Finite Free Information Theory
- Finite Free Information Theory is a framework that replaces asymptotic free-information constructs with exact finite structures such as polynomial zeros, matrix microstates, and rectangular transforms.
- It derives precise formulations of entropy, Fisher information, and additive inequalities by transitioning from limiting laws to finite-dimensional models.
- The theory unifies diverse approaches—from real-rooted polynomial analysis to microstate geometry and singular-value techniques—addressing challenges in noncommutative stochastic and quantum systems.
to=exa_search code 天天爱彩票怎么json {"query":"site:arxiv.org \"Finite Free Information Theory\" arXiv 2026 (Garza-Vargas et al., 17 Feb 2026, Jekel, 14 Apr 2026, Hashemi, 13 Apr 2026, Kearney, 3 May 2026)", "numResults": 5}:-------------</analysis to=exa_search code 天天爱彩票提现json {"query":"site:arxiv.org arXiv (Jekel, 14 Apr 2026) free information geometry model theory noncommutative stochastic processes", "numResults": 5}quotelevanalysis to=exa_search code 彩神争霸输钱json {"query":"site:arxiv.org arXiv (Garza-Vargas et al., 17 Feb 2026) \"Finite Free Information Inequalities\"", "numResults": 5} Finite Free Information Theory is not a single standardized formalism in the current literature. The term is used for several finite-dimensional or finite-horizon analogues of free-probabilistic, information-geometric, and correlation-theoretic constructions. In one major line, it studies entropy, Fisher information, and convolution directly on zeros of real-rooted polynomials; in another, it develops a microstate-based multivariate free information geometry for noncommutative stochastic processes; in a third, it constructs a rectangular calculus on singular values of matrix polynomials. Related uses of the phrase also occur in finite-temperature free scalar quantum field theory and in finite-horizon Maximum-Caliber formulations of information. This suggests that “Finite Free Information Theory” presently functions as an umbrella designation rather than a uniquely fixed doctrine (Garza-Vargas et al., 17 Feb 2026, Jekel, 14 Apr 2026, Gribinski, 2022, Katsinis et al., 2019, Kearney, 3 May 2026).
1. Scope, meanings, and recurring structures
Across its current uses, the term consistently denotes a passage from asymptotic or continuum information notions to explicitly finite objects: finite degree , finite matrix size, finite time horizon, finite temperature, or finite state space. The common pattern is the replacement of limiting laws by exact finite structures together with transforms, entropy-like functionals, Fisher-information–like quantities, or transport metrics that survive at finite scale.
| Formulation | Finite object | Representative result |
|---|---|---|
| Real-rooted polynomial theory | zeros of monic real-rooted degree- polynomials | finite free Stam inequality and entropy power inequality |
| Chronological microstate geometry | matrix microstates tested by chronological formulas | geodesic concavity of $\chi_{\chron}^{U}$ and EVI for heat flow |
| Rectangular finite free probability | singular-value polynomials for matrices | finite rectangular -transform linearizing additive convolution |
| Free Fisher regularity | polynomial evaluations under finite | Hölder CDFs, finite logarithmic energy, finite |
| Finite-temperature free fields | mutual information across a spatial bipartition | area law with finite classical high- remnant |
| MaxCal finite-horizon models | path ensembles over finite horizons | information as KL deviation from the constrained MaxCal ensemble |
The literature also indicates that the adjective “finite” is context dependent. In the polynomial and singular-value programs it refers to finite algebraic degree; in chronological entropy it refers to finite- matrix microstates and ultralimit constructions; in free scalar field theory it refers to finite temperature; and in the MaxCal framework it refers to a finite horizon and finite state spaces (Garza-Vargas et al., 17 Feb 2026, Jekel, 14 Apr 2026, Gribinski, 2022, Katsinis et al., 2019, Kearney, 3 May 2026).
2. Real-rooted polynomials, zeros, and finite free information inequalities
A central formulation of Finite Free Information Theory works on monic real-rooted polynomials
viewed through their root vector . For distinct roots, the score vector is
the finite free Fisher information is
and the finite free entropy is
Equivalently, is the normalized logarithmic energy of the zeros and , so the formalism has an explicit Coulomb-gas interpretation (Garza-Vargas et al., 17 Feb 2026).
