A Central Limit Theorem for Random Walks on the Dual of a Compact Grassmannian

We consider compact Grassmann manifolds G/K over the real, complex or quaternionic numbers whose spherical functions are Heckman-Opdam polynomials of type BC. From an explicit integral representation of these polynomials we deduce a sharp Mehler-Heine formula, that is an approximation of the Heckman...

Full description

Saved in:
Bibliographic Details
Published in:Symmetry, Integrability and Geometry: Methods and Applications
Date:2015
Main Authors: Rösler, M., Voit, M.
Format: Article
Language:English
Published: Інститут математики НАН України 2015
Online Access:https://nasplib.isofts.kiev.ua/handle/123456789/146999
Tags: Add Tag
No Tags, Be the first to tag this record!
Journal Title:Digital Library of Periodicals of National Academy of Sciences of Ukraine
Cite this:A Central Limit Theorem for Random Walks on the Dual of a Compact Grassmannian / M. Rösler, M. Voit // Symmetry, Integrability and Geometry: Methods and Applications. — 2015. — Т. 11. — Бібліогр.: 32 назв. — англ.

Institution

Digital Library of Periodicals of National Academy of Sciences of Ukraine
id nasplib_isofts_kiev_ua-123456789-146999
record_format dspace
spelling Rösler, M.
Voit, M.
2019-02-12T18:12:01Z
2019-02-12T18:12:01Z
2015
A Central Limit Theorem for Random Walks on the Dual of a Compact Grassmannian / M. Rösler, M. Voit // Symmetry, Integrability and Geometry: Methods and Applications. — 2015. — Т. 11. — Бібліогр.: 32 назв. — англ.
1815-0659
2010 Mathematics Subject Classification: 33C52; 43A90; 60F05; 60B15; 43A62; 33C80; 33C67
DOI:10.3842/SIGMA.2015.013
https://nasplib.isofts.kiev.ua/handle/123456789/146999
We consider compact Grassmann manifolds G/K over the real, complex or quaternionic numbers whose spherical functions are Heckman-Opdam polynomials of type BC. From an explicit integral representation of these polynomials we deduce a sharp Mehler-Heine formula, that is an approximation of the Heckman-Opdam polynomials in terms of Bessel functions, with a precise estimate on the error term. This result is used to derive a central limit theorem for random walks on the semi-lattice parametrizing the dual of G/K, which are constructed by successive decompositions of tensor powers of spherical representations of G. The limit is the distribution of a Laguerre ensemble in random matrix theory. Most results of this paper are established for a larger continuous set of multiplicity parameters beyond the group cases.
en
Інститут математики НАН України
Symmetry, Integrability and Geometry: Methods and Applications
A Central Limit Theorem for Random Walks on the Dual of a Compact Grassmannian
Article
published earlier
institution Digital Library of Periodicals of National Academy of Sciences of Ukraine
collection DSpace DC
title A Central Limit Theorem for Random Walks on the Dual of a Compact Grassmannian
spellingShingle A Central Limit Theorem for Random Walks on the Dual of a Compact Grassmannian
Rösler, M.
Voit, M.
title_short A Central Limit Theorem for Random Walks on the Dual of a Compact Grassmannian
title_full A Central Limit Theorem for Random Walks on the Dual of a Compact Grassmannian
title_fullStr A Central Limit Theorem for Random Walks on the Dual of a Compact Grassmannian
title_full_unstemmed A Central Limit Theorem for Random Walks on the Dual of a Compact Grassmannian
title_sort central limit theorem for random walks on the dual of a compact grassmannian
author Rösler, M.
Voit, M.
author_facet Rösler, M.
Voit, M.
publishDate 2015
language English
container_title Symmetry, Integrability and Geometry: Methods and Applications
publisher Інститут математики НАН України
format Article
description We consider compact Grassmann manifolds G/K over the real, complex or quaternionic numbers whose spherical functions are Heckman-Opdam polynomials of type BC. From an explicit integral representation of these polynomials we deduce a sharp Mehler-Heine formula, that is an approximation of the Heckman-Opdam polynomials in terms of Bessel functions, with a precise estimate on the error term. This result is used to derive a central limit theorem for random walks on the semi-lattice parametrizing the dual of G/K, which are constructed by successive decompositions of tensor powers of spherical representations of G. The limit is the distribution of a Laguerre ensemble in random matrix theory. Most results of this paper are established for a larger continuous set of multiplicity parameters beyond the group cases.
issn 1815-0659
url https://nasplib.isofts.kiev.ua/handle/123456789/146999
citation_txt A Central Limit Theorem for Random Walks on the Dual of a Compact Grassmannian / M. Rösler, M. Voit // Symmetry, Integrability and Geometry: Methods and Applications. — 2015. — Т. 11. — Бібліогр.: 32 назв. — англ.
work_keys_str_mv AT roslerm acentrallimittheoremforrandomwalksonthedualofacompactgrassmannian
AT voitm acentrallimittheoremforrandomwalksonthedualofacompactgrassmannian
AT roslerm centrallimittheoremforrandomwalksonthedualofacompactgrassmannian
AT voitm centrallimittheoremforrandomwalksonthedualofacompactgrassmannian
first_indexed 2025-11-27T04:31:13Z
last_indexed 2025-11-27T04:31:13Z
_version_ 1850800353791967232
fulltext Symmetry, Integrability and Geometry: Methods and Applications SIGMA 11 (2015), 013, 18 pages A Central Limit Theorem for Random Walks on the Dual of a Compact Grassmannian Margit RÖSLER † and Michael VOIT ‡ † Institut für Mathematik, Universität Paderborn, Warburger Str. 100, D-33098 Paderborn, Germany E-mail: roesler@math.upb.de ‡ Fakultät für Mathematik, Technische Universität Dortmund, Vogelpothsweg 87, D-44221 Dortmund, Germany E-mail: michael.voit@math.uni-dortmund.de Received October 14, 2014, in final form February 03, 2015; Published online February 10, 2015 http://dx.doi.org/10.3842/SIGMA.2015.013 Abstract. We consider compact Grassmann manifolds G/K over the real, complex or quaternionic numbers whose spherical functions are Heckman–Opdam polynomials of type BC. From an explicit integral representation of these polynomials we deduce a sharp Mehler–Heine formula, that is an approximation of the Heckman–Opdam polynomials in terms of Bessel functions, with a precise estimate on the error term. This result is used to derive a central limit theorem for random walks on the semi-lattice parametrizing the dual of G/K, which are constructed by successive decompositions of tensor powers of spherical representations of G. The limit is the distribution of a Laguerre ensemble in random matrix theory. Most results of this paper are established for a larger continuous set of multiplicity parameters beyond the group cases. Key words: Mehler–Heine formula; Heckman–Opdam polynomials; Grassmann manifolds; spherical functions; central limit theorem; asymptotic representation theory 2010 Mathematics Subject