l1 , . . . , lr .
Show that the Littlewood–Richardson coefficient cλµ,ν is the
number of puzzles of type (r, l, m, n), where λ is the partition
associated with (l1 , . . . , lr ), µ is the partition associated with
(m1 , . . . , mr ), and ν is the partition associated with (n1 , . . . , nr ).
10.12. An increasing edge-vertex labeling of a triangular lattice is a
labeling of the vertices and edges of the lattice with nonnegative
integers such that the following hold.
● The labels of the vertices are weakly increasing from left
to right along each row and each diagonal.
10.6. Notes 307
● The label on each edge is the nonnegative difference be-

tween the labels on the vertices it connects.
For any positive integer n and partitions µ, ν, and λ with ∣λ∣ =
∣µ∣ + ∣ν∣ and n ≥ max(l(µ), l(ν), l(λ)), an n-hive of type (µ, ν, λ)
is an increasing edge-vertex labeling of the (n+1)×(n+1)×(n+1)
triangular lattice such that the following hold.
● The edge labels on the northwest boundary from left to
right are µ1 , . . . , µn .
● The edge labels on the northeast boundary from left to
right are ν1 , . . . , νn .
● The edge labels on the south boundary from left to right
are λ1 , . . . , λn .
● In each rhombus as in Figure 10.20 we have b + c ≥ a + d.
Show that if ∣λ∣ = ∣µ∣+∣ν∣ and n ≥ max(l(µ), l(ν), l(λ)), then
the Littlewood–Richardson coefficient cλµ,ν is the number
of n-hives of type (µ, ν, λ).
a
a b c d b d
c d b a c
Figure 10.20. The rhombi in an n-hive in Problem 10.12
10.6. Notes
Our proof of the Littlewood–Richardson rule follows Fulton [Ful96],
but there are many other proofs available. These include proofs by
Berenstein and Zelevinsky [BZ88], Remmel and Shimozono [RS98],
Gasharov [Gas98], and Stembridge [Ste02]. As Problems 10.10,
10.11, and 10.12 suggest, there are also several combinatorial in-
terpretations of the Littlewood–Richardson coefficients. The inter-
pretations in these problems are based on work of Berenstein and
Zelevinsky [BZ92], Knutson and Tao [KT99], and Knutson, Tao,
and Woodward [KTW04].
Appendix A
Linear Algebra
Our approach to symmetric functions is combinatorial, but symmet-

ric functions form a vector space in a natural way, and many of the
questions we address arise from linear algebraic considerations. We
assume you’ve seen a variety of facts and ideas from linear algebra, in-
cluding linear transformations and bases of vector spaces. But we will
also need some general facts about inner products on vector spaces
and dual bases which you may not have seen. In this appendix we
quickly review the basic linear algebra facts we will use, and we de-
velop a couple of these results in detail. More details are available in
a variety of books, including the books of Axler [Axl14], Bretscher
[Bre12], and Lang [Lan04].
A.1. Fields and Vector Spaces

All of linear algebra is built on fields, so we start with these objects.
Definition A.1. A field F is a set together with two binary opera-

tions, addition and multiplication, with the following properties. (For
any a, b ∈ F , we write a + b to denote the sum of a and b, and we write
ab to denote their product.)
(1) Addition is associative: for all a, b, c ∈ F , we have (a+b)+c =
a + (b + c).
(2) Addition is commutative: for all a, b ∈ F , we have a+b = b+a.
309
310 A. Linear Algebra
(3) There is an additive identity, which is an element 0 ∈ F such

that a + 0 = a for all a ∈ F .
(4) Every element of the field has an additive inverse: for every
a ∈ F , there is an element −a ∈ F with a + (−a) = 0.
(5) Multiplication is associative and commutative.
(6) There is a multiplicative identity, which is an element 1 ∈ F
with 1 ≠ 0 and 1a = a for all a ∈ F .
(7) Every nonzero element of the field has a multiplicative in-
verse: for every a ∈ F , if a ≠ 0, then there is an element
a−1 ∈ F with aa−1 = 1.
(8) Multiplication distributes over addition: for all a, b, c ∈ F ,
we have a(b + c) = ab + ac.
Many familiar algebraic systems are fields: the set R of real num-
bers, the set Q of rational numbers, and the set C of complex numbers
are all fields under the usual addition and multiplication.
√ √ But there
are other fields, too. For example, the set Q[ 2] = {a+b 2 ∣ a, b ∈ Q}
is a field under the usual addition and multiplication, as is the set of
real numbers which are roots of polynomials with rational coefficients.
The set Q(t) of rational functions with rational coefficients is also a
field under the usual addition and multiplication.
All of the fields we have mentioned so far are infinite, but there
are also finite fields. For example, for any prime p, the set {0, 1, 2, . . . ,
p − 1} is a field under addition and multiplication modulo p. In fact,
for every prime p and every positive integer n, there is a field of order
pn , which is unique up to isomorphism. (However, when n ≥ 2 this
field is not isomorphic to the set {0, 1, 2, . . . , pn − 1} under addition
and multiplication modulo pn .) It turns out these are all of the finite
fields.
It is possible to study symmetric functions over any field, and
everything we do in this book will work over any field containing Q.
Indeed, many recent symmetric function results involve one or two
parameters, so they are most easily stated as results over Q(q) or
Q(q, t). For concreteness, though, we will do all of our work over Q.
Once we have a field, we can study vector spaces over that field.
A.1. Fields and Vector Spaces 311
Definition A.2. Suppose F is a field and V is a set, addition is

a binary operation on V , and scalar multiplication is an operation
which associates with every c ∈ F and every v⃗ ∈ V an element c⃗v∈V.
We say V , together with these two operations, is a vector space (over
F ) whenever the following hold.
(1) Addition is associative and commutative.
(2) There is an additive identity 0⃗ ∈ V .
(3) Every v⃗ ∈ V has an additive inverse −⃗
v.
(4) The multiplicative identity in F acts like the identity on V :
for all v⃗ ∈ V , we have 1⃗
v = v⃗.
(5) Scalar multiplication associates with multiplication in F : for
all a, b ∈ F and all v⃗ ∈ V , we have (ab)⃗
v = a(b⃗
v ).
(6) Scalar multiplication distributes over addition in F : for all
a, b ∈ F and v⃗ ∈ V , we have (a + b)⃗
v = a⃗
v + b⃗
v.
(7) Scalar multiplication distributes over addition in V : for all
a ∈ F and all v⃗, w
⃗ ∈ V , we have a(⃗ ⃗ = a⃗
v + w) ⃗
v + aw.
Examples of vector spaces abound. We give a few here, and we

leave it to you to check that our examples really are vector spaces.
Example A.3. Suppose F is a field and n is a positive integer. Set
V = F n , and define addition and scalar multiplication componentwise,
so that
(a1 , . . . , an ) + (b1 , . . . , bn ) = (a1 + b1 , . . . , an + bn )
and
c(a1 , . . . , an ) = (ca1 , . . . , can ).
Then V is a vector space over F , with ⃗0 = (0, 0, . . . , 0).
Example A.4. Suppose F is a field and n is a positive integer. Let
F [x1 , . . . , xn ] be the set of polynomials in x1 , . . . , xn with coefficients
in F . Then F [x1 , . . . , xn ] is a vector space over F with the usual
addition of polynomials and multiplication of polynomials by scalars.
The zero vector is the polynomial whose coefficients are all 0.
Example A.5. The set R of real numbers is a vector space over the
field Q, under the usual addition and multiplication of real numbers.
Many algebraic objects are like vector spaces, in that they are
sets with some additional structure, usually coming from operations
of various sorts. For these objects, we’re usually interested in subsets
with the same structure.
Definition A.6. Suppose V is a vector space over a field F . A

subspace W of V is a subset of V for which the following hold.
(1) W is nonempty.
(2) W is closed under addition: for every v⃗, w
⃗ ∈ W , we have
v⃗ + w
⃗ ∈ W.
(3) W is closed under scalar multiplication: for every c ∈ F and
every v⃗ ∈ W , we have c⃗
v ∈ W.
We can check that a subset W of a vector space is a subspace

exactly when it is a vector space under the same operations, with the
same additive identity, and with the same additive inverses.
A.2. Bases and Linear Transformations

