@article{SharmaZhang2021, author = {Sharma, Upanshu and Zhang, Wei}, title = {Non-reversible sampling schemes on submanifolds}, volume = {59}, journal = {SIAM Journal on Numerical Analysis}, number = {6}, arxiv = {http://arxiv.org/abs/2011.02835}, doi = {10.1137/20M1378752}, pages = {2989 -- 3031}, year = {2021}, abstract = {Calculating averages with respect to probability measures on submanifolds is often necessary in various application areas such as molecular dynamics, computational statistical mechanics and Bayesian statistics. In recent years, various numerical schemes have been proposed in the literature to study this problem based on appropriate reversible constrained stochastic dynamics. In this paper we present and analyse a non-reversible generalisation of the projection-based scheme developed by one of the authors [ESAIM: M2AN, 54 (2020), pp. 391-430]. This scheme consists of two steps - starting from a state on the submanifold, we first update the state using a non-reversible stochastic differential equation which takes the state away from the submanifold, and in the second step we project the state back onto the manifold using the long-time limit of a ordinary differential equation. We prove the consistency of this numerical scheme and provide quantitative error estimates for estimators based on finite-time running averages. Furthermore, we present theoretical analysis which shows that this scheme outperforms its reversible counterpart in terms of asymptotic variance. We demonstrate our findings on an illustrative test example.}, language = {en} } @article{LiuZhangSchuetteetal.2025, author = {Liu, Zichen and Zhang, Wei and Sch{\"u}tte, Christof and Li, Tiejun}, title = {Riemannian denoising diffusion probabilistic models}, journal = {Communications in Mathematical Sciences}, arxiv = {http://arxiv.org/abs/2505.04338}, year = {2025}, abstract = {We propose Riemannian Denoising Diffusion Probabilistic Models (RDDPMs) for learning distributions on submanifolds of Euclidean space that are level sets of functions, including most of the manifolds relevant to applications. Existing methods for generative modeling on manifolds rely on substantial geometric information such as geodesic curves or eigenfunctions of the Laplace-Beltrami operator and, as a result, they are limited to manifolds where such information is available. In contrast, our method, built on a projection scheme, can be applied to more general manifolds, as it only requires being able to evaluate the value and the first order derivatives of the function that defines the submanifold. We provide a theoretical analysis of our method in the continuous-time limit, which elucidates the connection between our RDDPMs and score-based generative models on manifolds. The capability of our method is demonstrated on datasets from previous studies and on new datasets sampled from two high-dimensional manifolds, i.e. SO(10) and the configuration space of molecular system alanine dipeptide with fixed dihedral angle.}, language = {en} } @article{LelievrePigeonStoltzetal.2024, author = {Leli{\`e}vre, Tony and Pigeon, Thomas and Stoltz, Gabriel and Zhang, Wei}, title = {Analyzing multimodal probability measures with autoencoders}, volume = {128}, journal = {J. Phys. Chem. B}, number = {11}, doi = {10.1021/acs.jpcb.3c07075}, pages = {2607 -- 2631}, year = {2024}, abstract = {Finding collective variables to describe some important coarse-grained information on physical systems, in particular metastable states, remains a key issue in molecular dynamics. Recently, machine learning techniques have been intensively used to complement and possibly bypass expert knowledge in order to construct collective variables. Our focus here is on neural network approaches based on autoencoders. We study some relevant mathematical properties of the loss function considered for training autoencoders, and provide physical interpretations based on conditional variances and minimum energy paths. We also consider various extensions in order to better describe physical systems, by incorporating more information on transition states at saddle points, and/or allowing for multiple decoders in order to describe several transition paths. Our results are illustrated on toy two dimensional systems and on alanine dipeptide.}, language = {en} } @article{ZhangSchuette2023, author = {Zhang, Wei and Sch{\"u}tte, Christof}, title = {Understanding recent deep-learning techniques for identifying collective variables of molecular dynamics}, volume = {23}, journal = {Proceedings in Applied Mathematics and Mechanics}, number = {4}, doi = {10.1002/pamm.202300189}, year = {2023}, abstract = {High-dimensional metastable molecular dynamics (MD) can often be characterised by a few features of the system, that is, collective variables (CVs). Thanks to the rapid advance in the area of machine learning and deep learning, various deep learning-based CV identification techniques have been developed in recent years, allowing accurate modelling and efficient simulation of complex molecular systems. In this paper, we look at two different categories of deep learning-based approaches for finding CVs, either by computing leading eigenfunctions of transfer operator associated to the underlying dynamics, or by learning an autoencoder via minimisation of reconstruction error. We present a concise overview of the mathematics behind these two approaches and conduct a comparative numerical study of these two approaches on illustrative examples.}, language = {en} } @article{GonnermannMuellerHaaseFackeldeyetal.2025, author = {Gonnermann-M{\"u}ller, Jana and Haase, Jennifer and Fackeldey, Konstantin and Pokutta, Sebastian}, title = {FACET: Teacher-Centred LLM-Based Multi-Agent Systems-Towards Personalized Educational Worksheets}, arxiv = {http://arxiv.org/abs/2508.11401}, year = {2025}, abstract = {The increasing heterogeneity of student populations poses significant challenges for teachers, particularly in mathematics education, where cognitive, motivational, and emotional differences strongly influence learning outcomes. While AI-driven personalization tools have emerged, most remain performance-focused, offering limited support for teachers and neglecting broader pedagogical needs. This paper presents the FACET framework, a teacher-facing, large language model (LLM)-based multi-agent system designed to generate individualized classroom materials that integrate both cognitive and motivational dimensions of learner profiles. The framework comprises three specialized agents: (1) learner agents that simulate diverse profiles incorporating topic proficiency and intrinsic motivation, (2) a teacher agent that adapts instructional content according to didactical principles, and (3) an evaluator agent that provides automated quality assurance. We tested the system using authentic grade 8 mathematics curriculum content and evaluated its feasibility through a) automated agent-based assessment of output quality and b) exploratory feedback from K-12 in-service teachers. Results from ten internal evaluations highlighted high stability and alignment between generated materials and learner profiles, and teacher feedback particularly highlighted structure and suitability of tasks. The findings demonstrate the potential of multi-agent LLM architectures to provide scalable, context-aware personalization in heterogeneous classroom settings, and outline directions for extending the framework to richer learner profiles and real-world classroom trials.}, language = {en} }