Generative modeling of conditional probability distributions on the level-sets of collective variables

2025/12/19 by Akhyar, Fatima-Zahrae, Zhang, Wei, Stoltz, Gabriel +1
#68T07 #FOS: Computer and information sciences #FOS: Mathematics #Machine Learning (stat.ML) #Optimization and Control (math.OC)

paper · doi:10.48550/arxiv.2512.17374

Abstract

Given a probability distribution μ in ℝ^d represented by data, we study in this paper the generative modeling of its conditional probability distributions on the level-sets of a collective variable ξ: ℝ^d → ℝ^k, where 1 ≤ k

Citations

• Improving the Euclidean Diffusion Generation of Manifold Data by Mitigating Score Function Singularity
• Riemannian Denoising Diffusion Probabilistic Models
• On finding optimal collective variables for complex systems by minimizing the deviation between effective and full dynamics
• Generative Modeling on Manifolds Through Mixture of Riemannian Diffusion Processes
• Analyzing multimodal probability measures with autoencoders
• Flow Matching on General Geometries
• Flow Matching for Generative Modeling
• Building Normalizing Flows with Stochastic Interpolants
• Flow Straight and Fast: Learning to Generate and Transfer Data with Rectified Flow
• Riemannian Diffusion Models
• Matching Normalizing Flows and Probability Paths on Manifolds
• Riemannian Score-Based Generative Modelling
• Moser Flow: Divergence-based Generative Modeling on Manifolds
• Score-Based Generative Modeling through Stochastic Differential Equations
• Non-reversible sampling schemes on submanifolds
• Denoising Diffusion Probabilistic Models
• Riemannian Continuous Normalizing Flows
• Multiple projection MCMC algorithms on submanifolds
• PyTorch: An Imperative Style, High-Performance Deep Learning Library
• Generative Modeling by Estimating Gradients of the Data Distribution
• Hybrid Monte Carlo methods for sampling probability measures on submanifolds
• Neural Ordinary Differential Equations
• Pathwise estimates for effective dynamics: the case of nonlinear vectorial reaction coordinates
• Ergodic SDEs on submanifolds and related numerical sampling schemes
• Monte Carlo on manifolds: sampling densities and integrating functions
• Pathwise estimates for an effective dynamics
• Pathwise estimates for an effective dynamics
• Deep Unsupervised Learning using Nonequilibrium Thermodynamics
• MDAnalysis: A toolkit for the analysis of molecular dynamics simulations
• Approximate Bayesian Computational methods
• Langevin dynamics with constraints and computation of free energy differences
• GROMACS: Fast, flexible, and free

Related

• Introduction to probability and statistics: a computational framework of randomness
• Statistical guarantees for generative models without domination
• A Tale of Three Probabilistic Families: Discriminative, Descriptive and Generative Models
• Conditional constrained and unconstrained quantization for probability distributions
• On The Statistical Representation Properties Of The Perturb-Softmax And The Perturb-Argmax Probability Distributions