⃗ and V are subspaces of V , but
For any vector space V , the subsets {0}
we would like to be able to build others. To see one way of doing this,
suppose W is a subspace of V , and w ⃗ ∈W is a vector in W . Since W is
closed under scalar multiplication, the set {cw∣c ⃗ ∈ F } must be con-
tained in W . Similarly, suppose w ⃗1 , w
⃗2 ∈ W . Since W is closed un-
der addition and scalar multiplication, the set {c1 w ⃗1 +c2 w
⃗2 ∣c1 , c2 ∈F }
must be contained in W . These examples suggest the following defi-
nitions.
Definition A.7. Suppose V is a vector space over a field F and

v⃗1 , . . . , v⃗n ∈ V . A linear combination of v⃗1 , . . . , v⃗n is a vector of the
form c1 v⃗1 + ⋅ ⋅ ⋅ + cn v⃗n , where c1 , . . . , cn ∈ F . The span of v⃗1 , . . . , v⃗n ,
written Span(⃗ v1 , . . . , v⃗n ), is the set of all linear combinations of v⃗1 , . . . ,
v⃗n . If V = Span(⃗ v1 , . . . , v⃗n ), then we say v⃗1 , . . . , v⃗n span V .
We can show that the span of any finite set of vectors is nonempty
(because it contains ⃗0) and closed under addition and scalar multipli-
cation, so the span of any finite set of vectors is a subspace.
A.2. Bases and Linear Transformations 313
When we take the span of a list of vectors, the resulting subspace

might be the span of a shorter list of vectors. For example, you can
check that Span(⃗ v1 , v⃗2 , v⃗1 + v⃗2 ) = Span(⃗
v1 , v⃗2 ) for any vectors v⃗1 and
v⃗2 . However, we can single out those sets of vectors which are most
efficient, in the sense of being a shortest list of vectors generating
their span.
Definition A.8. Suppose V is a vector space over a field F . We

say vectors v⃗1 , . . . , v⃗n ∈ V are linearly independent whenever the only
coefficients c1 , . . . , cn ∈ F with c1 v⃗1 + ⋅ ⋅ ⋅ + cn v⃗n = 0⃗ are c1 = c2 = ⋅ ⋅ ⋅ =
cn = 0. We say v⃗1 , . . . , v⃗n are linearly dependent whenever they are
not linearly independent.
You can check that a set X of vectors is linearly independent

if and only if Span(X) ≠ Span(Y ) for every proper subset Y of X.
Linearly independent sets of vectors which span their entire space are
especially useful, so we also name these sets.
Definition A.9. Suppose V is a vector space over a field F . A basis

for V is a set of vectors v⃗1 , . . . , v⃗n ∈ V which are linearly independent,
and which span V . We say V is finite dimensional whenever it has a
finite basis.
Having a basis for a vector space is useful in many ways, one of

which is that it gives us a natural “address” system for the space.
Proposition A.10. Suppose V is a vector space over a field F and

that v⃗1 , . . . , v⃗n is a basis for V . Then for each v⃗ ∈ V there is a unique
n-tuple (a1 , . . . , an ) ∈ F n with v⃗ = a1 v⃗1 + ⋅ ⋅ ⋅ + an v⃗n .
Proof. Since V = Span(⃗ v1 , . . . , v⃗n ), we know there exist a1 , . . . , an ∈

F with v⃗ = a1 v⃗1 + ⋅ ⋅ ⋅ + an v⃗n . To see these coefficients are unique, sup-
pose we also have b1 , . . . , bn ∈ F with v⃗ = b1 v⃗1 + ⋅ ⋅ ⋅ + bn v⃗n . Subtracting
these two equations, we find
⃗0 = (a1 − b1 )⃗
v1 + ⋅ ⋅ ⋅ + (an − bn )⃗
vn .
But v⃗1 , . . . , v⃗n are linearly independent, so the coefficients on the right
are all 0. Therefore, aj = bj for all j. □
Most vector spaces have numerous bases, but one can use facts
about ranks of rectangular matrices and solutions of the associated
systems of linear equations to show that any two bases for a finite-
dimensional vector space will be equinumerous. So it makes sense to
make the following definition.
Definition A.11. Suppose V is a finite-dimensional vector space over
a field F . We say V has dimension n, or V is n dimensional, and we
write dim V = n, whenever V has a basis with exactly n elements.
As we see next, if we are looking for a basis for a finite-dimensional

vector space and we have a set consisting of the right number of
vectors, then we only need to check one of the two conditions required
for our set to be a basis.
Proposition A.12. Suppose V is a finite-dimensional vector space
over a field F and that u ⃗1 , . . . , u
⃗n ∈ V . If any two of the following
conditions hold, then the third condition also holds, and u ⃗1 , . . . , u
⃗n is
a basis for V .
(i) dim V = n.
⃗1 , . . . , u
(ii) u ⃗n are linearly independent.
⃗1 , . . . , u
(iii) u ⃗n span V .
Proof. This is Problem A.4. □
We are often interested in relationships among algebraic objects

of the same type. When our objects are sets with some additional
structure, we can capture these relationships by studying functions
which preserve the structure of the objects. For vector spaces, these
functions are linear transformations.
Definition A.13. Suppose V and W are vector spaces over a field F .
A linear transformation is a function f ∶ V → W with the following
properties.
(1) f respects addition: for all v⃗1 , v⃗2 ∈ V , we have f (⃗
v1 + v⃗2 ) =
f (⃗
v1 ) + f (⃗
v2 ).
(2) f respects scalar multiplication: for all c ∈ F and all v⃗ ∈ V ,
we have f (c⃗
v ) = cf (⃗
v ).
A.2. Bases and Linear Transformations 315
It is worth noting that although the operations on each side of

the equations in parts (1) and (2) of Definition A.13 are represented
by the same notation, they are different operations. For example, in
v1 + v⃗2 ) = f (⃗
f (⃗ v1 ) + f (⃗
v2 ), the + on the left refers to addition in V ,
but the + on the right refers to addition in W . These might be very
different operations!
The zero vector and additive inverses are also part of the algebraic
structure of a vector space, so we might ask why we don’t insist
our linear transformations respect these parts of the structure too.
The answer is that if a function f is a linear transformation by our
definition, then it must also have f (⃗0) = 0⃗ and f (−⃗ v ) = −f (⃗
v ) for all
⃗ = f (⃗0+ 0)
v⃗ ∈ V . To prove the first claim, start with f (0) ⃗ = f (0)+f
⃗ ⃗
(0).
To prove the second, evaluate f (⃗ v + (−⃗
v )) in two different ways. The
net result is that our definition is simpler to use, and specifies exactly
the same set of functions.
We can now use bases to construct numerous linear transforma-
tions.
Proposition A.14. Suppose V and W are vector spaces over a field

F , v⃗1 , . . . , v⃗n is a basis for V , and w
⃗1 , . . . , w
⃗n are any vectors in W .
Then there is a unique linear transformation f ∶ V → W with f (⃗ vj ) =
⃗j for 1 ≤ j ≤ n.
w
Proof. To show f exists, suppose v⃗ ∈ V . By Proposition A.10 there

is a unique n-tuple (a1 , . . . , an ) ∈ F n with v⃗ = a1 v⃗1 + ⋅ ⋅ ⋅ + an v⃗n . Set
f (⃗ ⃗ 1 + ⋅ ⋅ ⋅ + an w
v ) = a1 w ⃗n .
Since a1 , . . . , an are unique, f is a function from V to W . In ad-
dition, f (⃗
vj ) = w ⃗j for all j by construction, so we just need to show f
respects addition and scalar multiplication. To show f respects addi-
tion, suppose u ⃗, v⃗ ∈ V , and we have n-tuples (a1 , . . . , an ), (b1 , . . . , bn ) ∈
F with u ⃗ = a1 v⃗1 + ⋅ ⋅ ⋅ + an v⃗n and v⃗ = b1 v⃗1 + ⋅ ⋅ ⋅ + bn v⃗n . Adding these
equations, we find
⃗ + v⃗ = (a1 + b1 )⃗
u v1 + ⋅ ⋅ ⋅ + (an + bn )⃗
vn .
Therefore, we have
u + v⃗) = (a1 + b1 )w
f (⃗ ⃗1 + ⋅ ⋅ ⋅ + (an + bn )w
⃗n
⃗1 + ⋅ ⋅ ⋅ + an w
= a1 w ⃗n + b1 w
⃗1 + ⋅ ⋅ ⋅ + bn w
⃗n
= f (⃗
u) + f (⃗
v ),
so f respects addition. The proof that f respects scalar multiplication
is similar.
The proof that f is unique is Problem A.9. □
A.3. Inner Products and Dual Bases

If you have seen vectors used in physics or calculus, then you may
have been told that a vector is an object characterized by its length
and direction. This seems quite different from the vectors we have
used so far, which are just elements of a vector space. But for vector
spaces over the field Q of rational numbers, we can use an analogue
of a dot product to make sense of the idea of the length of a vector,
or the angle between two vectors.
Definition A.15. For any vector space V over Q, an inner product
on V is a function ⟨ , ⟩ ∶ V × V → Q such that the following hold for
⃗, v⃗, w
all u ⃗ ∈ V and all c ∈ Q.
u, v⃗⟩ = ⟨⃗
(1) ⟨⃗ ⃗⟩.
v, u
u + v⃗, w⟩
(2) ⟨⃗ ⃗ = ⟨⃗ ⃗ + ⟨⃗
u, w⟩ ⃗
v , w⟩.
u, v⃗⟩ = c⟨⃗
(3) ⟨c⃗ u, v⃗⟩.
(4) ⟨⃗ ⃗⟩ ≥ 0.
u, u
(5) ⟨⃗ ⃗⟩ = 0 if and only if u
u, u ⃗
⃗ = 0.
The usual dot product on Qn is the canonical example of an inner

product, but we can construct inner products on other vector spaces
over Q in important and interesting ways.
Example A.16. Let V be the space of polynomials in x of degree at
most 2 which have coefficients in Q, and for any f (x), g(x) ∈ V , set
1
⟨f (x), g(x)⟩ = ∫ f (x)g(x) dx.
−1
Show ⟨ , ⟩ is an inner product on V .
A.3. Inner Products and Dual Bases 317
Solution. We show ⟨ , ⟩ satisfies Definition A.15(1)–(5).

