Full Ebook of Real Algebra A First Course 1St Edition Manfred Knebusch Claus Scheiderer Online PDF All Chapter

Real Algebra A First Course 1st Edition
Manfred Knebusch Claus Scheiderer

Visit to download the full and correct content document:
https://ebookmeta.com/product/real-algebra-a-first-course-1st-edition-manfred-knebu
sch-claus-scheiderer/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...
Abstract Algebra: A First Course 2nd Edition Stephen

Lovett
https://ebookmeta.com/product/abstract-algebra-a-first-
course-2nd-edition-stephen-lovett/
Introduction to Real Analysis 3rd Edition Manfred Stoll
https://ebookmeta.com/product/introduction-to-real-analysis-3rd-
edition-manfred-stoll/
Introduction to Real Analysis 3rd Edition Manfred Stoll
https://ebookmeta.com/product/introduction-to-real-analysis-3rd-
edition-manfred-stoll-2/
A First Course in Spectral Theory 1st Edition Milivoje

Luki■
https://ebookmeta.com/product/a-first-course-in-spectral-
theory-1st-edition-milivoje-lukic/
A First Course in Group Theory 1st Edition Bijan Davvaz
https://ebookmeta.com/product/a-first-course-in-group-theory-1st-
edition-bijan-davvaz/
Business Statistics: A First Course 8th Edition David

Levine
https://ebookmeta.com/product/business-statistics-a-first-
course-8th-edition-david-levine/
Business Statistics A First Course 8th Edition David

Levine
https://ebookmeta.com/product/business-statistics-a-first-
course-8th-edition-david-levine-2/
Course 1 - Real Estate Essentials - Humber Real Estate

Education 7th Edition Real Estate Council Of Ontario
(Reco)
https://ebookmeta.com/product/course-1-real-estate-essentials-
humber-real-estate-education-7th-edition-real-estate-council-of-
ontario-reco/
A First Course in Functional Analysis 1st Edition Orr

Moshe Shalit
https://ebookmeta.com/product/a-first-course-in-functional-
analysis-1st-edition-orr-moshe-shalit/
Universitext
Manfred Knebusch
Claus Scheiderer
Real Algebra
A First Course
Translated and with Contributions by
Thomas Unger
Universitext
Series Editors
Carles Casacuberta, Universitat de Barcelona, Barcelona, Spain
John Greenlees, University of Warwick, Coventry, UK
Angus MacIntyre, Queen Mary University of London, London, UK
Claude Sabbah, École Polytechnique, CNRS, Université Paris-Saclay, Palaiseau,
France
Endre Süli, University of Oxford, Oxford, UK
Universitext is a series of textbooks that presents material from a wide variety of
mathematical disciplines at master’s level and beyond. The books, often well class-
tested by their author, may have an informal, personal even experimental approach
to their subject matter. Some of the most successful and established books in the
series have evolved through several editions, always following the evolution of
teaching curricula, into very polished texts.
Thus as research topics trickle down into graduate-level teaching, first textbooks
written for new, cutting-edge courses may make their way into Universitext.
Manfred Knebusch • Claus Scheiderer
Real Algebra
A First Course
With Contributions by Thomas Unger

Manfred Knebusch Claus Scheiderer
Fakultät für Mathematik FB Mathematik
University of Regensburg University of Konstanz
Regensburg, Germany Konstanz, Germany
Translated by
Thomas Unger
School of Mathematics and Statistics
University College Dublin
Dublin, Ireland
ISSN 0172-5939 ISSN 2191-6675 (electronic)

Universitext
ISBN 978-3-031-09799-7 ISBN 978-3-031-09800-0 (eBook)
https://doi.org/10.1007/978-3-031-09800-0
Translation from the German language edition: “Einführung in die reelle Algebra” by Manfred Knebusch
et al., © Friedr. Vieweg & Sohn Verlagsgesellschaft mbH 1989. Published by Friedr. Vieweg & Sohn,
Braunschweig/Wiesbaden. All Rights Reserved.
Mathematics Subject Classification: 12D15, 14P05, 14P10, 12J15, 12J10, 13J30
© Springer Nature Switzerland AG 2022

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of
the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilms or in any other physical way, and transmission or information
storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology
now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors, and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, expressed or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.
This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface
More than 30 years after its publication, we are pleased and surprised with the
ongoing interest in Einführung in die reelle Algebra. Given the vibrant development
of real algebra and geometry in the previous decades, this seems by no means self-
evident.
Real algebra has grown in many directions, partially from within itself, and
exceedingly also through nudges and stimuli from without. While some of these
developments were already discernible at the end of the 1980s, many modern ones
that currently belong to the core of the area—such as the connections with tropical
geometry, or with semidefinite optimization—would have been difficult to predict.
Hence, it is not obvious that a text that was timely and modern in 1989 still serves
current needs in a reasonable manner.
Nevertheless, we are convinced that also from today’s perspective the choice
and exposition of the material in Einführung in die reelle Algebra still constitute a
solid first course on the basic principles and important techniques of real algebra.
For this reason, we decided to keep the text essentially unchanged in the English
translation—even though it was tempting to change the wording here and there, and
to drop some sections in favour of new ones.
A small number of typos, omissions and errors that were present in the German
original have been corrected, and some notation and terminology have been brought
up to date. Several new examples (some in the form of exercises) have been added.
All definitions and propositions have been numbered in a consistent manner. The
original bibliography has been augmented with a number of newer references.
A feature of this translation is the addition of a short fourth chapter that provides
a succinct overview of the most important developments and advances in those parts
of real algebra that are directly related to material covered in Einführung in die reelle
Algebra.
We are grateful to Springer Nature, and Rémi Lodh in particular, for their
endorsement and friendly and professional assistance.
We received support and encouragement from many quarters, and are pleased to
extend special thanks to Oliver Hien and Marcus Tressl, who made valuable remarks
on the German original.
v
vi Preface
Finally, our very special and heartfelt thanks go to Thomas Unger, who translated
the original text in a conscientious and competent manner, and co-authored the final
chapter. It was always a great pleasure to work with him.
Regensburg, Germany Manfred Knebusch

Konstanz, Germany Claus Scheiderer
May 2022
Preface to Einführung in die reelle
Algebra
Algebra textbooks that are currently in common use present real algebra only in
later chapters and then usually rather tersely. This is so for the influential works
of van der Waerden [113], Jacobson [55] and Lang [73], although Jacobson covers
a bit more than the others. Bourbaki’s substantial multivolume text Eléments de
Mathématique shows a similar picture: the volume Algèbre contains just a short
chapter (Chapter 6, Groupes et corps ordonnés) on real algebra. In contrast, a
complete volume (currently counting nine chapters) is dedicated to commutative
algebra, even though on the whole only the elementary part is covered, and not
more than what is absolutely essential for the foundation of the theory, both by
current standards as well as Bourbaki’s own standards.
That being the case, not too many algebraists today seem to perceive real algebra
as a proper branch of algebra at all. This has not always been so. In the nineteenth
century real algebra flourished. The study of real zeroes of a real polynomial in one
variable was at the centre of algebraic interest during the whole century, and was an
indispensable part of any higher mathematical education.
Heinrich Weber’s large three-volume textbook on algebra [114] provides evi-
dence for this fact. As Weber’s research interests were primarily in number theory,
especially in complex multiplication and class field theory, he focussed his textbook
mostly on these topics. Nevertheless he devoted well over a hundred pages in the
first volume to real zeroes of real polynomials.
The twentieth century witnessed a dramatic decline in interest in real algebra.
This trend only seems to have changed since the late 1970s, which is all the more
astonishing since the essential seeds of a modern real algebra, as we understand it
today, are already present in two works by Artin and Schreier from the 1920s [3, 4].
So, what is real algebra? Instead of giving a formal definition—which would
be difficult—we rather answer the question with an analogy that puts real algebra
in parallel with commutative algebra. Commutative algebra can be seen as that
part of algebra that contains the algebraic foundations that are typically important
for algebraic geometry (in particular in its modern, abstract form); and algebraic
geometry is ultimately the study of solution sets of systems of polynomial equations
F (x1 , . . . , xn ) = 0 and non-equations F (x1 , . . . , xn ) = 0. Correspondingly,
vii
viii Preface to Einführung in die reelle Algebra
real algebra provides algebraic methods that typically function as tools for real
algebraic geometry, and in particular semialgebraic geometry, which is the study
of the solution sets of systems of polynomial inequalities F (x1 , . . . , xn ) > 0 or
F (x1 , . . . , xn ) ≥ 0, where the coefficients classically come from the field of real
numbers and more generally from any ordered field.
This analogy provides an argument for why the interest has been so much
stronger in commutative than in real algebra during our century thus far. Indeed,
algebraic geometry experienced a continuous and ultimately triumphant upswing in
the twentieth century, while real algebraic geometry was only practiced in isolation
and then mostly with transcendental methods. Thus, from the geometric point of
view, there was for many decades no engine available that could have pushed real
algebra forward.
It was not until 1987 that the first textbook [11] on real algebraic geometry
was published. Its authors’ report (in the last section of the introduction to this
commendable work [11, p. 4 ff.]) on the strange sleeping beauty slumber of real
algebraic geometry (insofar as it was practiced with algebraic methods) is worth
considering.
In the meantime this slumber has given way to a lively development. We
consider the introduction—or better: discovery—of the real spectrum Sper A of a
commutative ring A by Michel Coste and Marie-Françoise Roy around the year
1979 as the most important trigger.
One can view Sper A as an analogue of the Zariski spectrum Spec A, intro-
duced by Grothendieck. It is well-known that the Zariski spectrum is the key
to Grothendieck’s abstract algebraic geometry. Likewise, the real spectrum is the
key to an abstract semialgebraic geometry. (There is one difference: in contrast to
Spec A, the real spectrum Sper A seems to carry (at least) two important structure
sheafs, the sheaf of abstract Nash functions, introduced by Coste and Roy [29], as
well as the sheaf of abstract semialgebraic functions, introduced by G. Brumfiel and
N. Schwartz [19, 107, 108].)
Real algebra is necessary to understand the latest developments in real algebraic
geometry and to meet the future intellectual challenges in this area. This brings us
to the objective of this book.
Our book is based on two insights: real algebra is a branch of algebra that is
largely autonomous in its foundations, with methods specific to this branch. These
foundations can be successfully taught with little more preparation than the standard
background from a typical one-semester algebra course on linear algebra, group
theory, field theory and ring theory.
We differentiate more precisely between an elementary and a higher real algebra.
The former can be developed without any special prior knowledge and used with
benefit in real algebraic geometry. The latter makes serious use of resources from
other branches of mathematics, especially real algebraic geometry, commutative
algebra, algebraic geometry, model theory and the theory of quadratic forms, but
occasionally also algebraic topology, real analysis and complex analysis. (This list
can certainly be extended.) An analogous distinction can be made in commutative
algebra. A demarcation between “elementary” and “higher” in either area is not
Preface to Einführung in die reelle Algebra ix
entirely objectively possible, but rather is subject to personal points of view and
taste. Furthermore, the higher one goes, the more fluid and arbitrary the boundary
between the two kinds of algebra and their corresponding geometries becomes.
Our book is dedicated to elementary real algebra in the above sense. Another
book on higher real algebra is planned.1 In the current book we do get by with
previous knowledge of the extent sketched above. By adding another twenty to
thirty pages, we could have reduced the requirements further and, for example,
developed everything that is needed from commutative algebra and the theory of
quadratic forms. We have not done so however, since students who are interested in
real algebra more than likely already mastered almost all of the prerequisites for the
current book, and would easily be able to find the few things they may be missing
elsewhere.
One specific feature of real algebra should be pointed out in particular: the major
role that general (Krull) valuation rings play. In most parts of commutative algebra,
valuation rings that are not discrete are only viewed as an aid that can often be
done without. In real algebra on the other hand, general valuation rings are a natural
and even central concept throughout. The reason is that every convex subring of
an ordered field is a valuation ring, but only in rare cases a discrete valuation
ring. This fact was already observed by Krull in the introduction to his pioneering
work Allgemeine Bewertungstheorie [67]. (It can already be found in embryonic
form, without the concept of valuation ring, in the work of Artin and Schreier,
cf. [4, p. 95].) Krull also identified the theory of ordered fields as an important
application area of general valuation theory, but then devoted only one section—
albeit substantial—of his great work to ordered fields [67, §12].
Our book is divided into three chapters. In addition to the Artin-Schreier theory
of ordered fields and the elementary relationships between orderings and quadratic
forms, the first chapter covers some aspects of real algebra from the nineteenth
century. Various methods for determining the number of real zeroes of a real
polynomial are treated (Sturm’s algorithm, Hermite’s method using quadratic forms,
Hurwitz’ Theorem). Another section is devoted to the relationship between the
Cauchy index and the Hankel form, as well as the Bézoutian of a rational function.
The second chapter deals with real valuation theory. Everything we need from
general valuation theory is developed from scratch. The chapter culminates in a
presentation of Artin’s solution of Hilbert’s 17th Problem.
Finally, the third chapter is devoted to the real spectrum. After a short crash
course on the Zariski spectrum, the real spectrum is examined in detail, but only
as a topological space (i.e., without the introduction of a structure sheaf). In the
geometric setting (affine algebras over real closed fields), the points as well as
certain subsets of the real spectrum are described in terms of filter sets. In the last
five sections of the chapter we come to some parts of real algebra in which the real
spectrum proves to be clarifying and helpful, such as the reduced Witt ring of a field,
1 This book was never written.

x Preface to Einführung in die reelle Algebra
Positivstellensätze, preorderings of rings, convex ideals, and the holomorphy ring of

a field. Many of the previously developed techniques and terms are used again.
Experts should note that the book manages without Tarski’s Principle throughout.
(That so much of real algebra can be developed in a meaningful way without using
this principle should hopefully come as a little surprise.) Although we consider
Tarski’s Principle to be extremely important and by no means particularly difficult,
we still count it as part of higher real algebra.
The book originates in a four-hour course by the first author in Regensburg in
the 1986/87 winter semester. This course was followed by an extensive course on
model theory, in which Tarski’s Principle was proved fairly quickly and painlessly
following Prestel [91], and then by a course on higher real algebra.
To the subject matter of the first course we added individual sections here and
there, so that the book now encompasses more than can be handled in one semester.
However, since some sections can be skipped without consequences for further
understanding, it can still be used as a guide for a one-semester course.
We want to guard against the misunderstanding that our pair of opposites “ele-
mentary–higher” is strongly correlated with the pair “easy–difficult”. We consider
elementary real algebra, as presented here, not to be always easy. Beginners may
need to overcome some pedagogical hurdles. Especially in the third chapter, some
things may seem strange at first glance. Our text is also rather terse and requires
the reader to participate intensively with paper and pencil. Our goal was to delight,
motivate and enable the beginner, but not to put all the obstacles out of the way with
lengthy statements, which after all could be quite tiresome.
We would like to thank the editor of the series “Aufbaukurs Mathematik”,
Gerd Fischer, as well as Vieweg-Verlag, and here in particular Ulrike Schmickler-
Hirzebruch, for her considerable understanding and empathy during the preparation
of the manuscript. Marina Franke TEXed numerous versions of the manuscript
with patience and competence; Uwe Helmke, Roland Huber and Michael Prechtl
proofread the manuscript and helped us a lot with comments and suggestions for
improvement. We would like to take this opportunity to thank all of them.
Regensburg
January 1989
Contents
1 Ordered Fields and Their Real Closures . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 1

1.1 Orderings and Preorderings of Fields . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 1
1.2 Quadratic Forms, Witt Rings, Signatures . . . . . . . .. . . . . . . . . . . . . . . . . . . . 5
1.3 Extension of Orderings . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 10
1.4 The Prime Ideals of the Witt Ring . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 13
1.5 Real Closed Fields and Their Field Theoretic
Characterization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 15
1.6 Galois Theoretic Characterization of Real Closed Fields .. . . . . . . . . . 17
1.7 Counting Real Zeroes of Polynomials (without
Multiplicities) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 19
1.8 Conceptual Interpretation of the Sylvester Form . . . . . . . . . . . . . . . . . . . . 26
1.9 Cauchy Index of a Rational Function, Bézoutian and
Hankel Forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 30
1.10 An Upper Bound for the Number of Real Zeroes
(with Multiplicities) .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 41
1.11 The Real Closure of an Ordered Field . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 46
1.12 Transfer of Quadratic Forms . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 48
2 Convex Valuation Rings and Real Places .. . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 53
2.1 Convex Subrings of Ordered Fields . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 53
2.2 Valuation Rings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 57
2.3 Integral Elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 60
2.4 Valuations, Ideals of Valuation Rings . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 63
2.5 Residue Fields and Subfields of Convex Valuation Rings . . . . . . . . . . 70
2.6 The Topology of Ordered and Valued Fields . . . .. . . . . . . . . . . . . . . . . . . . 73
2.7 The Baer–Krull Theorem . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 75
2.8 Places . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 78
2.9 The Orderings of R(t), R((t)) and Quot R{t} . . . .. . . . . . . . . . . . . . . . . . . . 82
2.10 Composition and Decomposition of Places . . . . . .. . . . . . . . . . . . . . . . . . . . 86
2.11 Existence of Real Places of Function Fields . . . . .. . . . . . . . . . . . . . . . . . . . 89
xi
xii Contents
2.12 Artin’s Solution of Hilbert’s 17th Problem and the Sign