Classification: 33C52; 43A90; 60F05; 60B15; 43A62; 33C80; 33C67 1 Introduction For Riemannian symmetric spaces G/K of the compact or non-compact type, there is a well- known contraction principle which states that under suitable scaling, the spherical functions ϕλ of G/K tend to the spherical functions ψλ of the tangent space of G/K in the base point, which is a symmetric space of the flat type: lim n→∞ ϕnλ(exp(x/n)) = ψλ(x). See [4] and, for a more recent account, [2]. This curvature limit, also known as Mehler–Heine formula, extends to the more general setting of hypergeometric functions associated with root systems, which converge under rescaling to generalized Bessel functions. This is proven in [6] by a limit transition in the Cherednik operators; see also [2] for a different approach. In the compact rank one case, the contraction principle is a weak version of the classical Hilb formula for Jacobi polynomials (see [27, Theorem 8.21.12]), which provides in addition a precise estimate on the rate of convergence. In this paper, we prove a Mehler–Heine formula with a precise estimate on the error term for a certain class of orthogonal polynomials associated with root systems, which in particular encompasses the spherical functions of compact Grassmannians. This result is a “compact” analogue of Theorem 5.4 in [26], which gives a scaling limit with error bounds for hypergeometric functions in the dual, non-compact setting. In the second part of the paper, we mailto:roesler@math.upb.de mailto:michael.voit@math.uni-dortmund.de http://dx.doi.org/10.3842/SIGMA.2015.013 2 M. Rösler and M. Voit shall use the Mehler–Heine formula 2.4 in order to establish a central limit theorem for random walks on the semi-lattice of dominant weights parametrizing the unitary dual of a compact Grassmannian. To become more precise, we consider the compact Grassmannians Gp,q(F) = G/K over one of the (skew-) fields F = R,C,H, with G = SU(p + q,F) and K = S(U(q,F)× U(p,F)), where p ≥ q ≥ 1. Via polar decomposition of G, the double coset space G//K = {KgK : g ∈ G} may be topologically identified with the fundamental alcove A0 := { x = (x1, . . . , xq) ∈ Rq : π 2 ≥ x1 ≥ x2 ≥ · · · ≥ xq ≥ 0 } , with x ∈ A0 being identified with the matrix ax = cosx − sinx 0 sinx cosx 0 0 0 Ip−q  . Here we use the diagonal matrix notation x = diag(x1, . . . , xq), and the functions sin, cos are understood component-wise. For details, see [23, Theorem 4.1]. The spherical functions of Gp,q(F) can be viewed as Heckman–Opdam polynomials of type BCq, which are also known as multivariable Jacobi polynomials. They may be described as follows: denote by FBC(λ, k; ·) the Heckman–Opdam hypergeometric function associated with the root system R = 2BCq = {±2ei,±4ei : 1 ≤ i ≤ q} ∪ {±2ei ± 2ej : 1 ≤ i < j ≤ q} ⊂ Rq, with spectral variable λ ∈ Cq and multiplicity parameter k = (k1, k2, k3) ∈ R3 corresponding to the roots ±2ei, ±4ei and 2(±ei ± ej). Fix the positive subsystem R+ = {2ei, 4ei, 1 ≤ i ≤ q} ∪ {2ei ± 2ej , 1 ≤ i < j ≤ q} and the associated semi-lattice of dominant weights, P+ = { λ ∈ (2Z)q : λ1 ≥ λ2 ≥ · · · ≥ λq ≥ 0 } . Then the set of spherical functions of Gp,q(F) is parametrized by P+ and consists of the functions ϕpλ(ax) = FBC(λ+ ρp, k(p), ix) =: Rpλ(x), λ ∈ P+ (1.1) with multiplicity parameter k(p) = (d(p− q)/2, (d− 1)/2, d/2), (1.2) where d = dimR F ∈ {1, 2, 4} and ρp = 1 2 ∑ α∈R+ kαα = q∑ i=1 ( d 2 (p+ q + 2− 2i)− 1 ) ei. The functions Rpλ are the Heckman–Opdam polynomials associated with the root system R (called Jacobi polynomials in the following) and with multiplicity k(p), normalized according to Rpλ(0) = 1. We refer to [12, 13, 22] for Heckman–Opdam theory in general, and to [23] and the references cited there for the connection with spherical functions in the compact BC case. Notice that our notion of FBC coincides with that of Heckman, Opdam and [25, 26], while it differs from the geometric notion in [23]. Theorem 4.3 of [23] corresponds to (1.1). In Theorem 4.2 of [23], the product formula for spherical functions of (G,K) was written as a formula on A0 and analytically extended to a product formula for the Jacobi polynomials Rpλ A Central Limit Theorem for Random Walks on the Dual of a Compact Grassmannian 3 with multiplicity k(p) corresponding to arbitrary real parameters p > 2q − 1. This led to three continuous series of positive product formulas for Jacobi polynomials corresponding to F = R,C,H and to associated commutative hypergroup structures on A0; see [15] and [3] for the notion of hypergroups. Using a Harish-Chandra-type integral representation for the Rpλ, we shall derive a Mehler–Heine formula with a precise asymptotic estimate for the Jacobi polynomials Rpλ in terms of Bessel functions associated with root system Bq on the Weyl chamber C = {x = (x1, . . . , xq) ∈ Rq : x1 ≥ · · · ≥ xq ≥ 0}. This Mehler–Heine formula will be the key ingredient for the main result of the present paper, a central limit theorem for random walks on the semi-lattice P+, which parametrizes the spherical unitary dual of G/K. To explain this CLT, let us first recall that via the GNS representation, the spherical functions ϕλ, λ ∈ P+ of (G,K), which are necessarily positive definite, are in a one- to-one correspondence with the (equivalence classes of) spherical representations (πλ, Hλ) of G, that is those irreducible unitary representations of G whose restriction to K contains the trivial representation with multiplicity one, see [9] or [14, Chapter IV]. The decomposition of tensor products of spherical representations into their irreducible components leads to a probability preserving convolution ∗d,p and finally a Hermitian hypergroup structure on the discrete set P+; see [7] and [18]. Following, e.g., [3, 29, 32], we introduce random walks (Sd,pn )n≥0 on P+ associated with ∗d,p and derive some limit theorems for n→∞. The main result of this paper will be the Central Limit Theorem 3.12. This CLT implies the following result for Gp,q(F) = G/K: Theorem 1.1. Let (πλ, Hλ) be a non-trivial spherical representation of G associated with λ ∈ P+ \ {0}. Let uλ ∈ Hλ be K-invariant with ‖uλ‖ = 1. For each n ∈ N, decompose the n-fold tensor power (π⊗,nλ , H⊗,nλ ) into its finitely many irreducible unitary components ( π⊗,nλ , H⊗,nλ ) = (⊕ τn πτn , ⊕ τn Hτn ) , where the components are counted with multiplicities. Consider the orthogonal projections pτn : H⊗,nλ → Hτn and a P+-valued random variable Xn,λ with the finitely supported distribution∑ τn ∥∥pτn(u⊗,nλ ) ∥∥2δτn ∈M1(P+) with the point measures δτn at τn. Then, for n→∞, Xn,λ m(λ) √ n tends in distribution to dρd,p(x) = c−1d,p q∏ j=1 x d(p−q+1)−1 j ∏ 1≤i<j≤q ( x2i − x2j )d e−(x 2 1+···+x2q)/2 dx ∈M1(C) with a suitable normalization cd,p. Notice that the probability measure dρd,p is the distribution of a Laguerre ensemble on C. The modified variance parameter m(λ) > 0 is a second order polynomial in λ and given explicitly in Lemma 3.3 below. For q = 1, the Central Limit Theorem 3.12 has a long history as a CLT for random walks on Z+ whose transition probabilities are related to product linearizations of Jacobi polynomials. This includes random walks on the duals of SU(2) and (SO(n),SO(n − 1)) in [8] and [11]. See also [29] for further one-dimensional cases. For q ≥ 2 our results are very closely related to the work [5] of Clerc and Roynette on duals of compact symmetric spaces. For a survey on limits for spherical functions and CLTs in the non-compact case for q = 1 we refer to [31]. 