Search

The well-posedness of a class of optimal control problems is analysed, where the state equation couples a nonlinear degenerate Fokker-Planck equation with a system of Ordinary Differential Equations (ODEs). Such problems naturally arise as mean-field limits of Stochastic Differential models for multipopulation dynamics, where a large number of agents (followers) is steered through parsimonious intervention on a selected class of leaders. The proposed approach combines stability estimates for measure solutions of nonlinear degenerate Fokker-Planck equations with a general framework of assumptions on the cost functional, ensuring compactness and lower semicontinuity properties. The Lie structure of the state equations allows one for considering non-Lipschitz nonlinearities, provided some suitable dissipativity assumptions are considered in addition to non-Euclidean H\"{o}lder and sublinearity conditions.

The logit transform is arguably the most widely-employed link function beyond linear settings. This transformation routinely appears in regression models for binary data and provides, either explicitly or implicitly, a core building-block within state-of-the-art methodologies for both classification and regression. Its widespread use, combined with the lack of analytical solutions for the optimization of general losses involving the logit transform, still motivates active research in computational statistics. Among the directions explored, a central one has focused on the design of tangent lower bounds for logistic log-likelihoods that can be tractably optimized, while providing a tight approximation of these log-likelihoods. Although progress along these lines has led to the development of effective minorize-maximize (MM) algorithms for point estimation and coordinate ascent variational inference schemes for approximate Bayesian inference under several logit models, the overarching focus in the literature has been on tangent quadratic minorizers. In fact, it is still unclear whether tangent lower bounds sharper than quadratic ones can be derived without undermining the tractability of the resulting minorizer. This article addresses such a challenging question through the design and study of a novel piece-wise quadratic lower bound that uniformly improves any tangent quadratic minorizer, including the sharpest ones, while admitting a direct interpretation in terms of the classical generalized lasso problem. As illustrated in a ridge logistic regression, this unique connection facilitates more effective implementations than those provided by available piece-wise bounds, while improving the convergence speed of quadratic ones.

Generalized linear models (GLMs) arguably represent the standard approach for statistical regression beyond the Gaussian likelihood scenario. When Bayesian formulations are employed, the general absence of a tractable posterior distribution has motivated the development of deterministic approximations, which are generally more scalable than sampling techniques. Among them, expectation propagation (EP) showed extreme accuracy, usually higher than many variational Bayes solutions. However, the higher computational cost of EP posed concerns about its practical feasibility, especially in high-dimensional settings. We address these concerns by deriving a novel efficient formulation of EP for GLMs, whose cost scales linearly in the number of covariates p. This reduces the state-of-the-art O(p^2 n) per-iteration computational cost of the EP routine for GLMs to O(p n min{p,n}), with n being the sample size. We also show that, for binary models and log-linear GLMs approximate predictive means can be obtained at no additional cost. To preserve efficient moment matching for count data, we propose employing a combination of log-normal Laplace transform approximations, avoiding numerical integration. These novel results open the possibility of employing EP in settings that were believed to be practically impossible. Improvements over state-of-the-art approaches are illustrated both for simulated and real data. The efficient EP implementation is available at https://github.com/niccoloanceschi/EPglm.

It is increasingly common in a wide variety of applied settings to collect data of multiple different types on the same set of samples. Our particular focus in this article is on studying relationships between such multiview features and responses. A motivating application arises in the context of precision medicine where multi-omics data are collected to correlate with clinical outcomes. It is of interest to infer dependence within and across views while combining multimodal information to improve the prediction of outcomes. The signal-to-noise ratio can vary substantially across views, motivating more nuanced statistical tools beyond standard late and early fusion. This challenge comes with the need to preserve interpretability, select features, and obtain accurate uncertainty quantification. We propose a joint additive factor regression model (JAFAR) with a structured additive design, accounting for shared and view-specific components. We ensure identifiability via a novel dependent cumulative shrinkage process (D-CUSP) prior. We provide an efficient implementation via a partially collapsed Gibbs sampler and extend our approach to allow flexible feature and outcome distributions. Prediction of time-to-labor onset from immunome, metabolome, and proteome data illustrates performance gains against state-of-the-art competitors. Our open-source software (R package) is available at https://github.com/niccoloanceschi/jafar.