Change Criterion .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 94
3 The Real Spectrum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 101
3.1 The Zariski Spectrum. Affine Varieties . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 101
3.2 Reality for Commutative Rings . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 110
3.3 Definition of the Real Spectrum . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 114
3.4 Constructible Subsets and Spectral Spaces . . . . . .. . . . . . . . . . . . . . . . . . . . 120
3.5 The Geometric Setting: Semialgebraic Sets and Filter
Theorems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 127
3.6 The Space of Closed Points . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 135
3.7 Specializations and Convex Ideals . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 140
3.8 The Real Spectrum and the Reduced Witt Ring of a Field .. . . . . . . . . 145
3.9 Preorderings of Rings and Positivstellensätze . . .. . . . . . . . . . . . . . . . . . . . 150
3.10 The Convex Radical Ideals Associated to a Preordering . . . . . . . . . . . . 158
3.11 Boundedness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 166
3.12 Prüfer Rings and the Real Holomorphy Ring of a Field . . . . . . . . . . . . 174
4 Recent Developments .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 187
4.1 Counting Real Solutions . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 187
4.2 Quadratic Forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 188
4.3 Stellensätze . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 189
4.4 Noncommutative Stellensätze . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 192
References .. .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 195
Symbol Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 201
Index . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 203
Chapter 1
Ordered Fields and Their Real Closures
Our starting point is the basic notion for the entire book, the general concept of
orderings of arbitrary fields. Conceived by Artin and Schreier in their foundational
1927 paper [4], it was successfully used by Artin in his solution of Hilbert’s 17th
Problem in the same year [3]. We introduce real closed fields and show that they
have the same algebraic properties as the field of real numbers. Moreover we prove
the existence and uniqueness of a real closure for any ordered field.
The second main topic in this chapter concerns methods for counting real roots
of real polynomials. Instead of trying to calculate or estimate the roots numerically,
we discuss purely algebraic methods that extend from the real numbers to any real
closed base field. These results were mostly found in the nineteenth century or even
earlier, and are connected to famous names such as Descartes, Sturm, Sylvester and
Hermite.
Parallel to these two subjects we give an introduction to the basic notions of
quadratic forms over fields and their algebraic theory. They will find applications in
Chaps. 2 and 3.
1.1 Orderings and Preorderings of Fields
Let K be a field.
Definition 1.1.1 An ordering of K is a subset P of K that satisfies the properties
(O1) P + P ⊆ P , P P ⊆ P ,
(O2) P ∩ (−P ) = {0},
(O3) P ∪ (−P ) = K,
where P + P := {a + b : a, b ∈ P } and P P := {ab : a, b ∈ P }. The pair (K, P )
is called an ordered field.
© Springer Nature Switzerland AG 2022 1

M. Knebusch, C. Scheiderer, Real Algebra, Universitext,
https://doi.org/10.1007/978-3-031-09800-0_1
2 1 Ordered Fields and Their Real Closures
Remark 1.1.2 If we assume (O1) and (O3), then (O2) is equivalent with
(O2’) −1 ∈ P .
Exercise 1.1.3 Let P be an ordering of K. Show that
a ≤P b :⇔ b − a ∈ P (a, b ∈ K)
defines a total order relation ≤P on the set K which satisfies the properties
(i) a ≤P b ⇒ a + c ≤P b + c,
(ii) a ≤P b, c ≥ 0 ⇒ ac ≤P bc
for all a, b, c ∈ K. Conversely, let ≤ be a total order relation on K which satisfies
(i) and (ii). Show that P := {a ∈ K : a ≥ 0} is an ordering of K.
Thus we clearly have a one-one correspondence between orderings of K and total
order relations on K that satisfy (i) and (ii). Usually the latter are called orderings
of K as well. When P is clear we simply write a ≤ b instead of a ≤P b.
Definition 1.1.4 A preordering of K is a subset T of K that satisfies the properties
(P1) T + T ⊆ T , T T ⊆ T ,
(P2) T ∩ (−T ) = {0},
(P3) a 2 ∈ T for all a ∈ K.
Exercise 1.1.5 If T is a preordering of K, show that
a ≤T b :⇔ b − a ∈ T (a, b ∈ K)
defines a partial order relation ≤ on K that is compatible with + and ·.

If T is clear, we simply write a ≤ b instead of a ≤T b.
Remarks 1.1.6
(1) If we assume (P1) and (P3), then (P2) is equivalent with
(P2’) −1 ∈ T .
Note that every ordering is a preordering. Indeed, (P1) is (O1), (P2) is (O2),
and (P3) is a weaker version of (O3) by (P1).
(2) If T = (Tα : α ∈ I ) is a nonempty family of preorderings of K, then α Tα
is also a preordering. If in addition T is upward directed
(i.e., for all α, β ∈ I
there exists γ ∈ I such that Tα ∪ Tβ ⊆ Tγ ), then α Tα is also a preordering.
In particular, if K has a preordering, it will have a smallest one.
We introduce the notation
K 2 := {a12 + · · · + an2 : n ∈ N, a1 , . . . , an ∈ K}.

1.1 Orderings and Preorderings of Fields 3
It is clear that K 2 is contained in every preordering of K, and is itself a preordering

if and only if −1 ∈ K 2 .
Definition 1.1.7 The field K is called real if −1 ∈ K 2 , i.e., when −1 cannot be
written as a sum of squares in K.
The previous discussion shows that a field K has a preordering if and only if it
is real. In this case, K 2 is the smallest preordering on K. Note that all real fields
have characteristic 0.
Preorderings are precisely the intersections of orderings, as we will show next.
Lemma 1.1.8 Let T be a preordering of K and let a ∈ K with a ∈ T . Then
T − aT = {b − ac : b, c ∈ T } is again a preordering of K.
Proof T −aT clearly satisfies (P1) and (P3). We show (P2’): assume on the contrary
that −1 ∈ T − aT . Then −1 = b − ac for some b, c ∈ T . But then c = 0 and so
a = c−2 · c(1 + b) ∈ T , a contradiction.
Lemma 1.1.9 For every preordering T of K there exists an ordering P of K such

that T ⊆ P .
Proof Let M := {T ⊆ K : T is a preordering and T ⊆ T }. Then M is ordered by
inclusion, M = ∅ and Zorn’s Lemma applies (Remark 1.1.6(2)). Thus there exists
a maximal preordering P of K with T ⊆ P . This P is an ordering of K. Indeed,
if a ∈ K, a ∈ P , then P = P − aP by Lemma 1.1.8 and the maximality of P , so
−a ∈ P .
Theorem 1.1.10 Every preordering T of K is an intersection of orderings of K.

Proof The inclusion T ⊆ {P : P is an ordering and T ⊆ P } is trivial. If a ∈ K,
a ∈ T , then T − aT is a preordering and thus contained in an ordering P of K
(Lemma 1.1.9). It follows that a ∈ P since −a ∈ P and a = 0.
Corollary 1.1.11 K has an ordering if and only if K is real.

(This follows already from Lemma 1.1.9.)
Corollary 1.1.12 (E. Artin [3]) Assume char K = 2 and let a ∈ K. Then a ≥ 0
for all orderings of K if and only if a is a sum of squares in K.
Proof If K is real, then K 2 is a preordering and the statement follows from
Theorem 1.1.10. If K is not real and char K = 2, then K 2 = K since
2 2
a+1 a−1
a= + (−1) ·
2 2
for all a ∈ K.
Naturally, we must assume that char K = 2, since otherwise K 2 is a subfield of

K. Next, we consider some examples of orderings. We will see many more examples
later on.
Examples 1.1.13
(1) The field R of real numbers (and thus each of its subfields) has the usual
ordering.
(2) Let Q(t) be the rational function field in one variable over Q and let ϑ ∈ R
be transcendental (over Q). Then f (ϑ) is a well-defined real number for every
f ∈ Q(t). The set
P = {f ∈ Q(t) : f (ϑ) ≥ 0}
is an ordering of Q(t). It is the ordering induced by the field embedding Q(t) →

R, f → f (ϑ).
(3) Let (F, ≤) be an ordered field and F (t) the rational function field in one variable
over F . For f ∈ F (t) the notation f (a) = ∞ signifies that f (t) has no pole
in t = a. Then f (a) ∈ F is well-defined. The notation f (a) = b for a, b ∈ F
implies in particular that f (a) = ∞. The sets

Pa,+ := {0} ∪ (t − a)r f (t) : r ∈ Z, f ∈ F (t) with f (a) = ∞ and f (a) > 0
and

Pa,− := {0} ∪ (a − t)r f (t) : r ∈ Z, f ∈ F (t) with f (a) = ∞ and f (a) > 0
are orderings of F (t) for every a ∈ F . Both orderings extend the ordering of F .
The reason for this notation is that a < t < b with respect to Pa,+ holds for all
b ∈ F with b > a. Thus F (t) is ordered in such a way that the transcendental t
is located “immediately to the right of a” on the “line” F . Similarly, t is located
“immediately to the left of a” with respect to Pa,− .
Notation 1.1.14 If (K, P ) is an ordered field, we denote the sign function with
respect to P by signP : K → {−1, 0, 1}. Thus, signP (0) = 0, and for a ∈ K ∗ we
have signP (a) = 1 if a ∈ P and signP (a) = −1 if a ∈ P .
As usual, | · |P : K → P , |a|P := a · signP (a) denotes the absolute value. If P
is clear, the index P may be omitted.
The meaning of the generalized intervals [a, b]P , [a, b[P , ]a, b]P , ]a, b[P for
a, b ∈ K ∪ {−∞, +∞} is also clear, namely [a, b[P = {x ∈ K : a ≤P x <P b},
]a, ∞[P = {x ∈ K : x >P a}, etc. Here too we will usually drop the index P , and
instead often write [a, b]K , etc. if several fields are considered.
1.2 Quadratic Forms, Witt Rings, Signatures 5
1.2 Quadratic Forms, Witt Rings, Signatures
It is not possible to give an in-depth treatment of the algebraic theory of quadratic

forms in this book. In the next sections we will therefore only develop what we need
from the basics of the theory for later applications in real algebra. We will focus in
particular on the relations between quadratic forms and orderings. For references
and for a detailed treatment of quadratic form theory we recommend the books of
Lam [69, 72] and Scharlau [98]. For Witt rings there is the smaller volume [62].
In this section we assume that K is a field of characteristic char K = 2. All vector
spaces are assumed finite dimensional. Let V be a K-vector space. A symmetric
bilinear form on V (over K) is a K-bilinear map b : V ×V → K such that b(v, w) =
b(w, v) (v, w ∈ V ). A quadratic form on V (over K) is a map q : V → K that
satisfies q(av) = a 2 q(v) (a ∈ K, v ∈ V ) and such that the map bq : V × V → K,
bq (v, w) = q(v + w) − q(v) − q(w) is a (symmetric) K-bilinear form. Every
symmetric bilinear form b on V defines a quadratic form qb on V via qb (v) :=
b(v, v). Since 2 = 0, the maps b → qb and q → 12 bq describe mutually inverse
bijections between the symmetric bilinear forms and the quadratic forms on V , and
as such these concepts are usually identified.
The pair ϕ = (V , b) is called a bilinear space, whereas the pair ϕ = (V , q)
is called a quadratic space. Two quadratic spaces (V , q) and (V , q ) are called
isomorphic (or isometric), denoted (V , q) ∼ = (V , q ), if there exists a vector space
isomorphism φ : V → V such that q = q ◦ φ. Quadratic spaces are most easily
described by symmetric matrices: If B = (bij ) ∈ Mn (K) is symmetric, then B
defines the quadratic space ϕB = (K n , qB ), where

n
qB (x) = bij xi xj x = (x1 , . . . , xn ) ∈ K n .
i,j =1
For symmetric matrices B, B ∈ Mn (K) it is straightforward to check that

ϕB ∼ = ϕB if and only if there exists S ∈ GLn (K) such that B = SBS t . If
B = diag(a1 , . . . , an ) is a diagonal matrix, we simply write ϕB as a1 , . . . , an
Every quadratic space is diagonalizable, i.e., isomorphic to a space a1 , . . . , an .
The hyperbolic plane H is the quadratic space defined by B = 01 10 . In diagonal
form, H ∼ = 1, −1. We write a1 , . . . , an K or HK when we want to emphasize the
base field K.
Given bilinear spaces ϕ = (V , b) and ϕ = (V , b ), we can construct new ones,
namely their orthogonal sum ϕ ⊥ ϕ = (V ⊕ V , b ⊥ b ) and their tensor product
ϕ ⊗ ϕ = (V ⊗ V , b ⊗ b ). Here b ⊥ b and b ⊗ b are defined by (b ⊥ b )(v +
v , w +w ) = b(v, w)+b (v , w ) and (b ⊗b )(v ⊗v , w ⊗w ) = b(v, w) b (v , w ),
respectively. If ϕ and ϕ are represented by matrices B and B , respectively, then
ϕ ⊥ ϕ is represented by B0 B0 and ϕ ⊗ ϕ is represented by the Kronecker product
B ⊗ B . In particular,
a1 , . . . , am ⊥ b1 , . . . , bn ∼
= a1 , . . . , am , b1 , . . . , bn
and
a1 , . . . , am ⊗ b1 , . . . , bn ∼
= ⊥a b .
i,j
i j
The n-fold sum ϕ ⊥ · · · ⊥ ϕ is abbreviated by n × ϕ and the n-fold tensor product

ϕ ⊗ · · · ⊗ ϕ by ϕ ⊗n . Finally, if ϕ = (V , b), we write −ϕ for (V , −b).
Let ϕ = (V , b) be a bilinear space and U ⊆ V a subset. Then U ⊥ := {v ∈
V : b(u, v) = 0 for all u ∈ U } is a linear subspace of V . If V ⊥ = 0, then ϕ (or
b, or q) is called nondegenerate (or regular), otherwise ϕ (or b, or q) is called
degenerate (or singular). For a symmetric matrix B, ϕB is degenerate if and only
if det B = 0. The space V ⊥ is also called the radical of ϕ, denoted Rad(ϕ), and
codimV (V ⊥ ) is called the rank of ϕ, denoted rank(ϕ). Given any quadratic space
ϕ there exists a nondegenerate quadratic space ϕ (unique up to isomorphism) such
that ϕ ∼= ϕ ⊥ Rad(ϕ). For this reason one usually only considers nondegenerate
spaces. Straightforward, yet important, is the following fact: If (V , q) is a quadratic
space and W ⊆ V a subspace such that (W, q|W ) is nondegenerate, then V = W ⊥
W ⊥.
The simplest invariants of nondegenerate quadratic spaces are the dimension and
determinant. We have det(V , q) := (det B)K ∗2 ∈ K ∗ /K ∗2 , where q is represented
by the matrix B. Note that det(ϕ ⊥ ϕ ) = det(ϕ) · det(ϕ ), and det(ϕ ⊗ ϕ ) =
(det ϕ)dim ϕ · (det ϕ )dim ϕ . Of fundamental importance is
Theorem 1.2.1 (Witt’s Cancellation Theorem) Let ϕ1 , ϕ2 and ψ be quadratic
= ϕ2 ⊥ ψ, then ϕ1 ∼
spaces. If ϕ1 ⊥ ψ ∼ = ϕ2 .
The proof can be found in any textbook on quadratic forms (e.g., [62, 69, 72, 98]),
as well as in a number of algebra textbooks (e.g., [54, 55, 73]).
A quadratic form q on V represents an element a ∈ K if there exists 0 = v ∈ V
such that q(v) = a. If q represents all nonzero a ∈ K, then q is called universal.
For example, the hyperbolic plane H is universal. A useful observation: If a1 ∈ K ∗
is represented by q, there exist a2 , . . . , an ∈ K such that q ∼ = a1 , a2 , . . . , an
(n = dim V ). The quadratic space (V , q) is called isotropic if q represents zero, and
anisotropic otherwise. Every nondegenerate quadratic space ϕ has an orthogonal
Witt decomposition ϕ ∼ = ϕ0 ⊥ n × H where n ≥ 0 and ϕ0 is anisotropic. By Witt’s
Cancellation Theorem, ϕ0 (up to isomorphism) and n are uniquely determined. We
call ϕ0 the kernel form and n the Witt index of ϕ. If ϕ0 = 0, i.e., ϕ ∼= n × H , then
ϕ is called hyperbolic. A quadratic space ϕ = (V , q) is hyperbolic if and only if it
is nondegenerate and there exists a subspace U of V such that U = U ⊥ . From Witt
decomposition it follows that every isotropic form is universal.
For the remainder of this section, all quadratic spaces are assumed to be
nondegenerate, unless stated otherwise.
Definition 1.2.2 Two quadratic spaces ϕ, ϕ are called Witt equivalent, denoted
ϕ ∼ ϕ , if their kernel forms are isomorphic. We denote the class of all quadratic
spaces that are Witt equivalent to ϕ by [ϕ] (the Witt class of ϕ), and the set of all
Witt classes of (nondegenerate) quadratic spaces over K by W (K).
Lemma 1.2.3 Let ϕ, ϕ , ψ, ψ be quadratic spaces.

