Free Segal–Bargmann Transform in Free Probability

• Free Segal–Bargmann Transform is a noncommutative operator extending the classical transform by encoding free unitary Brownian motion and trace polynomial calculus.
• It employs a two-parameter framework that unifies free stochastic analysis with holomorphic integral kernel representations and reproducing kernel Hilbert spaces.
• The transform bridges free probability theory with large‑N asymptotic analysis, elliptic Fock space constructions, and classical functional calculus.

The free Segal–Bargmann transform is a fundamental operator in free probability theory, providing a noncommutative analog of the classical Segal–Bargmann (or Segal–Bargmann–Hall) transform for the unitary group. In the free probability setting, this transform appears as the large-$N$ limit of its classical finite-dimensional counterpart on the unitary group $\mathbb{U}(N)$, acting naturally on spaces of Laurent polynomials with respect to free unitary Brownian motion. The construction admits a two-parameter generalization, $G_{s,t}$ or $\mathscr{G}_{s,t}$, interpolating between different regimes of free convolution and establishing deep connections with free stochastic analysis, elliptic Fock spaces, and conditional expectations in free probability (Ho, 2016, Driver et al., 2013).

1. Definition and Construction of the Free Segal–Bargmann Transform

Let $(A, \tau)$ be a $W^*$-probability space and $u_t$ a free unitary Brownian motion with law $\nu_t$ supported on the unit circle. Let $b_{s,t}$ denote the free multiplicative $(s, t)$-Brownian motion, freely independent from $\mathbb{U}(N)$0. The relevant Hilbert spaces are $\mathbb{U}(N)$1, defined as the GNS completion of Laurent polynomials in $\mathbb{U}(N)$2 with respect to $\mathbb{U}(N)$3, and $\mathbb{U}(N)$4, the completion under the standard noncommutative $\mathbb{U}(N)$5-norm.

The two-parameter free unitary Segal–Bargmann transform $\mathbb{U}(N)$6 acts as

$\mathbb{U}(N)$7

i.e., it is the conditional expectation of $\mathbb{U}(N)$8 given $\mathbb{U}(N)$9. Equivalently, for the trace-derivation operator $G_{s,t}$0 acting on Laurent polynomials,

$G_{s,t}$1

This operator is unitary onto a holomorphic subspace and encodes the action of the semigroup generated by $G_{s,t}$2 in a noncommutative setting (Ho, 2016, Driver et al., 2013).

2. Integral Kernel Formulation

The transform $G_{s,t}$3 admits a holomorphic integral kernel representation via a subordination map. Define $G_{s,t}$4 and $G_{s,t}$5. The subordination function is

$G_{s,t}$6

with the domain of holomorphy

$G_{s,t}$7

For $G_{s,t}$8 there is a unique probability measure $G_{s,t}$9 on the unit circle characterized by its moment generating function,

$\mathscr{G}_{s,t}$0

For $\mathscr{G}_{s,t}$1, $\mathscr{G}_{s,t}$2 is absolutely continuous with respect to $\mathscr{G}_{s,t}$3, and the density is explicit: $\mathscr{G}_{s,t}$4 The transform is then given by

$\mathscr{G}_{s,t}$5

This kernel formulation ensures the unitarity of $\mathscr{G}_{s,t}$6 onto a reproducing kernel Hilbert space $\mathscr{G}_{s,t}$7 (Ho, 2016).

3. Parameter Dependence, Special Cases, and Limiting Regimes

The parameters $\mathscr{G}_{s,t}$8 and $\mathscr{G}_{s,t}$9 govern the base measure and the "amount of transform," respectively:

• $(A, \tau)$0 determines the base measure $(A, \tau)$1 and its support; as $(A, \tau)$2 increases, $(A, \tau)$3 spreads over a larger subset of the circle.
• $(A, \tau)$4 interpolates from the identity ($(A, \tau)$5) to the one-parameter Biane transform ($(A, \tau)$6).

Special and limiting cases:

• For $(A, \tau)$7 (the one-parameter case), $(A, \tau)$8 is Biane’s free Hall transform.
• As $(A, \tau)$9, one recovers $W^*$0 on $W^*$1.
• In the classical limit $W^*$2 for fixed $W^*$3, the domain $W^*$4, with the kernel converging to the Cauchy kernel of the annulus.
• For $W^*$5 with $W^*$6, the domain collapses to the unit circle from inside (Ho, 2016, Driver et al., 2013).

4. Large-$W^*$7 Limit and Trace Polynomial Calculus

The free Segal–Bargmann transform arises as the large-$W^*$8 limit of the classical (two-parameter) Segal–Bargmann transform $W^*$9 on $u_t$0 with respect to heat kernel measure and its complexification. Consider the algebra of trace polynomials

$u_t$1

with $u_t$2 representing the evaluation variable and $u_t$3 encoding normalized traces. Finite-$u_t$4 infinitesimal generators of $u_t$5 reduce, in the large-$u_t$6 limit, to an explicit operator on Laurent polynomials. For each Laurent polynomial $u_t$7, there exist unique $u_t$8 with

$u_t$9

The limit operator is

$\nu_t$0

where $\nu_t$1 is a derivation on $\nu_t$2 and $\nu_t$3 denotes evaluation at moments of the free unitary Brownian motion $\nu_t$4 (Driver et al., 2013).

5. Elliptic Systems and Fock Space Realization

The free $\nu_t$5-Segal–Bargmann transform admits a Fock space construction via elliptic (circular) systems. Given a real Hilbert space $\nu_t$6 and two free semicircular systems $\nu_t$7 on $\nu_t$8, the elliptic system is

$\nu_t$9

with covariance $b_{s,t}$0 and cross-correlation $b_{s,t}$1. The $b_{s,t}$2-free SB transform $b_{s,t}$3 implements the correspondence

$b_{s,t}$4

and sends Wick polynomials and stochastic integrals appropriately. This construction realizes $b_{s,t}$5 via analytic continuation and second quantization (Ho, 2016).

6. Biane–Gross–Malliavin Identification and Functional Calculus

The two-parameter Biane–Gross–Malliavin theorem extends the identification between classical and free Segal–Bargmann transforms. Let $b_{s,t}$6 be the time-rescaled free unitary Brownian motion, and $b_{s,t}$7 the free multiplicative $b_{s,t}$8-Brownian motion. The main components are:

• Holomorphic functional calculus $b_{s,t}$9 defines a unitary from $(s, t)$0 onto $(s, t)$1.
• The map $(s, t)$2 is a unitary isomorphism $(s, t)$3.
• The commutative square realizes $(s, t)$4 via endpoint functional calculus and provides a precise identification between the free unitary and elliptic transforms (Ho, 2016).

7. Connections to Classical Theory and Structural Significance

The free Segal–Bargmann transform generalizes and extends the classical theory on compact Lie groups to free probability. In the limit $(s, t)$5, the transform recovers the Cauchy kernel on the annulus, parallel to the classical theory on the complexification of $(s, t)$6. For $(s, t)$7 and large $(s, t)$8, the domain approaches the unit circle, mirroring time-asymptotic behavior in the classical Hall transform. In the setting $(s, t)$9, the free transform degenerates to the identity, reflecting degeneration of the analytic continuation in the classical unitary group context.

This framework provides a robust link between stochastic processes in free probability, analysis on Fock spaces, and the asymptotic representation theory of Lie groups. The free Segal–Bargmann and free Hall transforms are recovered as explicit large-$\mathbb{U}(N)$00 limits and are characterized by their action on Biane polynomials, their norm convergence, and their connection to measure concentration phenomena (Ho, 2016, Driver et al., 2013).