---
title: 'Whittaker Measure: Integrable Probability'
url: https://www.emergentmind.com/topics/whittaker-measure
type: topic
---

# Whittaker Measure: Integrable Probability

The Whittaker measure refers to a class of probability measures intimately connected to symmetric functions, integrable probability, representation theory, and random polymer models. It arises as the non-discrete analogue of the Schur measure in various contexts: as a probability distribution on partitions or arrays derived from Whittaker/Givental-type functions, as a canonical measure on geometric crystals for complex reductive groups, and, in its $q$-deformations, as the underpinning measure for the integrable structure of random growth processes and interacting particle systems.

## 1. Scalar and $q$-Whittaker Measures: Symmetric Function and Probability Theory

The scalar Whittaker measure, for the group $\mathrm{GL}_N(\mathbb{R})$, is defined via the product of two class-one Whittaker functions—$\psi_\theta(y)$ and $\psi_\sigma(y)$—and the Haar measure in logarithmic coordinates, incorporating spectral parameters $\theta$ and $\sigma$. Explicitly, for $y=(y_1,\ldots,y_N)\in(0,\infty)^N$,
\[
\mathbf{WM}_N^{\mathrm{Whittaker}}(dy;\theta,\sigma) = \frac{1}{Z_N(\theta,\sigma)}\; \psi_\theta(y)\,\psi_\sigma(y)\;\prod_{i=1}^N\frac{d y_i}{y_i},
\]
where the normalization constant $Z_N(\theta,\sigma)$ is given by the Bump–Stade integral:
\[
Z_N(\theta,\sigma) = \prod_{i,j=1}^N\Gamma(\theta_i+\sigma_j) .
\]
Here, $\psi_\lambda(y)$ is most concretely realized by integrals over triangular (Gelfand–Tsetlin) arrays, as in the Givental formula, and the measure generalizes the classical Schur measure to the continuous, “positive-temperature” setting. The $q$-Whittaker measure is a deformation using $q$-Whittaker polynomials (Macdonald polynomials at $t=0$), taking the form
\[
M_{qW}\{\mu\} = \frac{P_\mu(a) Q_\mu(b)}{Z_{qW}},
\]
for partitions $\mu$, $q$-Whittaker symmetric polynomials $P_\mu(a)$, $Q_\mu(b)$, and $Z_{qW}$ the $q$-deformed Cauchy-type normalization:
\[
Z_{qW} = \prod_{i=1}^N\prod_{j=1}^M \frac{1}{(a_i b_j;q)_\infty} .
\]
The $q$-Whittaker measure emerges as a degeneration of Macdonald processes and underlies solvable models such as $q$-TASEP, the Higher Spin Six Vertex Model, and $q$-Hahn processes [2106.11913, 1901.08381, 1110.3489].

## 2. Whittaker Measures and Geometric RSK: Representation-Theoretic and Combinatorial Foundations

A foundational connection arises from the geometric Robinson–Schensted–Knuth (gRSK) correspondence, which maps arrays of positive weights to (tropical) Gelfand–Tsetlin patterns. When the weights are inverse-Gamma distributed, pushing forward the product measure under gRSK yields a probability law on the “shape” vector (i.e., the bottom row) whose density is the product of two Whittaker functions divided by normalization constants [1110.3489, 1210.5126]. The volume-preserving property of the gRSK mapping is critical: it enables rewriting the law of the shapes as
\[
\mathbb{P}\{Z\in dz\} = Z(\alpha,\beta;s)^{-1} V_{\alpha;s}(z)\,V_\beta(z)\,\prod_{i=1}^p \frac{dz_i}{z_i},
\]
with $V_{\alpha;s}(z)=e^{-s/z_p}\Psi_p^{\alpha}(z)$ and the normalization a product of Gamma functions. Analogue identities—Bump–Stade and generalized Cauchy–Littlewood integrals—ensure normalization and furnish Mellin–Barnes-type tools for moments and Laplace transforms [1210.5126].

## 3. Geometric and Lie-Theoretic Generalizations: Whittaker Measure on Geometric Crystals

In the setting of complex semisimple Lie groups, the Whittaker measure generalizes to geometric crystals. The geometric crystal $(\lambda)$ of highest weight $\lambda\in\mathfrak{a}$ is endowed with a canonical measure, which in Lusztig coordinates $(t_1,\ldots,t_m)\in(\mathbb{R}_{>0})^m$ has “toric” reference form $\prod_j dt_j / t_j$. This is then twisted by exponential weights involving the crystal weight map $\gamma(x)$ and the Landau–Ginzburg superpotential $f_B(x)$:
\[
\omega(dx) = e^{\langle \mu, \gamma(x) \rangle - f_B(x)}\, \prod_{j=1}^m \frac{dt_j}{t_j}.
\]
The Laplace transform of the pushforward by the weight map yields the Archimedean Whittaker function,
\[
\Psi_\mu(\lambda) = \int_{(\lambda)} e^{\langle \mu, \gamma(x) \rangle - f_B(x)}\, \omega_{(\lambda)}(dx),
\]
satisfying quantum Toda eigenfunction equations and providing a unifying representation-theoretic, analytic, and probabilistic object [1504.07321, 1302.0902].

