Feynman diagrams and Wick products associated with q-Fock space

Edward G. Effros and Mihai Popa

Department of Mathematics, University of California, Los Angeles, CA 90095-1555 Edited by Richard V. Kadison, University of Pennsylvania, Philadelphia, PA, and approved May 7, 2003 (received for review March 13, 2003)

It is shown that if one keeps track of crossings, Feynman diagrams can be used to compute the q-Wick products and usual operator products in terms of each other.

1. Introduction recurrent theme in noncommutative analysis is that one may use graphs to efficiently index the terms in complicated sums. One of the first to recognize this principle was Cayley (1), who introduced rooted trees in order to label differentials (see ref. 2 for additional examples). Currently, the best known example of graph-theoretic indexing may be found in perturbative quantum field theory. In this context one uses Feynman diagrams to index summands that arise when one evaluates the expectations of products of jointly Gaussian random variables (see refs. 3 and 4). The random variables of quantum field theory correspond to certain self-adjoint operators on symmetric or antisymmetric Fock spaces (see refs. 3 and 5). In 1991, Bozejko and Speicher (6) introduced a remarkable q version of the Fock space, which for q 1, 1, and 0 coincides with the symmetric (Boson), antisymmetric (Fermion), and full (Voiculescu) Fock spaces (some of these ideas had been considered in ref. 7; see also ref. 8). Bozejko and Speichers q versions of stochastic processes and second quantization have attracted the attention of a large number of researchers (see refs. 911). In ref. 8, Bozejko et al. introduced the q analogs of the Wick product. We show that some of the basic combinatorial calculations involving Wick products of Gaussian random variables have natural q versions. In particular, we use Feynman diagrams to express the Wick products in terms of the usual operator products and vice versa. We will explore q forms of the Hopf algebraic theory of Kreimer (12) and Connes and Kreimer (13) in a subsequent paper.

(we use # to indicate cardinality). We will generally delete the subscript q in the hermitian form. The q-Fock space Fq(H) is the completion of this pre-Hilbert space. If q 1 or q 1, we must first divide out by the null space, and we then obtain the usual symmetric and antisymmetric Fock spaces. If q 0 we obtain the full Fock space. Unless otherwise indicated, we restrict our attention to the case 1 q 1. We let H [0] and Hn  H  H Rn.

a and

a f f1   fn

f, f  f1   fn

[1]

a a f

0, a
n

f f1
1

f 1, f [2]

f1   fn
i 1

qi

f i, f f 1

  i   f n. f If q 1, a ( f ) and a ( f ) extend to bounded operators on Fq(H) satisfying a ( f ) a ( f )*. They satisfy the commutation relation a f a g qa g a f f, g I.

We refer to the operators f a f a f

2. q-Fock Spaces and Feynman Diagrams We begin by recalling the Bozejko and Speicher (6) theory. Let H 0 be a real Hilbert space and let H be its complexification. We write H Rn HH 1 q 1, we define

as q Gaussians. For q 1, these may be identified with jointly Gaussian random variables (see theorem I.11 in ref. 5). We define q(H 0) to be the von Neumann algebra on Fq(H) generated by these operators. The vector is separating and cyclic for q(H 0). We let :
q

H0 3

:b

b ,

for the algebraic tensor product, and given a hermitian form on the algebraic sum
alg Fq H

be the corresponding state on q(H 0). In particular, if (f) (g) for f, g H 0, then we have the covariance and a f a g , g, f f, g

 H  HR2  ,

where

is taken to be a unit vector, by letting f 1   f m, g 1   g n

mn Sn q n

(we recall that H 0 is a real Hilbert space). We wish to compute the multivariable moments
1

...

f 1, g

. . . f n, g

of q Gaussians i ( fi), 1 i m. As in the classical case, these are determined by polynomials of the covariances ( i j).
This paper was submitted directly (Track II) to the PNAS ofce.

and where # i, j : 1 i j n, i j
To

whom correspondence should be addressed. E-mail: ege@math.ucla.edu.

