Spectral distribution of the free Jacobi process

Free probability and free operator algebras (46L54) Linear symmetric and selfadjoint operators (unbounded) (47B25) Stochastic ordinary differential equations (aspects of stochastic analysis) (60H10)

Abstract: In this paper, we are interested in the free Jacobi process starting at the unit of the compressed probability space where it takes values and associated with the parameter values

l a m b d a = 1, h e t a = 1 / 2

. Firstly, we derive a time-dependent recurrence equation for the moments of the process (valid for any starting point and all parameter values). Secondly, we transform this equation to a nonlinear partial differential one for the moment generating function that we solve when

l a m b d a = 1, h e t a = 1 / 2

. The obtained solution together with tricky computations lead to an explicit expression of the moments which shows that the free Jacobi process is distributed at any time

t

as

(1 / 4) (2 + Y_{2 t} + Y_{2 t}^{s t a r})

where

Y

is a free unitary Brownian motion. This expression is recovered relying on enumeration techniques after proving that if

a

is a symmetric Bernoulli random variable which is free from

Y, Y^{s t a r}

, then the distributions of

Y_{2 t}

and that of

a Y_{t} a Y_{t}^{s t a r}

coincide. We close the exposition by investigating the spectral distribution associated with general sets of parameter values.

Review text

The authors characterize the spectral distribution of the free Jacobi process with parameter values \(\lambda = 1/2\) and \(\theta=1\). The case where \(0<\lambda<2\) and \(\theta = 1/2\) is to be discussed in a forthcoming paper. The starting point is the description of the free Jacobi process by the following free SDE NEWLINE\[NEWLINE dJ_t = \sqrt{\lambda\theta}\sqrt{J_t} \,dW_t\sqrt{P-J_t} + \sqrt{\lambda\theta}\sqrt{J_t} \,dW_t^*\sqrt{P-J_t} + (\theta P - J_t)\,dt,\quad t\in [0,T], NEWLINE\]NEWLINE where \(T\) is the infimum of all \(t\)'s such that \(J_t\) or \(P-J_t\) fails to be injective, \(W\) is a \(P\mathcal{A}P\)-complex free Brownian motion on the \(W^*\)-probability space \((\mathcal A, \tau)\) and \(P,Q\) are projections such that \(\tau(P)=\lambda\theta\), \(\tau(Q)=\theta\), \(PQ=QP=P\) if \(P\leq Q\) and \(=Q\) if \(P\geq Q\). This allows a recurrence relation of the moments \(m_n(t)\) of the free random variable \(J_t\) which can be explicitly solved using the generating function. In fact, NEWLINE\[NEWLINE m_n(t) = 2^{-2n}\binom{2n}{n} + 2^{-2n+1} \sum_{k=1}^n \binom{2n}{n-k} \frac 1k L_{k-1}^1(2kt) e^{-kt} NEWLINE\]NEWLINE with the \(n\)th Laguerre polynomial \(L_n^1\) of index \(1\). The main result of the paper states that the spectral distribution \(\mu_t\) of \(J_t\) in the compressed space fits the distribution of the free random variable \(1/4(Y_{2t}+Y_{2t}^*+2\mathbf{1})\) in \((\mathcal{A},\tau)\) (\(\mathbf 1\) is the unit and \(Y\) is the free unitary Brownian motion in the von Neumann algebra \(\mathcal A\)). Moreover, \(\mu_t\) is absolutely continuous with respect to the Lebesgue measure on \(\mathbb R\) and its support fills \((0,1)\) at time \(t=2\).

Recommendations

Cited in

(23)

This page was built for publication: Spectral distribution of the free Jacobi process

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2841859)