4 M. Rösler and M. Voit 2 A Mehler–Heine formula In this section we derive a Mehler–Heine formula for the Jacobi polynomials Rpλ(λ ∈ P+), describing the approximation of these polynomials in terms of Bessel functions with a precise error bound. Our result will be based on Laplace-type integrals for the Jacobi polynomials and the associated Bessel functions, where we treat the group cases with integers p ≥ q as well as the case p ∈ R with p ≥ 2q− 1 beyond the group case. The integral representation for Rpλ below is a special case of a more general Harish-Chandra integral representation for hypergeometric functions FBC in [26]. To start with, let us introduce some notation: Let Hq(F) = {x ∈ Mq(F) : x∗ := xt = x} denote the space of Hermitian matrices over F, and denote by ∆(x) the determinant of x ∈ Hq(F), which may be defined as the product of (right) eigenvalues of x. We mention that for F = H, this is just the Moore determinant, which coincides with the Dieudonné determinant if x is positive semi-definite, see, e.g., [1]. On Hq(F), we consider the power functions ∆λ(a) := ∆1(a)λ1−λ2 · · ·∆q−1(a)λq−1−λq∆q(a)λq , λ ∈ Cq, with the principal minors ∆r(a) = det((aij)1≤i,j≤r) of the matrix a = (aij)1≤i,j≤q ∈ Hq(F), see [10]. We introduce the matrix ball Bq := {w ∈ Mq(F) : w∗w < I}, where A < B means for matrices A,B ∈ Mq(F) that B − A is (strictly) positive definite. On Bq, we define the probability measures dmp(w) = 1 κpd/2 ∆(I − w∗w)pd/2−γdw ∈M1(Bq), with p ∈ R, p > 2q − 1. Here dw is the Lebesgue measure on the ball Bq, γ := d ( q − 1 2 ) + 1 and κpd/2 = ∫ Bq ∆(I − w∗w)pd/2−γ dw. According to Theorem 2.4 of [26], the Heckman–Opdam hypergeometric function FBC(λ, k(p), x) with λ ∈ Cq, x ∈ Rq and k(p) as in (1.2) has the following integral representation for p ∈ R with p > 2q − 1: FBC(λ, k(p), x) = ∫ Bq×U0(q,F) ∆(λ−ρp)/2(gx(u,w)) dmp(w)du, where U0(q,F) denotes the identity component of U(q,F) and gx(u,w) = u−1(coshx+ w∗ sinhx)(coshx+ sinhxw)u. It is easily checked that U0(q,F) may be replaced by U(q,F) in the domain of integration. Notice further that x 7→ gx(u,w) extends to a holomorphic function on Cq. As the principal minors ∆r(a) are polynomial in the entries of a ∈ Hq(F), it follows that x 7→ ∆λ/2(gx(u,w)) extends to a holomorphic function on Cq for each λ ∈ P+. In view of relation (1.1), this leads to the following integral representation for the Jacobi polynomials Rpλ: Proposition 2.1. Let p ∈ R with p > 2q − 1 and k(p) = (d(p − q)/2, (d − 1)/2, d/2) with d ∈ {1, 2, 4}. Then the Jacobi polynomials Rpλ, λ ∈ P+, have the integral representation Rpλ(x) = ∫ Bq×U(q,F) ∆λ/2(gix(u,w)) dmp(w)du for x ∈ A0 (2.1) A Central Limit Theorem for Random Walks on the Dual of a Compact Grassmannian 5 with gix(u,w) = u−1(cosx+ w∗i sinx)(cosx+ i sinxw)u. We next turn to the Bessel functions which will show up in the Mehler–Heine formula. They are given in terms of Bessel functions of Dunkl type which generalize the spherical functions of Cartan motion groups; see [6] and [21] for a general background. We denote by JBk the Bessel function which is associated with the rational Dunkl operators for the root system Bq = {±ei,±ei ± ej : 1 ≤ i < j ≤ q} and multiplicity k = (k1, k2) corresponding to the roots ±ei and ±ei ± ej . We shall be concerned with multiplicities which are connected as follows to the BCq multiplicities k(p) from (1.2): k = (k1, k2) with k1 = k(p)1 + k(p)2 = d(p− q + 1)/2− 1/2, k2 = k(p)3 = d/2. For such k on Bq, we use the notion ϕ̃pλ(x) := JBk (x, iλ), x ∈ C, λ ∈ Cq. It is well-known that for integers p ≥ q, the ϕ̃pλ are the spherical functions of the Euclidean symmetric spaces G0/K, where K = S(U(p,F)× U(q,F)) and G0 = K nMp,q(F) is the Cartan motion group associated with the Grassmannian Gp,q(F). Hereby the double coset space G0//K is identified with the Weyl chamber C such that x ∈ C corresponds to the double coset of (Ip × Iq, (Ip−q, x)) ∈ G0, and in this way, K-biinvariant functions on G0 may be considered as functions on C. Two functions ϕ̃pλ and ϕ̃pµ coincide if and only if λ and µ are in the same Weyl group orbit. Finally, the bounded spherical functions are exactly those ϕ̃pλ with λ ∈ C. The Bessel functions ϕ̃pλ with d = dimFR and not necessarily integral parameter p are closely related to Bessel functions on the symmetric cone of positive definite q×q-matrices over F, see Section 4 of [24]. It has been shown there that for p > 2q− 1, they have a positive product formula which generalizes the product formula in the Cartan motion group cases and leads to a commutative hypergroup structure on the Weyl chamber C. Lemma 2.2. For p ∈ R with p > 2q−1, the Bessel functions ϕ̃pλ with λ ∈ Rq have the following integral representation: ϕ̃pλ(x) = ∫ Bq ∫ U(q,F) eiRe tr(wxuλ)dmp(w)du. (2.2) Proof. This follows readily from equations (3.12) and (4.4) in [24]; see also Proposition 5.3 of [26]. � Remark 2.3. There are finitely many geometric cases which are not covered by the range p ∈ ]2q − 1,∞[, namely the indices p ∈ {q, q + 1, . . . , 2q − 1}. In these cases, the Jacobi polyno- mials Rpλ and the Bessel functions ϕ̃pλ both admit interpretations as spherical functions and have an integral representation similar to that above, by the following reasoning: According to Lemma 2.1 of [25], the measure mp ∈ M1(Bq) with p ∈ N, p ≥ 2q is just the pushforward measure of the normalized Haar measure on U(p,F) under the mapping v 7→ σ∗0vσ0, with σ0 = ( Iq 0(p−q)×q ) ∈Mp,q(F). For p ∈ {q, q + 1, . . . , 2q − 1}, we now define the measure mp ∈ M1(Bq) in the same way as a pushforward measure of the Haar measure on U(p,F). (But in contrast to the case p ≥ 2q, it will not have a Lebesgue density in these cases). From the integral representations (3.3) and (4.4) of [24] for the Bessel functions, as well as Theorem 2.1 of [26] and relation (1.1) between Jacobi polynomials and hypergeometric functions, one obtains that the integral representations of Proposition 2.1 and Lemma 2.2 extend to the case p ∈ {q, q + 1, . . . , 2p− 1}. 