www.pnas.org cgi doi 10.1073 pnas.1531460100

PNAS

July 22, 2003

vol. 100

no. 15

8629 8633

APPLIED MATHEMATICS

By an elementary tensor we mean an element in H Rn of the form f 1 R . . . R f n. For each f H 0, the creation and annihilation operators alg a ( f ) and a ( f ) are defined on F q (H) by

We begin by letting a 1( f ) a ( f ) and a 1( f ) first task is to compute expressions of the form m a

1

a ( f ). Our , {1, 1} 2n.

f1 a

f2 a

2n

f 2n

(Biane calls this the restricted crossing number in ref. 14). The total left crossings can be found by counting the intersections in the corresponding graph. In the diagram shown above, c l(1, 3) 0 and c l(2, 6) c l(4, 9) c l(8, 10) 1, and thus c c l 1, 3 c l 2, 6 c l 4, 9 c l 8, 10 3.

for sequences ( (1), . . . , (2n)) Given I(2n), we define

2n 2n 1

I(2n)

2n 2n ... 1 2n

2 1

2 1

2
1

2n 0, 2n .
2,

A simple induction shows that if a

k

...,
k

2n

0, then [3]

The general result is evident if one notes that an intersection in the graph will occur between an ascending line for one pair and a descending line for another (i, j) , which in (k, l) turn will correspond to the left crossing (k, l) of (i, j). Similarly, c r(i, j), where c r(i, j) is the number of right c( ) (i, j) . crossings i k j l where (k, l) , we define the gap g(i, j) to be the number If (i, j) of k with i k j, and we let a(i, j) g(i, j) c l(i, j) (we will not need the right version of this). We define g( ) g(i, j) and a( ) a(i, j) g( ) c( ). Given (i, j) (i, j) , it is evident that c l(i, j) c r(i, j) g(i, j), and thus (i, j) 2c g . [5]

fk a

2n

f 2n

. . . a (2n)(f n) 0. We say that and otherwise a k) I(2n) is a Catalan sequence if 2n 0, 2n 1 0, . . . , 1 0, and we let C(2n) be the set of such sequences. It is evident that m( ) 0 unless C(2n). We may associate a Dyck path (see ref. 15, p. 221) with each C(2n). For example, if ( 1, 1, 1, 1, 1, 1, 1, 1) C(8), then the corresponding Dyck path is given by
(k) (f

For some purposes, it is also useful to count degenerate crossings. These are the triples i k j, where k is not paired . We let d( ) be the number of such triples in , and (i, j) and we define the total crossing number to be tc( ) c( ) d( ). A Feynman diagram on S is complete if the there are no singletons (in which case S must have an even number of elements), and we let Fc(S) be the collection of all such diagrams. Given C(2n), we say that a complete Feynman diagram, i 1, j 1 , . . . , i n, j n ,

where each (k) is the slope of the corresponding line segment. If C(2n), we may evaluate m( ) in terms of Feynman diagrams on [2n] {1, . . . , 2n} (see refs. 3 and 4 for this terminology). Given a finite linearly ordered set S, a Feynman diagram on S is a partition of S into one- and two-element sets. It will be convenient to regard as a set of ordered pairs {(i 1, j 1), . . . , (i p, j p)} with i k j k and i k i l and j k j l for k l, and we refer to the unpaired indices as singletons. With this notation we will assume that i1 in [4]

on [2n] is compatible with if (i k) 1 and (j k) 1. We let Fc( ) be the set of all such Feynman diagrams on S. It is easy to see that Fc( ) is nonempty. For example, we may fix a vertex of maximum height and then pair the decending and ascending edges adjacent to that vertex. In the diagram shown above we begin by placing the pair (5, 6) in . Eliminating the two line segments and rejoinging the graph, we can continue by induction to appropriately pair all the ascending edges with descending edges. Conversely, each complete Feynman diagram on [2n] is compatible with the unique sequence C(2n), defined by 1 if k [2n] . letting (k) 1 if k [2n] and (k) (f i) (i 1, . . . , n) Given q-Gaussian random variables i and a Feynman diagram on [n], we let v
i1 j1

