Scribd
Upload
61 views17 pages

Matrix Airy Functions For Compact Lie Groups: V. S. Varadarajan University of California, Los Angeles, CA, USA

The document discusses generalizing the classical Airy function to matrix and compact Lie groups. It defines matrix and compact Lie group analogues of the Airy integral and shows they satisfy elliptic differential equations. Properties of these generalized Airy functions are explored, including determining the class of polynomials that give smooth Airy distributions and relating the functions to ones on Lie algebras.

Uploaded by

helmantico1970
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
61 views17 pages

Matrix Airy Functions For Compact Lie Groups: V. S. Varadarajan University of California, Los Angeles, CA, USA

The document discusses generalizing the classical Airy function to matrix and compact Lie groups. It defines matrix and compact Lie group analogues of the Airy integral and shows they satisfy elliptic differential equations. Properties of these generalized Airy functions are explored, including determining the class of polynomials that give smooth Airy distributions and relating the functions to ones on Lie algebras.

Uploaded by

helmantico1970
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 17

Matrix Airy functions for compact Lie groups

V. S. Varadarajan

University of California, Los Angeles, CA, USA

Los Angeles, November 12, 2008


Abstract

The classical Airy function and its many applications are


well known. A few years ago Kontsevich defined a matrix
generalization of the Airy integral and found that its theory
had many applications to intersection theory on the moduli
space of curves.

In my talk which will be as elementary as possible I shall


show that the Kontsevich integral has a very wide gener-
alization and discuss the properties of this generalization.
The work is joint with my former student Rahul Fernandez.
The original Airy function

First discovery due to Sir George Biddell Airy in 1838 in his


investigations of the intensity of light in the neighborhood
of a caustic:

∞ " " 3 ##
π ω
!
cos − mω dω
0 2 3

It is more convenient to work with

! ∞
3
A(x) = ei((1/3)y −xy)
dy
−∞

The integral is not convergent and so has to be interpreted


suitably, for instance as an improper Riemann integral.
The differential equation

A(x) is the (projectively)unique solution of the differential


equation

d2 u
+ xu = 0
dx2

which is of polynomial growth.

It extends to an entire function on C.

This is one of the simplest differential equations over C


with an irregular singularity at infinity.
Kontsevich’s generalization

In his work on intersection theory on the moduli space of


curves Kontsevich introduced a generalization of the Airy
integral:

!
3
A(X) = eiTr((1/3)Y −XY )
dY
H(N )

where H(N ) is the space of hermitian N × N matrices and


dY is Lebesgue measure.

With suitable interpretation it satisfies the elliptic equation

∆A(X) + Tr(X)A(X) = 0

and so is a smooth function.


The Airy distribution

Let V be a real finite dimensional vector space with a sym-


metric nondegenerate biliner form (x, y), p a real polyno-
mial on V . The Airy integral associated to p is:

!
A(x) = ei(p(y)−(x,y)) dy
V

We interpret this as a distribution, the Airy distribution:

Ap = F(eip ) F = Fourier transform

This makes sense because eip is bounded and so defines a


tempered distribution.
The Airy property

The polynomial p has the Airy property if Ap is a smooth


function of moderate growth, i.e., it and its derivatives are
of polynomial growth and it extends to an entire function
on VC = C ⊗ V .

If p is linear or depends only on a proper subset of the


coordinates on V , Ap will be a distribution supported on a
proper affine subspace of V and so will not be a function.
Generalization to compact Lie groups

The Kontsevich integral can be taken over iH(N ) instead


of H(N ) and so generalizes to the integral

!
A(X) = ei(p(Y )−(X,Y )) dY
g

where g is the Lie algebra of a compact Lie group G, (X, Y )


is the Cartan–Killing form, and p is a G-invariant real poly-
nomial.

For G = U (N ), the unitary group in N variables, g =


iH(N ), and p(Y ) = Tr((1/3)(Y 3 )) we obtain the Kontse-
vich integral.
Some general questions

The following are some of the questions that arise.

• Determine the class of p with the Airy property

• Find, if possible, explicit formulae for Ap in terms of


the Airy functions on one variable

• Study the differential equations satisfied by the Airy


functions

• The definitions make sense when we work not over R


but over any local field. What are the corresponding
questions and answers?
The case when the number of variables is 1

Theorem. Any polynomial in one variable of degree ≥ 2


has the Airy property.

Discussion. The Airy integral will become convergent


when we take the path of integration into the complex do-
main. The paths have to be chosen carefully depending on
whether the degree of p is odd or even.
The paths

Odd degree ≥ 3

The path is from −∞ + iη to ∞ + iη, η > 0.