We consider Kolmogorov-Fokker-Planck equations with unbounded drift terms which are only measurable in time and locally H\"older continuous in space. In particular, we extend the parametrix method to this setting and we prove existence and uniqueness of measure solutions to the associated Cauchy problem, as well as the equivalence with the corresponding stochastic formulation., Comment: 37 pages

We deal with the following wide class of kinetic equations, $$ \big[\partial_t \,+\, v\cdot\nabla_x \big] f \, = \, L_v f. $$ Above, the diffusion term $L_v$ is an integro-differential operator, whose nonnegative kernel is of fractional order $s\in(0,1)$ having merely measurable coefficients. Firstly, we obtain a general $L^\infty$-interpolation inequality with a natural nonlocal tail term in velocity, in turn giving local boundedness even for weak subsolutions $f$ without any sign assumption. This is a veritable novelty, being boundedness usually assumed apriori in such a setting. Then, provided that their nonlocal tail in velocity is $(2\!+\!\varepsilon)$-summable along the transport variables, we prove a general Strong Harnack inequality, which in the simpler case of globally nonnegative weak solutions $f$ reads as follows $$ \sup_{Q^-} f \ \leq \ c\,\inf_{Q^+} f \,+ \, c\,\| {\textrm {Tail}} (f)\|_{L^{2+\varepsilon}_{t,x}}\,, $$ where $Q^{\pm}$ are suitable slanted cylinders. This is the first strong Harnack inequality for kinetic integral equations under the aforementioned tail summability assumption, which is in fact naturally implied in literature, e.g., from the usual mass density boundedness (as for the Boltzmann equation without cut-off), and in clear accordance with the very recent surprising counterexample by Kassmann and Weidner; see arXiv:2405.05223. A new standalone result, a Besicovitch-type covering argument for very general kinetic geometries, independent on the involved equation is also needed, stated and proved., Comment: We provide a new formulation of the strong Harnack inequality for kinetic integral equations, in accordance with Kassmann-Weidner's counterexample arXiv:2405.05223, and with the assumptions in the previous related literature; see Remark 1.3. Also, we removed global positiveness both in the weak and the strong Harnack inequalities, so that a natural nonlocal term does appear in the estimates

We propose a systematic elementary approach based on trajectories to prove Poincar\'e inequalities for hypoelliptic equations with an arbitrary number of H\"ormander commutators, both in the local and in the non-local case. Our method generalises and simplifies the method introduced in Guerand-Mouhot (2022) in the local case with one commutator, and extended in Loher (2024) to the non-local case with one commutator. It draws inspiration from the paper Niebel-Zacher (2022), although we use different trajectories. We deduce Harnack inequalities and H\"older regularity along the line of the De Giorgi method. Our results recover those in Anceschi-Rebucci (2022) in the local case, and are new in the non-local case., Comment: 22 pages, 4 figures. Simplified the Lemma on the determinant of the Wronskian matrix in the construction of the trajectories (Lemma 7), and introduced the notion of solutions we work with

The smoothing distribution of dynamic probit models with Gaussian state dynamics was recently proved to belong to the unified skew-normal family. Although this is computationally tractable in small-to-moderate settings, it may become computationally impractical in higher dimensions. In this work, adapting a recent more general class of expectation propagation (EP) algorithms, we derive an efficient EP routine to perform inference for such a distribution. We show that the proposed approximation leads to accuracy gains over available approximate algorithms in a financial illustration.