(the j k will generally be out of order). We write S (respectively, S ) for the j S with (k, j) for some k [respectively, the i S such that (i, k) for some i]. We refer to the elements of S and S as creators and annihilators, respectively. We may specify a Feynman diagram with a simple graph of the following form

ip jp

h1

...

h r,

... where {(i k, j k) : k 1, . . . , p}, and h 1 h2 hr are the singletons. The following is due to Bozejko and Speicher (see proposition 2 in ref. 6). We have included an alternative proof for the convenience of the reader.
Theorem 2.1. For any

C(2n), m
Fc

This corresponds to the Feynman diagram {(1, 3), (2, 6), (4, 9), (8, 10)} on [10]. We let F(S) denote the set of all Feynman diagrams on S, and we let F(n) F([n]). We note that more general partitions and their crossings are analyzed by using a succession of semicircles to link the elements of an equivalence class (see ref. 14). We call the elements in S the vertices of the diagram. We is a left crossing for (i, j) if k i say that a pair (k, l) l j, and we define c l(i, j) to be the number of such left crossings. c l(i, j) as the crossing number of We refer to c( ) (i, j)
8630 www.pnas.org cgi doi 10.1073 pnas.1531460100

qc .

Proof: Given Fc( ), we will assume, as before, that ... i n. We define a sequence of {(i k, j k)}, where i 1 elementary tensors,
2n,

...,

0,

as follows. Let us suppose that i n define

in

...

jn

2n. We

Effros and Popa

2n 2n 1

a a ...

f 2n f 2n
1

f 2n
2n

Proof: We have that

f 2n

 f 2n

...

2n

f1 m

f1 a m
Cn

f 2n ,

f 2n

I 2n in 1

a f in

f in

2n k 1 2   f jn 1  f jn  .

1  f in

and thus the formula follows from the theorem. The result for odd moments is immediate from Eq. 3. 3. q-Wick Products Following ref. 8, we define the q-Wick product for f 1, . . . , f n H by . W f 1, . . . , f n a a f i1 a f jl q
I, J

From Eq. 2, a f in
in 1

f i n, f i n q jn

f in 1  f in f i n, f j n f i n


1

in 1

  f jn 

f ik a ,

f j1 a

f j2 [6]

We define in to be a particular elementary tensor summand in this expression:

in

q jn

in 1

f i n, f j n f i n

  f jn  .

in

c i n, j n

f i n, f j n f i n

1   f jn  .

qi
i 1

f, f i W f 1, . . . , i, . . . , f n f [7]

Let us suppose that we have defined elementary tensors 2n, k i p 1, and none 2n 1 , . . . , k in such a manner that i p of the factors f jp 1, . . . , f jn occur in k . If k ip 1, we let
k 1

a ip

fk

fk 

k.

(see proof of proposition 2.7 in ref. 8). It follows from a simple induction on Eq. 7 that W(f 1, . . . f n) is a polynomial of the (f i), (1 i n), and we will use noncommuting operators i the usual Wick product notation :
1

On the other hand, if k

k

1, let us suppose that

1

...

n:

W f1 , . . . , fn . :
1

f k1   f kr  f jp  f

for some scalar . We define

ip k 1

Because is separating for q(H 0), u operator in q(H 0) satisfying u

...

n:

is the unique

qr

f i p, f j p f k 1   f k r  f j p  f 1  ,

f 1   f n.

which is one of the elementary tensor summands of a (f ip) ip 1. Each k t lies in [2n] { j p 1, . . . , j n} and k t j p; hence if k t j h, then h p and therefore i h i p. It follows that (i h, k t) is a left crossing for (i p, j p). Because this holds for each t, we see that r c (i p, j p). Continuing in this manner, 0 v( )q c( ) . Because it is evident that every nonzero elementary tensor summand in a
1