Condition (1) is immediate, since multiplication of polynomials
is commutative.
To verify condition (2), note that for any f (x), g(x), h(x) ∈ V ,
we have
1
⟨f (x) + g(x), h(x)⟩ = ∫ (f (x) + g(x))h(x) dx
−1
1 1
=∫ f (x)h(x) dx + ∫ g(x)h(x) dx
−1 −1
= ⟨f (x), h(x)⟩ + ⟨g(x), h(x)⟩.
To verify condition (3), note that for any f (x), g(x) ∈ V and any
constant c, we have
1
⟨cf (x), g(x)⟩ = ∫ cf (x)g(x) dx
−1
1
= c∫ f (x)g(x) dx
−1
= c⟨f (x), g(x)⟩.
To verify conditions (4) and (5), first note that if f (x) = 0, then
1
∫−1 (f (x)) dx = 0, so ⟨f (x), f (x)⟩ = 0. Conversely, if f (x) ∈ V and
2
f (x) ≠ 0, then (f (x))2 ≥ 0 throughout the interval [−1, 1], and there
is a subinterval on which (f (x))2 > 0. Therefore, ⟨f (x), f (x)⟩ =
1
∫−1 (f (x)) dx > 0. □
2
Since an inner product ⟨ , ⟩ on a vector space V is essentially a

dot product on V , we can extend terminology and ideas related to
dot products to objects associated with V . For √ example, if u ⃗ ∈ V,
then the length of u ⃗ with respect to ⟨ , ⟩ is ∣u∣ ∶= ⟨⃗ ⃗⟩ (assuming this
u, u
square root is in our field), and if ∣⃗ u∣ = 1, then we say u ⃗ is a unit vector
with respect to ⟨ , ⟩. Similarly, if u ⃗, v⃗ ∈ V have ⟨⃗ u, v⃗⟩ = 0, then we say u ⃗
and v⃗ are orthogonal with respect to ⟨ , ⟩. Building on this, we say the
vectors u ⃗1 , . . . , u
⃗n ∈ V are orthonormal with respect to ⟨ , ⟩ whenever
⟨⃗ ⃗k ⟩ = δjk for 1 ≤ j, k ≤ n, and we say u1 , . . . , un is an orthonormal
uj , u
basis for V with respect to ⟨ , ⟩ whenever u ⃗1 , . . . , u
⃗n are orthonormal
with respect to ⟨ , ⟩ and are a basis for V . Given any basis for V , we
can try to use a natural analogue of the Gram–Schmidt process to
construct an orthonormal basis for V . Instead of recalling the Gram–

Schmidt process in complete detail, we apply it in a small example.
Since this process requires having certain square roots in our field, we
will use the real numbers instead of the rationals.
Example A.17. Let V be the space of polynomials in x of degree at

most 2 which have coefficients in R, and for any f (x), g(x) ∈ V , set
1
⟨f (x), g(x)⟩ = ∫ f (x)g(x) dx.
−1
Assuming ⟨ , ⟩ is an inner product on V , apply Gram–Schmidt or-

thogonalization to the basis 1, x, x2 to obtain an orthonormal basis
for V .
Solution. We first construct an orthogonal basis v⃗1 , v⃗2 , v⃗3 , and then
we divide each vector by its length to obtain an orthonormal basis
⃗1 , u
u ⃗2 , u
⃗3 .
We start with v⃗1 = 1, and we obtain v⃗2 via
⟨⃗
v1 , x⟩
v⃗2 = x − v⃗1
v1 , v⃗1 ⟩
⟨⃗
= x.
Similarly, we have
⟨⃗
v1 , x2 ⟩ ⟨⃗
v2 , x2 ⟩
v⃗3 = x2 − v⃗1 − v⃗2
v1 , v⃗1 ⟩
⟨⃗ v2 , v⃗2 ⟩
⟨⃗
1
= x2 − .
3
√ √
⃗1 =
Therefore, u √1 ,
2
⃗2 =
u √3 x,
2
⃗3 =
and u 3√ 5
2 2
(x2 − 13 ). □
Proposition A.18. Suppose V is a vector space and u ⃗1 , . . . , u

⃗n is a
basis for V . Then there exists a unique inner product ⟨ , ⟩ on V with
⃗1 , . . . , u
respect to which u ⃗n is an orthonormal basis of V .
A.3. Inner Products and Dual Bases 319
Proof. If such an inner product ⟨ , ⟩ exists and we have vectors v⃗ =

n
⃗j and w
∑j=1 aj u ⃗k in V , then we have
⃗ = ∑nk=1 bk u
n n
⟨⃗
v , w⟩ ⃗j , ∑ bk uk ⟩
⃗ = ⟨ ∑ aj u
j=1 k=1
n n
= ∑ ∑ aj bk ⟨⃗ ⃗k ⟩
uj , u
j=1 k=1
n
= ∑ aj bj .
j=1
Since every vector in V is a unique linear combination of u⃗1 , . . . , u

⃗n ,
this shows the inner product ⟨ , ⟩ is unique if it exists.
To show the desired inner product exists, we define a function
⟨ , ⟩ ∶ V × V → Q as follows. If v⃗, w
⃗ ∈ V , let a1 , . . . , an and b1 , . . . , bn
be the unique constants for which v⃗ = ∑nj=1 aj u ⃗j and w = ∑nk=1 bk u ⃗k .
Then we set ⟨⃗ ⃗ = ∑j=1 aj bj .
v , w⟩ n
To see ⟨ , ⟩ is an inner product, note that it is the dot product

on the vector space of coordinate vectors with respect to the basis
⃗1 , . . . , u
u ⃗n . Since the dot product is an inner product, ⟨ , ⟩ is also an
inner product. Moreover, for 1 ≤ j ≤ n, the coordinate vector for u ⃗j
is the jth standard basis vector for Qn . Since these standard basis
vectors form an orthonormal basis with respect to the dot product,
⃗1 , . . . , u
u ⃗n is an orthonormal basis for V with respect to ⟨ , ⟩. □
An orthonormal basis u ⃗1 , . . . , u
⃗n for a vector space is a basis which
has a certain relationship with itself, namely ⟨⃗ ⃗k ⟩ = δjk . It turns
uj , u
out that two different bases for a given vector space can have this rela-
tionship with each other. For example, if u ⃗1 = ⟨1, 1, 0⟩, u⃗2 = ⟨−1, 1, 1⟩,
⃗3 = ⟨1, 0, −1⟩, v⃗1 = ⟨1, 0, 1⟩, v⃗2 = ⟨−1, 1, −1⟩, and v⃗3 = ⟨−1, 1, −2⟩, then
u
⃗j ⋅ v⃗k = δjk . This situation happens often enough that we give it a
u
name.
Definition A.19. Suppose V is a vector space over Q, ⟨ , ⟩ is an

inner product on V , and u ⃗1 , . . . , u
⃗n and v⃗1 , . . . , v⃗n are bases for V .
We say u ⃗1 , . . . , u
⃗n and v⃗1 , . . . , v⃗n are dual with respect to ⟨ , ⟩ whenever
uj , v⃗k ⟩ = δjk for all j and k with 1 ≤ j, k ≤ n.
⟨⃗
For any bases u ⃗1 , . . . , u⃗n and v⃗1 , . . . , v⃗n of V , there is a unique

function ⟨ , ⟩ ∶ V × V → Q such that ⟨⃗ uj , v⃗k ⟩ = δjk . If u ⃗j = v⃗j for
1 ≤ j ≤ n, then this is the inner product we constructed in Proposition
A.18, with respect to which u ⃗1 , . . . , u
⃗n is orthonormal. So we might
hope that even when u ⃗1 , . . . , u
⃗n and v⃗1 , . . . , v⃗n are distinct bases for
V , the function ⟨ , ⟩ is an inner product on V with respect to which
⃗1 , . . . , u
u ⃗n and v⃗1 , . . . , v⃗n are dual. Unfortunately, this does not hold
in general, because for certain choices of u ⃗1 , . . . , u
⃗n and v⃗1 , . . . , v⃗n the
function ⟨ , ⟩ is not an inner product. We explore this in more detail
in Problems A.12 and A.13.
A.4. Problems
A.1. Suppose V is a finite-dimensional vector space over a field F .
Show that if v⃗1 , . . . , v⃗k ∈ V are linearly independent, then there
exist vectors v⃗k+1 , . . . , v⃗n for which v⃗1 , . . . , v⃗n is a basis for V .
A.2. Suppose V is a finite-dimensional vector space over a field F
and W is a subspace of V . Show W is finite-dimensional and
dim W ≤ dim V .
A.3. Suppose V is a finite-dimensional vector space over a field
F . Show that if v⃗1 , . . . , v⃗n span V , then there is a subset of
v⃗1 , . . . , v⃗n which is a basis for V .
A.4. Prove Proposition A.12.
A.5. Suppose V is a vector space over a field F and that W1 and W2
are subspaces of V . Prove or disprove: W1 ∪ W2 is a subspace
of V .
A.6. Suppose V is a vector space over a field F and that W1 and W2
are subspaces of V . Prove or disprove: W1 ∩ W2 is a subspace
of V .
A.7. Suppose V and W are vector spaces over a field F and
f ∶ V → W is a linear transformation. The kernel of f , writ-
ten ker f , is the set of v⃗ ∈ V with f (⃗ ⃗ The image of f ,
v ) = 0.
written Im f , is the set of w ⃗ ∈ W for which there exists v⃗ ∈ V
with f (⃗
v ) = w.⃗ Show ker f is a subspace of V and Im f is a
subspace of W .
A.4. Problems 321
A.8. Suppose V is a finite-dimensional vector space over a field F ,