The basic finite free additive operation is Walsh’s finite free convolution. If and , then
It preserves real-rootedness and has the symmetric-group average representation
Differentiation and finite free convolution are the two primary real-rootedness–preserving operators in the theory. The reverse heat flow
connects them to a Hermite benchmark, with the variance-$1$ monic Hermite polynomial (Garza-Vargas et al., 17 Feb 2026).
The central information inequalities are exact finite analogues of classical and free inequalities. The finite free Stam inequality states
The finite free entropy power inequality states
There is also monotonicity of finite free Fisher information under variance-normalized differentiation and monotonicity of finite free entropy under variance-normalized differentiation, with Hermite polynomials as the sharp equality cases. In the large-degree limit, these results recover corresponding inequalities in free probability (Garza-Vargas et al., 17 Feb 2026).
The proofs use a new link between score vectors and Jacobians of root maps. If maps to the roots of , then
A similar identity holds for the derivative root map . Together with double stochasticity of the Jacobian blocks and convexity results for hyperbolic polynomials, these identities play the role of a finite Blachman-type mechanism and yield the contraction estimates behind Stam and Fisher monotonicity (Garza-Vargas et al., 17 Feb 2026).
A later extension studies -generalizations under finite free additive convolution. For a root vector with distinct entries,
and the -Stam deficit is
At , FlowBoost numerically recovers the Hermite pair as the equality case and reveals a spectral structure for the linearized convolution map at the Hermite diagonal. Conditional on the conjecture that the singular values of the doubly stochastic coupling matrix on the mean-zero subspace are , independent of , the work derives a sharp local stability constant and an -uniform finite free CLT convergence rate. For , the Hermite pair itself violates the proposed inequality; for , the numerically extremal configurations bifurcate into non-matching pairs with bimodal root structure, converging back to the Hermite diagonal as (Hashemi, 13 Apr 2026).
3. Chronological entropy, matrix microstates, and free information geometry
A second major program develops a finite-, microstate-based “Finite Free Information Theory” for noncommutative stochastic processes. Its core object is a new multivariate free entropy $\chi_{\chron}^{U}$, defined from matrix microstates tested by chronological formulas. These formulas belong to a filtered metric language with domains , algebra operations, a metric, trace components, and constants coding increments of a non-selfadjoint free Brownian motion compatible with the filtration. Chronological formulas are built from quantifier-free trace-polynomial expressions in resolvents and Brownian increments, then closed under continuous connectives, partial suprema and infima in chronological order, and the heat-shift (Jekel, 14 Apr 2026).
For , , a finite set of restricted chronological formulas, and , the finite- microstate space is
The Gaussian chronological entropy is
$\tilde{\chi}_{\chron}^{U}(x \mid y) := \sup_{\Phi, \varepsilon} \lim_{n \to U} \Big( -\frac{1}{n^2}\log \sigma^{(n)}(\Gamma^{(n)}(x \mid Y^{(n)} \rightsquigarrow y; \Phi, \varepsilon)) \Big),$
and the Lebesgue version is
$\chi_{\chron}^{U}(x \mid y) := m \log(2\pi) + \tfrac{1}{2}\|x\|_2^2 - \tilde{\chi}_{\chron}^{U}(x \mid y).$
The construction is based on finite- pointwise functionals and an ultrafiber quotient that turns a random matrix ultraproduct into a factor with filtration and free Brownian motion (Jekel, 14 Apr 2026).
The information-geometric content is expressed on the space of conditional chronological types
$\mathbb{S}_{\chron,m}(M/\!y) := \{\, \mathrm{tp}_{\chron}^M(x,y): x \in M_0^m \,\},$
equipped with the free Wasserstein distance
$d_W(\mu_0,\mu_1) := \inf\big\{ \|x_0 - x_1\|_2 : \mathrm{tp}_{\chron}(x_j,y)=\mu_j \big\}.$
Optimal couplings exist, and there is a chronological Monge–Kantorovich duality with convex chronologically definable predicates of the form , where . If is an optimal coupling of and , then
$t\mapsto \chi_{\chron}^{U}(\mu_t) \quad\text{is concave on }[0,1].$
Thus the new entropy is displacement-concave along free Wasserstein geodesics (Jekel, 14 Apr 2026).