We may linearly extend this to define :p( 1, . . . , n): for any polynomial p( 1, . . . , n). The following provides an explicit (nonrecursive) expression for the q-Wick product analogous to the classical formula for q 1 (see theorem 3.4 in ref. 4). We let #( ) denote the number of pairs in .
Theorem 3.1. For any q-Gaussian random variables
i

(f i) (i [8]
0,

f1 a

2n

f 2n

1, . . . , n) :
1

corresponds to a unique Feynman diagram in Fc( ), m

Fc 0, Fc

n: Fn

qa v

qc ,

Proof: F(1) contains only the empty Feynman diagram

and

and we are done.

Corollary 2.1 (q-Wick Theorem). For any q-Gaussian random vari-

: 1:

W f1

a { a

f1
0,

a
1 },

f1 where a f1 a f2 a v
1

.
1

ables

i,

we have
1 2

On the other hand, F(2) {(1, 2)}. We have :

1 2:

is empty and f2

...

2n Fc 2n

qc ;

W f 1, f 2 qa a f1 f2 a a

f1 a f1 f1 I a v a

f2

f1 a f2 f2 .

and on the other hand,

1 2

1 2

...

2n 1

0.

1 2

1 2

Effros and Popa

PNAS

July 22, 2003

vol. 100

no. 15

8631

APPLIED MATHEMATICS

If i n k j n, then k [2n] , i.e., there is an h with (h, k) . Because h i n, it follows that (h, k) is a left crossing for (i n, j n), and we see that there are exactly c(i n, j n) jn in 1 terms between i n and j n. We conclude that

where the sum is taken over all families I {i 1, . . . , i k} and J [n], { j 1, . . . , j l} where i 1 . . . i k, j 1 . . . j l, and I J #({(p, q) : i p j q}. This operator is and we let (I, J) characterized by the recursion W f, f 1, . . . , f n f W f 1, . . . , f n

Let us suppose that we have proved Eq. 8 for n (f 0), then applying the formula for n and n recurrence relation,
0

1 and n. If 1 and the

If we let (p, i g) u ,

1 and (p, j h) m
Fc

1, then v qc q
I, J

W f 0, f 1, . . . , f n

0 Fn n

qa v

ql
l 1

f0 , fl
F n l

qa v

{(ik, jk)} F(n) trivially determines Each Feynman diagram a Feynman diagram {(ik, jk)} F({0} [n]}), for which v( ). Because #( ) #( ) and a( ) a( ), 0v( )
0

qa v

qa

where (I, J) (I k, J k). It should be noted that if Fc( ), then will not link elements of a block, because in the definition of v( ) a creator is always paired with an annihilator on its left. We may use a sequence of (I, J) transpositions of the index set S to transform S u into S lex. Retaining the same ordered pairs, each Feynman diagram on S with a given total ordering is mapped to a Feynman diagram on the reordered set. It is evident that v( ) v( ), but in general the number of crossings will change. If S u S lex, then i r(p) j 1 for some p. Our first step will be the transition from p, i 1 to p, i 1 p, j 1 p, i r p p, j s p . p, i r p p, j 1 p, j s p

Each Feynman diagram diagram {(0, l)} f 0, f l v

F([n] {l}) determines a Feynman F({0} [n]) for which v .

#( ) 1. Because c , (0, ) 0, It is evident that #( ) a (0, l) l 1. If (i, j) , then we may consider three cases. If l i, or j l, then it is evident that a (i, j) a (i, j). If i l j, then g (i, j) g (i, j) 1. On the other hand, 0 i l j introduces another left crossing when we regard (i, j) c (i, j) 1. It follows as an element of , and thus c (i, j) that a( ) l 1 a( ) and ql
1

Continuing in this manner, a series of (I, J) transpositions will give us a chain of Feynman diagrams 1 3 2 3 . . . 3 (I,J) on the permuted sets, which will return us to the lexicographic ordering. At each stage we will have an adjacent disordered pair (p, i k), (p, j l) with i k j l, and we perform the transposition p, i k 7 p, j l .