W is any vector space over F , and f ∶ V → W is a linear
transformation. Show Im f is finite dimensional and dim V =
dim ker f + dim Im f .
A.9. Prove that the linear transformation f in Proposition A.14 is
unique.
f ∶ V → W is a linear transformation. Prove or disprove: if
v⃗1 , . . . , v⃗n ∈ V are linearly independent, then f (⃗
v1 ), . . . , f (⃗
vn )
are linearly independent.
f ∶ V → W is a linear transformation. Prove or disprove: if
f (⃗ vn ) are linearly independent, then v⃗1 , . . . , v⃗n ∈ V
v1 ), . . . , f (⃗
are linearly independent.
A.12. Suppose V is a vector space over Q with bases v⃗1 , . . . , v⃗n and
⃗ ...,w
w, ⃗n . Define a function ⟨ , ⟩ ∶ V × V → Q so that if v⃗ =
∑j=1 aj v⃗j and w
⃗ = ∑nj=1 cj w
⃗j , then
n
n
⟨⃗ ⃗ = ∑ aj cj .
v , w⟩
j=1
Show conditions (1), (2), and (3) of Definition A.15 hold for
⟨ , ⟩.
A.13. Let V = Q2 , v⃗1 = ⟨1, 0⟩, v⃗2 = ⟨0, 1⟩, w
⃗1 = ⟨1, 1⟩, and w
⃗2 = ⟨1, −1⟩.
Define a function ⟨ , ⟩ ∶ V × V → Q as in Problem A.12.
(a) Show that for any v⃗ = ⟨a, b⟩ ∈ Q2 , we have
1 1
v , v⃗⟩ = a2 + ab − b2 .
⟨⃗
2 2
(b) Show ⟨ , ⟩ satisfies neither condition (4) nor condition (5)
of Definition A.15.
A.14. Let V be the vector space of continuous functions from the
closed interval [−π, π] to R, and define ⟨ , ⟩ ∶ V × V → R by
1 π
⟨f, g⟩ = ∫ f (x)g(x) dx.
π −π
(a) Show ⟨ , ⟩ is an inner product on V .
(b) Show that the functions in the set {sin(nx) ∣ n ∈ P} ∪

{cos(nx) ∣ n ∈ P} are mutually orthogonal unit vectors
with respect to ⟨ , ⟩. Here P is the set of positive integers.
Appendix B
Partitions
The combinatorics of symmetric functions rests on a foundation of

partitions: just about everything we do in this book is built on par-
titions, objects we build from partitions, and analogues of partitions.
In this appendix we review some facts we will need about partitions
and their generating functions, and we set our notation. More details
and information about partitions and their generating functions ap-
pear in several books, including the books of Andrews and Eriksson
[AE04] and Stanley [Sta11].
B.1. Partitions and a Generating Function

We start with the definition of a partition.
Definition B.1. For any nonnegative integer n, a partition λ of n

is a weakly decreasing sequence {λj }∞ j=1 of nonnegative integers such
that ∑∞j=1 λ j = n. If λ is a partition of n, then we often write λ ⊢ n
and ∣λ∣ = n.
Each partition λ has finitely many nonzero terms, so we generally

write only these terms when writing λ. We call these nonzero terms
the parts of λ, and we write l(λ) to denote the number of parts of
λ, which is also known as the length of λ. For any positive integer
k and any partition λ, the multiplicity of k in λ is the number of
parts of λ which are equal to k. We sometimes write a partition λ by
323
324 B. Partitions
writing each of its distinct parts once, with the multiplicity of each
part in that part’s exponent. In this notation we write the partition
(5, 5, 5, 4, 3, 3, 3, 3, 3, 1, 1, 1, 1) as (53 , 41 , 35 , 14 ).
It is often useful to think about partitions geometrically. For this
we generally use their Ferrers diagrams, which some call Young dia-
grams. There are several competing conventions; we use the French
convention. That is, for us the Ferrers diagram of a partition λ is a
stack of rows of 1 × 1 boxes in which each row is left-justified, the bot-
tom row has λ1 boxes, the next row up has λ2 boxes, and in general
the jth row has λj boxes. We number the rows in our Ferrers diagrams
from the bottom, so the first row is always the bottom row. In Figure
B.1 we have the Ferrers diagram of the partition (5, 3, 3, 3, 2, 1, 1).
Figure B.1. The Ferrers diagram of (5, 3, 3, 3, 2, 1, 1)
The reflection of any Ferrers diagram over the line y = x is another

Ferrers diagram, which gives us a natural symmetry on the set of
partitions of n. Specifically, for any λ ⊢ n, we write λ′ to denote
the conjugate of λ, which is the partition whose Ferrers diagram is
the reflection of the Ferrers diagram of λ over the line y = x. For
instance, we see from Figure B.1 that if λ = (5, 3, 3, 3, 2, 1, 1), then
λ′ = (7, 5, 4, 1, 1). The conjugate map on partitions is an example of
an involution, which is a function whose square is the identity map.
For each n ≥ 0, we write p(n) to denote the number of partitions
of n; in Figure B.2 we have the first nine values of p(n). No simple
formula for p(n) is known, but the sequence {p(n)}∞
n=0 satisfies several
recurrence relations from which one can efficiently compute p(n) for
B.2. Problems 325
n 0 1 2 3 4 5 6 7 8
p(n) 1 1 2 3 5 7 11 15 22
Figure B.2. The first nine partition numbers
large values of n. Several of these recurrence relations arise from the

fact that the generating function for {p(n)}∞
n=0 has an elegant product
expression, which we recall next.
Proposition B.2. The ordinary generating function for the sequence
{p(n)}∞
n=0 is given by
∞ ∞
1
∑ p(n)x = ∏
n
(B.1) .
n=0 j=1 1 − xj
Proof. Each partition is uniquely built from a partition using only

1’s, together with a partition using only 2’s, together with a partition
using only 3’s, etc. The generating function for partitions using only
j’s is 1 + xj + x2j + ⋅ ⋅ ⋅ = 1−x
1
j . Choosing a partition using only 1’s is
1
equivalent to choosing a term from 1−x , choosing a partition using
1
only 2’s is equivalent to choosing a term from 1−x 2 , and in general
choosing a partition using only j’s is equivalent to choosing a term

1
from 1−x j. Multiplying these terms corresponds to combining the
corresponding partitions into a single partition, so xn appears in the
product ∏∞ 1
j=1 1−xj exactly once for each partition of n. □
B.2. Problems
B.1. Find an expression, as an infinite product, for the generating
function for partitions using only odd parts.
B.2. For which partitions is the infinite product ∏∞
j=1 (1 + x + x )
j 3j
the generating function?

B.3. We say a partition λ is self-conjugate whenever λ′ = λ. Find
and prove a formula for the number of self-conjugate partitions
with largest part k.
B.4. For any partition λ, set
∞
n(λ) = ∑ (j − 1)λj .
j=1
326 B. Partitions
Show that for every partition λ, we have

∞
λ′k
n(λ) = ∑ ( ).
k=1 2
B.5. For any partition λ and any box x in the Ferrers diagram of
λ, the hook-length h(x) of x is given by
h(x) = λj + λ′k − j − k + 1,
where x is in row j and column k of the Ferrers diagram of λ.
′
∑ h(x) = n(λ) + n(λ ) + ∣λ∣.
x∈λ
B.6. For any partition λ and any box x in the Ferrers diagram of
λ, the content c(x) of x is given by
c(x) = k − j,
where x is in row j and column k of the Ferrers diagram of λ.
′
∑ c(x) = n(λ ) − n(λ).
x∈λ
B.7. Show that for every partition λ, we have
∑ (h(x) − c(x) ) = ∣λ∣ .
2 2 2
x∈λ
Appendix C
Permutations
The combinatorics of symmetric functions is built on partitions, but

symmetric functions themselves come from permutations, and how
they act on polynomials. We assume you have seen permutations
from a combinatorial point of view, as sequences in which each of
1, 2, . . . , n appears exactly once. This view of permutations is impor-
tant to our study of symmetric functions, but permutations play a
more central role as bijections from [n] ∶= {1, 2, . . . , n} to itself. And
at a key moment we will want to understand a certain determinant
as a sum over permutations. In this appendix we collect some use-
ful combinatorial facts about permutations, and we develop in detail
some algebraic facts about permutations.
C.1. Permutations as Bijections

We start with the idea of a permutation of a set.
Definition C.1. For any finite set A, a permutation of A is a bi-

jection from A to A. We often take A = [n], so a permutation is a
permutation of [n]. For each n ≥ 0, we write Sn to denote the set of
all permutations of [n].
In combinatorics, permutations are often introduced as lists

rather than functions, and in fact permutations can be represented
in a variety of useful ways. For instance, since each permutation is a
327
328 C. Permutations
1 2 3 4 5
4 1 5 2 3
Figure C.1. Two-line notation for a permutation
function on a finite set, we can represent permutations as tables of in-

puts and outputs. The resulting notation is sometimes called two-line
notation, and in Figure C.1 we see the two-line notation for the per-
mutation π with π(1) = 4, π(2) = 1, π(3) = 5, π(4) = 2, and π(5) = 3.
If we agree to list the elements of the domain of a permutation in in-
creasing order, then we need only give the second line in the two-line
notation to specify a permutation. This gives us the one-line notation
for a permutation π ∈ Sn , which is the sequence π(1), π(2), . . . , π(n).
For example, the one-line notation for the permutation in Figure C.1
is 41523.
By drawing a certain picture of a permutation π, we can reach
a fruitful compromise between writing the domain and range of π
separately (as in two-line notation) and not writing the domain at all
(as in one-line notation). In particular, for any permutation π, we
can construct a directed graph for π whose vertices are the numbers
1, 2, . . . , n, and for which we have an edge j → k whenever π(j) = k. In
Figure C.2 we have the directed graph corresponding to the permuta-
tion 7162534. Since each permutation is a bijection, the graph of each
permutation will be a collection of disjoint cycles. In cycle notation
we record the elements of each cycle in order, in parentheses. For
example, in cycle notation the permutation 7162534 can be written
(7421)(36)(5). We can begin each cycle at any of its entries, and we
can write disjoint cycles in any order, so 7162534 can also be written
1 7 3
2 4 6 5
Figure C.2. The directed graph for the permutation 7162534