Binary regression models represent a popular model-based approach for binary classification. In the Bayesian framework, computational challenges in the form of the posterior distribution motivate still-ongoing fruitful research. Here, we focus on the computation of predictive probabilities in Bayesian probit models via expectation propagation (EP). Leveraging more general results in recent literature, we show that such predictive probabilities admit a closed-form expression. Improvements over state-of-the-art approaches are shown in a simulation study.

Bayesian binary regression is a prosperous area of research due to the computational challenges encountered by currently available methods either for high-dimensional settings or large datasets, or both. In the present work, we focus on the expectation propagation (EP) approximation of the posterior distribution in Bayesian probit regression under a multivariate Gaussian prior distribution. Adapting more general derivations in Anceschi et al. (2023), we show how to leverage results on the extended multivariate skew-normal distribution to derive an efficient implementation of the EP routine having a per-iteration cost that scales linearly in the number of covariates. This makes EP computationally feasible also in challenging high-dimensional settings, as shown in a detailed simulation study.

After carrying out an overview on the non Euclidean geometrical setting suitable for the study of Kolmogorov operators with rough coefficients, we list some properties of the functional space $\mathcal{W}$, mirroring the classical $H^1$ theory for uniformly elliptic operators. Then we provide the reader with the proof of a new Sobolev embedding for functions in $\mathcal{W}$. Additionally, after reviewing recent results regarding weak regularity theory, we discuss some of their recent applications to real life problems arising both in Physics and in Economics. Finally, we conclude our analysis stating some recent results regarding the study of nonlinear nonlocal kinetic Kolmogorov-Fokker-Planck operators.

This work is devoted to the study of the obstacle problem associated to the Kolmogorov-Fokker-Planck operator with rough coefficients through a variational approach. In particular, after the introduction of a proper anisotropic Sobolev space and related properties, we prove the existence and uniqueness of a weak solution for the obstacle problem by adapting a classical perturbation argument to the convex functional associated to the case of our interest. Finally, we conclude this work by providing a one-sided associated variational inequality, alongside with an overview on related open problems.

We investigate local properties of weak solutions to nonlocal and nonlinear kinetic equations whose prototype is given by $$ \partial_t u +v\cdot\nabla_x u +(-\Delta_v)^s_p u = f(u). $$ We consider equations whose diffusion part is a (possibly degenerate) integro-differential operator of differentiability order $s \in (0,1)$ and summability exponent $p\in (1,\infty)$. Amongst other results, we provide an explicit local boundedness estimate by combining together a suitable Sobolev embedding theorem and a fractional Caccioppoli-type inequality with tail. For this, we introduce in the kinetic framework a new definition of nonlocal tail of a function and of its related tail spaces, also by establishing some useful estimates for the tail of weak solutions. Armed with the aforementioned results we give a precise control of the long-range interactions arising from the nonlocal behaviour of the involved diffusion operator.

We consider a class of second order degenerate kinetic operators $\mathscr{L}$ in the framework of special relativity. We first describe $\mathscr{L}$ as an H\"ormander operator which is invariant with respect to Lorentz transformations. Then we prove a Lorentz-invariant Harnack type inequality, and we derive accurate asymptotic lower bounds for positive solutions to $\mathscr{L} f = 0$. As a consequence we obtain a lower bound for the density of the relativistic stochastic process associated to $\mathscr{L}$., Comment: 23 pages

Soil texture is key information in agriculture for improving soil knowledge and crop performance, so the accurate mapping of this crucial feature is imperative for rationally planning cultivations and for targeting interventions. We studied the relationship between radioelements and soil texture in the Mezzano Lowland (Italy), a 189 $km^2$ agricultural plain investigated through a ded-icated airborne gamma-ray spectroscopy survey. The K and Th abundances were used to retrieve the clay and sand content by means of a multi-approach method. Linear (simple and multiple) and non-linear (machine learning algorithms with deep neural networks) predictive models were trained and tested adopting a 1:50,000 scale soil texture map. The comparison of these approaches highlighted that the non-linear model introduces significant improvements in the prediction of soil texture fractions. The predicted maps of the clay and of the sand content were compared with the regional soil maps. Although the macro-structures were equally present, the airborne gam-ma-ray data permits us shedding light on finer features. Map areas with higher clay content were coincident with paleo-channels crossing the Mezzano Lowland in Etruscan and Roman periods, confirmed by the hydrographic setting of historical maps and by the geo-morphological features of the study area., Comment: 18 pages, 9 figures, 3 tables, correspondence to Andrea Maino via email at maino@fe.infn.it