f0 , fl

qa v

qa

. if and . Thus [9]

A Feynman diagram in F({0} [n]) has the form only if 0 is a singleton in , and the form if (0, l) :
0 1

n: F 0 n

qa v

Let us consider the crossings in the corresponding Feynman diagrams and . Suppose that a b c d is a crossing in either or . If none of these four vertices coincides with i k or j l, the crossing will be left invariant under the transpositions 7 . We have three other cases [remember that c, d and (p, j l) are creators, and the terms (p, i k), (p, j l) are adjacent]: a a a b b b p, i k p, i k p, i k c c p, j l p, j l c p, j l d d d.

and we are done. Conversely, we may express products 1 . . . n in terms of q-Wick products. For this purpose we need to generalize the q-Wick theorem to products of q-Wick products. Given qGaussian random variables { p,k} with 1 p t and 1 k n p, we may regard the index set S {(p, k)} as partitioned by the first integer, and we refer to each partition as a block. We let S lex denote S with the lexicographic ordering 1, 1 1, 2 2, 1 t, n t .

Theorem 3.2. Suppose that we are given q-Gaussian random varip t and 1 k n p. Then if Y p ables { p,k} with 1 : p,1 . . . p,np:, we have

Y1 Yt

qc , on

In the first and second cases, the crossing a b c d will remain unaffected under these transpositions. From our construction, the third case will occur precisely when the noncross(p, j ) d occurs in with ik j l, ing sequence a (p, i k) and we obtain by the transposition. It follows that c( ) c( ) 1, and thus q c( )q h q c( )q h 1. Starting with a Feynman diagram 0 on S u , we obtain a c( ) q (I,J) Feynman diagram (I,J) on S lex with v( )q v( )q c( ). It is easy to see that all complete diagrams on S lex that do not link elements within blocks arise in this fashion; hence taking the sum of terms u, we obtain the desired result.
Theorem 3.3. Let Y i

where the sum is taken over all complete Feynman diagrams S lex that do not link vertices within blocks. Proof: A typical summand of Y 1 . . . Y t has the form
t

i1

...

in i:.

Then : q tc , [11]

Y1 Yt
I p, J p

:v

u
p 1

f i1p a

f irpp a

f j1p a

f j2p . . . a

f jspp q

. . . and i 1 .... where for each p, j 1 j2 i2 We may use Theorem 2.1 to compute u , provided we reorder the index set. We let S u denote S with the total ordering (p, k) (p , k ) if p p , and p, i 1
8632

where the sum is taken over all Feynman diagrams on S lex that do not link vertices within blocks. Proof: Let A Y 1 . . . Y t and B :v( ): q tc( ). As in the proof of theorem 3.15 in ref. 4, it suffices to show that for any Wick product W : 1 . . . u:, (AW) (BW). From Theorem 3.2, AW v qc , on the

p, i 2

p, j 1

p, j 2

[10]

where we sum over all complete Feynman diagrams partitioned ordered set

www.pnas.org cgi doi 10.1073 pnas.1531460100

Effros and Popa

1, 1 , . . . , 1, n 1 ; 2, 1 , . . . , t, n t ; 1, . . . , u

BW

:v

: q tc

qc

AW .

(where the semicolons separate blocks), which do not link vertices in the same block. Each such diagram determines a (generally incomplete) diagram (S S) on S lex with the same property. Because is complete, the singletons (p 1, k 1) . . . (p u, k u) in are linked with elements of [u] {1, . . . , u}. It follows that v
i, j i j p1,k1 g1

The one-variable case of the following result was proved by Anshelevich in remark 6.15 in ref. 11.
Corollary 3.1. Given
1

(f j) as above, we have
n Fn

:v

: qtc .

pu,ku

gu

i j i, j

4. The Free Case (q 0) and Noncrossing Diagrams We say that a Feynman diagram on a finite totally ordered set S is noncrossing if c( ) 0, strongly noncrossing if tc( ) 0, and gap-free if g( ) 0. We let NC(S), SNC(S), and GF(S) denote the corresponding diagrams on S. It is evident that NC S SNC S GF S .