C.1. Permutations as Bijections 329
in cycle notation as (63)(2174). Here we have omitted the cycle with

a single element; this is a common convention in permutation circles,
which we generally follow.
Since permutations are bijections, it is natural to consider the
inverse of a permutation. In particular, if π ∈ Sn , then π −1 is the
permutation in Sn such that, for all 1 ≤ j, k ≤ n, we have π(j) = k if
and only if π −1 (k) = j.
Example C.2. Suppose π and σ are permutations in S9 such that

in one-line notation we have π = 718236954, and in cycle notation we
have σ = (718)(3695). Write π −1 and σ −1 in both one-line and cycle
notation.
Solution. Since π(1) = 7, we must have π −1 (7) = 1. Similarly, since

π(2) = 1, we must have π −1 (1) = 2. Continuing in this way, we find
π −1 = 245986137. In cycle notation, π −1 = (12497)(358).
Since σ(7) = 1, we must have σ −1 (1) = 7. Similarly, since σ(1) = 8,
we must have σ −1 (8) = 1. Continuing in this way, we find σ −1 =
(817)(5963). In one-line notation, σ −1 = 725493816. □
As our solution to Example C.2 suggests, if π ∈ Sn is given in

cycle notation, then we can compute π −1 by reversing the entries in
each cycle. On the other hand, if π is given in one-line notation, then
we can compute π −1 by writing down the positions of the elements
of π. That is, for each j, the entry π −1 (j) is the position of j in π.
Finally, if π is given in two-line notation, then we can compute π −1 by
permuting the columns of π so its bottom row is in increasing order,
and then switching the two rows.
The fact that permutations are bijections also means function
composition gives us a natural product on Sn . That is, if π, σ ∈ Sn ,
then πσ = π ○ σ, so (πσ)(j) = π(σ(j)) for all j, 1 ≤ j ≤ n. We will
usually apply this product when our permutations are written in cycle
notation.
Example C.3. If π = (32)(51) and σ = (24)(153), then find πσ and

σπ in both cycle notation and one-line notation. Do the permutations
π and σ commute?
330 C. Permutations
Solution. First note that we have (πσ)(1) = π(σ(1)) = π(5) = 1.

Similarly, (πσ)(2) = 4, (πσ)(3) = 5, (πσ)(4) = 3, and (πσ)(5) = 2.
Combining these, we find πσ = 14532 in one-line notation and πσ =
(2435) in cycle notation.
In the same way we find (σπ)(1) = 3, (σπ)(2) = 1, (σπ)(3) = 4,
(σπ)(4) = 2, and (σπ)(5) = 5. Therefore, σπ = 31425 in one-line
notation and σπ = (1342) in cycle notation.
Since σπ ≠ πσ, the permutations π and σ do not commute. □
As is the case for integers, expressing permutations as products is

more useful and interesting than just computing products. We have
seen that every permutation is a product of disjoint cycles, and it
is not difficult to show that disjoint cycles commute. But we can
also write every permutation as a product of transpositions, which
are cycles involving just two elements.
Proposition C.4. Every permutation is a product of transpositions.
Proof. Since every permutation is a product of disjoint cycles, it is

sufficient to show that every cycle is a product of transpositions. But
this follows from the fact that
(a1 a2 ⋅ ⋅ ⋅ ak ) = (a1 a2 )(a2 a3 ) ⋅ ⋅ ⋅ (ak−1 ak ). □
We can even write every permutation as a product of adjacent

transpositions, which are transpositions of the form (j, j + 1).
Proposition C.5. Every permutation is a product of adjacent trans-

positions.
Proof. Since every permutation is a product of transpositions, it is

sufficient to show that every transposition is a product of adjacent
transpositions. But this follows from the fact that if a1 < a2 , then
(a1 a2 ) = αβ,
where
α = (a1 , a1 + 1)(a1 + 1, a1 + 2) ⋅ ⋅ ⋅ (a2 − 1, a2 )
and
β = (a2 − 1, a2 − 2)(a2 − 2, a2 − 3) ⋅ ⋅ ⋅ (a1 + 1, a1 ). □
C.2. Determinants and Permutations 331
Proposition C.5 gives us a useful technique for showing every

permutation has a given property: we just need to show the adjacent
transpositions have the property, and that if α and β are permutations
with the property, then their product αβ also has the property.
C.2. Determinants and Permutations

Once we have constructed several families of symmetric functions,
we’ll be interested in expressing the symmetric functions in one fam-
ily in terms of those from another family. In some important cases
the easiest way to do this will be with determinants. If you have
seen determinants before, perhaps in an introductory linear algebra
course, then you may have seen them defined recursively, in terms
of expansions along a row or column. However, we can also express
the determinant of an n × n matrix as a sum over the permutations
in Sn . Writing determinants this way doesn’t usually provide any
computational advantage, but it does allow us to view determinants
combinatorially, and to express certain combinatorial quantities in
terms of determinants.
To recall the recursive definition of the determinant, we first set
some notation. For any n × n matrix A and any j, k with 1 ≤ j, k ≤ n,
we write Ajk or Aj,k to denote the entry of A in row j and column k,
and we write A(j, k) to denote the (n−1)×(n−1) matrix obtained by
removing row j and column k from A. The determinant of A(j, k) is
sometimes called the jkth minor of A. If A is 1 × 1, then det A = A11 ,
and if A is 2 × 2, then det A = A11 A22 − A12 A21 . To compute det A
for n ≥ 3, we can use the Laplace expansion, which says that if n ≥ 2,
then
n
(C.1) det A = ∑ (−1)j+1 A1j det(A(1, j)).
j=1
If we use Laplace expansion to write the terms in the determinant

of a general 3 × 3 matrix A, then we find
det(A) = A11 A22 A33 − A11 A23 A32 − A12 A21 A33
+ A12 A23 A31 + A13 A21 A32 − A13 A22 A31 .
332 C. Permutations
This determinant has 6 = 3! terms, and we can turn each term into a
different permutation in two-line notation by writing the subscripts
of its factors as the columns in our permutation. Alternatively, if we
view a permutation π as a bijection from [n] to [n], then each term in
our determinant has the form ±A1π(1) A2π(2) A3π(3) for some π ∈ S3 .
For example, the term A12 A23 A31 corresponds to the permutation
231. Using a similar computation, we can see that the same pattern
holds for the determinant of a general 4 × 4 matrix A, so we can
completely describe these determinants in terms of permutations if
we can describe their coefficients in terms of permutations. To do
this, we recall the sign of a permutation.
Definition C.6. For any n ≥ 0 and any permutation π ∈ Sn , an

inversion in π is an ordered pair (j, k) with 1 ≤ j < k ≤ n and π(j) >
π(k). We write inv(π) to denote the number of inversions in π, and
we write sgn(π) = (−1)inv(π) . We say π is even whenever sgn(π) = 1
and we say π is odd whenever sgn(π) = −1.
As our 3×3 example suggests, the coefficient of each term in det A

is the sign of its corresponding permutation. That is, we can express
det A in terms of permutations as follows.
Proposition C.7. For any n ≥ 1 and any n × n matrix A, we have
n
(C.2) det A = ∑ sgn(π) ∏ Ajπ(j) .
π∈Sn j=1
Proof. For any n × n matrix A, set
n
(C.3) D(A) = ∑ sgn(π) ∏ Ajπ(j) ;
π∈Sn j=1
we show D(A) = det(A). This is easy to check for n = 1 and n = 2, so

suppose n ≥ 3; we argue by induction on n.
C.2. Determinants and Permutations 333
If we group the terms on the right side of (C.3) according to the

value of π(1), then we find
n n
D(A) = ∑ ∑ sgn(π) ∏ Ajπ(j)
k=1 π∈Sn j=1
π(1)=k
n n
= ∑ A1k ∑ sgn(π) ∏ Ajπ(j) .
k=1 π∈Sn j=2
π(1)=k
We now observe that for each π ∈ Sn with π(1) = k, there is a corre-

sponding σ ∈ Sn−1 defined by
⎧
⎪
⎪π(j + 1) if π(j + 1) < k,
σ(j) = ⎨
⎪
⎪π(j + 1) − 1 if π(j + 1) > k,
⎩
and it is not difficult to show this map is a bijection. We also note
that if π ∈ Sn has π(1) = k, then π has k −1 inversions of the form 1, j,
so sgn(π) = (−1)k−1 sgn(σ) = (−1)k+1 sgn(σ). Moreover, if π ∈ Sn has
π(1) = k and j ≥ 2, then Ajπ(j) = A(1, k)j−1,σ(j−1) , so ∏nj=2 Ajπ(j) =
n−1
∏j=1 A(1, k)jσ(j) . Combining these observations, we find
n n−1
D(A) = ∑ (−1)k+1 A1k ∑ sgn(σ) ∏ A(1, k)jσ(j)
k=1 σ∈Sn−1 j=1
n
= ∑ (−1)k+1 A1k D(A(1, k)).
k=1
By induction we have D(A(1, k)) = det(A(1, k)), so our result follows

from (C.1). □
By adapting our proof of Proposition C.7, we can show that we

can compute det A by expanding along any row or column. In par-
ticular, for any k with 1 ≤ k ≤ n, when we expand along the kth row
we find
n
det A = ∑ (−1)j+k Akj det(A(k, j)),
j=1
and when we expand along the kth column we find

