Distributional Statistical Models: Weak Moments, Cumulants, and a Central Limit Theorem
R. Labouriau · 2026
Read Distributional Statistical Models: Weak Moments, Cumulants, and a Central Limit Theorem by R. Labouriau (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-20634.txt.
Introduction
Labouriau proposes a way to show a probability law by a tempered distribution, like a density or characteristic. Information comes from testing the distribution on test functions smoothed by a positive Schwartz kernel. That yields weak moments, weak characteristic functions, and weak cumulants. Main results include an algebra of weak cumulants and a weak moment problem. A weak central limit theorem gives a Gaussian limit. The weak first moment gives a steady estimator of location for Cauchy data, which lacks any classical moment-based estimator.
Catalog record
| CREATOR | R. Labouriau |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 74445 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.20634
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Distributional Statistical Models: Weak Moments, Cumulants, and a Central Limit Theorem R. Labouriau Abstract Many important statistical models fall outside classical moment-based methods due to the non-existence of moments or moment generating functions. We propose a generalised probabilistic framework in which the classical requirement of a density with respect to a dominating measure is replaced by a pair ( T , φ ) (T,\varphi) , where T ∈ 𝒮 ′ ( ℝ ) T\in\mathcal{S}^{\prime}(\mathbb{R}) is a tempered distribution and φ ∈ 𝒮 ( ℝ ) \varphi\in\mathcal{S}(\mathbb{R}) is a Schwartz kernel. Expectations are defined via the action of distributions on regularised test functions, yielding well-defined weak moments , weak characteristic functions , and weak cumulants of all orders. These extend classical quantities and retain key algebraic properties such as additivity under independence and natural affine transformation rules. Despite dispensing with classical densities, the framework recovers ordinary event probabilities: the weak expectation determines a kernel-weighted distribution function, and under the uniqueness conditions developed here, the classical distribution function is fully recovered. The main results are: (i) a systematic algebra of weak cumulants; (ii) a weak moment problem where existence of all moments holds unconditionally and uniqueness depends on the kernel, with uniqueness results under Gaussian kernels (via Hermite completeness, including a multivariate extension to ℝ d \mathbb{R}^{d} ), positive Schwartz kernels with square-integrable densities (via a Carleman-type criterion), and kernels with exponential decay (via Denjoy–Carleman quasi-analyticity); (iii) a weak central limit theorem formulated as convergence of weak characteristic functions to a Gaussian limit, covering cases where the classical theorem fails; and (iv) a distributional CLT showing that, in the density case, the kernel-weighted measures converge in distribution to the standard normal—thereby bridging weak transform convergence and classical convergence in distribution. The framework is illustrated with Student’s t t , stable, and hyperbolic distributions. As a statistical consequence, the weak first moment yields a consistent estimator of the location parameter in the Cauchy model, where no classical moment-based estimator exists. A full statistical treatment is given in a companion paper [ 6 ] . 1 Introduction Classical statistical theory is largely built upon probability measures that admit densities with respect to a reference measure. In this setting, expectations, moments, cumulants, and transform-based methods are defined via integration against such densities [ 1 , 16 ] . However, many important models—heavy-tailed distributions such as the Cauchy distribution, low-degree Student t distributions, and \alpha -stable laws—may fail to possess moments of sufficiently high order, thereby limiting the applicability of moment-based methods [ 3 , 14 ] . We propose a framework in which densities are replaced by pairs (T,\varphi) where T\in\mathcal{S}^{\prime}(\mathbb{R}) is a tempered distribution and \varphi\in\mathcal{S}(\mathbb{R}) is a Schwartz kernel. Expectations are defined via the pairing \mathbb{E}_{T,\varphi}[\psi]=\langle T,\psi\varphi\rangle , and the closure of \mathcal{S}(\mathbb{R}) under multiplication by polynomials immediately gives that weak moments of all orders exist for every pair (T,\varphi) . When the ordinary moments of the underlying distribution exist, they coincide with the corresponding weak moments up to the kernel factor; when they do not (as in the Cauchy case), the weak moments are still well defined. The classical approach to statistical inference rests on the existence of a probability density