(a) ϕ ∼ ψ ⇔ there exist m, n ≥ 0 such that ϕ ⊥ m · H ∼ = ψ ⊥ n · H.
(b) ϕ ⊗ H ∼= ϕ ⊥ (−ϕ) is a hyperbolic space.
(c) Witt equivalence is compatible with ⊥ and ⊗, i.e., ϕ ∼ ϕ and ψ ∼ ψ imply
ϕ ⊥ ψ ∼ ϕ ⊥ ψ and ϕ ⊗ ψ ∼ ϕ ⊗ ψ .
Proof (a) follows from the uniqueness of Witt decomposition.
(b) Since H ∼ = 1, −1 we have ϕ ⊗ H ∼ = ϕ ⊥ (−ϕ). Since ϕ is diagonalizable,
it suffices to settle the case ϕ = a (a ∈ K ∗ ). Assume that ϕ ⊗ H is represented by
the matrix a0 a0 with respect to a basis (e, f ). Then it is represented by 01 10 with
respect to the basis (e, a −1 f ). This shows ϕ ⊗ H ∼ = H.
(c) follows from (a) and (b).
Theorem 1.2.4 The set W (K) is made into a commutative unitary ring with unit
[1] by the (well-defined) pairings
[ϕ] + [ψ] := [ϕ ⊥ ψ] and [ϕ] · [ψ] := [ϕ ⊗ ψ].
W (K) is called the Witt ring of the field K.

Proof That the pairings are well-defined follows from Lemma 1.2.3(c). That [ϕ] +
[−ϕ] = 0 in W (K) for all ϕ follows from Lemma 1.2.3(b). The other ring axioms
are trivially satisfied.
In order not to overload our notation, we will often write a1 , . . . , an instead of
a1 , . . . , an . Thus, a1 , . . . , an may describe a quadratic form as well as its Witt
class.
Example 1.2.5 With reference to Exercises 1.2.17 and 1.2.18 below we have
W (R) ∼
= Z and W (C) ∼
= Z/2Z.
Exercise 1.2.6 Let K be a finite field of characteristic not 2.

(a) Show that |K ∗ /K ∗2 | = 2.
(b) Show that any 2-dimensional form is universal, and conclude that |W (K)| = 4.
(c) Distinguish the two cases |K| ≡ 1 or − 1 mod 4, and conclude that W (K) ∼ =
Z/2Z[t] ∼ Z/4Z in the second case.
in the first case, and W (K) =
(t 2 )
Example 1.2.7 (Milnor’s Exact Sequence) For every prime number p, we denote
the finite field with p elements by Fp . Then W (Q) can be computed explicitly as an
additive group by the split exact sequence
where i : Z → W (Q) is the unique homomorphism that sends 1 to 1, and ∂2 and
the ∂p for odd primes p are certain group homomorphisms, cf. [72, VI, §4] for the
details. Witt groups of number fields are considerably harder to understand, cf. [72,
VI, §3]. There is a similar split exact sequence for Witt groups of rational function
fields in one variable, cf. [72, IX, §3].
Next we study the relations between the Witt ring of K and orderings of K.
Definition 1.2.8 Let (K, P ) be an ordered field and let ϕ = (V , q) be a quadratic
space over K. Then ϕ (or q) is called positive definite with respect to P if q(v) >P 0
for all 0 = v ∈ V . If −q is positive definite, then q is called negative definite.
The following theorem is well-known:
Theorem 1.2.9 (Sylvester’s Inertia Theorem) Let ϕ be a possibly degenerate
quadratic space over K and P an ordering of K. Then there is an orthogonal
decomposition
ϕ∼
= ϕ+ ⊥ ϕ− ⊥ Rad(ϕ),
where ϕ+ is positive definite with respect to P and ϕ− is negative definite with

respect to P . Furthermore, dim ϕ+ and dim ϕ− only depend on ϕ (but not on ϕ+
and ϕ− ).
Proof We may assume that Rad(ϕ) = 0. The existence of ϕ+ and ϕ− follows from
diagonalizing ϕ. Assume that ϕ ∼ = ϕ+ ⊥ ϕ− ∼ = ψ+ ⊥ ψ− , with ϕ+ , ψ+ positive
definite and ϕ− , ψ− negative definite. Assume that ϕ (resp. ϕ± , resp. ψ± ) is defined
on V (resp. on V± , resp. on W± ). We interpret V± and W± as subspaces of V .
If dim V+ < dim W+ , then V− ∩ W+ = 0 since dim V− + dim W+ > dim V .
This is a contradiction since ϕ is positive and negative definite on V− ∩ W+ . If
dim V+ > dim W+ we get a similar contradiction.
Definition 1.2.10 Let ϕ be a quadratic space over K, P an ordering of K, and

ϕ∼
= ϕ+ ⊥ ϕ− ⊥ Rad(ϕ) a decomposition as in Theorem 1.2.9. Then the integer
signP ϕ := dim(ϕ+ ) − dim(ϕ− )
is called the Sylvester signature of ϕ at P .

Lemma 1.2.11 Let (K, P ) be an ordered field. Then [ϕ] → signP ϕ defines a ring
homomorphism signP from W (K) to Z.
Proof The following equalities are easily verified:
signP (ϕ ⊥ ψ) = signP ϕ + signP ψ,

signP (ϕ ⊗ ψ) = (signP ϕ) · (signP ψ),
signP H = 0.
The statement follows.

Now a critical observation is that, conversely, every ring homomorphism
W (K) → Z arises in this manner. In other words, orderings of K and
homomorphisms W (K) → Z are “the same”:
Theorem 1.2.12 The Sylvester signature defines a bijection sign : P → signP from
the set of orderings of K to the set of ring homomorphisms W (K) → Z.
Proof The map sign is injective since P = {0} ∪ {a ∈ K ∗ : signP [a] = 1}. Let
φ : W (K) → Z be a homomorphism. Define χ : K ∗ → Z by χ(a) := φ[a]. Then
χ : K ∗ → {±1} is a group homomorphism since χ(1) = 1 and χ(ab) = χ(a)χ(b)
(a, b ∈ K ∗ ). We must show that P := {0} ∪ Ker χ is an ordering of K. Then
φ = signP follows immediately. Since [1] + [−1] = 0 in W (K), we have
φ[−1] = χ(−1) = −1, i.e., −1 ∈ P . Then P ∪ (−P ) = K is also clear, and
P P ⊆ P follows since χ is a homomorphism. In order to show P + P ⊆ P
(which completes the proof), we use Lemma 1.2.13 below. Assume that a, b ∈ P .
We may assume a, b, a + b = 0. By Lemma 1.2.13, applying φ to [a, b] gives
χ(a)+χ(b) = χ(a +b)·(1 + χ(a)χ(b)), and χ(a) = χ(b) = 1 gives χ(a +b) = 1.
Lemma 1.2.13 If a, b ∈ K ∗ with a + b = 0, then
a, b ∼
= a + b, (a + b)ab .
Proof Since a + b is represented by a, b, there exists c ∈ K ∗ such that a, b ∼
=
a + b, c. Comparing determinants gives abK ∗2 = (a + b)cK ∗2, i.e., cK ∗2 =
(a + b)abK ∗2. The statement follows.
Theorem 1.2.12 justifies:
Definition 1.2.14 A signature of a field K is a ring homomorphism W (K) → Z.
Remark 1.2.15 We see that a field admits a signature if and only if it is real.
Note that for every field K (with char K = 2) there exists a ring homomorphism
e : W (K) → Z/2Z, defined by e[ϕ] := dim(ϕ) + 2Z. Every signature σ yields a
commutative diagram
Definition 1.2.16 The map e : W (K) → Z/2Z is called the dimension index. Its
kernel is denoted I (K) and is called the fundamental ideal of W (K). (Thus I (K)
consists of the Witt classes of even dimensional quadratic spaces.)
Exercise 1.2.17 Show that if K ∗2 ∪ {0} = {a 2 : a ∈ K} is an ordering of K, then

it is the only ordering of K, and the Sylvester signature gives a ring isomorphism
W (K) → Z.
This is the case for example when K = R or, more generally, when K is any real
closed field, cf. Sect. 1.5.
Exercise 1.2.18 Show that if K is a quadratically closed field (i.e., K ∗2 = K ∗ )
then, up to isomorphism, every nondegenerate quadratic form is determined by its
dimension, and e is a ring isomorphism W (K) → Z/2Z.
1.3 Extension of Orderings
In this section, K denotes a field of characteristic = 2.

Let L/K be a field extension. A bilinear space ϕ = (V , b) over K gives rise to
a bilinear space ϕL = (VL , bL ) over L via extension of scalars: VL = V ⊗K L and
bL is defined by bL (v ⊗ a, v ⊗ a ) = b(v, v )aa for a, a ∈ L, v, v ∈ V . If q
is the quadratic form (over K) associated to ϕ, then qL denotes the quadratic form
(over L) associated to ϕL . If ϕ is represented by a matrix, then ϕL is represented by
the same matrix, but this time considered as a matrix over L.
The form ϕL is nondegenerate if and only if the form ϕ is nondegenerate. If
ϕ is hyperbolic, then ϕL is also hyperbolic, but the converse is in general false
(example?). Consequently, ϕ → ϕL induces a map on Witt classes, iL/K : W (K) →
W (L), which is a ring homomorphism. If M/L is a further field extension, then
iM/K = iM/L ◦ iL/K .
1.3 Extension of Orderings 11
Definition 1.3.1 Let L/K be a field extension.

(a) An ordering P of K extends to an ordering Q of L if P ⊆ Q (and so, P =
K ∩ Q).
(b) A signature σ of K extends to a signature τ of L if σ = τ ◦ iL/K . We also write
τ | σ.
One should convince oneself that for orderings P of K and Q of L, P extends to
Q if and only if signP extends to signQ .
Proposition 1.3.2 Let L/K be a field extension and P an ordering of K. The
following statements are equivalent:
(i) P extends to an ordering Q of L;
(ii) −1 is not contained in the semiring T generated by P and L2 in L (i.e., the
smallest subset T of L such that P ∪ L2 ⊆ T , T + T ⊆ T and T T ⊆ T );
(iii) every quadratic form p1 , . . . , pn with p1 , . . . , pn ∈ P ∗ := P \ {0} is
anisotropic over L.
A further equivalent statement will be listed in Proposition 1.4.4.
Proof (i) ⇒ (iii): If p1 b12 + · · · + pn bn2 = 0, with b1 , . . . , bn ∈ L, then b1 = · · · =
bn = 0 since Q is an ordering.
(iii) ⇒ (ii): We can write T = {p1 b12 + · · · + pm bm 2 : m ≥ 1, p ∈ P ∗ , b ∈ L}.
i i
If −1 ∈ T , then −1 = p1 b12 + · · · + pm bm 2 and the form 1, p , . . . , p is isotropic
1 m
over L.
(ii) ⇒ (i): By assumption T is a preordering of L. Every ordering Q of L such
that Q ⊇ T (cf. Lemma 1.1.9) extends P .
√
Proposition 1.3.3 Let P be an ordering of K, d ∈ K and L = K( d). Then P
extends to L if and only if d ∈ P .
Proof If d ∈ P , then P has no extension to L since d ∈ L∗2 . Assume thus that
d ∈ P and, without loss of generality, that K = L. We verify Proposition 1.3.2(ii).
Assume for the sake of contradiction that there is an equality

m
√
−1 = pi (ai + bi d)2
i=1
in L with m ≥ 1, pi ∈ P and ai , bi ∈ K. It follows in particular (by comparing

√ m
coefficients with respect to the K-basis (1, d) of L) that −1 = pi (ai2 + dbi2 ),
and so −1 ∈ P , a contradiction. i=1
√
We will see later that if√d ∈ P (and d ∈ K), then there are precisely two
extensions of P to L = K( d) (cf. Corollary 1.11.7).
Example 1.3.4 The unique
√ ordering of Q has precisely two extensions,
√ P+ and P− ,
to√the field L = Q( 2). They are distinguished by the fact that 2 ∈ P+ and
− 2 ∈ P− . They are the only orderings of L and correspond to the two possible
embeddings of L into R.
In the following results we consider field extensions for which every ordering
extends:
Proposition 1.3.5 Let L/K be a finite field extension of odd degree. Then every
ordering of K extends to L.
By Proposition 1.3.2, this proposition follows from
Theorem 1.3.6 (T.A. Springer) If L/K is a finite field extension of odd degree and
q is an anisotropic quadratic form over K, then qL is anisotropic over L.
Proof We may assume without loss of generality that L = K(α) is a simple field
extension. Let f ∈ K[t] be the minimal polynomial of α over K. We proceed
by induction on n = [L : K] = deg f . Assume that n > 1 is odd and that the
statement is true for all smaller degrees. Let q = a1 , . . . , am be an anisotropic
form over K. Assume for the sake of contradiction that qL is isotropic. Then there
exist g1 , . . . , gm , h ∈ K[t] with deg gi < n (i = 1, . . . , m) and not all gi = 0, such
that
a1 g1 (t)2 + · · · + am gm (t)2 = f (t)h(t) (1.1)
in K[t]. We may assume that gcd(g1 , . . . , gm ) = 1.

Let d be the maximum of the degrees of the gi . Then n + deg h = 2d since q is
anisotropic. Since d < n we have deg h < n and deg h is odd. In particular, h has an
irreducible factor h1 = h1 (t) of odd degree. For the extension field E = K[t]/(h1 )
of K we thus have [E : K] < n and [E : K] is odd. On the other hand, qE is
isotropic by (1.1). This contradicts the induction hypothesis.
Proposition 1.3.7 If L/K is a purely transcendental field extension (not necessar-

ily finitely generated), then every ordering of K extends to L.
Using Zorn’s Lemma, we may assume that L = K(t) is simple transcendental
and specify explicit extensions (cf. Example 1.1.13(3)). A different proof (again
using Proposition 1.3.2) can be obtained from:
Proposition 1.3.8 If L/K is purely transcendental and q is an anisotropic
quadratic form over K, then qL is also anisotropic.
Proof We may assume without loss of generality that L = K(t) is simple
transcendental. If q = a1 , . . . , am (with ai ∈ K ∗ ) and qL is isotropic, then there
exist polynomials g1 , . . . , gm ∈ K[t], not all 0, with a1 g1 (t)2 + · · · + am gm (t)2 = 0
in K[t]. By considering again the maximal occurring degree, it follows that q is
isotropic.
Remark 1.3.9 From Theorem 1.3.6 and Proposition 1.3.8 it follows in particular
that the homomorphism iL/K : W (K) → W (L) is injective in case L/K is finite of
odd degree or purely transcendental.
1.4 The Prime Ideals of the Witt Ring 13
1.4 The Prime Ideals of the Witt Ring
In this section, K denotes a field of characteristic = 2.