n
det A = ∑ (−1)j+k Ajk det(A(j, k)).
j=1
334 C. Permutations
C.3. Problems
C.1. For each n ≥ 1, set σj = (j, j + 1) for 1 ≤ j ≤ n − 1. Show that
for all j with 1 ≤ j ≤ n − 2, we have σj2 = 1 and σj+1 σj σj+1 =
σj σj+1 σj = (j, j + 2). These identities are known as the braid
relations.
C.2. The cycle type of a permutation π ∈ Sn is the partition λ ⊢ n
such that in cycle notation π consists of a λ1 -cycle, a λ2 -cycle,
etc. For example, if π ∈ S10 is given by π = (47)(5132)(68),
then the cycle type of π is (4, 2, 2, 1, 1). How many permuta-
tions in S11 have cycle type (3, 3, 3, 2)?
C.3. List, in one-line notation, all 18 permutations π ∈ S6 with cycle
type (4, 2) which also have π(1) = 3.
C.4. For each partition λ ⊢ n, let cλ be the number of permutations
in Sn with cycle type λ. Then there exists a constant zλ such
that cλ zλ = n!. Find and prove a formula for zλ in terms of the
multiplicities nj of the parts of λ. The numbers zλ will appear
in Proposition 7.14.
C.5. Fill in the blank, and prove the resulting statement.
For all n ≥ 0, a permutation π ∈ Sn is an invo-
lution (that is, satisfies π = π −1 ) if and only if
the cycle type of π .
C.6. For each n ≥ 0, let in be the number of involutions in Sn . Show
that for all n ≥ 2, we have in = in−1 + (n − 1)in−2 .
C.7. For each n ≥ 2, show that the number of even permutations in
Sn is equal to the number of odd permutations in Sn .
C.8. For each k with 0 ≤ k ≤ (n2 ), find a bijection between the
set of permutations π ∈ Sn with inv(π) = k and the set of
permutations in Sn with inv(π) = (n2 ) − k.
C.9. Find and prove a relationship between inv(π) and inv(π −1 ) for
π ∈ Sn .
C.10. Show that for all π, σ ∈ Sn , we have sgn(πσ) = sgn(π) sgn(σ).
C.11. For each n ≥ 2, let Bn be the set of permutations in Sn in
which 1 and 2 are in different cycles, and let Cn be the set
of permutations in Sn in which 1 and 2 are in the same cycle.
C.3. Problems 335
Show that ∣Bn ∣ = ∣Cn ∣ for all n ≥ 2 by giving a bijection between

these two sets.
C.12. In this problem let
⎛1 2 3 4⎞
⎜5 6 7 8⎟
A=⎜ ⎟
⎜ 9 10 11 12⎟ .
⎜ ⎟
⎝13 14 15 16⎠
Find the term in det A associated with the permutation 3142.
C.13. In this problem let
⎛0 3 −2 0 0⎞
⎜1 −4 0 0 6⎟
⎜ ⎟
A=⎜
⎜0 0 0 −2 0 ⎟
⎟.
⎜ ⎟
⎜5 0 0 0 −1⎟
⎝0 0 4 1 0⎠
(a) Find all π ∈ S5 for which the term corresponding with π
in det A is nonzero.
(b) Use equation (C.2) and your result from part (a) to com-
pute det A.
Bibliography
[AE04] George E. Andrews and Kimmo Eriksson, Integer partitions, Cambridge

University Press, Cambridge, 2004. MR2122332
[Axl14] Sheldon Axler, Linear algebra done right, 3rd ed., Undergraduate Texts in
Mathematics, Springer, Cham, 2015. MR3308468
[BD07] Arthur T. Benjamin and Gregory P. Dresden, A combinatorial proof of
Vandermonde’s determinant, Amer. Math. Monthly 114 (2007), no. 4,
338–341, DOI 10.1080/00029890.2007.11920421. MR2281930
[BQ03] Arthur T. Benjamin and Jennifer J. Quinn, Proofs that really count:
The art of combinatorial proof, The Dolciani Mathematical Expositions,
vol. 27, Mathematical Association of America, Washington, DC, 2003.
MR1997773
[Bre99] David M. Bressoud, Proofs and confirmations: The story of the alternat-
ing sign matrix conjecture, MAA Spectrum, Mathematical Association of
America, Washington, DC; Cambridge University Press, Cambridge, 1999.
MR1718370
[Bre12] Otto Bretscher, Linear algebra: With applications, Prentice Hall, Inc., Up-
per Saddle River, NJ, 1997. MR1429616
[BRT19] S. C. Billey, B. Rhoades, and V. Tewari, Boolean product polynomials,
Schur positivity, and Chern plethysm, arXiv:1902.11165 [math.CO], 2019.
[BTvW06] Louis J. Billera, Hugh Thomas, and Stephanie van Willigenburg, Decom-
posable compositions, symmetric quasisymmetric functions and equality
of ribbon Schur functions, Adv. Math. 204 (2006), no. 1, 204–240, DOI
10.1016/j.aim.2005.05.014. MR2233132
[Buc02] Anders Skovsted Buch, A Littlewood–Richardson rule for the K-
theory of Grassmannians, Acta Math. 189 (2002), no. 1, 37–78, DOI
10.1007/BF02392644. MR1946917
[BZ88] A. D. Berenstein and A. V. Zelevinsky, Involutions on Gel′fand-Tsetlin
schemes and multiplicities in skew GLn -modules (Russian), Dokl. Akad.
Nauk SSSR 300 (1988), no. 6, 1291–1294; English transl., Soviet Math.
Dokl. 37 (1988), no. 3, 799–802. MR950493
[BZ92] A. D. Berenstein and A. V. Zelevinsky, Triple multiplicities for sl(r + 1)
and the spectrum of the exterior algebra of the adjoint representation, J.
337
338 Bibliography
Algebraic Combin. 1 (1992), no. 1, 7–22, DOI 10.1023/A:1022429213282.

MR1162639
[CRK95] Joaquin O. Carbonara, Jeffrey B. Remmel, and Andrius Kulikauskas, A
combinatorial proof of the equivalence of the classical and combinatorial
definitions of Schur function, J. Combin. Theory Ser. A 72 (1995), no. 2,
293–301, DOI 10.1016/0097-3165(95)90066-7. MR1357775
[CvW16] Soojin Cho and Stephanie van Willigenburg, Chromatic bases for sym-
metric functions, Electron. J. Combin. 23 (2016), no. 1, Paper 1.15, 7.
MR3484720
[dBR38] G. de B. Robinson, On the Representations of the Symmetric Group,
Amer. J. Math. 60 (1938), no. 3, 745–760, DOI 10.2307/2371609.
MR1507943
[ER90] Ömer Eğecioğlu and Jeffrey B. Remmel, A combinatorial interpretation
of the inverse Kostka matrix, Linear and Multilinear Algebra 26 (1990),
no. 1-2, 59–84, DOI 10.1080/03081089008817966. MR1034417
[FK96] Sergey Fomin and Anatol N. Kirillov, The Yang-Baxter equation, sym-
metric functions, and Schubert polynomials, Proceedings of the 5th Con-
ference on Formal Power Series and Algebraic Combinatorics (Florence,
1993), Discrete Math. 153 (1996), no. 1-3, 123–143, DOI 10.1016/0012-
365X(95)00132-G. MR1394950
[Fom95] Sergey Fomin, Schensted algorithms for dual graded graphs, J. Algebraic
Combin. 4 (1995), no. 1, 5–45, DOI 10.1023/A:1022404807578. MR1314558
[Ful96] William Fulton, Young tableaux: With applications to representation the-
ory and geometry, London Mathematical Society Student Texts, vol. 35,
Cambridge University Press, Cambridge, 1997. MR1464693
[Gas98] Vesselin Gasharov, A short proof of the Littlewood–Richardson rule, Eu-
ropean J. Combin. 19 (1998), no. 4, 451–453, DOI 10.1006/eujc.1998.0212.
MR1630540
[Ges79] Ira Gessel, Tournaments and Vandermonde’s determinant, J. Graph The-
ory 3 (1979), no. 3, 305–307, DOI 10.1002/jgt.3190030315. MR542554
[GGL16] Pavel Galashin, Darij Grinberg, and Gaku Liu, Refined dual stable
Grothendieck polynomials and generalized Bender–Knuth involutions,
Electron. J. Combin. 23 (2016), no. 3, Paper 3.14, 28. MR3558051
[GNW79] Curtis Greene, Albert Nijenhuis, and Herbert S. Wilf, A probabilistic proof
of a formula for the number of Young tableaux of a given shape, Adv.
in Math. 31 (1979), no. 1, 104–109, DOI 10.1016/0001-8708(79)90023-9.
MR521470
[GS01] David D. Gebhard and Bruce E. Sagan, A chromatic symmetric function
in noncommuting variables, J. Algebraic Combin. 13 (2001), no. 3, 227–
255, DOI 10.1023/A:1011258714032. MR1836903
[GV85] Ira Gessel and Gérard Viennot, Binomial determinants, paths, and
hook length formulae, Adv. in Math. 58 (1985), no. 3, 300–321, DOI
10.1016/0001-8708(85)90121-5. MR815360
[Hag07] James Haglund, The q,t-Catalan numbers and the space of diagonal har-
monics, University Lecture Series, vol. 41, American Mathematical Society,
Providence, RI, 2008. With an appendix on the combinatorics of Macdonald
polynomials. MR2371044
[Hai01] Mark Haiman, Hilbert schemes, polygraphs and the Macdonald positiv-
ity conjecture, J. Amer. Math. Soc. 14 (2001), no. 4, 941–1006, DOI
10.1090/S0894-0347-01-00373-3. MR1839919
[Hum11] Brandon Humpert, A quasisymmetric function generalization of the chro-
matic symmetric function, Electron. J. Combin. 18 (2011), no. 1, Paper
31, 13. MR2776807
Bibliography 339
[Jam09] G. James, The representation theory of the symmetric group, Cambridge