A broad class of models that routinely appear in several fields can be expressed as partially or fully discretized Gaussian linear regressions. Besides including basic Gaussian response settings, this class also encompasses probit, multinomial probit and tobit regression, among others, thereby yielding to one of the most widely-implemented families of models in applications. The relevance of such representations has stimulated decades of research in the Bayesian field, mostly motivated by the fact that, unlike for Gaussian linear regression, the posterior distribution induced by such models does not seem to belong to a known class, under the commonly-assumed Gaussian priors for the coefficients. This has motivated several solutions for posterior inference relying on sampling-based strategies or on deterministic approximations that, however, still experience computational and accuracy issues, especially in high dimensions. The scope of this article is to review, unify and extend recent advances in Bayesian inference and computation for this class of models. To address such a goal, we prove that the likelihoods induced by these formulations share a common analytical structure that implies conjugacy with a broad class of distributions, namely the unified skew-normals (SUN), that generalize Gaussians to skewed contexts. This result unifies and extends recent conjugacy properties for specific models within the class analyzed, and opens avenues for improved posterior inference, under a broader class of formulations and priors, via novel closed-form expressions, i.i.d. samplers from the exact SUN posteriors, and more accurate and scalable approximations from VB and EP. Such advantages are illustrated in simulations and are expected to facilitate the routine-use of these core Bayesian models, while providing a novel framework to study theoretical properties and develop future extensions.

Multinomial probit (mnp) models are fundamental and widely-applied regression models for categorical data. Fasano and Durante (2022) proved that the class of unified skew-normal distributions is conjugate to several mnp sampling models. This allows to develop Monte Carlo samplers and accurate variational methods to perform Bayesian inference. In this paper, we adapt the abovementioned results for a popular special case: the discrete-choice mnp model under zero mean and independent Gaussian priors. This allows to obtain simplified expressions for the parameters of the posterior distribution and an alternative derivation for the variational algorithm that gives a novel understanding of the fundamental results in Fasano and Durante (2022) as well as computational advantages in our special settings.

The aim of this work is to prove the existence of a fundamental solution associated to the Kolmogorov equation L u = f with measurable coefficients in the dilation invariant case. Moreover, we prove Gaussian upper and lower bounds for it, and other related properties., Comment: 20 pages

We establish spatial a priori estimates for the solution u to a class of dilation invariant Kolmogorov equation, where u is assumed to only have a certain amount of regularity in the diffusion's directions. The result is that u is also regular with respect to the remaining directions. The approach we propose is based on the commutators identities and allows us to obtain a Sobolev exponent that does not depend on the integrability assumption of the right-hand side. Lastly, we provide an alternative proof to that of Theorem 1.5 of [9] for the optimal spatial regularity., Comment: 10 pages

A wide class of Bayesian models involve unidentifiable random matrices that display rotational ambiguity, with the Gaussian factor model being a typical example. A rich variety of Markov chain Monte Carlo (MCMC) algorithms have been proposed for sampling the parameters of these models. However, without identifiability constraints, reliable posterior summaries of the parameters cannot be obtained directly from the MCMC output. As an alternative, we propose a computationally efficient post-processing algorithm that allows inference on non-identifiable parameters. We first orthogonalize the posterior samples using Varimax and then tackle label and sign switching with a greedy matching algorithm. We compare the performance and computational complexity with other methods using a simulation study and chemical exposures data. The algorithm implementation is available in the infinitefactor R package on CRAN.