where

is a complete Feynman diagram on T0 p 1, k 1 , . . . , p u, k u ; 1, . . . , u on S that does

that does not link vertices in the two blocks. On the other hand, given a Feynman diagram not link elements in any of the blocks, v
i, j i j p1,k1

If S [n], we will simply write NC(n), etc. If q 0, we have the commutation relation a ( f )a (g) f, g I. The convention of ref. 6 is that q c 0 for c 0, and q 0 1. Thus we may drop terms in the 0-Wick theorem for which q is raised to a positive power:
1 2

2n NC 2n

pu,ku

and thus :v :
i, j i j

Turning to Wick products, we delete terms with inversions in the definition of the free Wick product:
n

p 1,k 1

p u,k u

:.

W f 1, . . . , f n
k 0

f1 a

fk a

fk

fn .

From Theorem 3.2, :v :W

i, j i j

p1,k1

pu,ku

:W

For the alternative formula (Eq. 8) we need only consider terms with a( ) g( ) c( ) 0. It follows from Eq. 5 that g( ) 0, and thus is gap-free. We conclude that :
1

:
GF n

[12]

i j i, j G

qc ,

We have, for example, that :

1 2 3: 1 2 3 1 2 3 2 3 1.

where G is the set of all complete Feynman diagrams on T 0 that do not link vertices of the two blocks. Given such a , is a generic complete Feynman diagram on T extending , which does not link internal vertices. It is evident that c c c d ,

[13]

Turning to Eq. 11, we have Y1 Yt :v :, [14]

( i j) will hide precisely because forming the term (i, j) d( ) intersections arriving from the pairs in . It follows that :v :W q
ct i, j i j G

where the sum is over strongly noncrossing diagrams that do not link elements of a block. Thus, in particular,
1

n SNC S

:v

:.

[15]

ct

where we sum over nonlinking extensions

of , and thus

We are indebted to M. Anshelevich, P. Biane, and R. Speicher for helpful suggestions regarding the literature. This work was partially supported by the National Science Foundation (E.G.E.).
9. Bozejko, M. & Speicher, R. (1992) in Quantum Probability and Related Topics VII, ed. Accardi, L. (World Scientific, Singapore), pp. 219236. 10. Bozejko, M. & Speicher, R. (1994) Math. Ann. 300, 97120. 11. Anshelevich, M. (2001) Doc. Math. 6, 343384. 12. Kreimer, D. (1998) Adv. Theor. Math. Phys. 2.2, 303384. 13. Connes, A. & Kreimer, D. Commun. Math. Phys. (2001) Commun. Math. Phys. 216, 215241. 14. Biane, P. (1997) Discrete Math. 175, 4153. 15. Stanley, R. (1999) Enumerative Combinatorics (Cambridge Univ. Press, Cambridge, U.K.), Vol. 2.

1. Cayley, A. (1857) Philos. Mag. 13, 172176. 2. Brouder, C. (2000) Eur. Phys. J. C 12, 521534. 3. Glimm, J. & Jaffe, A. (1987) Quantum Physics: A Functional Integral Point of View (Springer, New York), 2nd Ed. 4. Janson, S. (1997) Gaussian Hilbert Spaces (Cambridge Univ. Press, Cambridge, U.K.), Vol. 129. 5. Simon, B. (1974) The P( 2) Euclidean Field Theory (Princeton Univ. Press, Princeton). 6. Bozejko, M. & Speicher, R. (1991) Commun. Math. Phys. 137, 519531. 7. Frisch, U. & Bourret, R. (1970) J. Math. Phys. 11, 364390. 8. Bozejko, M., Kummerer, B. & Speicher, R. (1997) Commun. Math. Phys. 185, 129154.

Effros and Popa

PNAS

July 22, 2003

vol. 100

no. 15

8633

APPLIED MATHEMATICS