University Press, 2009.
[Knu70] Donald E. Knuth, Permutations, matrices, and generalized Young
tableaux, Pacific J. Math. 34 (1970), 709–727. MR272654
[Knu92] Donald E. Knuth, Two notes on notation, Amer. Math. Monthly 99 (1992),
no. 5, 403–422, DOI 10.2307/2325085. MR1163629
[KT99] Allen Knutson and Terence Tao, The honeycomb model of GLn (C) tensor
products. I. Proof of the saturation conjecture, J. Amer. Math. Soc. 12
(1999), no. 4, 1055–1090, DOI 10.1090/S0894-0347-99-00299-4. MR1671451
[KTW04] Allen Knutson, Terence Tao, and Christopher Woodward, The honeycomb
model of GLn (C) tensor products. II. Puzzles determine facets of the
Littlewood–Richardson cone, J. Amer. Math. Soc. 17 (2004), no. 1, 19–48,
DOI 10.1090/S0894-0347-03-00441-7. MR2015329
[Lan04] Serge Lang, Linear algebra, 3rd ed., Undergraduate Texts in Mathematics,
Springer-Verlag, New York, 1989. MR996636
[Lin73] Bernt Lindström, On the vector representations of induced matroids, Bull.
London Math. Soc. 5 (1973), 85–90, DOI 10.1112/blms/5.1.85. MR0335313
[Loe11] Nicholas A. Loehr, Bijective combinatorics, Discrete Mathematics and its
Applications (Boca Raton), CRC Press, Boca Raton, FL, 2011. MR2777360
[Loe17] Nicholas A. Loehr, Combinatorics, second ed., Chapman and Hall/CRC,
2017.
[LP07] Thomas Lam and Pavlo Pylyavskyy, Combinatorial Hopf algebras and K-
homology of Grassmannians, Int. Math. Res. Not. IMRN 24 (2007), Art.
ID rnm125, 48, DOI 10.1093/imrn/rnm125. MR2377012
[Mac92] I. G. Macdonald, Schur functions: theme and variations, Séminaire
Lotharingien de Combinatoire (Saint-Nabor, 1992), Publ. Inst. Rech. Math.
Av., vol. 498, Univ. Louis Pasteur, Strasbourg, 1992, pp. 5–39, DOI
10.1108/EUM0000000002757. MR1308728
[Mac15] I. G. Macdonald, Symmetric functions and Hall polynomials, 2nd ed.,
Oxford Classic Texts in the Physical Sciences, The Clarendon Press, Oxford
University Press, New York, 2015. With contribution by A. V. Zelevinsky
and a foreword by Richard Stanley; Reprint of the 2008 paperback edition
[ MR1354144]. MR3443860
[MMW08] Jeremy L. Martin, Matthew Morin, and Jennifer D. Wagner, On distin-
guishing trees by their chromatic symmetric functions, J. Combin. The-
ory Ser. A 115 (2008), no. 2, 237–253, DOI 10.1016/j.jcta.2007.05.008.
MR2382514
[MR15] Anthony Mendes and Jeffrey Remmel, Counting with symmetric functions,
Developments in Mathematics, vol. 43, Springer, Cham, 2015. MR3410908
[Niv69] Ivan Niven, Formal power series, Amer. Math. Monthly 76 (1969), 871–
889, DOI 10.2307/2317940. MR252386
[NPS97] Jean-Christophe Novelli, Igor Pak, and Alexander V. Stoyanovskii, A direct
bijective proof of the hook-length formula, Discrete Math. Theor. Comput.
Sci. 1 (1997), no. 1, 53–67. MR1605030
[OS14] Rosa Orellana and Geoffrey Scott, Graphs with equal chromatic
symmetric functions, Discrete Math. 320 (2014), 1–14, DOI
10.1016/j.disc.2013.12.006. MR3147202
[Pat19] Rebecca Patrias, What is Schur positivity and how common is it?, Math.
Intelligencer 41 (2019), no. 2, 61–64, DOI 10.1007/s00283-018-09862-8.
MR3953520
[Pro89] Robert A. Proctor, Equivalence of the combinatorial and the classical
definitions of Schur functions, J. Combin. Theory Ser. A 51 (1989), no. 1,
135–137, DOI 10.1016/0097-3165(89)90086-1. MR993658
340 Bibliography
[Red27] J. Howard Redfield, The Theory of Group-Reduced Distributions, Amer.

J. Math. 49 (1927), no. 3, 433–455, DOI 10.2307/2370675. MR1506633
[Rom15] Dan Romik, The surprising mathematics of longest increasing subse-
quences, Institute of Mathematical Statistics Textbooks, vol. 4, Cambridge
University Press, New York, 2015. MR3468738
[RS98] Jeffrey B. Remmel and Mark Shimozono, A simple proof of the Littlewood–
Richardson rule and applications, Discrete Math. 193 (1998), no. 1-3,
257–266, DOI 10.1016/S0012-365X(98)00145-9. Selected papers in honor
of Adriano Garsia (Taormina, 1994). MR1661373
[Sag01] Bruce E. Sagan, The symmetric group: Representations, combinatorial
algorithms, and symmetric functions, 2nd ed., Graduate Texts in Mathe-
matics, vol. 203, Springer-Verlag, New York, 2001. MR1824028
[Sch61] C. Schensted, Longest increasing and decreasing subsequences, Canadian
J. Math. 13 (1961), 179–191, DOI 10.4153/CJM-1961-015-3. MR121305
[Sta95] Richard P. Stanley, A symmetric function generalization of the chro-
matic polynomial of a graph, Adv. Math. 111 (1995), no. 1, 166–194, DOI
10.1006/aima.1995.1020. MR1317387
[Sta99] Richard P. Stanley, Enumerative combinatorics, vol. 2, Cambridge Univer-
sity Press, 1999.
[Sta11] Richard P. Stanley, Enumerative combinatorics, 2nd ed., vol. 1, Cambridge
University Press, 2011.
[Ste02] John R. Stembridge, A concise proof of the Littlewood–Richardson rule,
Electron. J. Combin. 9 (2002), no. 1, Note 5, 4. MR1912814
[SW16] John Shareshian and Michelle L. Wachs, Chromatic quasisymmetric func-
tions, Adv. Math. 295 (2016), 497–551, DOI 10.1016/j.aim.2015.12.018.
MR3488041
[vL96] Marc A. A. van Leeuwen, The Robinson-Schensted and Schützenberger
algorithms, an elementary approach: The Foata Festschrift, Electron. J.
Combin. 3 (1996), no. 2, Research Paper 15, approx. 32. MR1392500
[vW05] Stephanie van Willigenburg, Equality of Schur and skew Schur functions,
Ann. Comb. 9 (2005), no. 3, 355–362, DOI 10.1007/s00026-005-0263-5.
MR2176598
[Wil05] Herbert S. Wilf, generatingfunctionology, 3rd ed., A K Peters, Ltd., Welles-
ley, MA, 2006. MR2172781
[Wil07] Ellison-Anne Williams, A two parameter chromatic symmetric function,
Electron. J. Combin. 14 (2007), no. 1, Research Paper 22, 17. MR2285826
Index
coefficient elementary symmetric, 24–38,

q-binomial, 64–71 72, 76–77, 149–152, 171–175,
binomial, 57–60, 189 177–178, 200
Littlewood–Richardson, 304–307 forgotten symmetric, 182–183,
content 189, 200
of a semistandard skew tableau, monomial symmetric, 17–18,
121 25–28, 40–41, 181–183,
of a semistandard tableau, 82 191–193, 199–200
correspondence power sum symmetric, 44–48,
RSK, 228–231, 241, 251, 272 71–72, 180, 191–193, 201–202,
256–257, 265–268
descent, 73 Schur, 79, 86–87, 89, 125–129,
134–137, 142–144, 158–159,
equivalence 170–172, 177–178, 191–193,
Knuth, 278–285, 301 199, 220, 248–257, 265–268,
302–303
formal power series, 13–17, 48, skew Schur, 120–129, 188, 291,
193–194 302–303
formula stable Grothendieck, 130–137
Cauchy’s, 199
hook-length, 246 growth diagrams, 232–242
function
chromatic symmetric, 146–153 hook-length, 326
complete homogeneous
symmetric, 39–44, 72, 77, identity
158–159, 164, 170–171, 178, first Jacobi–Trudi, 157–171
191–193, 199–200, 248–254 hockey stick, 70
complete homogeneous second Jacobi–Trudi, 171–178
symmetrix, 160 independent
dual stable Grothendieck, algebraically, 36–38, 44, 47–49
139–144 inner product
341
342 Index
Hall, 196–197, 200–201, 205–207, monomial symmetric, 8–13,