6 M. Rösler and M. Voit We shall now compare the integral representations of Proposition 2.1 and Lemma 2.2, which will lead to the following quantitative Mehler–Heine (or Hilb-type) formula. Theorem 2.4. There exist constants C1, C2 > 0 such that for all p ∈ {q, q + 1, . . . , 2q − 1}∪ ]2q − 1,∞[, all λ ∈ P+, and x ∈ A0,∣∣Rpλ(x)− ϕ̃pλ(x) ∣∣ ≤ C1x 2 1λ1e C2x21λ1 . Thus in particular,∣∣∣Rpnλ (xn)− ϕ̃pλ(x) ∣∣∣ ≤ C1 n x21λ1e C2x21λ1/n → 0 for n→∞. Notice that the estimate of Theorem 2.4 is uniform in p, a fact which was to our knowledge so far not even noticed in the rank-one case. We conjecture that the statement of this theorem remains correct for p ∈ [q,∞[. Proof. We only consider the case p > 2q − 1 where the proof is based on Proposition 2.1 and Lemma 2.2. By the previous remark, the cases p = q, q + 1, . . . , 2q − 1 can be treated in the same way. Notice that it suffices to check uniformity in p for p > 2q − 1. We substitute w 7→ u∗w∗ in the integral (2.2) and obtain ϕ̃pλ(x) = ∫ Bq×U(q,F) eiRe tr(u∗w∗xuλ)dmp(w)du. Denoting the trace of the upper left (r × r)-block of a (q × q)-matrix by trr, we have Re tr(u∗w∗xuλ) = 1 2 q∑ r=1 (u∗((xw)∗ + xw)u)rrλr = q∑ r=1 [ trr(u ∗((xw)∗ + xw)u)− trr−1(u ∗((xw)∗ + xw)u) ] λr/2 = q∑ r=1 trr(u ∗((xw)∗ + xw)u)(λr − λr+1)/2 with λq+1 := 0. Hence ϕ̃pλ(x) = ∫ Bq×U(q,F) q∏ r=1 ei trr(u ∗((xw)∗+xw)u)(λr−λr+1)/2dmp(w)du. Furthermore, by Proposition 2.1, Rpλ(x) = ∫ Bq×U(q,F) q∏ r=1 ∆r(gix(u,w))(λr−λr+1)/2dmp(w)du. Telescope summation yields the well-known estimate∣∣∣∣∣ q∏ r=1 ar − q∏ r=1 br ∣∣∣∣∣ ≤ (max(|a1|, . . . , |aq|, |b1|, . . . , |bq|) )q−1 q∑ r=1 |ar − br| for a1, . . . , aq, b1, . . . , bq ∈ C. We thus obtain ∣∣Rpλ(x)− ϕ̃pλ(t) ∣∣ ≤ q∑ r=1 ∫ Bq×U(q,F) M(x, u, w, λ) (2.3) A Central Limit Theorem for Random Walks on the Dual of a Compact Grassmannian 7 × ∣∣∆r(gx(u,w))(λr−λr+1)/2 − ei trr(u∗((tw)∗+tw)u)(λr−λr+1)/2 ∣∣dmp(w)du with M(x, u, w, λ) := max ( 1, max r=1,...,q ∣∣∆r(gx(u,w))(λr−λr+1)/2 ∣∣q−1). We now investigate ∆r(gix(u,w))(λr−λr+1)/2 more closely. As x, u, w run through compacta, we obtain that uniformly in x, u, w, gix(u,w) = u−1(cosx+ w∗i sinx)(cosx+ i sinxw)u = u−1 ( Iq + w∗ix+O ( x2 ))( Iq + ixw +O ( x2 )) u = Iq + u−1(ixw + w∗ix)u+O ( x2 ) , and thus ∆r(gx(u,w)) = 1 + trr ( u−1(ixw + w∗ix)u ) +O ( x2 ) . (2.4) Using the power series for ln(1 + z), we further have ∆r(gix(u,w))(λr−λr+1)/2 = exp [ 1 2(λr − λr+1) ln ( 1 + trr ( u−1(ixw + w∗ix)u ) +O ( x2 ))] = exp [ 1 2(λr − λr+1) trr ( u−1(ixw + w∗ix)u ) +O ( x2 ) (λr − λr+1) ] . Notice that y := u−1(ixw + w∗ix)u is skew-Hermitian, that is y∗ = −y. Therefore trr(y) = − trr(y), which implies that Re(trr(y)) = 0. It follows that∣∣∆r(gix(u,w))(λr−λr+1)/2 ∣∣ = exp [ 1 2(λr − λr+1) Re ( trr(y) +O ( x2 ))] = e(λr−λr+1)O(x2). Note that these considerations apply for all fields F = R,C,H. It follows that there exists a constant C3 > 0 (independent of x, u, w, λ) such that M(x, u, w, λ) ≤ eC3x21λ1 for all x ∈ A0, λ ∈ P+, u ∈ U(q,F), w ∈ Bq. From this inequality we obtain by the mean value theorem that for all x ∈ A0 and λ ∈ P+,∣∣∆r(gx(u,w))(λr−λr+1)/2 − ei trr(u∗((xw)∗+xw)u)(λr−λr+1)/2 ∣∣ ≤ eC3x21λ1 − 1 ≤ C3x 2 1λ1e C3x21λ1 . These estimates together with (2.3) imply the assertion. � Example 2.5 (the rank one case). For q = 1 the Jacobi polynomials Rpλ associated with root system BC1 = {±e1,±2e1} are classical one-variable Jacobi polynomials in trigonometric parametrization. For integers p, the associated Grassmannians are the projective spaces Pp(F). For the details, recall that the classical Jacobi polynomials R (α,β) n with the normalization R (α,β) n (1) = 1 are given by R(α,β) n (x) = 2F1(α+ β + n+ 1,−n;α+ 1; (1− x)/2), (2.5) where n ∈ Z+, α, β > −1. It is easily derived from the example on p. 17 of [22] that Rpλ(x) = R (α,β) λ/2 (cos 2x) (2.6) for λ ∈ 2Z+, with α = (dp− 2)/2, β = (d− 2)/2; 8 M. Rösler and M. Voit see also [23, Section 5]. In the rank one case, the U(1,F) integral in representation (2.1) cancels by invariance of ∆ under unitary conjugation. Thus (2.1) reduces to Rpλ(x) = 1 κpd/2 ∫ B1 ((cosx+ wi sinx)(cosx+ i sinxw))λ/2 ( 1− |w|2 )d(p−1)/2−1 dw for λ ∈ Z+, p > 1. In particular, if F = R, then d = 1 and B1 = [−1, 1]. Thus R(p/2−1,−1/2) n (cos 2x) = 1 κp/2 ∫ 1 −1 (cosx+ it sinx)2n ( 1− t2 )(p−3)/2 dt. If F = C, then d = 2 and B1 = {z ∈ C : |z| ≤ 1}. Using polar coordinates z = teiθ, one obtains R(p−1,0) n (cos 2x) = 1 κp ∫ 1 −1 ∫ π 0 ( (cosx+ iteiθ sinx)(cosx+ ite−iθ sinx) )2n( 1− t2 )p−2 t dtdθ. The quaternionic case can be treated in a similar way. These formulas are just special cases of a well-known Laplace-type integral representation for Jacobi polynomials with general indices α ≥ β ≥ −1/2; see, e.g., [20, Section 18.10]. Let us finally mention that the Mehler–Heine formula 2.4 corresponds to [27, Theo- rem 8.21.12] and that in the case of rank two (q = 2), the Jacobi polynomials of type BC were first studied by Koornwinder [16, 17]. 3 Random walks on the dual of a compact Grassmannian and on P+ Recall that for integers p ≥ q the functions ϕλ := ϕpλ, λ ∈ P+ form the spherical functions of the compact Grassmannians G/K = Gp,q(F). As functions on G, they are positive-definite. In other words, the Jacobi polynomials (Rpλ)λ∈P+ are just the hypergroup characters of the compact double coset hypergroups G//K ∼= A0. We now recapitulate the associated dual hypergroup structures on P+. 3.1 Dual hypergroup structures on P+ As mentioned in the introduction, there is a one-to-one correspondence between the set of (positive definite) spherical functions of G/K, which is parametrized by P+, and the spherical unitary dual of G/K, i.e., the set ĜK of all equivalence classes of irreducible unitary representa- tions (π,H) of G whose restriction to K contains the trivial representation with multiplicity one. Here a representation (π,H) ∈ ĜK and its spherical function ϕπ are related by ϕπ(x) = 〈u, π(x)u〉 for x ∈ G with some K-invariant vector u ∈ H with ‖u‖ = 1, which is determined uniquely up to a complex constant of absolute value 1. Now consider λ, µ ∈ P+ with associated spherical functions ϕλ, ϕµ and the corresponding representations (πλ, Hλ), (πµ, Hµ) ∈ ĜK with K-invariant vectors uλ, uµ. The tensor product (πλ ⊗ πµ, Hλ ⊗Hµ) is a finite-dimensional unitary representation of G which decomposes into a finite orthogonal sum (⊕kτk = πλ ⊗ πµ, ⊕kH̃k = Hλ ⊗Hµ) A Central Limit Theorem for Random Walks on the Dual of a Compact Grassmannian 9 of irreducible unitary representations (τk, H̃k) where some of them may appear several times. Consider the orthogonal projections pk : Hλ ⊗Hµ → H̃k. Then the vectors pk(uλ ⊗ uµ) ∈ H̃k are K-invariant, and for pk(uλ ⊗ uµ) 6= 0, we obtain (τk, H̃k) ∈ ĜK , i.e., (τk, H̃k) is equal to some (πτ , Hτ ), τ ∈ P+. Moreover, for g ∈ G, ϕλ(g)ϕµ(g) = 〈uλ ⊗ uµ, (πλ ⊗ πµ)(g)uλ ⊗ uµ〉 = ∑ k 〈pk(uλ ⊗ uµ), τk(g) pk(uλ ⊗ uµ)〉 = ∑ k ‖pk(uλ ⊗ uµ)‖2ϕτk(g) with ∑ k ‖pk(uλ⊗uµ)‖2 = 1. For λ, µ, τ ∈ P+ we now define cλ,µ,τ ≥ 0 as ‖pk(uλ⊗uµ)‖2, whenever (τk, H̃k) = (πτ , Hτ ) appears above with a positive part, and 0 otherwise. As ϕλ(g) ∈ R for all λ, g in our case, these nonnegative linearization coefficients also satisfy cλ,µ,τ = dimHτ ∫ G ϕλ(g)ϕµ(g)ϕτ (g) dg. For λ, µ ∈ P+ we define the probability measure δλ ∗d,p δµ := ∑ τ∈P+ cλ,µ,τδτ ∈M1(P+) (3.1) with finite support. By its very construction, this convolution can be extended uniquely in a weakly continuous, bilinear way to a probability preserving, commutative, and associative convolution on the Banach space Mb(P+) of all bounded, signed measures on P+. Moreover, as all spherical functions are R-valued in our specific examples, the contragredient representation of any element in ĜK is the same representation, i.e., the canonical involution .