$ ≤ Ce−ηDξ2k
$ Im(p(ξ+iη)) $
$e (deg(p) = 2k + 1, D > 0)

Even degree ≥ 4

The above method fails because the leading term in


Im(p(ξ + iη)) is like −ηξ 2k−1 and exp(−ηξ 2k−1 ) is not in-
tegrable for ξ < 0. So the path starts out from −∞ − iη
and then changes over to ∞ + iη in a finite part of C. The
estimate is

$ ≤ Ce−ηD|ξ|2k−1
$ Im(p(ξ+iη)) $
$e (deg(p) = 2k, D > 0)
The case of n variables

Theorem. If m is even, m = 2k, let pm = |α|=k cα y 2α ,


%
where the coefficients cα are non-negative and the coeffi-
cients of yj2k are strictly positive for 1 ≤ j %
≤ n.
If m is odd, m = 2k + 1, let pm = |α|=2k+1 cα y α ,
where all the coefficients cα are non-negative, and the co-
efficients of yj2k+1 are strictly positive for 1 ≤ j ≤ n.
Then p = pm + q, where q is arbitrary and deg(q) < m,
has the Airy property. In particular, p = c1 y1m +. . .+cn ynm +
q has the Airy property if all ci are > 0 and deg(q) < m.

The paths (or cycles) are generalizations of the ones when


we have only one variable. The integral is independent of
the path and is identified with the Airy distribution by let-
ting η → 0. Since η → 0 we cannot use the above estimates
but rather some Payley-Wiener estimates.

The derivatives of Ap of order r are O(|x|(r+n)/(m−1) ) as


|x| → ∞
Invariant Airy functions on Lie algebras of compact type

Let G be a compact Lie group with Lie algebra g. Let h


be a Cartan subalgebra and π the product of roots of a
positive system. Let ∂(π) be the corresponding differential
operator on h.

Theorem. Let p be a real invariant polynomial and ph its


restriction to h. If ph has the Airy property on h, then p
has the Airy property on g. If Aph , Ap are the respective
Airy functions, then

$
$ 1
Ap $$ = ∂(π)Aph
h π

The proofs use the theory of invariant integrals and differ-


ential operators on semi simple Lie algebras, developed by
Harish-Chandra in the 1950’s.
Example

For G = U (N ) we have g = H(N ), and h is the diag-


onal subalgebra. If yj is the linear function on h taking
diag(ia1 , . . . , iaN ) to aj , then
& &
π= (yk − y$ ), ∂(π) = (∂k − ∂$ )
k<$ k<$

where
∂
∂k = .
∂yk
A formula of Kontsevich

For the case when G = U (N ) we work over H(N ) rather


than iH(N ). Let

p(Y ) = Tr(Y r ) (Y ∈ H(N ))

Let Am be the one-dimensional Airy function for the poly-


nomial y m . Let y1 , . . . , yn be the linear coordinates on h.
Then the theorem above leads to the following formula
(j−1)
for Ap which was obtained by Kontsevich (Am is the
th
(j − 1) derivative of Am ):

& ' (j−1) (


(yk − y$ )Ap (diag(y1 , . . . , yn )) = det Am (yi ) .
k>$

The Kontsevich formula has some additional constants


which appear because he works with standard Lebesgue
measure. The constants disappear when we use the self-
dual Lebesgue measure.
Airy functions on local fields

The definitions of Airy functions and Airy property make


sense when R is replaced by a local field K. Let V be a
finite dimensional vector space over K with a symmetric
nondegenerate bilinear form. Then for a polynomial func-
tion p : V −→ K the Airy distribution Ap is the Fourier
transform of ψ(p) where ψ is a non-trivial additive charac-
ter of K. It appears that there is a corresponding theory
in this case. For example we have the following result.

Theorem. Let K be of characteristic 0. Then for any


polynomial p : K −→ K of degree ≥ 2 the Airy distribution
is a locally constant function which is at most O(|x|) as
|x| → ∞.

There is a corresponding result when K has positive char-


acteristic.
Compact p-adic Lie groups

Let D be a division algebra over K and G the compact


p-adic ie group of units (elements of norm 1) of D. The
computation of the Airy functions for D is a very inter-
esting problem. It is also interesting to ask if these Airy
functions have some geometric interpretation.

The work on p-adic Airy functions is ongoing and is joint


with Rahul Fernandez and David Weisbart.

You might also like