253–254 79–81
inversion, 65–67, 104–105, 166–168, power sum symmetric, 45
332 Schur, 79–81, 86–87, 95–96,
involution, 43, 56, 71, 96, 99, 115, 110–111
140–143, 153, 165, 166, 176, skew Schur, 120–129
187–188, 245, 324, 334 stable Grothendieck, 130–137
ω, 178–183, 196–197
Bender–Knuth, 82–86, 111–113, rectification, 273–278
121, 132, 134, 140 rule
first Pieri, 248–252
jeu de taquin, 274 Littlewood–Richardson, 291–303
Murnaghan–Nakayama, 256–269
major index, 73 second Pieri, 254–255
number sequence
Kostka, 87–89, 113–114, 116, bumping, 215–218
182, 189, 192, 207, 253–255 strip
skew Kostka, 121–122, 252–253, border, 258
255 horizontal, 249, 260–265
Stirling of the first kind, 60–64 vertical, 249, 260–265
Stirling of the second kind, 60, subsequence
62–64 longest increasing, 243, 245
order tableau
dominance, 49 border strip, 266–269
lexicographic, 30–32, 87–89 elegant, 136–137, 220–223
semistandard, 78–81, 120,
partition 168–170, 176–177, 187,
integer, 323–325 209–231, 251, 254, 272–278
reverse plane, 137–138, 140–142, semistandard skew, 120, 124
220 set-valued semistandard,
set, 60 129–130, 132–133
path standard, 244–245
bumping, 215–218 tournament, 114–115
permutation, 327–333 tree, 146, 152
generalized, 197–199, 214, 226,
word
232
Littlewood–Richardson, 301–303
polynomial
reading, 99–100
alternating, 89–96, 115
chromatic, 145–147, 153–154
complete homogeneous
symmetric, 39–40, 53–54,
57–58, 62–63, 69, 71–72, 77
dual stable Grothendieck,
139–144
elementary symmetric, 24–27,
53–54, 57–58, 62–63, 69–70,
72, 76–77
Selected Published Titles in This Series
91 Eric S. Egge, An Introduction to Symmetric Functions and Their
Combinatorics, 2019
90 Nicholas A. Scoville, Discrete Morse Theory, 2019
89 Martin Hils and François Loeser, A First Journey through Logic, 2019
88 M. Ram Murty and Brandon Fodden, Hilbert’s Tenth Problem, 2019
87 Matthew Katz and Jan Reimann, An Introduction to Ramsey Theory,
2018
86 Peter Frankl and Norihide Tokushige, Extremal Problems for Finite
Sets, 2018
85 Joel H. Shapiro, Volterra Adventures, 2018
84 Paul Pollack, A Conversational Introduction to Algebraic Number
Theory, 2017
83 Thomas R. Shemanske, Modern Cryptography and Elliptic Curves, 2017
82 A. R. Wadsworth, Problems in Abstract Algebra, 2017
81 Vaughn Climenhaga and Anatole Katok, From Groups to Geometry
and Back, 2017
80 Matt DeVos and Deborah A. Kent, Game Theory, 2016
79 Kristopher Tapp, Matrix Groups for Undergraduates, Second Edition,
2016
78 Gail S. Nelson, A User-Friendly Introduction to Lebesgue Measure and
Integration, 2015
77 Wolfgang Kühnel, Differential Geometry: Curves — Surfaces —
Manifolds, Third Edition, 2015
76 John Roe, Winding Around, 2015
75 Ida Kantor, Jiřı́ Matoušek, and Robert Šámal, Mathematics++,
2015
74 Mohamed Elhamdadi and Sam Nelson, Quandles, 2015
73 Bruce M. Landman and Aaron Robertson, Ramsey Theory on the
Integers, Second Edition, 2014
72 Mark Kot, A First Course in the Calculus of Variations, 2014
71 Joel Spencer, Asymptopia, 2014
70 Lasse Rempe-Gillen and Rebecca Waldecker, Primality Testing for
Beginners, 2014
69 Mark Levi, Classical Mechanics with Calculus of Variations and Optimal
Control, 2014
68 Samuel S. Wagstaff, Jr., The Joy of Factoring, 2013
67 Emily H. Moore and Harriet S. Pollatsek, Difference Sets, 2013
For a complete list of titles in this series, visit the

AMS Bookstore at www.ams.org/bookstore/stmlseries/.
This book is a reader-friendly introduction
to the theory of symmetric functions, and
Sara Rubinstein, Rubinstein Photo

it includes fundamental topics, such as the
monomial, elementary, homogeneous, and
Schur function bases; the skew Schur functions;
the Jacobi–Trudi identities; the involution Ω;
the Hall inner product; Cauchy’s formula; the
RSK correspondence and how to implement
it with both insertion and growth diagrams; the Pieri rules; the
Murnaghan–Nakayama rule; Knuth equivalence; jeu de taquin; and
the Littlewood–Richardson rule. The book also includes glimpses
of recent developments and active areas of research, including
Grothendieck polynomials, dual stable Grothendieck polyno-
mials, Stanley’s chromatic symmetric function, and Stanley’s
chromatic tree conjecture. Written in a conversational style,
the book contains many motivating and illustrative examples.
Whenever possible it takes a combinatorial approach, using bijec-
tions, involutions, and combinatorial ideas to prove algebraic
results.
The prerequisites for this book are minimal—familiarity with

linear algebra, partitions, and generating functions is all one needs
to get started. This makes the book accessible to a wide array of
undergraduates interested in combinatorics.
45 5 4
33 ӷ 24 = 3
2
35
34
12 13 1123
For additional information

and updates on this book, visit
www.ams.org/bookpages/stml-91
STML/91

Dokumen - Pub An Introduction To Symmetric Functions and Their Combinatorics

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Dokumen - Pub An Introduction To Symmetric Functions and Their Combinatorics

Uploaded by

Copyright:

Available Formats

STUDENT MATHEMATICAL LIBRARY

2010 Mathematics Subject Classiﬁcation. Primary 05E05, 05A05, 05A19.

For additional information and updates on this book, visit

Library of Congress Cataloging-in-Publication Data

Chapter 1. Symmetric Polynomials, the Monomial Symmetric

Chapter 2. The Elementary, Complete Homogeneous, and

Chapter 3. Interlude: Evaluations of Symmetric Functions 53

§3.4. q-Binomial Coeﬃcients 64

Chapter 4. Schur Polynomials and Schur Functions 75

Chapter 5. Interlude: A Rogues’ Gallery of Symmetric

Chapter 6. The Jacobi–Trudi Identities and an Involution

Chapter 7. The Hall Inner Product 191

Chapter 8. The Robinson–Schensted–Knuth Correspondence 209

Chapter 9. Special Products Involving Schur Functions 247

Chapter 10. The Littlewood–Richardson Rule 271

Appendix A. Linear Algebra 309

Appendix B. Partitions 323

Appendix C. Permutations 327

Many excellent books are devoted entirely or in part to the founda-

bases and the relationship between determinants and permutations,

Many of the main results in the theory of symmetric functions

’14, and Tessa Whalen-Wagner ’15. Their interest and enthusiasm

Symmetric polynomials and symmetric functions arise naturally in

topology, graph theory, combinatorics, and elsewhere. But symmet-

1.1. Symmetric Polynomials

Example 1.1. Suppose that in one-line notation we have π = 2413

Solution. By deﬁnition we have

π(f ) = f (xπ(1) , xπ(2) , xπ(3) , xπ(4) )

Since π(1) = 2, π(2) = 4, π(3) = 1, and π(4) = 3, we ﬁnd

π(f ) = x22 + 3x4 x3 − 2x21 x3 .

Similarly, π(g) = 3x1 x23 .

As we can already see in Example 1.1, we will be considering

Proposition 1.2. Suppose f (Xn ) and g(Xn ) are polynomials, c is

(i) π(cf ) = cπ(f );

π(f g) = π(abxa1 1 +b1 ⋅ ⋅ ⋅ xnan +bn )

which is what we wanted to prove.

(iv) By deﬁnition we have

After computing π(f ) for various permutations π and polynomi-

Solution. Every polynomial is invariant under the identity permuta-

Every permutation has its own set of invariant polynomials, but

We leave it as an exercise to use Proposition 1.2 to show Λ(Xn )

Solution. (a) This polynomial is a symmetric polynomial in x1 , x2 , x3 ,

When we apply a permutation π to a polynomial f , it may alter

Deﬁnition 1.6. We say a polynomial f (Xn ) in which every term has

Note that if f is any polynomial in x1 , . . . , xn , then we can group

Example 1.7. We saw in Example 1.5 that 3x1 x2 x3 + x1 x2 + x2 x3 +

Solution. In general fk is the sum of all of the terms of total degree

As we did for Λ(Xn ), we leave it as an exercise to show Λk (Xn )

1.2. The Monomial Symmetric Polynomials

Solution. If f is a symmetric polynomial, then it must be invariant

When we look more closely at our work in Example 1.8, we see we

this monomial under the permutations in Sn . If one monomial is the

Deﬁnition 1.9. Suppose n ≥ 1 and λ is a partition. Then the mono-

We will often have partitions as subscripts, as we do for the mono-

Example 1.10. Compute the four monomial symmetric polynomials

We saw in our solution to Example 1.10 that the polynomials

Proposition 1.11. For all positive integers n and all partitions λ,

Note that if λ ⊢ k, then the monomial symmetric polynomial

Proposition 1.12. If n ≥ 1, k ≥ 0, and n ≥ k, then the set

of monomial symmetric polynomials is a basis for Λk (Xn ). In par-

Proof. First observe that for any monomial xα 1 ⋅ ⋅ ⋅ xn , there is only

one partition λ whose parts are a rearrangement of α1 , . . . , αn , and

tions with λ ≠ µ, then mλ (Xn ) and mµ (Xn ) have no terms in com-