The aim of this work is to prove a Harnack inequality and the H\"older continuity for weak solutions to the Kolmogorov equation $\mathscr{L} u = f$ with measurable coefficients, integrable lower order terms and nonzero source term. We introduce a functional space $\mathcal{W}$, suitable for the study of weak solutions to $\mathscr{L}u = f$, that allows us to prove a weak Poincar\'e inequality. More precisely, our goal is to prove a weak Harnack inequality for non-negative super-solutions by considering their Log-transform and following S. N. Kruzkov (1963). Then this functional inequality is combined with a classical covering argument (Ink-Spots Theorem) that we extend for the fist time to the case of ultraparabolic equations., Comment: 36 pages

We investigate the Cauchy problem and the diffusion asymptotics for a spatially inhomogeneous kinetic model associated to a nonlinear Fokker-Planck operator. We derive the global well-posedness result with instantaneous smoothness effect, when the initial data lies below a Maxwellian. The proof relies on the hypoelliptic analog of classical parabolic theory, as well as a positivity-spreading result based on the Harnack inequality and barrier function methods. Moreover, the scaled equation leads to the fast diffusion flow under the low field limit. The relative phi-entropy method enables us to see the connection between the overdamped dynamics of the nonlinearly coupled kinetic model and the correlated fast diffusion. The global in time quantitative diffusion asymptotics is then derived by combining entropic hypocoercivity, relative phi-entropy and barrier function methods., Comment: 39 pages

We prove the existence and uniqueness of the fundamental solution for Kolmogorov operators associated to some stochastic processes, that arise in the Black & Scholes setting for the pricing problem relevant to path dependent options. We improve previous results in that we provide a closed form expression for the solution of the Cauchy problem under weak regularity assumptions on the coefficients of the differential operator. Our method is based on a limiting procedure, whose convergence relies on some barrier arguments and uniform a priori estimates recently discovered., Comment: 34 pages

We present a survey on the regularity theory for classic solutions to subelliptic degenerate Kolmogorov equations. In the last part of this note we present a detailed proof of a Harnack inequality and a strong maximum principle., Comment: 32 pages, 3 figures

We prove the local boundedness of the solutions to degenerate second order partial differential equations of Kolmogorov type with measurable coefficients in divergence form, under minimal integrability assumption on the lower order coefficients., Comment: 22 pages

The evolutionary and ecological processes behind the origin of species are among the most fundamental problems in biology. In fact, many theoretical hypothesis on different type of speciation have been proposed. In particular, models of sympatric speciation leading to the formation of new species without geographical isolation, are based on the niche hypothesis: the diversification of the population is induced by the competition for a limited set of the available resources. On the other hand, neutral models of evolution have shown that stochastic forces are sufficient to generate coexistence of different species. In this work, we bring this dichotomy to the context of species formation, and we study how neutral and niche forces contribute to sympatric speciation in a model ecosystem. In particular, we study the evolution of a population of individuals with asexual reproduction whose inherited characters or phenotypes are specified by both niche-based and neutral traits. We analyse the stationary state of the dynamics, and study the distribution of individuals in the whole space of possible phenotypes. We show, both by numerical simulations and analytics, that there is a non-trivial coupling between neutral and niche forces induced by stochastic effects in the evolution of the population that allows the formation of clusters (i.e., species) in the phenotypic space. Our framework can be generalised also to sexual reproduction or other type of population dynamics., Comment: 8 pages, 5 figures

We consider weak solutions of degenerate second order partial differential equations of Kolmogorov-Fokker-Planck type with measurable coefficients in divergence form. We give a geometric statement of the Harnack inequality recently proven by Golse, Imbert, Mouhot and Vasseur. As a corollary we obtain a strong maximum principle., Comment: 16 pages, 3 figures