with respect to a dominating measure: one writes down a likelihood, derives moments and cumulants from the density, and builds estimators and tests on these quantities. When moments do not exist or the density is not available in closed form, the classical machinery breaks down. The framework proposed here represents a genuine change of paradigm: it replaces the density–dominating-measure pair with a distribution–kernel pair (T,\varphi) , and it replaces ordinary expectations with the distributional pairing \langle T,\psi\varphi\rangle . The crucial point is that this substitution does not sacrifice probabilistic content . In Section 2.1 we show that the weak expectation determines a kernel-weighted distribution function by approximating indicator functions with Schwartz functions (Proposition 2.9 ), and that the ordinary distribution function is recovered whenever the underlying density is identifiable from the weak representation (Corollary 2.10 ). The weak framework thus supports the same kinds of probabilistic statements—event probabilities, quantiles, confidence intervals—and even allows statistical inference, just as classical density-based methods do, while extending seamlessly to models where no classical moments exist or densities are not available. Moreover, in the classical density setting, the framework admits an explicit reconstruction interpretation: weak data determine a regularised version of the density, from which the original density can be recovered by stable inverse methods (see Appendix C ). The paper develops five blocks of results: (i) Algebra of weak cumulants (Sections 3 – 4 ): we show that weak characteristic functions factorise under independence, that weak cumulants are additive, and that affine transformation rules carry over from the classical theory. (ii) The weak moment problem (Section 6 ): existence of weak moments of all orders is automatic; uniqueness depends on the kernel. We establish three levels of uniqueness: (a) for any positive Schwartz kernel, among square-integrable densities via Carleman’s condition; (b) for kernels with exponential-type decay, in all of \mathcal{S}^{\prime}(\mathbb{R}) via the Denjoy–Carleman theorem; (c) for Gaussian kernels, via the completeness of Hermite functions in \mathcal{S}(\mathbb{R}) . The Gaussian uniqueness theorem is extended to \mathbb{R}^{d} using the multivariate Hermite basis. (iii) Weak central limit theorem (Section 7 ): we prove convergence of the weak characteristic function of normalised sums to a Gaussian limit, together with a quantitative (Berry–Esseen type) refinement. (iv) Distributional CLT (Section 7.5 ): in the density case, the weak CLT is shown to imply classical convergence in distribution of the kernel-weighted measures to the standard normal. This bridges the weak framework and classical probability, and connects to the Tikhonov reconstruction of Appendix C . (v) A statistical consequence (Section 8 ): a short illustration showing that even a single weak moment yields a consistent estimator for the location parameter of a Cauchy distribution, where no classical moment-based estimator exists. Section 2 introduces the framework, basic definitions, and the recovery of event probabilities from weak expectations. Section 3 develops independence and additivity. Section 4 studies weak transforms and cumulants. Section 5 presents illustrative examples. Section 6 is the centrepiece: the weak moment problem and its uniqueness theorems. Section 7 establishes the weak CLT, its quantitative refinement, and the distributional CLT for kernel-weighted measures. Section 8 gives the statistical illustration. Section 9 concludes. Appendix A presents a technical estimate used in the proof of Theorem 6.6 . Additional proofs omitted from the main text for brevity are collected in Appendix B . Appendix C develops Tikhonov reconstruction from weak data. Notation. Throughout the paper, weak objects carry a superscript \varphi denoting the kernel in play: {}^{(\varphi)}m_{n} for weak moments, {}^{(\varphi)}\phi(t) for the weak characteristic function, {}^{(\varphi)}K(t) for the weak cumulant generating function, and {}^{(\varphi)}\kappa_{n} for weak cumulants. Classical objects (when they exist) carry no superscript. 2 Preliminaries and generalised probability framework 2.1 Distribution–kernel pairs and weak expectation Let \mathcal{S}(\mathbb{R}^{d}) denote the Schwartz space of rapidly decreasing smooth functions, and \mathcal{S}^{\prime}(\mathbb{R}^{d}) its dual, the space of tempered distributions. We work primarily in the tempered setting, since it is stable under Fourier transform and accommodates the weak characteristic functions introduced below. The closure of \mathcal{S}(\mathbb{R}^{d}) under multiplication by polynomials—if \varphi\in\mathcal{S}(\mathbb{R}^{d}) and p is a