The heat semigroup is defined on chronologically definable predicates by
Its evolution satisfies the metric EVI inequality
$\frac{1}{2}\Big[ d_W(\nu_t,\sigma)^2 - d_W(\nu_s,\sigma)^2 \Big] \le (t-s)\Big[ \chi_{\chron}^{U}(\nu_t) - \chi_{\chron}^{U}(\sigma) \Big],\qquad 0\le s\le t,$
so heat evolution is the Wasserstein gradient flow of $-\chi_{\chron}^{U}$ in the metric sense. The corresponding minimizing-movement scheme is the free JKO iteration
$\mu_{k+1} \in \operatorname{argmin}_\mu \Big\{ \frac{1}{2\tau} d_W(\mu,\mu_k)^2 - \chi_{\chron}^{U}(\mu)\Big\}.$
This places the framework squarely inside an information-geometric and optimal-transport paradigm (Jekel, 14 Apr 2026).
The same theory also proves a true chain rule under iterated conditioning,
$\chi_{\chron}^{U}(x,y \mid w) = \chi_{\chron}^{U}(x \mid y,w) + \chi_{\chron}^{U}(y \mid w),$
invariance under definable closure of the conditioning variable, and a stochastic-control representation. For $\varphi\in F_{\chron,m+m'}^0$, the ultralimit pressure is
with adapted bounded controls . The Gaussian chronological entropy admits the variational formula
$\tilde{\chi}_{\chron}^{U}(x \mid y) = \sup_{\varphi \in \overline{F}_{\chron,m+m'}^0} \Big[ \mathcal{P}^{U}(\varphi)(y) - \varphi^{Q}(x,y) \Big].$
This gives the theory a finite-/control-theoretic bridge analogous, in the paper’s terms, to Borell/Boué–Dupuis and Schrödinger-bridge/Benamou–Brenier structures (Jekel, 14 Apr 2026).
4. Free Fisher information, distributional regularity, and operator-algebraic rigidity
A related strand studies the consequences of finite free Fisher information itself. In the tracial, non-microstates setting, if has finite and is a selfadjoint noncommutative polynomial of degree , then the cumulative distribution function is Hölder continuous with explicit exponent
If admits Lipschitz conjugate variables, the exponent improves to
For linear polynomials, the exponent $2/3$ under finite is optimal, while Lipschitz conjugate variables yield Lipschitz continuity and hence absolute continuity with bounded density. These regularity estimates imply finite logarithmic energy and therefore finite non-microstates free entropy for every selfadjoint nonconstant polynomial , partially resolving a conjecture of Charlesworth–Shlyakhtenko under the stronger assumption (Banna et al., 2018).
The same work supplies an explicit route from weak convergence to rates in Kolmogorov distance. If the limiting law has Hölder CDF with exponent and the Cauchy transforms satisfy suitable strip estimates, then
with a compact-support refinement
Applications include convergence in Kolmogorov distance for polynomial eigenvalue laws of Gibbs ensembles and explicit GUE rates such as
in the block-GUE semi-flat case, and
for selfadjoint polynomial GUE models (Banna et al., 2018).
In the non-tracial setting, finite free Fisher information for eigenvectors of a modular operator has much stronger structural consequences. If is generated by a finite selfadjoint set of eigenoperators of with finite free Fisher information, then
In particular, is a factor, and if is the closed subgroup generated by the eigenvalues of , then is a factor of type , (), or according as , , or . The same hypotheses imply that does not have property , and if is type with , then is full (Nelson, 2016).
These results rely on -modular derivations, conjugate variables entire for the modular group, Dirichlet forms on the centralizer, contraction resolvents, and non-tracial -homology estimates. In this line of work, finite free Fisher information is not merely a regularity parameter; it becomes a rigidity hypothesis controlling diffuseness, factoriality, and type classification (Nelson, 2016).
5. Rectangular finite free probability and singular-value information
Another formulation replaces eigenvalues by singular values of rectangular matrices. For with , the basic polynomial is
Equivalently, if is a univariate polynomial with nonnegative roots, its rectangular polynomial extension of order is . This encodes the squared singular values of directly at finite dimension (Gribinski, 2022).
