HAL Id: hal-00002150
https://hal.archives-ouvertes.fr/hal-00002150
Preprint submitted on 25 Jun 2004HAL is a multi-disciplinary open access
archive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
A short note about Morozov’s formula
Bertrand Eynard
To cite this version:
ccsd-00002150, version 1 - 25 Jun 2004
SPhT-T04/077
A short note about Morozov’s formula
B. Eynard1
Service de Physique Th´eorique de Saclay, F-91191 Gif-sur-Yvette Cedex, France.
Abstract
The purpose of this short note, is to rewrite Morozov’s formula for correlation functions over the unitary group, in a much simpler form, involving the computation of a single determinant.
1
Introduction
The main result of this paper is given in 2.13.
In [1], A. Morozov proposed a formula for correlation functions of unitary matrices with Itzykson-Zuber’s type measure:
h|Uji|2iU(N ) := 1 I(X, Y ) Z U(N ) dU|Uji|2etr (XU †Y U) (1.1) where dU is the Haar measure over the unitary group U(N) (appropriately normalized, so that 1.4 below holds with prefactor 1), and X and Y are two given diagonal complex matrices:
X = diag(x1, . . . , xN) , Y = diag(y1, . . . , yN) (1.2)
and the normalization factor I(X, Y ) is the so-called Harish-Chandra-Itzykson-Zuber integral: I(X, Y ) := Z U(N ) dUetr (XU†Y U) (1.3) 1E-mail: eynard@spht.saclay.cea.fr
It is well-known [4, 5] that:
I(X, Y ) = det E
∆(X)∆(Y ) (1.4)
where E is the matrix with entries:
Eij := exiyj (1.5)
and where ∆(X) and ∆(Y ) are the Vandermonde determinants: ∆(X) :=Y i<j (xi− xj) , ∆(Y ) := Y i<j (yi− yj) . (1.6)
Here we shall assume that det E 6= 0 and ∆(X) 6= 0 and ∆(Y ) 6= 0.
Morozov’s formula was proven in [2], from Shatashvili’s formula [3]. Morozov’s for-mula was originaly written as follows, for any arbitrary sequences of complex numbers ai, bj one has: N X i,j=1 aibj Z U(N ) dU|Uji|2etr (XU †Y U) = 1 ∆(X)∆(Y ) X ρ∈SN ǫ(ρ)eP xℓyρ(ℓ) N −1 X n=0 (−1)n X i1<i2<...<in+1 det 1 · · · 1 xi1 · · · xin+1 ... ... xn−1i 1 · · · x n−1 in+1 ai1 · · · ain+1 det 1 · · · 1 xi1 · · · xin+1 ... ... xn i1 · · · x n in+1 det 1 · · · 1 yρ(i1) · · · yρ(in+1) ... ... yρ(in−1 1) · · · y n−1 ρ(in+1) bρ(i1) · · · bρ(in+1) det 1 · · · 1 yρ(i1) · · · yρ(in+1) ... ... yn ρ(i1) · · · y n ρ(in+1) , (1.7)
where SN is the smmetric group of rank N.
2
Rewriting Morozov’s formula
The purpose of this short note is to rewrite this formula in a much simpler form, in a way very similar to what was done in [2].
For any two complex numbers x and y (such that |x| > max |xi| and |y| > max |yi|),
consider the choice: ai = 1 x− xi = ∞ X r=0 xr i xr+1 , bi = 1 y− yi = ∞ X s=0 ys i ys+1 (2.8)
Introduce the Schur polynomials (corresponding to hook diagramms): Sr(xi1, . . . , xin+1) := det 1 · · · 1 xi1 · · · xin+1 ... ... xn−1i1 · · · xn−1in+1 xr i1 · · · x r in+1 det 1 · · · 1 xi1 · · · xin+1 ... ... xn i1 · · · x n in+1 = X a1≤a2≤...≤ar−n r−n Y k=1 xiak = X j1+···+jn+1=r−n xj1 i1 · · · x jn+1 in+1 . (2.9)
The formal generating function of these Schur polynomials is:
∞ X r=0 1 xr+1Sr(xi1, . . . , xin+1) = n+1 Y k=1 1 x− xik (2.10) Inserting that into 1.7, one gets:
I(X, Y ) X i,j 1 x− xi 1 y− yj h|Uji|2iU(N ) = Z U(N ) dU tr 1 x− XU 1 y− Y U † etr (XU†Y U) = 1 ∆(X)∆(Y ) ∞ X r,s=0 X ρ∈SN ǫ(ρ)eP xℓyρ(ℓ) N −1 X n=0 (−1)nX i1<i2<...<in+1 Sr(xi1, . . . , xin+1) xr+1 Ss(yρ(i1), . . . , yρ(in+1)) ys+1 = 1 ∆(X)∆(Y ) X ρ∈SN ǫ(ρ)eP xℓyρ(ℓ) N −1 X n=0 (−1)nX i1<i2<...<in+1 n+1 Y k=1 1 x− xik n+1 Y l=1 1 y− yρ(il) = 1 ∆(X)∆(Y ) X ρ∈SN ǫ(ρ)eP xℓyρ(ℓ) " 1 − N Y i=1 1 − 1 x− xi 1 y− yρ(i) # = 1 ∆(X)∆(Y ) X ρ∈SN ǫ(ρ) " N Y i=1 exiyρ(i) − N Y i=1 exiyρ(i) − 1 x− xi exiyρ(i) 1 y− yρ(i) # = 1 ∆(X)∆(Y ) det ( exiyj) − det exiyj − 1 x− xi exiyj 1 y− yj (2.11)
Thus, Morozov’s formula can be rewritten: Z U(N ) dU tr 1 x− XU 1 y− Y U † etr (XU†Y U) = det E − det E−x−X1 Ey−Y1 ∆(X)∆(Y ) (2.12)
or: tr 1 x− XU 1 y− Y U † = 1 − detE−x−X1 Ey−Y1 det E = 1 − det 1 − 1 x− XE 1 y− YE −1 (2.13)
3
Concluding remarks
From that expression of Morozov’s formula, it is rather easy to recover any individual correlator by taking residues:
D UijUji† E = Res x→xi Res y→xj tr 1 x− XU 1 y− Y U † (3.14) Notice also that 2.13 is very similar to what was found in [2] after integration over X and Y .
Notice that by expanding the determinant in 2.13 along its last column, one can find a recursion relation relating U(N) to U(N − 1) integrals, which is equivalent to Shatashvili’s approach [3].
Aknowledgements: the author wants to thank M. Bertola and J. Harnad for discus-sions about that topic. The author thinks that this short note should be seen as an addendum to the article [2].
References
[1] A. Morozov, “Pair correlator in the Itzykson–Zuber Integral”, Modern Phys. Lett. A 7, no. 37 3503–3507 (1992).
[2] M. Bertola, B. Eynard, ”Mixed Correlation Functions of the Two-Matrix Model”, SPHT T03/028, CRM-2961 (2003). J. Phys. A36 (2003) 7733-7750, xxx, hep-th/0303161.
[3] S. L. Shatashvili, “Correlation Functions in the Itzykson–Zuber Model”, Comm. Math. Phys. 154 421–432 (1993).
[4] C. Itzykson and J.B. Zuber, “The planar approximation (II)”, J. Math. Phys. 21, 411 (1980).