∗ on P+ ∼= ĜK is the identity. In summary, (Mb(P+), ∗d,p) is a commutative Banach-∗-algebra with the complex conjugation µ∗(A) := µ(A) as involution. Moreover, (P+, ∗d,p) becomes a so-called Hermitian hypergroup in the sense of Dunkl, Jewett and Spector; see [3, 7, 15]. The Haar measure on this hypergroup, which is unique up to a multiplicative constant, is the positive measure ω = ∑ λ∈P+ h(λ)δλ with h(λ) = c−1λ,λ,0 = ∫ G ϕ2 λ(g) dg = 1 dimHλ , where the first two equations follow from general hypergroup theory (see [15]) and the last one from the theory of Gelfand pairs (see, e.g., [9]). The coefficients cλ,µ,τ of the convolution ∗d,p on P+ are related to the unique product li- nearization RpλR p µ = ∑ τ∈P+ cλ,µ,τR p τ of the Jacobi polynomials Rpλ. It is clear by our construction that for integers p ≥ q, all cλ,µ,τ are nonnegative with ∑ τ∈P+ cλ,µ,τ = 1. Clearly, as Rpλ(0) = 1 for all λ, the normalization also holds for all real p ∈ [q,∞[. We conjecture that actually all cλ,µ,τ are nonnegative for all p ∈ [q,∞[ or at least for all p ∈ [2q − 1,∞[. Suppose that for fixed p ∈ [q,∞[ the linearization coefficients cλ,µ,τ are all nonnegative. Then equation (3.1) defines a commutative discrete hypergroup structure (P+, ∗d,p) with the convolution δλ ∗d,p δµ := ∑ τ∈P+ cλ,µ,τδτ ∈M1(P+) 10 M. Rösler and M. Voit of point measures. For instance, in the rank one case of Example 2.5 the linearization coeffi- cients cλ,µ,τ are explicitly known and nonnegative for all p ≥ 1 as the product linearization coefficients of the associated one-dimensional Jacobi polynomials. 3.2 Random walks on P+ We next introduce certain random walks on P+, i.e., time-homogeneous Markov chains on P+ whose transition probabilities are given in terms of the product linearizations coefficients cλ,µ,τ for some fixed p ∈ [q,∞[. This concept even works, under a suitable restriction, in the case where some of the cλ,µ,τ are negative. To describe the restriction, we fix p ∈ [q,∞[ and define the set P p+ := {λ ∈ P+ : cλ,µ,τ ≥ 0 for all µ, τ ∈ P+} as well as M1 p (P+) := { ν ∈M1(P+) : supp ν ⊂ P p+ } . We shall call the probability measures ν ∈M1 p (P+) admissible. Clearly, for integers p ≥ q, as well as for q = 1 we have M1 p (P+) = M1(P+). Unfortunately, it seems difficult to find further examples. To illustrate the problem, consider the measure δ(1,0,...,0) ∈ M1(P+) for q ≥ 2 and p ∈ [q,∞[. The explicit Pieri-type formula (6.4) of [28] then leads to a concrete product linearization formula for RpλR p (1,0,...,0). It can be easily derived from [28] that for all q ≥ 2, p ∈ [q,∞[, and all λ 6= µ, we have cλ,(1,0,...,0),µ ≥ 0 as desired. However, we are so far not able to check from [28] that cλ,(1,0,...,0),λ ≥ 0 holds, which would be necessary for δ(1,0,...,0) ∈M1 p (P+). As before, we fix d = 1, 2, 4 and p ∈ [q,∞[. We also fix an admissible probability measure ν = ∑ µ pµδµ ∈ M1 p (P+) and consider a time-homogeneous Markov chain (Sd,pn )n≥0 in discrete time on P+ starting at time 0 in the origin 0 ∈ P+ and with transition probability P ( Sd,pn+1 = τ | Sd,pn = λ ) = (ν ∗d,p δλ)({τ}), λ, τ ∈ P+, n ∈ N, where ν ∗d,p δλ := ∑ τ (∑ µ pµcλ,µ,τ ) δτ ∈M1(P+). Such Markov-chains are called random walks on (P+, ∗d,p) associated with ν. It is well-known and can be easily checked by induction that for all n the n-fold convolution power ν(n) := ν ∗d,p · · · ∗d,p ν ∈M1(P+) exists, and that ν(n) is just the distribution P Sd,p n of Sd,pn . In view of the Central Limit Theorem 1.1, we give an interpretation of these convolution powers δ (n) λ for integers p ≥ q and λ ∈ P+ with λ 6= 0 in terms of representation theory. We expect that this result is well-known, but we do not know an explicit proof in the literature. Lemma 3.1. Let (πλ, Hλ) be the non-trivial irreducible unitary representation of G associated with λ ∈ P+, λ 6= 0 and with a K-invariant vector uλ ∈ Hλ with ‖uλ‖ = 1. For each n ∈ N, decompose the n-th tensor power (π⊗,nλ , H⊗,nλ ) into its irreducible components( π⊗,nλ , H⊗,nλ ) = (⊕ τn πτn , ⊕ τn Hτn ) (3.2) and consider the orthogonal projections pτn : H⊗,nλ → Hτn. Then for all n ∈ N, δ (n) λ = ∑ τn ∥∥pτn(u⊗,nλ )∥∥2δτn . A Central Limit Theorem for Random Walks on the Dual of a Compact Grassmannian 11 Proof. We proceed by induction. In fact, the case n = 1 is trivial. For n → n + 1, we start with (3.2) and the associated orthogonal projections pτn : H⊗,nλ → Hτn . We now decompose the products πτn ⊗ πλ and obtain( π⊗,n+1 λ , H⊗,n+1 λ ) = (⊕ τn (πτn ⊗ πλ), ⊕ τn (Hτn ⊗Hλ) ) = (⊕ τn (⊕ µk,n πµk,n ) , ⊕ τn (⊕ µk,n Hµk,n )) . Notice that here the sum ⊕ τn ⊕ µk,n corresponds to the sum ⊕ τn+1 of the lemma with n + 1 instead of n. We now consider the orthogonal projections pµk,n : H⊗,n+1 λ → Hµk,n . Then pτn ( u⊗,nλ ) = c ∥∥pτn(u⊗,nλ )∥∥uτn , where |c| = 1, and thus∥∥pµk,n(u⊗,n+1 λ )∥∥2 = ∥∥pµk,n(pτn(u⊗,nλ ) ⊗ uλ )∥∥2 = ∥∥pτn(u⊗,nλ )∥∥2∥∥pµk,n(uτn ⊗ uλ)∥∥2. This fact, the assumption of our induction and the definition of the convolution now readily imply the assertion of the lemma for n+ 1. � We shall prove below that under a natural moment condition on a probability measure ν ∈ M1(P+), the C-valued random variables 1√ n Sd,pn converge in distribution for n → ∞. In order to identify the limit distribution µ = µ(d, p, ν) ∈ M1(C) in this central limit theorem, we need some further preparations. 3.3 Bessel convolutions on C and Laguerre ensembles As described in Section 2, the Bessel functions ϕ̃pλ with λ ∈ C make up the set of bounded spherical functions of the Euclidean symmetric space G0/K with K = U(p,F) × U(q,F) and G0 = K nMp,q(F). Thus by the notation of [3] and [15], the chamber C ∼= G0//K with the associated double coset convolution •d,p is a commutative double coset hypergroup with the functions ϕ̃pλ as (bounded) hypergroup characters. We now introduce the probability measures dρd,p(x) := c−1d,p q∏ j=1 x d(p−q+1)−1 j ∏ 1≤i<j≤q ( x2i − x2j )d e−(x 2 1+···+x2q)/2 dx on the Weyl chamber C, with the normalization constant cd,p = ∫ C q∏ j=1 x d(p−q+1)−1 j ∏ 1≤i<j≤q ( x2i − x2j )d e−(x 2 1+···+x2q)/2 dx. The measure ρd,p ∈ M1(C) is well-known in the random matrix theory of so-called Laguerre- or β-ensembles as it is the distribution of the singular values of a Mp,q(F)-valued random variable for which the real and imaginary parts of all entries are i.i.d. and standard normally distributed. This fact is well-known; it can also be derived from the Haar measure of the double coset hypergroups (C, •d,p) in [24, Section 4.1]. Moreover, having this group theoretic interpretation in mind, one easily obtains the following well-known relation from the Fourier transform of a multivariate standard normal distribution:∫ C ϕ̃pλ(x) dρd,p(x) = e−(λ 2 1+···+λ21)/2 for λ ∈ C, (3.3) see [10, Proposition XV.2.1] or [30]. This identification of the spherical Fourier-transform of ρd,p will be essential in the following for the central limit theorem. 