The finite rectangular additive convolution is defined by averaging over bi-orthogonal rotations:
where , are Haar and . Setting yields a univariate real-rooted polynomial with nonnegative roots. The operation is bilinear, associative, and preserves real-rootedness. The theory also gives an explicit coefficient formula and a differential-operator expression for the convolution (Gribinski, 2022).
This program introduces finite analogues of classical rectangular free-probability transforms. A -point random variable is associated to a polynomial so that its power sums are fixed linear functionals of the coefficients, and the finite rectangular -transform is defined by
It linearizes finite rectangular additive convolution:
A modified finite -transform converges, as , to the classical rectangular free -transform, so the finite theory converges to asymptotic rectangular free probability (Gribinski, 2022).
The same framework produces explicit finite-dimensional LLN and CLT analogues for polynomials. After $1/N$ rescaling, iterated finite rectangular convolution converges to the zero polynomial, while after rescaling it converges, up to scaling, to a generalized Laguerre polynomial. In transform terms,
This identifies generalized Laguerre polynomials as the Gaussian analogues of the theory (Gribinski, 2022).
Because the roots of the convolved polynomial approximate squared singular values, the construction interfaces directly with spectral information functionals. For a nonnegative spectral law , the Shannon transform is
and for a positive matrix it is . In the finite rectangular framework, one computes , extracts its nonnegative roots , and evaluates
The paper explicitly notes that a finite multiplicative convolution or finite -transform is not introduced there, so multiplicative problems remain asymptotic in the current rectangular theory (Gribinski, 2022).
6. Physical and finite-horizon extensions
In free scalar quantum field theory at finite temperature, Finite Free Information Theory refers to the mutual information across a spatial bipartition. For a free real scalar in $3+1$ dimensions, the mutual information between a region and its complement satisfies an area law
with computable in an inverse-mass expansion. The high-temperature limit is finite and purely classical in origin:
For general coupled harmonic systems, the high- expansion has no term, and in the field-theory setting this cancellation is recovered once a uniform angular cutoff is imposed. The low-temperature correction is exponentially small, proportional to , while the limit matches the classical thermal calculation (Katsinis et al., 2019). A complementary numerical study of free scalar theory on concentric spherical shells established the same area-law behavior, reporting in the massless $3$-dimensional spherical setup
thereby exhibiting the finite classical remnant directly in the lattice computation (Katsinis et al., 2019).
A conceptually different finite-horizon usage defines information as deviation from a constrained Maximum-Caliber ensemble. For a one-step transition network, if is the MaxCal input marginal
then information is
For longer horizons, large-deviation theory yields
Within this framework, Integrated Information Theory repertoires are re-derived from constrained MaxEnt/MaxCal posteriors, and partition-based integration is defined by
The same paper states a duality to active-inference free-energy functionals and gives CLT- and LDP-based predictive-coding reductions in which becomes a Bayesian-surprise or transition-accuracy term (Kearney, 3 May 2026).
The open-problem landscape is correspondingly plural. In the chronological microstate program, open directions include ultrafilter independence of , a chronological free Fisher functional, and EVI for free Ornstein–Uhlenbeck semigroups (Jekel, 14 Apr 2026). In the polynomial program, the main unresolved issues are proof of the dyadic singular-value conjecture for , proof of the -Stam inequality for all , uniqueness of the Hermite equality case at , and characterization of the bifurcating extremizers (Hashemi, 13 Apr 2026). In the rectangular program, the missing finite multiplicative analogue remains explicit (Gribinski, 2022). In the non-tracial Fisher-information line, an important question is how much of the factoriality and fullness theory survives without an eigenoperator generating set (Nelson, 2016). In the MaxCal/IIT/FEP program, open questions include explicit non-equilibrium current formulations, non-Gaussian regimes beyond the CLT, and empirical validation relating , , and PCI (Kearney, 3 May 2026).
Taken together, these programs show that Finite Free Information Theory currently names a family of finite analogues rather than a single invariant package. What unifies them is a shared strategy: replace asymptotic free-information objects by exact finite structures—zeros, microstate sets, singular-value polynomials, path ensembles, or finite-temperature Gaussian subsystems—and recover entropy, Fisher information, transport, or mutual-information phenomena before passing to a limit.