We already know some of the prime ideals of the Witt ring W (K), namely the
fundamental ideal I (K), as well as the kernels of the signatures of K. We will see
that these are essentially the only prime ideals. Thus, Witt rings have a very simple
and clear prime ideal structure, a fact that was observed surprisingly late (J. Leicht,
F. Lorenz in [77] and D.K. Harrison in [44], both in 1970).
Theorem 1.4.1 (The Prime Ideals of W (K))
(a) If K is not real, then I (K) is the only prime ideal of W (K).
(b) If K is real, then the kernels pσ of the signatures σ of K are precisely the
minimal prime ideals of W (K). Every other prime ideal is either equal to I (K),
or is of the form p = pσ + p · W (K) for some signature σ and some prime
number p = 2, both uniquely determined by p.
Thus, pσ + 2 · W (K) = I (K) for every signature σ , as already observed in
Sect. 1.2.
Proof Let p be a prime ideal of W (K), κ(p) = Quot W (K)/p and π : W (K) →
π
W (K)/p the residue map. The composition φ : Z → W (K) − → W (K)/p is
∗
surjective since the Witt classes ξ = [a] (a ∈ K ) generate W (K) as a ring
and ξ 2 − 1 = 0 in W (K), hence ξ ≡ 1 or ξ ≡ −1 mod p.
If char κ(p) = 0, then φ is also injective, and thus an isomorphism, and p is the
kernel of the signature φ −1 ◦ π.
If char κ(p) = p > 0, then φ induces an isomorphism φ : Z/pZ → ∼ W (K)/p.
Assume that p = 2. Then p = I (K) since there is only one ring homomorphism
W (K) → Z/2Z (the generators [a], a ∈ K ∗ , are units in W (K)). Finally, assume
−1
that p > 2. Then +1 = −1 (in Z/pZ), and χ : a → φ ◦ π[a] (a ∈ K ∗ )
defines a group homomorphism χ : K ∗ → {±1}. (Note that χ(K ∗ ) ⊆ {±1} holds
by the argument above.) This map satisfies χ(−1) = −1. Following the proof of
Theorem 1.2.12 verbatim, it follows that P := {0} ∪ Ker χ is an ordering of K. Let
σ := signP . Since φ is an isomorphism, it follows from the commutative diagram
σ

that p = Ker W (K) − → Z → Z/pZ , i.e., that p = pσ + p · W (K). It is clear that
σ and p are uniquely determined by p. This establishes the proof.
Since in every (commutative) ring the nilradical (i.e., the set of all nilpotent
elements) is equal to the intersection of all prime ideals ([55, 68, 73], see also
Proposition 3.1.11), we obtain immediately
Corollary 1.4.2 Let ϕ be a nondegenerate quadratic space over K. The following

statements are equivalent:
(i) ϕ ⊗n is hyperbolic for some n ∈ N;
(ii) dim ϕ is even, and signP ϕ = 0 for every ordering P of K.
If K is ordered, the parity condition on dim ϕ can of of course be omitted
from (ii).
Applying Corollary 1.4.2 to ϕ = 1, 1 = 2 × 1 gives
Corollary 1.4.3 If K is not real, then there exists n ∈ N such that 2n · W (K) = 0.
Having determined the prime ideals of W (K), we can state a further criterion for
when orderings extend:
Proposition 1.4.4 Let L/K be a field extension. A signature of K extends to L if
and only if it vanishes on the kernel of iL/K : W (K) → W (L).
For the proof we require a simple fact from commutative algebra:
Lemma 1.4.5 If φ : A → B is an injective ring homomorphism and p a minimal
prime ideal of A, then there exists a prime ideal q of B such that p = φ −1 (q).
Proof Since S := φ(A\p) does not contain zero, we have that S −1 B = 0 and every
prime ideal q of S −1 B satisfies p = φ −1 j −1 (q ) , where j : B → S −1 B denotes
the canonical homomorphism.
Proof of Proposition 1.4.4 Let I := Ker iL/K and let σ be a signature of K. If σ
has an extension τ , then σ = τ ◦ iL/K , i.e., σ (I ) = 0.
Conversely, assume that σ (I ) = 0, i.e., I ⊆ pσ := Ker σ . Then pσ /I is a
minimal prime ideal of W (K)/I by Theorem 1.4.1. By Lemma 1.4.5 (applied to
−1
W (K)/I → W (L)) there exists a prime ideal q of W (L) such that pσ = iL/K (q).
Since Z ∼= W (K)/pσ → W (L)/q and again by Theorem 1.4.1, it follows that
W (K)/pσ → W (L)/q is an isomorphism, i.e., q determines an extension τ of σ .
In Sect. 1.3 we proved the injectivity of the map iL/K for extensions L/K that are
of odd degree or that are purely transcendental. Next, we will determine the kernel
of iL/K for quadratic extensions (and so with Proposition 1.4.4 give a new proof of
Proposition 1.3.3):
√
Proposition 1.4.6 Let a ∈ K ∗ \ K ∗2 and L = K( a). Then the kernel of iL/K is
the ideal of W (K) generated by 1, −a.
A more precise statement is:
Proposition 1.4.7 Let q be an anisotropic quadratic form over K. There exist
quadratic forms q , q over K with q ∼ = q ⊥ 1, −a ⊗ q such that qL is
anisotropic.
1.5 Real Closed Fields and Their Field Theoretic Characterization 15
Proof By induction on dim q. The case dim q = 1 is clear. Assume thus that
dim q ≥ 2 and without loss of generality that qL is isotropic. If (V , b) denotes
the bilinear space associated to q, then there exist v, w ∈ V , not both zero, such that
√ √
0 = qL (v + a w) = q(v) + a · q(w) + a · b(v, w).
It follows that q(v)+a q(w) = 0 = b(v, w). Since q is anisotropic, q(v), q(w) = 0.
Let W := Kv + Kw ⊆ V . Then dim W = 2 (if v, w were linearly dependent, then
we would have b(v, w) = 0), and q|W ∼= −ac, c = 1, −a⊗c for c := q(w). In
particular, q|W is nondegenerate. Since V = W ⊥ W ⊥ , we can apply the induction
hypothesis to q|W ⊥ and so conclude the proof.
1.5 Real Closed Fields and Their Field Theoretic

Characterization
After our excursion into the theory of quadratic forms we return to real algebra in
the narrower sense. This section is fundamental for everything that follows.
Definition 1.5.1 A field is called real closed if it is real and has no proper algebraic
extension that is real.
The field R of real numbers is a well-known example of a real closed field.
Proposition 1.5.2 Let K be a field. The following statements are equivalent:
(i) K is real closed;
(ii) there exists an ordering P of K that cannot be extended to any proper algebraic
extension of K;
(iii) K ∗2 ∪ {0} = {a 2 : a ∈ K} is an ordering of K, and every polynomial (in one
variable) of odd degree over K has a zero in K.
Moreover, if these conditions are satisfied, then K ∗2 ∪ {0} is the only ordering of K.
Proof The final comment is clear by (iii) (every ordering P satisfies K ∗2 ∪{0} ⊆ P ).
(i) ⇒ (ii) is clear since a field has an ordering if (and only if) it is real.
(ii)√⇒ (iii): If it were true that K ∗2 ∪ √{0} = P , then there would exist d ∈ P
with d ∈ K, and P would extend to K( d) (Proposition 1.3.3), a contradiction.
Furthermore, K has no proper extensions of odd degree by Proposition 1.3.5.
(iii) ⇒ (i): Let L ⊃ K be a finite proper field extension of K. We will show that
L is not real. By (iii), [L : K] is a power of 2. Since K has an ordering by (iii), it
follows that char K = 0 and in particular that L/K is separable. Let L1 be the Galois
L over K, G = Gal(L1 /K) and H = Gal(L1 /L). Then H is a subgroup
closure of
of G, and g∈G H g = {1} by Galois theory, −1
where H g := g H g. By elementary
g
group theory it follows that |G| divides g∈G [G : H ] . By assumption (iii) this
number is a power of 2, and so G is a 2-group. From Sylow’s Theorems (elementary
group theory again) it follows that there exists a subgroup H of G of index 2 with
H ⊂ H . Let F be the fixed field of H . Then K ⊂ F ⊂ L and [F : K] = 2.
√
There exists a ∈ K with F = K( a). Since K ∗2 ∪ {0} is an √ ordering of K and
a∈ / K ∗2 ∪{0}, there exists b ∈ K ∗ with a = −b2 . Hence −1 = ( a/b)2 is a square
in F , and so F is not real. We conclude that L is not real either.
Remark 1.5.3 Condition (iii) in Proposition 1.5.2 is clearly equivalent to

(iii’) K is real, a or −a is a square in K for every a ∈ K, and every polynomial of
odd degree over K has a zero in K.
In the literature it is common to denote real closed fields with capital letters
R, S, . . . . We will usually follow this convention.
Theorem√ 1.5.4 (Fundamental Theorem of Algebra) If R is a real closed field,
then R( −1) is algebraically closed.
Proof (Gauss) We use Proposition 1.5.2(iii). Every
√ finite extension of R (and so
also every finite extension of C := R(i), i := −1) has 2-power degree. Assume
for the sake of contradiction that C = R(i) is not algebraically closed. As in the
proof of Proposition 1.5.2 it follows that C has an extension F with [F : C] = 2.
Thus, there exist a, b ∈ R such that α := a + bi is not a square in C. Hence, b = 0.
Since R 2 is an ordering of R, there exists c ∈ R with a 2 + b 2 = c2 and c > 0, as
well as x, y ∈ R with x > 0, sign y = sign b and x 2 = (c + a)/2, y 2 = (c − a)/2.
(Note that c ± a are positive and that all signs are of course understood to be with
respect to the unique ordering of R.) Now we have (x + iy)2 = (x 2 − y 2 ) + 2xyi,
and also x 2 − y 2 = a and (2xy)2 = b2 . Since sign
√ (xy) = sign b we have 2xy = b,
and it follows that (x + iy)2 = α, contradicting α ∈ C.
A very strong converse to this theorem is given by Theorem 1.6.1.
Corollary 1.5.5 Let R be a real closed field and let K ⊆ R be a subfield. Then K
is real closed if and only if K is (relatively) algebraically closed in R.
√
Proof Assume that K is algebraically closed in R and let i = −1. Then K(i) is
also algebraically closed in R(i). Indeed, if α = a + bi ∈ R(i) is algebraic over
K(i) (a, b ∈ R), then also over K, and the same holds for α = a − bi, thus also for
a = (α + α)/2 and b = i(α − α)/2. It follows that a, b ∈ K, hence α ∈ K(i). Thus
K(i) is algebraically closed by Theorem 1.5.4 and it follows immediately that K is
real closed.
The converse direction is trivial by definition.
1.6 Galois Theoretic Characterization of Real Closed Fields 17
Another characteristic of real closed fields is their relative rigidity:

Proposition 1.5.6 Let R be a real closed field.
(a) Every (ring) endomorphism ϕ of R is order preserving, i.e., a ≤ b ⇒ ϕ(a) ≤
ϕ(b) for all a, b ∈ R.
(b) If K ⊆ R is a subfield and R/K is algebraic, then the identity is the only K-
automorphism of R.
Proof (a) This follows immediately from R 2 = {a ∈ R : a ≥ 0}.
(b) Let ϕ ∈ AutK (R) and a ∈ R. Since R/K is algebraic, {ϕ n (a) : n ∈ N} is a
finite set. If it were true that ϕ(a) = a, thus for instance ϕ(a) > a, it would follow
from (a) by induction that a < ϕ(a) < ϕ 2 (a) < · · · , a contradiction.
If R is real closed, we will henceforth denote the uniquely determined ordering
of R by ≤ without further comment (as we already did several times above).
1.6 Galois Theoretic Characterization of Real Closed Fields
Let K be a field of arbitrary characteristic with algebraic closure K. This section is

dedicated to a proof of the following beautiful result of Artin and Schreier [4]:
Schreier, 1927) If K = K and [K : K] < ∞, then
Theorem 1.6.1 (E. Artin, O.√
K is real closed and K = K( −1).
Proof The elementary proof that we present here goes back to J. Leicht [75]. We
break it down into several steps. Let n = [K : K].
(1) The field K is perfect.
Assume K is not perfect. Then char K = p > 0, and there exists a ∈ K with
α := a 1/p ∈ K. But then also α 1/p ∈ K(α). Indeed, if α = β p with β ∈ K(α), then
p
NK(α)/K (β) = NK(α)/K (α) = (−1)p−1 a = a since (−1)p−1 = 1. Iterating
this step, we obtain field extensions of K of degree p, p2 , p3 , . . . , contradicting
[K : K] < ∞.
It follows that K/K is a finite Galois extension. Let L be a maximal intermediate
field of K/K, different from K, and let q := [K : L]. (Note that q is prime!)
(2) char K = q.
Assume for the sake of contradiction that char K = p = q. By Artin–Schreier
theory for Galois extensions of degree p in characteristic p we have that K is
L-isomorphic to L[t]/(t p − t − a) for some a ∈ L (see, for example, [96,
Theorem 12.2.1], [73, VI, Theorem 6.4] or [55, Vol. II, §8.11]). We only need the
consequence that ψ(L) = L for the map ψ : K → K defined by ψ(b) := bp − b.
Note that ψ is additive, i.e., ψ(b + b ) = ψ(b) + ψ(b ). Consider the trace
tr = trK/L : K → L. Then tr ◦ψ = (ψ|L )◦tr. Indeed, if b ∈ K and if b1 , . . . , bp are
the L-conjugates of b in K, then ψ(b1 ), . . . , ψ(bp ) are the L-conjugates of ψ(b),
and it follows that tr (ψ(b)) = ψ(bi ) = ψ( bi ) = ψ(tr b). Hence the diagram
i i
commutes. Since ψ and tr are surjective (the second map since K/L is separable),
the map ψ|L : L → L is also surjective, a contradiction.
(3) q = 2 (and char K = 2).
L contains the q-th roots of unity since [K : L] = q and the q-th cyclotomic
polynomial is of degree q − 1. Hence, since char K = q, there exists a ∈ L with
2
K = L(a 1/q ) (see for instance [55, Vol. I, §4.7] or [73, VI, §6]). For β := a 1/q we
obtain as in (1) that (NK/L (β))q = (−1)q−1 a. Since a is not a q-th power in L, q
must be even, thus q = 2.
√
(4) K = L( −1).
√
Let i = −1 ∈ K, and let a ∈ L and β ∈ K as in (3), thus with √ β = a
4
and K = L(β ). Since NK/L (β) = −β , we have β /NK/L (β) = −1, and so
2 2 4 2
√
K = L( −1).
(5) L = K, thus K = K(i).
If K = L, then also K(i) = K. Carrying out steps (1) to (4) for K(i) instead of
K leads to a contradiction by (4).
(6) K is real (and thus real closed).
Since i ∈ K it suffices to show that the sum of two squares in K is a square
(Proposition 1.5.2). Thus, let a, b ∈ K ∗ . Since K is algebraically closed, there are
c, d ∈ K with a+bi = (c+di)2. Hence, a−bi = (c−di)2 and a 2 +b 2 = (c2 +d 2 )2 .
This proves the theorem.
Let Ks denote the separable closure of K.
Corollary 1.6.2 If char K = 0 and [Ks : K] < ∞, then K = Ks .
Proof Steps (2)–(6) of the proof can be carried out for Ks just as for K.
We finish this section with a consequence of the theorem of Artin and Schreier
for absolute Galois groups of fields. These are profinite groups (i.e., projective
limits of finite groups). For the basic aspects of infinite Galois theory, [54, Vol. III,
§IV.2], [55, Vol. II, §8.6] or [12, Ch. V, §10], for example, can be consulted, but for
understanding what follows this is not required.
1.7 Counting Real Zeroes of Polynomials (without Multiplicities) 19
Corollary 1.6.3 Let K be a field with separable algebraic closure Ks and absolute
Galois group = Gal (Ks /K). All elements of finite order > 1 in are involutions
(i.e., have order 2), and the map τ → Fix (τ ) (the fixed field of τ in Ks = K) is a
bijection from the set of involutions in to the set of real closed overfields of K in
K. In particular, contains elements of finite order > 1 if and only if K is real.
1.7 Counting Real Zeroes of Polynomials (without