12 M. Rösler and M. Voit 3.4 A central limit theorem for random walks on P+ The probability measure ρd,p appears in the CLT below as limit up to some scaling parameter σ2 = σ2(ν, p, d), which admits an interpretation as a variance parameter. For the description of σ2, we need the so-called moment functions on (P+, ∗d,p) up to order two. For the general theory of such moment functions and their applications to limit theorems for random walks on hypergroups we refer to [3, Chapter 7] and [32], and the references there. The moment functions are characterized by additive functional equations similar to the multiplicative ones for hypergroup characters. They are usually defined in terms of (partial) derivatives with respect to the spectral variables at the identity character. In our examples, the identity corresponds to x = 0 ∈ C. This motivates the following definition. Definition 3.2. Let p ∈ [q,∞[ be fixed. For k, l = 1, . . . , q we define the moment functions mk,mk,l : P+ → R of the first and second order by mk(λ) := i ∂ ∂xk Rλ(x) ∣∣∣∣ x=0 and mk,l(λ) := − ∂2 ∂xk∂xl Rλ(x) ∣∣∣∣ x=0 . Let us collect some properties of these moment functions. Lemma 3.3. (1) For all k, l = 1, . . . , q with k 6= l and all λ ∈ P+, mk(λ) = mk,l(λ) = 0. (2) The functions mk,k are independent of k = 1, . . . , q, and the function m := m1,1 : P+ → R is a second order polynomial of the form m(λ) = 1 4 q∑ r,s=1 ar,s(λr − λr+1)(λs − λs+1) + 1 2 q∑ r=1 br(λr − λr+1) with suitable coefficients ar,s, br. In particular, m(0) = 0. (3) For all λ, τ ∈ P+, ∫ P+ m d(δλ ∗d,p δτ ) = m(λ) +m(τ). Proof. The Jacobi polynomials Rλ(x) are invariant under the Weyl group of type B acting in the variable x. In particular, Rλ(x1, . . . , xq) is even in each xi, and this gives part (1). Moreover, as Rλ(x1, . . . , xq) is invariant under permutations of the xi, the function mk,k is independent of k. We now study m := m1,1 more closely. We start with the case p > 2q − 1. In this case we obtain from the integral representation (2.1) that m(λ) = − ∫ Bq×U(q,F) ∂2 ∂x21 ( ∆λ/2(gix(u,w)) ) ∣∣∣∣ x=0 dmp(w)du (3.4) with the power function ∆λ/2(gix(u,w)) = q∏ r=1 ∆r(gix(u,w))(λr−λr+1)/2 with λq+1 = 0. A short calculation, using that ∆λ/2(g0(u,w)) = 1, gives ∂2 ∂x21 ∆λ/2(gix(u,w)) ∣∣∣∣ x=0 = 1 4 ( q∑ r=1 ∂ ∂x1 (∆r(gix(u,w))) ∣∣∣∣ x=0 (λr − λr+1) )2 + 1 2 q∑ r=1 ∂2 ∂x21 (ln ∆r(gix(u,w))) ∣∣∣∣ x=0 (λr − λr+1). (3.5) A Central Limit Theorem for Random Walks on the Dual of a Compact Grassmannian 13 Formulas (3.4) and (3.5) now imply that m is a second order polynomial as claimed, with linear terms br = ∫ Bq×U(q,F) ∂2 ∂x21 (ln ∆r(gix(u,w))) ∣∣∣∣ x=0 dmp(w)du. The coefficients ar,s are obtained from the Taylor expansion (2.4) of ∆r(gix(u,w)). They are given by ar,s := ∫ Bq×U(q,F) trr(u ∗(w∗P1 + P1w)u) trs(u ∗(w∗P1 + P1w)u) dmp(w)du (3.6) with the diagonal matrix P1 := diag(1, 0, . . . , 0) ∈Mq(F). This proves that m is a second order polynomial for p > 2q − 1. The case p ≥ q follows by analytic continuation. Finally, for the proof of part (3) we observe that for λ, τ ∈ P+,∫ P+ m d(δλ ∗d,p δτ ) = − ∫ P+ ∂2 ∂x21 Rκ(x) ∣∣∣∣ x=0 d(δλ ∗d,p δτ )(κ) = − ∂2 ∂x21 (∫ P+ Rκ(x) d(δλ ∗d,p δτ ) ) ∣∣∣∣ x=0 = − ∂2 ∂x21 (Rλ(x)Rτ (x)) ∣∣∣∣ x=0 = m(λ) +m(τ). Notice that the last equality follows from part (1) and Rλ(0) = 1. � Example 3.4 (the rank one case). For q = 1, the moment function m is given by m(λ) = λ(λ+ dp+ d− 2) dp , λ ∈ 2Z+. In fact, this can be easily derived from the definition of m and the explicit formulas for the classical Jacobi polynomials in (2.5) and (2.6). Moreover, it can be also derived from the proof of part (2) of the preceding lemma and a direct elementary computation of a1,1 and b1 for q = 1 there. We shall need the following variant of Lemma 3.3(2) on the growth of m for p ∈ [q,∞[: Lemma 3.5. For all x ∈ Rq, λ ∈ P+ and k, l = 1, . . . , q,∣∣∣∣ ∂∂xkRpλ(x) ∣∣∣∣ ≤ λ1 and ∣∣∣∣ ∂2 ∂xk∂xl Rpλ(x) ∣∣∣∣ ≤ λ21. In particular, m(λ) ≤ λ21 for λ ∈ P+. Proof. Let again W denote the Weyl group of type BCq. We introduce the normalized W - invariant orbit sums M̃λ(x) := 1 |Wλ| ∑ µ∈W.λ ei〈λ,x〉, λ ∈ P+. Then the Jacobi polynomials Rpλ can be written as linear combinations of such orbit sums. It follows from the considerations in [19, Section 11] that for non-negative multiplicity values, the expansion coefficients are all non-negative. That is, Rpλ = ∑ µ∈P+:µ≤λ cλµM̃µ 14 M. Rösler and M. Voit with cλµ = cpλµ ≥ 0, ∑ µ≤λ cλµ = 1. Here ≤ denotes the dominance order on P+ given by µ ≤ λ⇐⇒ r∑ i=1 µi ≤ r∑ i=1 λi for r = 1, . . . , q. We have ∂xkM̃λ(x) = 1 |Wλ| ∑ µ∈W.λ iµke i〈µ,x〉, ∂xkxlM̃λ(x) = −1 |Wλ| ∑ µ∈W.λ µkµle i〈µ,x〉. Notice that |µk| ≤ λk ≤ λ1 for each µ ∈Wλ. We thus obtain, independently of x ∈ A0,∣∣∂xkM̃λ(x) ∣∣ ≤ λ1, ∣∣∂xk∂xlM̃λ(x) ∣∣ ≤ λ21. Further, if µ ∈ P+ with µ ≤ λ, then µ1 ≤ λ1 and therefore∣∣∂xkRpλ(x) ∣∣ ≤∑ µ≤λ cλµ ∣∣∂xkM̃µ(x) ∣∣ ≤∑ µ≤λ cλµµ1 ≤ λ1. In the same way,∣∣∂xk∂xlRpλ(x) ∣∣ ≤∑ µ≤λ cλµ ∣∣∂xk∂xlM̃µ(x) ∣∣ ≤∑ µ≤λ cλµµ 2 1 ≤ λ21. � We also need some further properties of the moment function m. We here have to restrict our attention to the case p ∈ {q, q+ 1, . . . , 2q− 1}∪]2q− 1,∞[. We shall assume this restriction from now on. We expect that the results below are also valid for all p ∈ [q,∞[. Lemma 3.6. (1) The matrix A = (ar,s)r,s=1,...,q ∈Mq(R) is positive definite. (2) For all λ ∈ P+ \ {0}, m(λ) > 0. (3) There exists a constant C1 > 0 such that for all λ ∈ P+, C1λ 2 1 ≤ m(λ). Proof. For the proof of (1), we first consider the case p ∈ ]2q − 1,∞[ and conclude from the definition of the ar,s in the proof in Lemma 3.3 that A is symmetric, and that for all τ ∈ Rq, τTAτ = ∫ Bq×U(q,F) ( q∑ r=1 τr trr(·u∗(w∗P1 + P1w)u) )2 dmp(w)du, where the functions (w, u) 7→ trr(·u∗(w∗P1 + P1w)u), Bq × U(q,F)→ R, are linearly independent for r = 1, . . . q. This shows that τTAτ > 0 for all τ ∈ Rq with τ 6= 0 as claimed. The case of integers p ≥ q can be handled in a similar way by using a modified version of integral representation (3.6) for ar,s which is based on Remark 2.3 instead of Proposition 2.1. For the proof of part (2) we proceed as in the proof of Lemma 3.5 and write Rpλ = ∑ µ∈P+:µ≤λ cλµM̃µ with cλµ ≥ 0, cλλ > 0, ∑ µ≤λ cλµ = 1. A Central Limit Theorem for Random Walks on the Dual of a Compact Grassmannian 15 Thus for λ ∈ P+ \ {0}, m(λ) = −∂2x1R p λ(x) ∣∣ x=0 = ∑ µ∈P+:µ≤λ cλµ |Wµ| ∑ τ∈Wµ τ21 > 0. For the proof of part (3), we use (1) and write m(λ) as m(λ) = λT Ãλ− bTλ with some positive definite matrix à and some b ∈ Rq. We thus find constants c, d > 0 such that m(λ)− cλ21 > 0 holds for all λ ∈ P+ with λ1 ≥ d. As there are only finitely many λ ∈ P+ with λ1 < d, we conclude from part (2) that there exists some C1 > 0 with m(λ)− cλ21 > 0 for all λ ∈ P+ with λ 6= 0. � Remark 3.7. The nonnegativity of m(λ) in Lemma 3.6(2) can be easily established directly. In fact, assume that m(λ) < 0 for some λ ∈ P+. Then the Taylor formula Rλ(x) = 1− m(λ) 2 ( x21 + · · ·+ x2q ) +O ( ‖x‖3 ) implies that Rλ(x) > 1 for some x close to 0, and thus by the Weyl group invariance of Rλ, for some x ∈ A0. But this contradicts the fact that ‖Rλ‖∞ ≤ 1 on A0, which is a consequence of Proposition 2.1. However, we have no different proof for the strict positivity of m(λ) for λ 6= 0 than the one given in Lemma 3.6(2). We next use the moment function m in order to define a modified variance of measures ν ∈M1(P+) depending on the underlying convolution ∗d,p. This modified variance will appear in the CLT below. Definition 3.8. Let ν ∈M1(P+) be a probability measure with finite second moments, meaning that ∑ λ∈P+ λ21ν({λ}) <∞. Then the modified second moment σ2 := σ2(ν) of ν is defined as σ2 := ∑ λ∈P+ m(λ)ν({λ}). Notice that by Lemmas 3.3 and 3.6, σ2 is finite and non-negative where σ2 = 0 holds precisely for ν = δ0. Lemma 3.9. Let ν ∈M1(P+) be a probability measure with finite second moments and with the modified variance σ2 ≥ 0. Then the spherical Fourier transform Fν : Rq → R with Fν(x) := ∑ λ∈P+ Rpλ(x)ν({λ}) is twice continuously differentiable on Rq with Fν(0) = 1, ∇Fν(0) = 0, and Hesse matrix D2(Fν)(0) = −σ2Iq. Proof. As ν has finite second moments, we conclude form Lemma 3.5 that for all k, l = 1, . . . , q, the series∑ λ∈P+ ν({λ}) ∂ ∂xk Rpλ(x), ∑ λ∈P+ ν({λ}) ∂2 ∂xk∂xl Rpλ(x) converge uniformly with respect to x ∈ Rq. This implies that Fν(x) is twice continuously differentiable on Rq, and the partial derivatives commute with the summation. The derivatives at x = 0 are now obtained by Lemma 3.3. � 16 M. Rösler and M. Voit We now turn to random walks (Sd,pn )n≥0 on (P+, ∗d,p) associated with some admissible ν ∈ M1 p (P+). It is well-known (see, e.g., [3, Section 7.3]) that the additive functional equation for m in Lemma 3.3(3) leads to relations between the modified variance of ν and random walks associated with ν. Lemma 3.10. (1) For all ν1, ν2 ∈M1 p (P+) with finite second moments, the measure ν1 ∗d,p ν2 ∈M1(P+) has also finite second moments, and σ2(ν1 ∗d,p ν2) = σ2(ν1) + σ2(ν2). (2) Let (Sd,pn )n≥0 be a random walk on (P+, ∗d,p) associated with the measure ν ∈M1 p (P+) with finite second moments. Then, for all integers n ≥ 0, the expectation of m(Sd,pn ) satisfies E(m(Sd,pn )) = nσ2(ν), and the process (m(Sd,pn )− nσ2(ν))n≥0 is a martingale with respect to the canonical filtration of (Sd,pn )n≥0. Proof. Part (1) follows easily from Lemma 3.3(3); c.f. [3, Section 7.3.7]. Moreover, the first assertion of (2) follows from (1) by induction. For the proof of the second statement in (2) we refer to [3, Proposition 7.3.19]. � Lemma 3.11. For a ∈ ]0,∞[, define the finite set Ka := {λ ∈ P+ : λ1 ≤ a}. Let (Sd,pn )n≥0 be a random walk on (P+, ∗d,p) as described above associated with some admissible ν ∈M1 p (P+) with finite second moments. Then, for each ε > 0 there exists some a ≥ 1 such that for all n ∈ N, P (Sd,pn 6∈ K√na) ≤ ε. Proof. By Lemma 3.6, we find c > 0 with m(λ) ≥ cλ21 for λ ∈ P+. Therefore, m(λ) ≥ cλ21 ≥ ca2n for all λ ∈ P+ \K√na and all n. Hence, P ( Sd,pn 6∈ K√na ) ≤ P ( m ( Sd,pn ) ≥ ca2n ) ≤ E(m(Sd,pn )) ca2n = σ2n ca2n = σ2 ca2 with the finite modified variance σ2. This implies the claim. � We are now ready to prove the main result of this section: Theorem 3.12. Let p ∈ {q, q + 1, . . . , 2q − 1}∪]2q − 1,∞[. and ν ∈ M1 p (P+) be an admissible probability measure with ν 6= δ0 and with finite second moments. Let σ2 ∈ ]0,∞[ be the modified modified variance of ν, and (Sd,pn )n≥0 be a random walk on (P+, ∗d,p) associated with ν. Then Sd,pn / √ nσ2 converges in distribution to the distribution ρd,p ∈ M1(C) of a Laguerre ensemble in C. Proof. Fix x ∈ C. Let n ∈ N be large enough such that x/ √ n ∈ A0. By Section 3.2, Sd,pn has the distribution ν(n). We thus obtain from the multiplicativity of the spherical Fourier transform of probability measures on (P+, ∗d,p), Lemma 3.9, the qualitative Taylor formula, and from the properties of the moment functions in Lemma 3.3 that for n→∞, E ( R Sd,p n (x/ √ n) ) = ( Fν(x/ √ n) )n = ( 1− σ2 2n ( x21 + · · ·+ x2q ) + o(1/n) )n −→ e−(x 2 1+···+x2q)σ2/2. (3.7) Now fix ε > 0. By Lemma 3.11, there is so a > 0 such that for all n ∈ N, P (Sd,pn 6∈ K√na) ≤ ε. We now conclude from the Mehler–Heine formula (2.4) that for all λ ∈ K√na, that is, λ ∈ P+ with λ1 ≤ a √ n,∣∣Rpλ(x/ √ n)− ϕ̃pλ(x/ √ n) ∣∣ ≤ C1x 2 1λ1e C2x21λ1/ √ n ≤ ε A Central Limit Theorem for Random Walks on the Dual of a Compact Grassmannian 17 whenever n is sufficiently large. As |Rpλ(x/ √ n)| ≤ 1 and |ϕ̃pλ(x/ √ n)| ≤ 1, we thus have∣∣E(R Sd,p n (x/ √ n) ) − E ( ϕ̃ Sd,p n (x/ √ n) )∣∣ ≤ E (∣∣R Sd,p n (x/ √ n)− ϕ̃ Sd,p n (x/ √ n) ∣∣1{Sd,p n ∈K√na} ) + 2P ( Sd,pn 6∈ K√na ) ≤ 3ε for n sufficiently large. Together with (3.7) and the identity ϕ̃pcy(x) = ϕ̃py(cx) for c > 0 and x, y ∈ C, this implies that for all x ∈ C, lim n→∞ E ( ϕ̃p Sd,p n / √ σ2n (x) ) = lim n→∞ E ( ϕ̃p Sd,p n ( x/ √ σ2n )) = lim n→∞ E ( R Sd,p n ( x/ √ σ2n )) = e−(x 2 1+···+x2q)/2. From this limit, equation (3.3) and Levy’s continuity theorem for the spherical Fourier transform on the double coset hypergroup (C, •d,p) (see, e.g., [3, Section 4.2]), we now infer that Sd,pn / √ σ2n converges in distribution to ρd,p as claimed. � Theorem 1.1 in the introduction is an immediate consequence from Theorem 3.12 and Lem- ma 3.1. We also remark that the methods of the preceding proof lead with some additional technical effort to rates of convergence in the CLT; see [11, 29] for the rank one case. We finish this paper with a strong law of large numbers; it follows easily from the preceding properties of the moment function m, in combination with strong laws of large numbers for random walks on commutative hypergroups in [3, Section 7.3] and [32]. Theorem 3.13. Let ν ∈ M1 p (P+) be admissible with with finite second moments, and let (Sd,pn )n≥0 be an associated random walk on (P+, ∗d,p). Then for all ε > 1/2, Sn/n ε → 0 al- most surely. Proof. Consider first the hypergroup case with an integer p ≥ q. By Lemmas 3.3 and 3.6, all conditions of Theorem 7.3.26 in [3] are satisfied for the time-homogeneous random walk (Sd,pn )n≥0, the sequence (rn := n2ε)n≥1, and the moment function m instead of m2 in [3]. This theorem now yields that m(Sd,pn )/n2ε tends to 0 almost surely, and Lemma 3.6 proves the