## 4. Matrix Whittaker Measures and Noncommutative Extensions

The matrix Whittaker measure provides a noncommutative extension: it is defined on tuples of real, symmetric positive-definite $d\times d$ matrices $z=(z_1,\ldots,z_N)\in P_d^N$, using integrals over arrays of matrices,
\[
W^{N,n}_{\lambda,\theta}(dz) = \Big[ \prod_{\ell=1}^n \prod_{i=1}^N \Gamma_d(\lambda_\ell+\theta_i)^{-1} \Big]\, \psi^{N,n}_{\lambda;I_d}(z)\, \psi^N_\theta(z)\, \mu^{\otimes N}(dz),
\]
where $\Gamma_d$ is the multivariate Gamma function, and $\psi^{N,n}$, $\psi^N$ are matrix-valued Whittaker functions obtained from recursive integrals over triangular matrix arrays. For $d=1$, this construction recovers the classical scalar Whittaker measure, establishing continuity between scalar and matrix-valued frameworks [2203.14868]. The measure arises as the fixed-time law for the bottom edge in certain random matrix-valued Markovian processes driven by inverse Wishart random variables, and under Laplace approximations concentrates on energy-minimizing configurations solvable by saddle point equations.

## 5. Integrable Probabilistic Models, Scaling Limits, and Asymptotics

Whittaker measures occupy a central role in integrable probability, as they encode the law of partition functions in log-gamma and directed polymer models, stochastic vertex models, and their $q$-deformations. Their Laplace transforms and one-point distributions admit explicit contour or Fredholm determinant formulas, facilitating rigorous analysis of fluctuations and Tracy–Widom-type asymptotics. A central result is that, under suitable scaling, the one-point fluctuations of the height function in models such as the Higher Spin Six Vertex Model or $q$-TASEP converge to Baik–Rains distributions with $1/3$ exponent, as established in [1901.08381]. The Whittaker measure forms the scaling limit (as $q\to1$) of $q$-Whittaker measures and lies at the core of asymptotic analysis in random interface growth, non-intersecting polymers, and last passage percolation [1909.03219].

## 6. Degenerations, Limiting Regimes, and Representation-Theoretic Connections

Degenerations of the Whittaker measure relate it to uniform measures on polytopes (e.g., string polytopes under tropicalization), and to classical objects in representation theory such as Harish-Chandra characters. For example, as $q\to 1$, $q$-Whittaker polynomials become rational Whittaker functions, and the $q$-Whittaker measure converges to the classical Whittaker measure. Within the geometric crystal framework, the canonical measure deforms continuously to the uniform measure on the string polytope—reflecting a crystal-to-tropical transition mirroring the degeneration of Markov processes to Brownian motion conditioned to remain in a Weyl chamber. Representation-theoretically, the Whittaker measure encapsulates the spectral theory of the quantum Toda lattice and encodes tensor product decompositions via crystal combinatorics [1504.07321, 1302.0902].

## 7. Summary Table: Key Variants of the Whittaker Measure

| Setting                    | Measure Definition           | Probabilistic Model          |
|----------------------------|-----------------------------|-----------------------------|
| Scalar (GL$_N$)            | Product of Whittaker functions times Haar measure; Bump–Stade norm. | Directed polymer with inverse-Gamma weights; geometric RSK. |
| $q$-Whittaker (Macdonald)  | Product of $q$-Whittaker polynomials, $q$-Pochhammer normalization. | $q$-TASEP, Higher Spin Six Vertex, $q$-Hahn processes, partitions. |
| Geometric crystal / Lie type | Integration on geometric crystal with toric measure, spectral twisting, superpotential. | Brownian motion in Cartan with Doob transform; geometric Pitman map; string polytopes. |
| Matrix-valued              | Higher-dimensional integration over positive-definite matrices; matrix Whittaker functions. | Matrix log-gamma polymer, inverse Wishart arrays. |

The Whittaker measure thus provides a central, unifying structure at the interface of integrable systems, symmetric functions, stochastic processes, and algebraic geometry, encoding in its variants a broad class of exactly solvable models, representation-theoretic transformations, and scaling limits underpinning modern integrable probability [1110.3489, 1210.5126, 1504.07321, 2106.11913, 1302.0902, 2203.14868, 1901.08381, 1909.03219].

Source: https://www.emergentmind.com/topics/whittaker-measure