Multiplicities)
One of the oldest problems in real analysis consists of determining the number
and location of the real zeroes of a polynomial f (t) with real coefficients. The
first general answer was formulated by J. C. F. Sturm (1803–1855) in the form of
an algorithm, the so-called Sturm sequence, that determines the number of zeroes
of f in any given interval, counted without multiplicities. Another solution of
the problem, based on the examination of certain quadratic forms, was found by
Ch. Hermite (1822–1901). Both methods will be presented in this section.
It should be remarked that these results used to be part of the mathematics
curriculum at the turn of the nineteenth century (see for example H. Weber’s
Lehrbuch der Algebra [114]).
Sturm’s solution, found in 1829, initiated an historically very interesting and to
some extent stormy period in real algebra. (Sturm published the full proof only
in 1835.) Hermite’s theorem (Theorem 1.7.17) was anticipated by C.G.J. Jacobi
(1804–1851) and his student and friend C.W. Borchardt (1817–1880) (Theo-
rem 1.7.15), and was proved in 1853 by Hermite and—essentially independently—
also by J.J. Sylvester (1814–1897). In addition, Sylvester, A. Cayley (1821–1895)
and Sturm established interesting links between the theorems of Sturm and Hermite.
In addition to the original literature, the reader can find full particulars of this history
in the substantial text [65] of Krein and Naimark, as well as in [39].
We start with a number of classical results, well-known from real analysis, that
stay valid for rational functions over arbitrary real closed fields.
In the remainder of this section R denotes a real closed field.
Proposition 1.7.1 Let f (t) = t n +a1 t n−1 +· · ·+an ∈ R[t] be a monic polynomial.
Then all real zeroes of f in R) are in the interval [−M, M],
of f (i.e., all zeroes
where M := max 1, |a1 | + · · · + |an | .
Proof If 0 = a ∈ R is such that f (a) = 0, then a = −(a1 + a2a −1 + · · ·+ an a 1−n ),
and so |a| ≤ 1 or |a| ≤ |a1 | + · · · + |an | by the triangle inequality.
Proposition 1.7.2 (Intermediate Value Theorem) Let f ∈ R(t) and a, b ∈ R

with a < b, such that f has no pole in [a, b]. If f (a)f (b) < 0, then f has an odd
number of zeroes in ]a, b[, counted with multiplicities.
Proof If f = g/ h with polynomials g, h, where h is nonzero on [a, b], then
we may replace f by f h2 = gh, and thus assume that 0 = f ∈ R[t]. Let
a1 ≤ · · · ≤ ar be the real roots of f , counted with multiplicities (r ≥ 0).
By the Fundamental Theorem of Algebra (Theorem 1.5.4) there exist c ∈ R ∗
and irreducible monic quadratic polynomials p1 , . . . , ps (s ≥ 0) with f (t) =
c · (t − a1 ) · · · (t − ar ) p1 (t) · · · ps (t). Since the pk only take positive values, we
have
f (a)
r
a − aj
−1 = sign = sign .
f (b) b − aj
j =1
The number of j with a < aj < b is therefore odd.
Lemma 1.7.3 Let 0 = f ∈ R(t) be a rational function and a ∈ R a zero of f . The

logarithmic derivative f /f of f changes sign from minus to plus at t = a (it has a
pole at t = a), i.e., there exists ε > 0 such that f /f is negative on ]a − ε, a[ and
positive on ]a, a + ε[.
Proof There exist g, h ∈ R[t] with g(a)h(a) = 0 and f (t) = (t − a)d g(t)/ h(t)
for some d ∈ N. Then
f (t) d g (t) h (t)

= + − ,
f (t) t −a g(t) h(t)
from which the statement follows.
Proposition 1.7.4 Let 0 = f ∈ R(t) and a, b ∈ R with a < b and f (a) = f (b) =
0. If f has neither poles nor zeroes in (a, b), then the number of zeroes of f in
]a, b[, counted with multiplicities, is odd.
Corollary 1.7.5 (Rolle’s Theorem) If f ∈ R(t), f (a) = f (b) ∈ R for a, b ∈

R, a < b, and if f has no poles in [a, b], then f has a zero in ]a, b[.
Proof of Proposition 1.7.4 Let ε > 0 be such that f neither has zeroes in
]a, a + ε], nor in [b − ε, b[, and such that a + ε < b − ε. By Lemma 1.7.3 we
have
f (a + ε) f (b − ε)
· < 0,
f (a + ε) f (b − ε)
and by Proposition 1.7.2 the number of zeroes of f /f in ]a + ε, b − ε[ is odd. This

number equals the number of zeroes of f in ]a, b[.
In what follows we assume that f ∈ R[t] is a fixed non-constant polynomial and

that a, b ∈ R with a < b and f (a)f (b) = 0. Unless stated otherwise, zeroes will
be counted without multiplicities.
We consider
Sturm’s Problem Determine the number of distinct real zeroes of the polynomial
f in the interval [a, b].
Definition 1.7.6 The Sturm sequence of f is the sequence (f0 , f1 , . . . , fr ) of

polynomials, defined recursively as follows: let f0 = f, f1 = f and
f0 = q1 f1 − f2 ,
f1 = q2 f2 − f3 ,
..
.
fr−2 = qr−1 fr−1 − fr ,
fr−1 = qr fr ,
where r ≥ 1, f0 , . . . , fr , q1 , . . . , qr ∈ R[t], fi = 0 and deg fi < deg fi−1 for

i = 2, . . . , r. These conditions determine r, as well as the fi (and qi ) uniquely.
If we change the minus signs on the right hand side into plus signs, we obtain
the usual version of Euclid’s algorithm for the pair f, f . This sign difference is
essential for the determination of the Sturm sequence, but plays no role in the
computation of the greatest common divisor. In particular, fr = gcd(f, f ).
Notation 1.7.7 For (c0 , . . . , cr ) ∈ R r+1 we denote by Var(c0 , . . . , cr ) the num-
ber of sign changes in the given sequence, after cancelling all zeroes. We let
Var(0, . . . , 0) := −1. For example, for the sequence (1, 0, −2, 1, 0, 4, −1, 1, 0)
we obtain Var = 4. For x ∈ R we let V (x) := Var (f0 (x), f1 (x), . . . , fr (x)), where
(f0 , . . . , fr ) is the Sturm sequence of f .
Theorem 1.7.8 (Sturm, 1829) Let f ∈ R[t] be a non-constant polynomial, and

a, b ∈ R with a < b and f (a)f (b) = 0. Then the number of distinct zeroes of f in
the interval ]a, b[ equals V (a) − V (b).
Proof We proceed in two steps.
(1) Let (g0 , . . . , gr ) be a sequence in R[t] \ {0} that satisfies properties (a)–(d)
below:
(a) g0 (a)g0 (b) = 0;
(b) gr (a)gr (b) = 0, and gr is semidefinite on [a, b] (i.e., either gr (x) ≥ 0 or
gr (x) ≤ 0 for all x ∈ [a, b]);
(c) if c ∈ [a, b] and g0 (c) = 0, then g0 (x)g1 (x) changes sign from minus to
plus at x = c;
(d) if c ∈ [a, b] and gi (c) = 0, 0 < i < r, then gi−1 (c)gi+1 (c) < 0.
For x ∈ R let W (x) := Var (g0 (x), . . . , gr (x)). We claim that g0 has exactly W (a)−
W (b) distinct zeroes in [a, b].
To prove the claim, we first note that for every c ∈ R with g0 (c) · · · gr (c) = 0
the function W (x) is constant on a neighbourhood of x = c. Let g := g0 · · · gr , and
let c ∈ [a, b] be a zero of g. We distinguish three cases:
If g0 (c) = 0, then a < c < b, and the function x → Var (g0 (x), g1 (x)) has value
1 on ]c − ε, c[ and value 0 on [c, c + ε[, for some ε > 0. This follows from (c).
If gi (c) = 0, 0 < i < r, then by (d) there exists ε > 0, such that gi−1 (x)gi+1 (x)
is negative on ]c − ε, c + ε[. Hence, x → Var (gi−1 (x), gi (x), gi+1 (x)) is constant
equal to 1 on ]c − ε, c + ε[ (note that for every α ∈ R, Var(−1, α, 1) =
Var(1, α, −1) = 1 !).
If gr (c) = 0, and g0 (c) = 0 or r > 1, then gr−1 (c) = 0 by (d), and x →
Var (gr−1 (x), gr (x)) is constant on {x : 0 < |x − c| < ε} for some ε > 0, since gr
is semidefinite on [a, b].
From these three cases we obtain for all c ∈ [a, b]: If g0 (c) gr (c) = 0, then W (x)
is constant on a neighbourhood of x = c; if g0 (c) = 0, then W (x) is constant on
a punctured neighbourhood of x = c; if g0 (c) = 0, there exists ε > 0 and N ∈ Z
with W (x) = N for c − ε < x < c and W (x) = N − 1 for c < x < c + ε.
The claim then follows since g0 and gr do not vanish at the boundary points a, b.
(2) Now let f be as in the statement of the theorem and let (f0 , . . . , fr ) be the
Sturm sequence of f . We let gi := fi /fr ∈ R[t] (i = 0, . . . , r) and W (x) :=
Var (g0 (x), . . . , gr (x)) (x ∈ R). Then the sequence (g0 , . . . , gr ) satisfies (a)–(d)
from (1). Indeed, since gi−1 = hi gi − gi+1 (i = 1, . . . , r − 1) and gr−1 = hr gr for
suitable hi ∈ R[t], and since gr = 1, it follows that gi−1 and gi have no zeroes in
common (i = 1, . . . , r), and so (d) follows recursively. (a) and (b) are clear, and (c)
follows from Lemma 1.7.3 since g0 g1 = ff /fr2 .
Furthermore, g0 and f0 = f have the same zeroes (if fr (c) = 0, then also
f (c) = 0, and the zero c occurs with larger multiplicity in f than in fr !). It thus
follows from (1) that f has precisely W (a) − W (b) distinct zeroes in [a, b]. The
proof then follows from the observation that V (a) = W (a) and V (b) = W (b) since
fr (a) fr (b) = 0.
We state a more general version of Sturm’s Theorem, also due to Sturm. We make
use of
Definition 1.7.9 Let f ∈ R[t] and a, b ∈ R with a < b and f (a)f (b) = 0. A
generalized Sturm sequence of f on [a, b] is a sequence (f0 , . . . , fr ) in R[t] \ {0}
with r ≥ 1 that satisfies the properties
(1) f0 = f ;
(2) fr (a)fr (b) = 0, and fr is semidefinite on [a, b];
(3) for 0 < i < r and c ∈ [a, b] with fi (c) = 0, we have fi−1 (c)fi+1 (c) < 0.
(Warning: The Sturm sequence of f is usually not a generalized Sturm sequence of
f in the sense of Definition 1.7.9!)
Let f ∈ R[t] and a, b as in Definition 1.7.9. Let (f0 , . . . , fr ) be a generalized

Sturm sequence of f on [a, b] and, once again, V (x) := Var (f0 (x), . . . , fr (x))
(x ∈ R).
Theorem 1.7.10 (Sturm)

V (a) − V (b) = # c ∈ ]a, b[ : f (c) = 0 and ff1 changes sign from − to + at c

− # c ∈ ]a, b[ : f (c) = 0 and ff1 changes sign from + to − at c .
The statement is clear by the proof of Theorem 1.7.8! (We only have to count the
different types of sign changes of ff1 .)
Remark 1.7.11 From the proof we see that simplifications of the original Sturm
algorithm are possible. For example:
(a) The original Sturm sequence f0 = f, f1 = f , . . . may be terminated at fs
provided that fs (a) fs (b) = 0 and fs is semidefinite on [a, b].
(b) For a remainder fi+1 in the modified Euclidean algorithm (Definition 1.7.6) it
is allowed to leave out those factors that are semidefinite on [a, b] and that do
not vanish at a, b.
Exercise 1.7.12 Let (K, ≤) be an ordered field, and let

n
n
f = tn + ai t n−i = (t − ξj )
i=1 j =1
be a monic polynomial in K[t] that splits over K (with ai , ξj ∈ K). Prove that
ξ1 , . . . , ξn ≥ 0 ⇔ (−1)i ai ≥ 0 for i = 1, . . . , n,
and similarly with strict instead of non-strict inequalities.

Next, we discuss Hermite’s method. Let K be an arbitrary field and f (t) =
t n + a1 t n−1 + · · · + an ∈ K[t] a monic polynomial (n ≥ 1). Let α1 , . . . , αn be the
zeroes of f in the algebraic closure of K. For r = 0, 1, 2, . . . , let sr = sr (f ) :=
α1r + · · · + αnr . Then sr (f ) ∈ K since sr (f ) is symmetric in the αi . For instance,
s0 (f ) = n, s1 (f ) = −a1 , s2 (f ) = a12 − 2a2 , s3 (f ) = −a13 + 3a1 a2 − 3a3, and so
on.
Exercise 1.7.13 Prove the Newton identity
sk + sk−1 a1 + sk−2 a2 + · · · + s1 ak−1 + kak = 0
for all k ≥ 0, where we put ak = 0 for k > n. Conclude for r ≥ 0 that sr =

sr (a1 , . . . , an ) is a polynomial in a1 , . . . , an with integer coefficients.
Definition 1.7.14 The Sylvester form S(f ) of f is the n-dimensional quadratic

form over K defined by the n × n-matrix
sj +k−2 (f ) 1≤j,k≤n
.
n
In other words, S(f )(x1 , . . . , xn ) = sj +k−2 (f )xj xk .
j,k=1
Theorem 1.7.15 (Jacobi, Borchardt, Hermite) Let R be a real closed field and
f ∈ R[t] a monic non-constant polynomial. Then the rank of √ S(f ) equals the
number of distinct roots of f in the algebraic closure C = R( −1) of R, and
the signature of S(f ) equals the number of distinct real roots of f .
Note that therefore S(f ) always has nonnegative signature!
Proof Let β1 , . . . , βr be the pairwise distinct real roots of f , and γ1 , γ1 , . . . , γs , γs
the pairwise distinct non-real roots of f , where r, s ≥ 0 and α → α denotes
the nontrivial R-automorphism of√C. Denote the multiplicity of βj by mj and the
multiplicity of γj by nj . Let i = −1 ∈ C, and let Re α, Im α for α ∈ C have their
usual meaning. Then

S(f ) (x1 , . . . , xn ) = αjk+l−2 xk xl
1≤j,k,l≤n
n
n 2
= αjk−1 xk
j =1 k=1

r 2
s 2 2
= mj βjk−1 xk + nj γjk−1 xk + γj k−1 xk
j =1 k j =1 k k

r 2
= mj βjk−1 xk
j =1 k

s 2 2
+2 nj Re(γjk−1 )xk − Im(γjk−1 )xk ,
j =1 k k
and we obtain S(F ) in diagonal form. Indeed, letting

uj (x) := βjk−1 xk (j = 1, . . . , r),
k

vj (x) := Re(γjk−1 )xk , wj (x) := Im(γjk−1 )xk (j = 1, . . . , s),
k k
we have

r
s
S(f ) = mj u2j + 2 nj (vj2 − wj2 ),
j =1 j =1
and the linear forms u1 , . . . , ur , v1 , w1 , . . . , vs , ws are linearly independent.

(The transition matrix arises from the Vandermonde matrix associated to
β1 , . . . , βr , γ1 , γ1 , . . . , γs , γs via two elementary transformations.) It follows that
S(f ) ∼
= r × 1 ⊥ s × H ⊥ (n − r − 2s) × 0,
from which we immediately see that rank S(f ) = r + 2s, sign S(f ) = r.
The following modification makes it possible to also count real zeroes in
intervals:
Definition 1.7.16 Let K be a field, f ∈ K[t] a monic polynomial of degree n ≥ 1,
and λ ∈ K. The Sylvester form of f with parameter λ is the quadratic form

(x1 , . . . , xn ) → λ sj +k−2 (f ) − sj +k−1 (f ) xj xk
1≤j,k≤n
over K, denoted by Sλ (f ).
Theorem 1.7.17 (Hermite, Sylvester, 1853) Let R be real closed and f ∈ R[t] a
monic polynomial of degree n ≥ 1. For λ ∈ R we have
√
rank Sλ (f ) = # a ∈ R( −1) : f (a) = 0 and a = λ ,

sign Sλ (f ) = # a ∈ R : f (a) = 0, a < λ − # a ∈ R : f (a) = 0, a > λ .
Proof We use the notation from the previous proof.

Sλ (f ) (x1 , . . . , xn ) = (λ − αj )αjk+l−2 xk xl
1≤j,k,l≤n

n 2
= (λ − αj ) αjk−1 xk
j =1 k

r
= mj (λ − βj )u2j
j =1

s
+ nj (λ − γj )(vj + iwj )2 + (λ − γj )(vj − iwj )2 .
j =1
We rewrite [· · · ] as follows:
[· · · ] = (λ − γj + λ − γj )(vj2 − wj2 ) + 2i(λ − γj − λ + γj )vj wj
= 2 (λ − Re γj )(vj2 − wj2 ) + 4 Im (γj )vj wj

=: 2ϕj (vj , wj ).
Hence,

r
s
Sλ (f ) = mj (λ − βj )u2j + 2nj ϕj (vj , wj ).
j =1 j =1
The forms ϕj are hyperbolic (j = 1, . . . , s). Indeed, if γj = aj + ibj with aj , bj ∈

R, then ϕj is represented by the matrix

λ − aj bj
bj −(λ − aj )
which has negative determinant since bj = 0. We conclude that
Sλ (f ) ∼
= λ − β1 , . . . , λ − βr ⊥ s × H ⊥ (n − r − 2s) × 0.
1.8 Conceptual Interpretation of the Sylvester Form
Let K be a field of characteristic not 2.