claim. An inspection of the details in the proof of Theorem 7.3.26 in [3] shows that this theorem is also available for all p and admissible ν ∈M1 p (P+) which proves the theorem in general. � References [1] Aslaksen H., Quaternionic determinants, Math. Intelligencer 18 (1996), 57–65. [2] Ben Säıd S., Ørsted B., Analysis on flat symmetric spaces, J. Math. Pures Appl. 84 (2005), 1393–1426. [3] Bloom W.R., Heyer H., Harmonic analysis of probability measures on hypergroups, de Gruyter Studies in Mathematics, Vol. 20, Walter de Gruyter & Co., Berlin, 1995. [4] Clerc J.L., Une formule asymptotique du type Mehler–Heine pour les zonoles d’un espace riemannien symétrique, Studia Math. 57 (1976), 27–32. [5] Clerc J.L., Roynette B., Un théorème central-limite, in Analyse harmonique sur les groupes de Lie (Sém., Nancy-Strasbourg, 1976–1978), II, Lecture Notes in Math., Vol. 739, Springer, Berlin, 1979, 122–131. [6] de Jeu M., Paley–Wiener theorems for the Dunkl transform, Trans. Amer. Math. Soc. 358 (2006), 4225– 4250, math.CA/0404439. [7] Dunkl C.F., The measure algebra of a locally compact hypergroup, Trans. Amer. Math. Soc. 179 (1973), 331–348. [8] Eymard P., Roynette B., Marches aléatoires sur le dual de SU(2), in Analyse harmonique sur les groupes de Lie, Lecture Notes in Math., Vol. 497, Editors P. Eymard, J. Faraut, G. Schiffmann, R. Takahashi, Springer-Verlag, Berlin – New York, 1975, 108–152. http://dx.doi.org/10.1007/BF03024312 http://dx.doi.org/10.1016/j.matpur.2005.01.006 http://dx.doi.org/10.1515/9783110877595 http://dx.doi.org/10.1515/9783110877595 http://dx.doi.org/10.1007/BFb0062490 http://dx.doi.org/10.1090/S0002-9947-06-03960-2 http://arxiv.org/abs/math.CA/0404439 http://dx.doi.org/10.1090/S0002-9947-1973-0320635-2 http://dx.doi.org/10.1007/BFb0078014 18 M. Rösler and M. Voit [9] Faraut J., Analyse harmonique sur les paires de Guelfand et les espaces hyperboliques, in Analyse har- monique (Université de Nancy I, 1980), Editors J.L. Clerc, P. Eymard, J. Faraut, M. Räıs, R. Takahasi, Les Cours du C.I.M.P.A., Nice, France, 1983, 315–446. [10] Faraut J., Korányi A., Analysis on symmetric cones, Oxford Mathematical Monographs, Oxford Science Publications, The Clarendon Press, Oxford University Press, New York, 1994. [11] Gallardo L., Comportement asymptotique des marches aléatoires associées aux polynômes de Gegenbauer et applications, Adv. in Appl. Probab. 16 (1984), 293–323. [12] Heckman G., Dunkl operators, Astérisque (1997), exp. No. 828, 223–246. [13] Heckman G., Schlichtkrull H., Harmonic analysis and special functions on symmetric spaces, Perspectives in Mathematics, Vol. 16, Academic Press, Inc., San Diego, CA, 1994. [14] Helgason S., Groups and geometric analysis: integral geometry, invariant differential operators, and spherical functions, Mathematical Surveys and Monographs, Vol. 83, Amer. Math. Soc., Providence, RI, 2000. [15] Jewett R.I., Spaces with an abstract convolution of measures, Adv. Math. 18 (1975), 1–101. [16] Koornwinder T.H., Two-variable analogues of the classical orthogonal polynomials, in Theory and Applica- tion of Special Functions (Proc. Advanced Sem., Math. Res. Center, Univ. Wisconsin, Madison, Wis., 1975), Academic Press, New York, 1975, 435–495. [17] Koornwinder T.H., Harmonics and spherical functions on Grassmann manifolds of rank two and two-variable analogues of Jacobi polynomials, in Constructive Theory of Functions of Several Variables (Proc. Conf., Math. Res. Inst., Oberwolfach, 1976), Lecture Notes in Math., Vol. 571, Springer, Berlin, 1977, 141–154. [18] Koornwinder T.H., Positive convolution structures associated with quantum groups, in Probability Measures on Groups, X (Oberwolfach, 1990), Plenum, New York, 1991, 249–268. [19] Macdonald I.G., Orthogonal polynomials associated with root systems, Sém. Lothar. Combin. 45 (2000), Art. B45a, 40 pages, math.QA/0011046. [20] Olver F.W.J., Lozier D.W., Boisvert R.F., Clark C.W., NIST handbook of mathematical functions, U.S. De- partment of Commerce, National Institute of Standards and Technology, Washington, DC; Cambridge Uni- versity Press, Cambridge, 2010. [21] Opdam E.M., Dunkl operators, Bessel functions and the discriminant of a finite Coxeter group, Compositio Math. 85 (1993), 333–373. [22] Opdam E.M., Harmonic analysis for certain representations of graded Hecke algebras, Acta Math. 175 (1995), 75–121. [23] Remling H., Rösler M., Convolution algebras for Heckman–Opdam polynomials derived from compact Grass- mannians, J. Approx. Theory , to appear. [24] Rösler M., Bessel convolutions on matrix cones, Compos. Math. 143 (2007), 749–779, math.CA/0512474. [25] Rösler M., Positive convolution structure for a class of Heckman–Opdam hypergeometric functions of type BC, J. Funct. Anal. 258 (2010), 2779–2800, arXiv:0907.2447. [26] Rösler M., Voit M., Integral representation and uniform limits for some Heckman–Opdam hypergeometric functions of type BC, Trans. Amer. Math. Soc., to appear, arXiv:1402.5793. [27] Szegö G., Orthogonal polynomials, American Mathematical Society Colloquium Publications, Vol. 23, Amer. Math. Soc., Providence, R.I., 1959. [28] van Diejen J.F., Properties of some families of hypergeometric orthogonal polynomials in several variables, Trans. Amer. Math. Soc. 351 (1999), 233–270, q-alg/9604004. [29] Voit M., Central limit theorems for a class of polynomial hypergroups, Adv. in Appl. Probab. 22 (1990), 68–87. [30] Voit M., Bessel convolutions on matrix cones: algebraic properties and random walks, J. Theoret. Probab. 22 (2009), 741–771, math.CA/0603017. [31] Voit M., Central limit theorems for hyperbolic spaces and Jacobi processes on [0,∞[, Monatsh. Math. 169 (2013), 441–468, arXiv:1201.3490. [32] Zeuner H., Moment functions and laws of large numbers on hypergroups, Math. Z. 211 (1992), 369–407. http://dx.doi.org/10.2307/1427071 http://dx.doi.org/10.1090/surv/083 http://dx.doi.org/10.1016/0001-8708(75)90002-X http://dx.doi.org/10.1007/BFb0086570 http://dx.doi.org/10.1007/978-1-4899-2364-6_18 http://arxiv.org/abs/math.QA/0011046 http://dx.doi.org/10.1007/BF02392487 http://dx.doi.org/10.1016/j.jat.2014.07.005 http://dx.doi.org/10.1112/S0010437X06002594 http://arxiv.org/abs/math.CA/0512474 http://dx.doi.org/10.1016/j.jfa.2009.12.007 http://arxiv.org/abs/0907.2447 http://arxiv.org/abs/1402.5793 http://dx.doi.org/10.1090/S0002-9947-99-02000-0 http://arxiv.org/abs/q-alg/9604004 http://dx.doi.org/10.2307/1427597 http://dx.doi.org/10.1007/s10959-008-0186-2 http://arxiv.org/abs/math.CA/0603017 http://dx.doi.org/10.1007/s00605-012-0460-3 http://arxiv.org/abs/1201.3490 http://dx.doi.org/10.1007/BF02571436 1 Introduction 2 A Mehler–Heine formula 3 Random walks on the dual of a compact Grassmannian and on P+ 3.1 Dual hypergroup structures on P+ 3.2 Random walks on P+ 3.3 Bessel convolutions on C and Laguerre ensembles 3.4 A central limit theorem for random walks on P+ References