Searcharxiv⌕ Search

arXiv · 2611.12107

Geometry and Stochastic Analysis of Flow and Diffusion Models: A Survey

Abstract

Flow-based and diffusion-based models have become the principal means of learning and sampling high-dimensional probability distributions, from images and video to proteins, molecules and physical fields, and increasingly language. Each carries a simple law to the law of the data along a curve of probability measures, and behind the variety of their names lie three geometries, those of the state space on which the data live, of the space of probability measures through which the curve passes, and of the space of paths traced by the generating process. A single analytic idea runs through the subject. Every score identity used to train a diffusion model is an integration by parts, on the noise, on the data or on Wiener space, and Malliavin calculus extends this to nonlinear and infinite-dimensional diffusions. We develop these foundations with full definitions and follow them through normalising flows, flow matching and few-step maps, diffusion models and their convergence theory, conditioning, Schrödinger bridges and stochastic control, and their extension to manifolds, Lie groups, discrete and function spaces and jump processes. We close with statistical physics, renormalisation and, more speculatively, holography, in which the radial direction of anti-de Sitter space can play the role of flow time. A table of the main methods and a collection of open problems complete the survey.

Explore related subjects

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Ehsan Mirafzali, Razvan Marinescu. 2026-11-09. Geometry and Stochastic Analysis of Flow and Diffusion Models: A Survey. https://searcharxiv.com/abs/2611.12107

Citations retain the original authors and DOI and include this site’s record URL. If this discovery service contributed to your research, you can also cite Searcharxiv by Indiciae: BibTeX · RIS.

KEEP EXPLORING

Related papers

Approximation of the Lévy-driven stochastic heat equation on the sphere

The stochastic heat equation on the sphere driven by an additive square-integrable Lévy process is approximated by a spectral method in space and forward and backward Euler-Maruyama schemes in time. New sharp regularity results are proven for its solution. The spectral approximation is based on a truncation of the series expansion with respect to the spherical harmonic functions. For a given regularity of the initial condition and two different settings of regularity for the driving noise, strong convergence rates for the spectral approximation and for the Euler-Maruyama methods are proven. Moreover, weak rates of up to twice the strong rates are shown for Wiener noise and for Lévy noise with independent and up to a deterministic decay rate factor identically distributed components, while for general square-integrable Lévy noise the weak rate depends explicitly on the polynomial order of the test function. Numerical simulations confirm the theoretical results.

math.PR↗

A Multivariate Generalized Cox Model for Dependent and Simultaneous Default Times

We propose a framework for dependent default times where each name defaults when an increasing driver crosses an independent unit exponential threshold. As in Gueye and Jeanblanc (2022), the drivers may jump, so defaults may occur at stopping times of the reference filtration, including predictable ones, and a common jump may cause no, one or several simultaneous defaults. For general drivers, we derive the conditional joint survival law by a backward recursion; it is an explicit product of survival functions if and only if, for every set of names, the predictable reduction of the compensator of their first default time is deterministic. If, moreover, defaults avoid predictable stopping times, the law is a fatal shock law with hazards identified from the drivers; predictable default atoms can rule this out. Examples include Lévy subordinators and, with random reductions but explicit laws, Liu's model and shot-noise drivers with memory.

math.PR↗

Boundary Feller-Dynkin processes associated with Laguerre processes and Pickrell diffusions

In this paper, we construct a Feller-Dynkin boundary process by applying the method of intertwiners to the coherent family, introduced in our previous work, of Laguerre processes with a fixed parameter. The corresponding boundary process is computed explicitly, and it turns out to be a deterministic dynamical system. Since the Pickrell diffusions that leave the Pickrell measure invariant are coherent with respect to the same projective system, we also obtain the stochastic boundary process associated with the coherent family of Pickrell diffusions.

math.PR↗