In this section we will present a more conceptual interpretation of the Sylvester
form of a polynomial, namely as the trace form of a certain algebra. We start with
generalities about trace forms of algebras.
By a K-algebra we mean a commutative (and associative) unital finite-
dimensional algebra over K. Let A be a K-algebra. For a ∈ A, let λ(a) be the
K-linear endomorphism b → ab of A, and trA/K (a) the trace of λ(a) (over K).
The K-linear form trA/K : A → K is called the trace of A over K.
The following rules are easily verified:
Lemma 1.8.1
(a) If K /K is a field extension and A = A ⊗K K , then
trA/K (a) = trA /K (a ⊗ 1) for all a ∈ A.

1.8 Conceptual Interpretation of the Sylvester Form 27
(b) If A1 , . . . , Ar are K-algebras and A = A1 × · · · × Ar , then

r
trA/K (a1 , . . . , ar ) = trAi /K (ai ) for all (a1 , . . . , ar ) ∈ A.
i=1
(c) trA/K (Nil A) = 0 (where Nil A = {a ∈ A : a n = 0 for some n ∈ N} is the

nilradical of A).
Assume now that A = K[t]/(f ), where f ∈ K[t] is a monic polynomial of
degree n ≥ 1. We denote the image of t in A by τ . Let α1 , . . . , αn be the roots of f
in the algebraic closure K of K.
n
Proposition 1.8.2 For g ∈ K[t] we have trA/K (g(τ )) = g(αj ).
j =1
Proof By Lemma 1.8.1(a) we may assume K = K. Let β1 , . . . , βr be the distinct

roots of f and e1 , . . . , er their multiplicities, i.e., f (t) = (t − β1 )e1 · · · (t − βr )er .
Let Ai = K[t]/(t −βi )ei (i = 1, . . . , r). Then A → A1 ×· · ·×Ar is an isomorphism
by the Chinese Remainder Theorem, and by Lemma 1.8.1(b) we may assume r = 1,
i.e., f (t) = (t − β)e . There exists h ∈ K[t] such that g(t) = g(β) + (t − β)h(t).
Since (τ − β)h(τ ) ∈ Nil A, and using Lemma 1.8.1(c), we conclude trA/K (g(τ )) =
trA/K (g(β)) = e · g(β).
Notation 1.8.3
(a) Let A be a K-algebra and a ∈ A. Then [A, a]K denotes the quadratic space
over K defined by the quadratic form x → trA/K (ax 2 ) on A.
(b) For polynomials f, g ∈ K[t], f = 0, let
SylK (f ; g) := K[t]/(f ), g(τ ) K
and
SylK (f ) := SylK (f ; 1).
We usually drop the index K in the notation.

Lemma 1.8.1 immediately yields the following computational rules for these
forms:
Lemma 1.8.4 Let A, A , A1 , . . . , Ar be K-algebras.
(a) If ϕ : A → A is a K-algebra isomorphism, then
[A, a] ∼
= A , ϕ(a) for all a ∈ A.
(b) If K /K is a field extension then, for all a ∈ A,
[A, a]K ⊗K K ∼
= [A ⊗K K , a ⊗ 1]K (over K ).
(c) A1 × · · · × Ar , (a1 , . . . , ar ) ∼
= [A1 , a1 ] ⊥ · · · ⊥ [Ar , ar ] (ai ∈ Ai ).
(d) Nil A ⊆ Rad[A, a] for all a ∈ A.
Lemma 1.8.5 Let f, g ∈ K[t] with f = 0.

(a) If K /K is a field extension, then
SylK (f ; g) ⊗K K ∼
= SylK (f ; g) (over K ).
(b) Let f1 , . . . , fr ∈ K[t] be nonzero and pairwise relatively prime, then
Syl(f1 · · · fr ; g) ∼
= Syl(f1 ; g) ⊥ · · · ⊥ Syl(fr ; g).
(c) Syl(f e ; g) ∼
= e ⊗ Syl(f ; g) ⊥ 0, . . . , 0 (e ∈ N).
(d) Syl(af ; g) = Syl(f ; g) and Syl(f ; ag) ∼ = a ⊗ Syl(f ; g) for any a ∈ K ∗ .
(e) Syl (t − a; g(t)) = g(a) for any a ∈ K.
Proof (a) and (b) follow from Lemma 1.8.4, and (d) and (e) are clear. We show (c):
let A = K[t]/(f e ), A = K[t]/(f ), and π : A → A the residue class map. By
Proposition 1.8.2 we have trA/K (a) = e · trA /K (πa) for all a ∈ A, from which the
proof follows.
Corollary 1.8.6 Let f, g ∈ K[t] with f = 0. The following statements are

equivalent:
(i) SylK (f ; g) is nondegenerate;
(ii) f is separable, and f and g are relatively prime.
Proof Since (i) and (ii) are not affected by field extensions, we may assume that
K = K is algebraically closed. Let α1 , . . . , αr be the distinct zeroes of f , with
respective multiplicities e1 , . . . , er ∈ N. By Lemma 1.8.5 we have
r
Syl(f ; g) ∼
= ⊥ Syl (t − α )
j =1
j
ej
; g(t)
r
∼
= ⊥
j =1
ej ⊗ Syl t − αj ; g(t) ⊥ (ej − 1) × 0
r
∼
= ⊥
j =1
ej g(αj ) ⊥ (ej − 1) × 0 ,
from which the statement follows.

1.8 Conceptual Interpretation of the Sylvester Form 29
Proposition 1.8.7 Let f ∈ K[t] be a monic polynomial of degree n ≥ 1. The

Sylvester form S(f ) of f satisifies S(f ) ∼ = Syl(f ), whereas the Sylvester form
Sλ (f ) of f with parameter λ satisfies Sλ (f ) ∼
= Syl(f ; λ − t).
Proof Let α1 , . . . , αn be the zeroes of f in K and let g ∈ K[t] be arbitrary. With
respect to the basis 1, τ, . . . , τ n−1 of A := K[t]/(f ), the quadratic space Syl(f ; g)
is represented by the matrix (bj k )1≤j,k≤n , with bj k = trA/K τ j +k−2 g(τ ) . We have
j +k−2
bj k = ni=1 αi g(αi ) by Proposition 1.8.2, from which the statements follow
with g = 1 and g = λ − t, respectively.
We finish this section by showing that the theorems of Sylvester and Hermite
(Theorems 1.7.15 and 1.7.17) are quite easily obtained by means of trace form
calculus.
To this end, assume that K = R is real closed, and that f ∈ R[t] is monic
of degree n ≥ 1 with distinct real roots β1 , . . . , βr and distinct non-real roots
γ1 , γ1 , · · · , γs , γs . Assume again that the βj and γj have multiplicities mj and nj ,
respectively. We let qj (t) := (t − γj )(t − γj ) (j = 1, . . . , s). Then

r
s
f (t) = (t − βj )mj qj (t)nj ,
j =1 j =1
and by Lemma 1.8.5 it follows that for all g ∈ R[t],

r s
Syl(f ; g) ∼
= ⊥
(b) j =1
Syl (t − βj )mj ; g ⊥ ⊥ Syl(q
j =1
nj
j ; g)
r s
∼
= ⊥ Syl(t − β ; g) ⊥ ⊥ Syl(q ; g) ⊥ 0, . . . , 0
(c) j =1
j
j =1
j
s
∼
= g(β1 ), . . . , g(βr ) ⊥
(e)
⊥ Syl(q ; g) ⊥ 0, . . . , 0.
j =1
j
The forms√ Syl(qj ; g) are all hyperbolic (or zero in case qj | g) since R[t]/(qj ) and
C = R( −1) are isomorphic as R-algebras, and [C; α]R ∼ = 1, −1 = H for all
α ∈ C ∗ . (Choosing β ∈ C with 2αβ 2 = i, the form [C; α]R is represented by the
matrix 01 10 with respect to the basis β, −iβ.)
It follows that
Syl(f ; g) ∼
= g(β1 ), . . . , g(βr ) ⊥ t × H ⊥ (n − r − 2t) × 0
for some 0 ≤ t ≤ s, and in particular that
S(f ) ∼
= Syl(f ) = Syl(f ; 1) ∼
= r × 1 ⊥ s × H ⊥ (n − r − 2s) × 0
and
Syl(f ; λ − t) ∼
= λ − β1 , . . . , λ − βr ⊥ s × H ⊥ (n − r − 2s) × 0,
from which the statements of the theorems of Sylvester and Hermite can be read off
directly.
In addition to the works [65] and [39], mentioned before, the reader may want
to compare the current topic and related themes with the article [10] that contains
interesting historical remarks about Sylvester.
1.9 Cauchy Index of a Rational Function, Bézoutian and

Hankel Forms
This section, and the next one, will not be used in the remainder of this book and can
therefore be skipped during a first reading. Having said that, its content is a classical
part of real algebra, and is essential for understanding the historical development
of mathematics in the nineteenth century. Furthermore, it has become important in
many applications such as the stability of motion or control theory.
We will follow the paper [65] of Krein and Naimark, but can only cover a small
part for lack of space. The interested reader may therefore want to consult the
original source.
R will always denote a real closed field.
Definition 1.9.1 Let ϕ(t) = g(t)/f (t) ∈ R(t) be a rational function with g and f
relatively prime polynomials over R.
(a) Let α ∈ R be a pole of ϕ (i.e., a zero of f ). The Cauchy index indα (ϕ) of ϕ at
α is defined as follows:
⎧
⎪
⎨+1 if the value of ϕ(t) jumps from −∞ to +∞ at t = α
⎪
indα (ϕ) := −1 if the value of ϕ(t) jumps from +∞ to −∞ at t = α .
⎪
⎪
⎩ 0 otherwise
(b) The Cauchy index of ϕ in an open interval ]a, b[ ⊆ R (a and b are allowed to be
±∞), denoted Iab (ϕ), is defined as the sum of the Cauchy indices of ϕ at those
poles of ϕ that are in ]a, b[. If a = −∞ and b = +∞, we speak of the global
Cauchy index of ϕ. (Note that if ϕ has a pole at ∞, this pole is ignored for the
Cauchy index, no matter what a and b are.)
1.9 Cauchy Index of a Rational Function, Bézoutian and Hankel Forms 31
The Cauchy index of ϕ(t) = f (t)/g(t) at the pole t = α is thus +1, −1 or 0,

depending on whether the polynomial f (t)g(t) changes sign from minus to plus,
from plus to minus, or not at all, at t = α.
Examples 1.9.2
(1) Let f ∈ R[t] be a non-constant polynomial with logarithmic derivative
ϕ = f /f . Let α1 , . . . , αr be the distinct real zeroes of f with respective
multiplicities m1 , . . . , mr . Then

r
mi
ϕ(t) = ψ(t) + ,
t − αi
i=1
for some rational function ψ without real poles. Thus ϕ = f /f has precisely
r real poles α1 , . . . , αr and its Cauchy index at each one of them is +1.
(2) More generally: let f, h ∈ R[t] with f = 0. For ϕ = hf /f we have
+∞
I−∞ (ϕ) = # α ∈ R : f (α)= 0 and h(α) > 0 − # α ∈ R : f (α)= 0 and h(α)< 0 ,
as can easily be verified.

How can we determine the integer Iab (ϕ) from the coefficients of ϕ? A solution
was already presented in Sect. 1.7: Carry out Sturm’s algorithm for the polynomials
f, g (instead of f, f ) and apply Theorem 1.7.10 to the generalized Sturm sequence
thus obtained. This procedure was already developed in the nineteenth century as
a practical method for finding Iab (ϕ). For an important special case, the so-called
Routh–Hurwitz Problem, we refer to [40, §16].
Next we will establish a connection between the global Cauchy index and certain
quadratic forms. This will require some preparation. For ϕ = g/f we may always
assume that deg g ≤ deg f since the Cauchy indices do not change after adding a
polynomial.
Consider thus two polynomials
f (t) = a0 t n + a1 t n−1 + · · · + an and g(t) = b0 t n + b1 t n−1 + · · · + bn
with coefficients in a field K (char K = 2) and let a0 = 0. To the pair (f, g) we

associate two quadratic forms over K as follows.
Let x and y be new variables. The polynomial f (x)g(y) − f (y)g(x) is divisible
by x − y. More precisely,
f (x)g(y) − f (y)g(x)
n−1
= cij x i y j ,
x−y
i,j =0
where

i
cij = dk,i+j +1−k with dij := an−j bn−i − an−i bn−j .
k=0
Independently from these explicit formulas (for which we will have no further use),
it is clear that cij = cj i for all i, j = 0, . . . , n − 1.
Definition 1.9.3 The symmetric n × n-matrix (cj i )0≤i,j ≤n−1 is called the Bézout
matrix of f and g, and is denoted B(f, g). The associated quadratic form over K is
called the Bézoutian of f , also denoted B(f, g). Thus,

n−1
B(f, g) (x0 , . . . , xn−1 ) = cij xi xj .
i,j =0
Since deg g ≤ deg f we can write the rational function ϕ(t) = g(t)/f (t) as a
formal power series in t −1 :
ϕ(t) = s−1 + s0 t −1 + s1 t −2 + · · · .
Definition 1.9.4 The symmetric n × n-matrix (si+j )0≤i,j ≤n−1 is called the Hankel
matrix of f and g and is denoted H (f, g). (In general, any quadratic matrix (aij )
whose entry aij only depends on i +j is called a Hankel matrix.) Again we interpret
H (f, g) as a quadratic form,

n−1
H (f, g) (x0 , . . . , xn−1 ) = si+j xi xj ,
i,j =0
and then refer to it as the Hankel form of f and g. For 1 ≤ p ≤ n the truncated
matrix (si+j )0≤i,j ≤p−1 is denoted Hp (f, g).
Remark 1.9.5 H (f, g) clearly only depends on n and the rational function ϕ =
g/f . More generally, a Hankel matrix Hn (ϕ) as in Definition 1.9.4 can be associated
to any formal power series ϕ(t) ∈ K[[t −1 ]] in t −1 and any n ≥ 0. A well-known
(elementary) theorem of Frobenius says that ϕ(t) is rational (i.e., ϕ(t) ∈ K(t)) if
and only if there exists n ≥ 0 such that det Hm (ϕ) = 0 for all m ≥ n. Furthermore,
in this case the smallest such n is the degree of the denominator of ϕ, written as
an irreducible fraction, and every matrix Hm (ϕ) with m > n has rank n. See, for
example, [40, §16.10].
Another random document with
no related content on Scribd:
None of them said a word, but all of them wore bright, knowing
smiles.
By Monday morning, however, Burbank was himself again. The
rebuff given him by Don Channing had worn off and he was
sparkling with ideas. He speared Franks with the glitter in his eye
and said: "If our beams are always on the center, why is it necessary
to use multiplex diversity?"
Franks smiled. "You're mistaken," he told Burbank. "They're not
always on the button. They vary. Therefore, we use diversity
transmission so that if one beam fails momentarily, one of the other
beams will bring the signal in. It is analogous to tying five or six
ropes onto a hoisted stone. If one breaks, you have the others."
"You have them running all the time, then?"
"Certainly. At several minutes of time-lag in transmission, to try and
establish a beam failure of a few seconds' duration is utter
foolishness."
"And you disperse the beam to a thousand miles wide to keep the
beam centered at any variation?" Burbank shot at Channing.
"Not for any variation. Make that any normal gyration and I'll buy it."
"Then why don't we disperse the beam to two or three thousand
miles and do away with diversity transmission?" asked Burbank
triumphantly.
"Ever heard of fading?" asked Channing with a grin. "Your signal
comes and goes. Not gyration, it just gets weaker. It fails for want of
something to eat, I guess, and takes off after a wandering cosmic
ray. At any rate, there are many times per minute that one beam will
be right on the nose and yet so weak that our strippers cannot clean
it enough to make it usable. Then the diversity system comes in
handy. Our coupling detectors automatically select the proper signal
channel. It takes the one that is the strongest and subdues the rest
within itself."
"Complicated?"
"It was done in the heyday of radio—1935 or so. Your two channels
come in to a common detector. Automatic volume control voltage
comes from the single detector and is applied to all channels. This
voltage is proper for the strongest channel, but is too high for the
ones receiving the weaker signal; blocking them by rendering them
insensitive. When the strong channel fades and the weak channel
rises, the detector follows down until the two signal channels are
equal and then it rises with the stronger channel."
"I see," said Burbank, "Has anything been done about fading?"
"It is like the weather, according to Mark Twain," smiled Channing.
"'Everybody talks about it, but nobody does anything about it.' About
all we've learned is that we can cuss it out and it doesn't cuss back."
"I think it should be tried," said Burbank.
"If you'll pardon me, it has been tried. The first installation at Venus
Equilateral was made that way. It didn't work, though we used more
power than all of our diversity transmitters together. Sorry."
"Have you anything to report?" Burbank asked Channing.
"Nothing. I've been more than busy investigating the trouble we've
had in keeping the beams centered."
Burbank said nothing. He was stopped. He hoped that the secret of
his failure was not generally known, but he knew at the same time
that when three hundred men are aware of something interesting,
some of them will see to it that all the others involved will surely
know. He looked at the faces of the men around the table and saw
suppressed mirth in every one of them. Burbank writhed in inward
anger. He was a good poker player. He didn't show it at all.
He then went on to other problems. He ironed some out, others he
shelved for the time being. Burbank was a good business man. But
like so many other businessmen, Burbank had the firm conviction
that if he had the time to spare and at the same time was free of the
worries and paper work of his position, he could step into the
laboratory and show the engineers how to make things hum. He was
infuriated every time he saw one of the engineering staff sitting with
hands behind head, lost in a gazy, unreal land of deep thought.
Though he knew better, he was often tempted to raise hell because
the man was obviously loafing.
But give him credit. He could handle business angles to perfection.
In spite of his tangle over the beam control, he had rebounded
excellently and had ironed out all of the complaints that had poured
in. Ironed it out to the satisfaction of the injured party as well as the
Interplanetary Communications Commission, who were interested in
anything that cost money.
He dismissed the conference and went to thinking. And he assumed
the same pose that infuriated him in other men under him; hands
behind head, feet upon desk.
The moving picture theater was dark. The hero reached longing
arms to the heroine, and there was a sort of magnetic attraction.
They approached one another. But the spark misfired. It was blacked
out with a nice slice of utter blackness that came from the screen
and spread its lightlessness all over the theater. In the ensuing
darkness, there were several osculations that were more personal
and more satisfying than the censored clinch. The lights flashed on
and several male heads moved back hastily. Female lips smiled
happily. Some of them parted in speech.
One of them said: "Why, Mr. Channing!"
"Shut up, Arden," snapped the man. "People will think that I've been
kissing you."
"If someone else was taking advantage of the situation," she said,
"you got gypped. I thought I was kissing you and I cooked with gas!"
"Did you ever try that before?" asked Channing interestedly.
"Why?" she asked.
"I liked it. I merely wondered, if you'd worked it on other men, what
there was about you that kept you single."
"They all died after the first application," she said. "They couldn't
take it."
"Let me outta here! I get the implication. I am the first bird that hasn't
died, hey?" He yawned luxuriously.
"Company or the hour?" asked Arden.
"Can't be either," he said. "Come on, let's break a bottle of beer
open. I'm dry!"
"I've got a slight headache," she told him, "From what, I can't
imagine."
"I haven't a headache, but I'm sort of logy."
"What have you been doing?" asked Arden. "Haven't seen you for a
couple of days."
"Nothing worth mentioning. Had an idea a couple of days ago and
went to work on it."
"Haven't been working overtime or missing breakfast?"
"Nope."
"Then I don't see why you should be ill. I can explain my headache
away by attributing it to eyestrain. Since Billyboy came here, and
censored the movies to the bone, the darned things flicker like
anything. But eyestrain doesn't create an autointoxication. So, my
fine fellow, what have you been drinking?"
"Nothing that I haven't been drinking since I first took to my second
bottlehood some years ago."
"You wouldn't be suffering from a hangover from that hangover you
had a couple of weeks ago?"
"Nope. I swore off. Never again will I try to drink a whole quart of Two
Moons in one evening. It got me."
"It had you for a couple of days," laughed Arden. "All to itself."
Don Channing said nothing. He recalled, all too vividly, the rolling of
the tummy that ensued after that session with the only fighter that
hadn't yet been beaten: Old John Barleycorn.
"How are you coming on with Burbank?" asked Arden. "I haven't
heard a rave for—well, ever since Monday morning's conference.
Three days without a nasty dig at Our Boss. That's a record."
"Give the devil his due. He's been more than busy placating irate
citizens. That last debacle with the beam control gave him a real
Moscow winter. His reforms came to a stop whilst he entrenched.
But he's been doing an excellent job of squirming out from under. Of
course, it has been helped by the fact that even though the service
was rotten for a few hours, the customers couldn't rush out to some
other agency to get communications with the other planets."
"Sort of: 'Take us, as lousy as we are?'"
"That's it."
Channing opened the door to his apartment and Arden went in.
Channing followed, and then stopped cold.
"Great Jeepers!" he said in an awed tone. "If I didn't know—"
"Why, Don! What's so startling?"
"Have you noticed?" he asked. "It smells like the inside of a chicken
coop in here!"
Arden sniffed. "It does sort of remind me of something that died and
couldn't get out of its skin." Arden smiled. "I'll hold my breath. Any
sacrifice for a drink."
"That isn't the point. This is purified air. It should be as sweet as a
baby's breath."
"Some baby," whistled Arden. "What's baby been drinking?"
"It wasn't cow-juice. What I've been trying to put over is that the air
doesn't seem to have been changed in here for nine weeks."
Channing went to the ventilator and lit a match. The flame bent over,
flickered, and went out.
"Air intake is O.K.," he said. "Maybe it is I. Bring on that bottle,
Channing; don't keep the lady waiting."
He yawned again, deeply and jaw-stretchingly. Arden yawned, too,
and the thought of both of them stretching their jaws to the breaking-
off point made both of them laugh foolishly.
"Arden, I'm going to break one bottle of beer with you, after which I'm
going to take you home, kiss you good night, and toss you into your
own apartment. Then I'm coming back here and I'm going to hit the
hay!"
Arden took a long, deep breath. "I'll buy that," she said. "And tonight,
it wouldn't take much persuasion to induce me to snooze right here
in this chair!"
"Oh, fine," cheered Don. "That would fix me up swell with the
neighbors. I'm not going to get shotgunned into anything like that!"
"Don't be silly," said Arden.
"From the look in your eye," said Channing, "I'd say that you were
just about to do that very thing. I was merely trying to dissolve any
ideas that you might have."
"Don't bother," she said pettishly. "I haven't any ideas. I'm as free as
you are, and I intend to stay that way!"
Channing stood up. "The next thing we know, we'll be fighting," he
observed. "Stand up, Arden. Shake."
Arden stood up, shook herself, and then looked at Channing with a
strange light in her eyes. "I feel sort of dizzy," she admitted. "And
everything irritates me."
She passed a hand over her eyes wearily. Then, with a visible effort,
she straightened. She seemed to throw off her momentary ill feeling
instantly. She smiled at Channing and was her normal self in less
than a minute.
"What is it?" she asked. "Do you feel funny, too?"
"I do!" he said. "I don't want that beer. I want to snooze."
"When Channing would prefer snoozing to boozing he is sick," she
said. "Come on, fellow, take me home."
Slowly they walked down the long hallway. They said nothing. Arm in
arm they went, and when they reached Arden's door, their good-
night kiss lacked enthusiasm. "See you in the morning," said Don.
Arden looked at him. "That was a little flat. We'll try it again—
tomorrow or next week."
Don Channing's sleep was broken by dreams. He was warm. His
dreams depicted him in a humid, airless chamber, and he was forced
to breathe that same stale air again and again. He awoke in a hot
sweat, weak and feeling—lousy!
He dressed carelessly. He shaved hit-or-miss. His morning coffee
tasted flat and sour. He left the apartment in a bad mood, and
bumped into Arden at the corner of the hall.
"Hello," she said. "I feel rotten. But you have improved. Or is that
passionate breathing just a lack of fresh air?"
"Hell! That's it!" he said. He snapped up his wrist watch, which was
equipped with a stop-watch hand. He looked about, and finding a
man sitting on a bench, apparently taking it easy while waiting for
someone, Channing clicked the sweep hand into gear. He started to
count the man's respiration.
"What gives?" asked Arden, "What's 'It'? Why are you so excited?
Did I say something?"
"You did," said Channing after fifteen seconds. "That bird's
respiration is better than fifty! This whole place is filled to the gills
with carbon dioxide. Come on, Arden, let's get going!"
Channing led the girl by several yards by the time that they were
within sight of the elevator. He waited for her, and then sent the car
upward at a full throttle. Minutes passed, and they could feel that
stomach-rising sensation that comes when gravity is lessened.
Arden clasped her hands over her middle and hugged. She
squirmed and giggled.
"You've been up to the axis before," said Channing. "Take long, deep
breaths."
The car came to a stop with a slowing effect. A normal braking stop
would have catapulted them against the ceiling. "Come on," he
grinned at her, "here's where we make time!"
Channing looked up at the little flight of stairs that led to the
innermost level. He winked at Arden and jumped. He passed up
through the opening easily. "Jump," he commanded. "Don't use the
stairs."
Arden jumped. She sailed upward, and as she passed through the
opening, Channing caught her by one arm and stopped her flight. "At
that speed you'd go right on across," he said.
She looked up, and there about two hundred feet overhead she
could see the opposite wall.
Channing snapped on the lights. They were in a room two hundred
feet in diameter and three hundred feet long. "We're at the center of
the station," Channing informed her. "Beyond that bulkhead is the air
lock. On the other side of the other bulkhead, we have the air plant,
the storage spaces, and several rooms of machinery."
"Come on," he said. He took her by the hand and with a kick he
propelled himself along on a long, curving course to the opposite
side of the inner cylinder. He gained the opposite bulkhead as well.
"Now, that's what I call traveling," said Arden. "But my tummy goes
whoosh, whoosh every time we cross the center."
Channing operated a heavy door. They went in through rooms full of
machinery and into rooms stacked to the center with boxes; stacked
from the wall to the center and then packed with springs. Near the
axis of the cylinder, things weighed so little that packing was
necessary to keep them from floating around.
"I feel giddy," said Arden.
"High in oxygen," said he. "The CO2, drops to the bottom, being
heavier. Then, too, the air is thinner up here because centrifugal
force swings the whole out to the rim. Out there we are so used to
'down' that here, a half mile above—or to the center, rather—we
have trouble in saying, technically, what we mean. Watch!"
He left Arden standing and walked rapidly around the inside of the
cylinder. Soon he was standing on the steel plates directly over her
head. She looked up, and shook her head.
"I know why," she called, "but it still makes me dizzy. Come down
from up there or I'll be sick."
Channing made a neat dive from his position above her head. He did
it merely by jumping upward from his place toward her place,
apparently hanging head down from the ceiling. He turned a neat
flip-flop in the air and landed easily beside her. Immediately, for both
of them, things became right-side-up again.
Channing opened the door to the room marked: "Air Plant." He
stepped in, snapped on the lights, and gasped in amazement.
"Hell!" he groaned. The place was empty. Completely empty.
Absolutely, and irrevocably vacant. Oh, there was some dirt on the
floor and some trash in the corners, and a trail of scratches on the
floor to show that the life giving air plant had been removed, hunk by
hunk, out through another door at the far end of the room.
"Whoa, Tillie!" screamed Don. "We've been stabbed! Arden, get on
the type and have ... no, wait a minute until we find out a few more
things about this!"
They made record time back to the office level. They found Burbank
in his office, leaning back, and talking to someone on the phone.
Channing tried to interrupt, but Burbank removed his nose from the
telephone long enough to snarl, "Can't you see I'm busy? Have you
no manners or respect?"
Channing, fuming inside, swore inwardly. He sat down with a show
of being calm and folded his hands over his abdomen like the famed
statue of Buddha. Arden looked at him, and for all the trouble they
were in, she couldn't help giggling. Channing, tall, lanky, and strong,
looked as little as possible like the popular, pudgy figure of the Sitting
Buddha.
A minute passed.
Burbank hung up the phone.
"Where does Venus Equilateral get its air from?" snapped Burbank.
"That's what I want—"
"Answer me, please. I'm worried."
"So am I. Something—"
"Tell me first, from what source does Venus Equilateral get its fresh
air?"
"From the air plant. And that is—"
"There must be more than one," said Burbank thoughtfully.
"There's only one."
"There must be more than one. We couldn't live if there weren't,"
said the Director.
"Wishing won't make it so. There is only one."
"I tell you, there must be another. Why, I went into the one up at the
axis day before yesterday and found that instead of a bunch of
machinery, running smoothly, purifying air, and sending it out to the
various parts of the station, all there was was a veritable jungle of
weeds. Those weeds, Mr. Channing, looked as though they must
have been put in there years ago. Now, where did the air-purifying
machinery go?"
Channing listened to the latter half of Burbank's speech with his chin
at half-mast. He looked as though a feather would knock him clear
across the office.
"I had some workmen clear the weeds out. I intend to replace the air
machinery as soon as I can get some new material sent from Terra."
Channing managed to blink. It was an effort. "You had workmen toss
the weeds out—" he repeated dully. "The weeds—"
There was silence for a minute, Burbank studied the man in the chair
as though Channing were a piece of statuary. Channing was just as
motionless. "Channing, man, what ails you—" began Burbank. The
sound of Burbank's voice aroused Channing from his shocked
condition.
Channing leaped to his feet. He landed on his heels, spun, and
snapped at Arden: "Get on the type. Have 'em slap as many oxy-
drums on the fastest ship as they've got! Get 'em here at full throttle.
Tell 'em to load up the pilot and crew with gravanol and not to spare
the horsepower! Scram!"
Arden gasped. She fled from the office.
"Burbank, what did you think an air plant was?" snapped Channing.
"Why, isn't it some sort of purifying machinery?" asked the wondering
Director.
"What better purifying machine is there than a plot of grass?"
shouted Channing. "Weeds, grass, flowers, trees, alfalfa, wheat, or
anything that grows and uses chlorophyll. We breathe oxygen,
exhale CO2. Plants inhale CO2, and exude oxygen. An air plant
means just that. It is a specialized type of Martian sawgrass that is
more efficient than anything else in the system for inhaling dead air
and revitalizing it. And you've tossed the weeds out!" Channing
snorted in anger. "We've spent years getting that plant so that it will
grow just right. It got so good that the CO2 detectors weren't even
needed. The balance was so adjusted that they haven't even been
turned on for three or four years. They were just another source of
unnecessary expense. Why, save for a monthly inspection, that room
isn't even opened, so efficient is the Martian sawgrass. We, Burbank,
are losing oxygen!"
The Director grew white. "I didn't know," he said.
"Well, you know now. Get on your horse and do something. At least,
Burbank, stay out of my way while I do something."
"You have a free hand," said Burbank. His voice sounded beaten.
Channing left the office of the Director and headed for the chem lab.
"How much potassium chlorate, nitrate, sulphate, and other oxygen-
bearing compounds have you?" he asked. "That includes mercuric
oxide, spare water, or anything else that will give us oxygen if broken
down."
There was a ten-minute wait until the members of the chem lab took
a hurried inventory.
"Good," said Channing. "Start breaking it down. Collect all the
oxygen you can in containers. This is the business! It has priority!
Anything, no matter how valuable, must be scrapped if it can
facilitate the gathering of oxygen. God knows, there isn't by half
enough—not even a tenth. But try, anyway."
Channing headed out of the chemistry laboratory and into the

electronics lab. "Jimmie," he shouted, "get a couple of stone jars and
get an electrolysis outfit running. Fling the hydrogen out of a
convenient outlet into space and collect the oxygen. Water, I mean.
Use tap water, right out of the faucet."
"Yeah, but—"
"Jimmie, if we don't breathe, what chance have we to go on
drinking? I'll tell you when to stop."
"O.K., Doc," said Jimmie.
"And look. As soon as you get that running, set up a CO2 indicator
and let me know the percentage at the end of each hour! Get me?"
"I take it that something has happened to the air plant?"
"It isn't functioning," said Channing shortly. He left the puzzled
Jimmie and headed for the beam-control room. Jimmie continued to
wonder about the air plant. How in the devil could an air plant cease
functioning unless it were—dead! Jimmie stopped wondering and
began to operate on his electrolysis set-up furiously.
Channing found the men in the beam-control room worried and ill at
ease. The fine co-ordination that made them expert in their line was
ebbing. The nervous work demanded perfect motor control, excellent
perception, and a fine power of reasoning. The perceptible lack of
oxygen at this high level was taking its toll already.
"Look, fellows, we're in a mess. Until further notice, take five-minute
shifts. We've got about thirty hours to go. If the going gets tough,
drop it to three-minute shifts. But, fellows, keep those beams
centered until you drop!"
"We'll keep 'em going if we have to call our wives up here to run 'em
for us," said one man. "What's up?"
"Air plant's sour. Losing oxy. Got a shipload coming out from Terra,
be here in thirty hours. But upon you fellows will rest the
responsibility of keeping us in touch with the rest of the system. If
you fail, we could call for help until hell freezes us all in—and no one
would hear us!"
"We'll keep 'em rolling," said a little fellow who had to sit on a tall
stool to get even with the controls.
Channing looked out of the big, faceted plexiglass dome that

covered the entire end of the Venus Equilateral Station. "Here
messages go in and out," he mused. "The other end brings us things
that take our breath away."
Channing was referring to the big air lock at the other end of the
station, three miles away, right through the center.
At the center of the dome, there was a sighting 'scope. It kept Polaris
on a marked circle, keeping the station exactly even with the
Terrestrial North. About the periphery of the dome, looking out
across space, the beam-control operators were sitting, each with a
hundred-foot parabolic reflector below his position, outside the
dome, and under the rim of the transparent howl. These reflectors
shot the interworld signals across space in tight beams, and the
men, half the time anticipating the vagaries of space-warp, kept them
centered on the proper, shining speck in that field of stars.
Above his head the stars twinkled. Puny man, setting his will against
the monstrous void. Puny man, dependent upon atmosphere.
"'Nature abhors a vacuum,' said Torricelli," groaned Channing. "Nuts!
If nature abhorred a vacuum, why did she make so much of it?"
Arden Westland entered the apartment without knocking. "I'd give

my right arm up to here for a cigarette," she said, marking above the
elbow with the other hand.
"Na-hah," said Channing. "Can't burn oxygen."
"I know. I'm tired, I'm cold, and I'm ill. Anything you can do for a
lady?"
"Not as much as I'd like to do," said Channing. "I can't help much.
We've got most of the place stopped off with the air-tight doors.
We've been electrolyzing water, baking KClO3 and everything else
we can get oxy out of. I've a crew of men trying to absorb the CO2
content and we are losing. Of course, I've known all along that we
couldn't support the station on the meager supplies we have on
hand. But we'll win in the end. Our microcosmic world is getting a
shot in the arm in a few hours that will re-set the balance."
"I don't see why we didn't prepare for this emergency," said Arden.
"This station is well balanced. There are enough people here and
enough space to make a little world of our own. We can establish a
balance that is pretty darned close to perfect. The imperfections are
taken care of by influxes of supplies from the system. Until Burbank
upset the balance, we could go on forever, utilizing natural
purification of air and water. We grow a few vegetables and have
some meat critters to give milk and steak. The energy to operate
Venus Equilateral is supplied from the uranium pile. Atomic power, if
you please. Why should we burden ourselves with a lot of cubic feet
of supplies that would take up room necessary to maintain our
balance? We are not in bad shape. We'll live, though we'll all be a
bunch of tired, irritable people who yawn in one another's faces."
"And after it is over?"
"We'll establish the balance. Then we'll settle down again. We can
take up where we left off," said Don.
"Not quite. Venus Equilateral has been seared by fire. We'll be
tougher and less tolerant of outsiders. If we were a closed
corporation before, we'll be tighter than a vacuum-packed coffee can
afterwards. And the first bird that cracks us will get hissed at."
Three superliners hove into sight at the end of thirty-one hours. They
circled the station, signaling by helio. They approached the air lock
end of the station and made contact. The air lock was opened and
space-suited figures swarmed over the South End Landing Stage. A
stream of big oxygen tanks was brought into the air lock, admitted,
and taken to the last bulwark of huddled people on the fourth level.
From one of the ships there came a horde of men carrying huge
square trays of dirt and green, growing sawgrass.
For six hours, Venus Equilateral was the scene of wild, furious
activity. The dead air was blown out of bad areas, and the hissing of
oxygen tanks was heard in every room. Gradually the people left the
fourth level and returned to their rightful places. The station rang with
laughter once more, and business, stopped short for want of breath,
took a deep lungful of fresh air and went back to work.
The superliners left. But not without taking a souvenir. Francis
Burbank went with them. His removal notice was on the first ship,
and Don Channing's appointment as Director of Venus Equilateral
was on the second.
Happily he entered the Director's office once more. He carried with
him all the things that he had removed just a few short weeks before.
This time he was coming to stay.
Arden entered the office behind him. "Home again?" she asked.
"Yop," he grinned at her. "Open file B, will you, and break out a
container of my favorite beverage?"
"Sure thing," she said.
There came a shout of glee. "Break out four glasses," she was told
from behind. It was Walt Franks and Joe.
It was Arden that proposed the toast. "Here's to a closed
corporation," she said. They drank on that.
She went over beside Don and took his arm. "You see?" she said,
looking up into his eyes. "We aren't the same. Things have changed
since Burbank came, and went. Haven't they?"
"They have," laughed Channing. "And now that you are my
secretary, it is no longer proper for you to shine up to me like that.
People will talk."
"What's he raving about?" asked Joe.
Channing answered, "It is considered highly improper for a secretary
to make passes at her boss. Think of what people will say; think of
his wife and kids."
"You have neither."
"People?" asked Channing innocently.
"No—you ape—the other."
"Maybe so," nodded Don, "but it is still in bad taste for a secretary—"
"No man can use that tone of voice on me!" stormed Arden with a
glint in her eye. "I resign! You can't call me a secretary!"
"But Arden—darling—"
Arden relaxed into the crook of Channing's arm. She winked at Walt
and Joe. "Me—," she said, "I've been promoted!"
Interlude:
Maintaining Communications through the worst of interference was a
type of problem in which dire necessity demanded a solution. Often
there are other problems of less demanding nature. These are
sometimes called "projects" because they may be desirable but are
not born of dire necessity.
Barring interference, the problem of keeping communication with
another planet across a hundred million miles of interplanetary space
is partially solved by the fact that you can see your target! Keeping
the cross-hairs in a telescope properly centered is a technical job
more arduous than difficult.
But seeing a spacecraft is another problem. Consider the relative
sizes of spacecraft and planet. Where Terra is eight thousand miles
in diameter, the largest of spacecraft is eight hundred feet long.
Reduced to a common denominator and a simple ratio, it reads that
the earth is 50,000 times as large as the largest spacecraft. Now go
outside and take a look at Venus. At normal distances, it is a mote in
the sky. Yet Venus is only slightly smaller than the earth. Reduce
Venus by fifty thousand times, and no astronomer would ever
suspect its existence.
Then take the invisible mote and place it in a volume of
1,000,000,000,000,000,000 cubic miles and he who found the
needle in a haystack is a piker by comparison.
It could have been lives at stake that drove the job out of the
"project" class and into the "necessity" stage. The fact that it was
ebb and flow of a mundane thing like money may lower the quality of
glamor.
But there it was—a problem that cried out for a solution; a man who
was willing to pay for the attempt; and a group of technicians more
than happy to tackle the job.
CALLING THE EMPRESS
The chart in the terminal building at Canalopsis Spaceport, Mars,
was a huge thing that was the focus of all eyes. It occupied a thirty-
by-thirty space in the center of one wall, and it had a far-flung iron
railing about it to keep the people from crowding it too close, thus
shutting off the view. It was a popular display, for it helped to drive
home the fact that space travel was different from anything else.
People were aware that their lives had been built upon going from
one fixed place to another place, equally immobile. But on
interplanet travel, one left a moving planet for another planet, moving
at a different velocity. You found that the shortest distance was not a
straight line but a space curve involving higher mathematics.
The courses being traveled at the time were marked, and those that
would be traversed in the very near future were drawn upon the
chart, too; all appropriately labeled. At a glance, one could see that
in fifty minutes and seventeen seconds the Empress of Kolain would
take off from Mars, which was the red disk on the right, and she
would travel along the curve so marked to Venus, which was almost
one hundred and sixty degrees clockwise around the Sun. People
were glad of the chance to go on this trip because the Venus
Equilateral Relay Station would come within a telescope's sight on
the way.
The Empress of Kolain would slide into Venus on the day side; and a
few hours later she would lift again to head for Terra, a few degrees
ahead of Venus and about thirty million miles away.
Precisely on the zero-zero, the Empress of Kolain lifted upward on
four tenuous pillars of dull-red glow and drove a hole in the sky. The
glow was almost lost in the bright sunshine, and soon it died. The
Empress of Kolain became a little world in itself, and would so
remain until it dropped onto the ground at Venus, almost two
hundred million miles away.
Driving upward, the Empress of Kolain could not have been out of
the thin Martian atmosphere when a warning bell rang in the
telephone and telespace office at the terminal. The bell caught
official ears, and all work was stopped as the personnel of the
communications office ran to the machine to see what was so
important that the "immediate attention" signal was rung.
Impatiently the operator waited for the tape to come clicking from the
machine. It came, letter by letter, click by click, at fifty words per
minute. The operator tore the strip from the machine and read aloud:
"Hold Empress of Kolain. Reroute to Terra direct. Will be quarantined
at Venus. Whole planet in epidemic of Venusian Fever."
"Snap answer," growled the clerk. "Tell 'em: 'Too little and too late.
Empress of Kolain left thirty seconds before warning bell. What do
we do now?'"
The operator's fingers clicked madly over the keyboard. Across
space went the signal, across the void to the Relay Station. It ran
through the Station's mechanism and went darting to Terra. It clicked
out as sent in the offices of Interplanet Transport. A vice president
read the message and swore roundly. He swore in three Terran
languages, in the language of the Venusians, and even managed to
visualize a few choice remarks from the Martian Pictographs that
were engraved on the Temples of Canalopsis.
"Miss Deane," he yelled at the top of his voice. "Take a message!
Shoot a line to Channing on Venus Equilateral. Tell him: 'Empress of
Kolain on way to Venus. Must be contacted and rerouted to Terra
direct. Million dollars' worth of Martian Line Moss aboard; will perish
under quarantine. Spare no expense.' Sign that 'Keg Johnson,
Interplanet.'"
"Yes, Mr. Johnson," said the secretary. "Right away."
More minutes of light-fast communication. Out of Terra to Luna,
across space to Venus Equilateral. The machines clicked and tape
cleared away from the slot. It was pasted neatly on a sheet of official
paper, stamped rush, and put in a pneumatic tube.
As Don Channing began to read the message, Williams on Mars was
chewing worriedly on his fourth fingernail, and Vice President Keg
Johnson was working on his second. But Williams had a head start
and therefore would finish first. Both men knew that nothing more
could be done. If Channing couldn't do it, nobody could.
Channing finished the 'gram and swore. It was a good-natured
swear-word, far from downright vilification, though it did consign
certain items to the nether regions. He punched a button with some
relish, and a rather good-looking woman entered. She smiled at him
with more intimacy than a secretary should, and sat down.
"Arden, call Walt, will you?"
Arden Westland smiled. "You might have done that yourself," she
told him. She reached for the call button with her left hand, and the
diamond on her finger glinted like a pilot light.
"I know it," he answered, "but that wouldn't give me the chance to
see you."
"Baloney," said Arden. "You just wait until next October. I'll be in your
hair all the time then."
"By then I may be tired of you," said Channing with a smile. "But until
then, take it or leave it." His face grew serious, and he tossed the
message across the table to her. "What do you think of that?"
Arden read, and then remarked: "That's a huge order, Don. Think
you can do it?"
"It'll cost plenty. I don't know whether we can contact a ship in space.
It hasn't been done to date, you know, except for short distances."
The door opened without a knock and Walt Franks walked in. "Billing
and cooing?" he asked. "Why do you two need an audience?"
"We don't," answered Don. "This was business."
"For want of evidence, I'll believe that. What's the dope?"

Full Ebook of Real Algebra A First Course 1St Edition Manfred Knebusch Claus Scheiderer Online PDF All Chapter

Uploaded by

Copyright:

Available Formats

You might also like

Full Ebook of Real Algebra A First Course 1St Edition Manfred Knebusch Claus Scheiderer Online PDF All Chapter

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

Full Ebook of Real Algebra A First Course 1St Edition Manfred Knebusch Claus Scheiderer Online PDF All Chapter

Uploaded by

Copyright:

Available Formats

Real Algebra A First Course 1st Edition

Manfred Knebusch Claus Scheiderer

Abstract Algebra: A First Course 2nd Edition Stephen

Introduction to Real Analysis 3rd Edition Manfred Stoll

Introduction to Real Analysis 3rd Edition Manfred Stoll

A First Course in Spectral Theory 1st Edition Milivoje

Business Statistics: A First Course 8th Edition David

Business Statistics A First Course 8th Edition David

Course 1 - Real Estate Essentials - Humber Real Estate

A First Course in Functional Analysis 1st Edition Orr

With Contributions by Thomas Unger

ISSN 0172-5939 ISSN 2191-6675 (electronic)

Mathematics Subject Classification: 12D15, 14P05, 14P10, 12J15, 12J10, 13J30

© Springer Nature Switzerland AG 2022

Regensburg, Germany Manfred Knebusch

1 This book was never written.

Positivstellensätze, preorderings of rings, convex ideals, and the holomorphy ring of

1 Ordered Fields and Their Real Closures . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 1

2.12 Artin’s Solution of Hilbert’s 17th Problem and the Sign

1.1 Orderings and Preorderings of Fields

© Springer Nature Switzerland AG 2022 1

Exercise 1.1.3 Let P be an ordering of K. Show that

Exercise 1.1.5 If T is a preordering of K, show that

defines a partial order relation ≤ on K that is compatible with + and ·.

K 2 := {a12 + · · · + an2 : n ∈ N, a1 , . . . , an ∈ K}.

It is clear that K 2 is contained in every preordering of K, and is itself a preordering

Lemma 1.1.9 For every preordering T of K there exists an ordering P of K such

Theorem 1.1.10 Every preordering T of K is an intersection of orderings of K.

Corollary 1.1.11 K has an ordering if and only if K is real.

Naturally, we must assume that char K = 2, since otherwise K 2 is a subfield of

is an ordering of Q(t). It is the ordering induced by the field embedding Q(t) →

1.2 Quadratic Forms, Witt Rings, Signatures

It is not possible to give an in-depth treatment of the algebraic theory of quadratic

For symmetric matrices B, B ∈ Mn (K) it is straightforward to check that

The n-fold sum ϕ ⊥ · · · ⊥ ϕ is abbreviated by n × ϕ and the n-fold tensor product

Lemma 1.2.3 Let ϕ, ϕ , ψ, ψ be quadratic spaces.

[ϕ] + [ψ] := [ϕ ⊥ ψ] and [ϕ] · [ψ] := [ϕ ⊗ ψ].

W (K) is called the Witt ring of the field K.

Exercise 1.2.6 Let K be a finite field of characteristic not 2.

where ϕ+ is positive definite with respect to P and ϕ− is negative definite with

Definition 1.2.10 Let ϕ be a quadratic space over K, P an ordering of K, and

signP ϕ := dim(ϕ+ ) − dim(ϕ− )

is called the Sylvester signature of ϕ at P .

signP (ϕ ⊥ ψ) = signP ϕ + signP ψ,

The statement follows.

Lemma 1.2.13 If a, b ∈ K ∗ with a + b = 0, then

Exercise 1.2.17 Show that if K ∗2 ∪ {0} = {a 2 : a ∈ K} is an ordering of K, then

1.3 Extension of Orderings

In this section, K denotes a field of characteristic = 2.

Definition 1.3.1 Let L/K be a field extension.

in L with m ≥ 1, pi ∈ P and ai , bi ∈ K. It follows in particular (by comparing

a1 g1 (t)2 + · · · + am gm (t)2 = f (t)h(t) (1.1)

in K[t]. We may assume that gcd(g1 , . . . , gm ) = 1.

Proposition 1.3.7 If L/K is a purely transcendental field extension (not necessar-

1.4 The Prime Ideals of the Witt Ring

In this section, K denotes a field of characteristic = 2.

Corollary 1.4.2 Let ϕ be a nondegenerate quadratic space over K. The following

1.5 Real Closed Fields and Their Field Theoretic

K 2 := {a12 + · · · + an2 : n ∈ N, a1 , . . . , an ∈ K}.

It is clear that K 2 is contained in every preordering of K, and is itself a preordering

Naturally, we must assume that char K = 2, since otherwise K 2 is a subfield of