ML Researcher @DEShawResearch. Former PhD student @FU_Berlin, Visiting Researcher @MSFTResearch, and ML Intern @DEShawResearch.
Joined January 2021
- Tweets142
- Following312
- Followers646
- Likes1.8K
Pinned Tweet
(1/n) Can diffusion models simulate molecular dynamics instead of generating independent samples?
In our NeurIPS2025 paper, we train energy-based diffusion models that can do both:
- Generate independent samples
- Learn the underlying potential ๐ผ
๐งต๐
arxiv.org/abs/2506.17139
Leon Klein retweeted
Ever get tired of tiny timesteps bottlenecking your MD simulations?
We show how to train a model for large-timestep Hamiltonian dynamics directly on standard MLFF datasets. ๐ก๐ผ ๐ฟ๐ฒ๐ณ๐ฒ๐ฟ๐ฒ๐ป๐ฐ๐ฒ ๐๐ฟ๐ฎ๐ท๐ฒ๐ฐ๐๐ผ๐ฟ๐ถ๐ฒ๐, ๐ป๐ผ ๐๐ป๐ฟ๐ผ๐น๐น๐ถ๐ป๐ด, ๐ป๐ผ ๐๐ฒ๐ฎ๐ฐ๐ต๐ฒ๐ฟ needed!๐งต๐
Leon Klein retweeted
Excited to share our latest preprint: ๐๐ฒ๐ฎ๐ฟ๐ป๐ถ๐ป๐ด ๐๐ฎ๐บ๐ถ๐น๐๐ผ๐ป๐ถ๐ฎ๐ป ๐๐น๐ผ๐ ๐ ๐ฎ๐ฝ๐: ๐ ๐ฒ๐ฎ๐ป ๐๐น๐ผ๐ ๐๐ผ๐ป๐๐ถ๐๐๐ฒ๐ป๐ฐ๐ ๐ณ๐ผ๐ฟ ๐๐ฎ๐ฟ๐ด๐ฒ-๐ง๐ถ๐บ๐ฒ๐๐๐ฒ๐ฝ ๐ ๐ผ๐น๐ฒ๐ฐ๐๐น๐ฎ๐ฟ ๐๐๐ป๐ฎ๐บ๐ถ๐ฐ๐ ๐
Ever get tired of tiny timesteps bottlenecking your MD simulations?
We show how to train a model for large-timestep Hamiltonian dynamics directly on standard MLFF datasets. ๐ก๐ผ ๐ฟ๐ฒ๐ณ๐ฒ๐ฟ๐ฒ๐ป๐ฐ๐ฒ ๐๐ฟ๐ฎ๐ท๐ฒ๐ฐ๐๐ผ๐ฟ๐ถ๐ฒ๐, ๐ป๐ผ ๐๐ป๐ฟ๐ผ๐น๐น๐ถ๐ป๐ด, ๐ป๐ผ ๐๐ฒ๐ฎ๐ฐ๐ต๐ฒ๐ฟ needed!๐งต๐
Leon Klein retweeted
I'm excited to open the new year by sharing a new perspective paper
I give a informal outline of MD and how it can interact with Generative AI. Then, how far the field has come since the seminal contributions, such as Boltzmann Generators, and what is still missing
Leon Klein retweeted
How can ML assist sampling from equilibrium distributions?
On the coming Tuesday (Nov 25th), 4pm-5pm UK time, we will have Leon Klein @leonklein26 to talk about โEfficient Generation of Molecular Boltzmann Distributions with Normalizing Flowsโ ๐ฅ
Zoom link and more info ๐
Leon Klein retweeted
Fantastic work by Michael Plainer and friends. Energy-based diffusion models to ensure that the denoising distribution equals exp(-u(x)), with the energy u(x). A keystone for connecting molecular dynamics, statistical mechanics and generative AI.
(1/n) Can diffusion models simulate molecular dynamics instead of generating independent samples?
In our NeurIPS2025 paper, we train energy-based diffusion models that can do both:
- Generate independent samples
- Learn the underlying potential ๐ผ
๐งต๐
arxiv.org/abs/2506.17139
(6/n) With this, we can run coarse-grained Langevin dynamics directly, without the need for any priors or force labels.
This works across biomolecular systems including fast-folding proteins like Chignolin and BBA.
Here is a comparison with and without our regularization:
ALT Comparison of equilibrium distributions obtained by iid sampling and Langevin simulation (sim) across different systems and methods. While classical iid sampling recovers the reference equilibrium distribution, performing simulation with the learned score reveals inconsistencies when models are not trained with Fokker-Planck regularization, i.e., p(x) != p_0(x). Regularized models achieve consistent behavior across systems.
(7/n) Check out our paper and code:
Paper: arxiv.org/abs/2506.17139
Code + models: github.com/noegroup/ScoreMD
And also, our self-contained notebooks!
Colab (JAX):
colab.research.google.com/drโฆ
Colab (PyTorch):
colab.research.google.com/drโฆ
Leon Klein retweeted
After a 4-year journey, we are super happy to see this paper out in @NatureComms - @ElezKatarina et al: High-throughput molecular dynamics + active @machinelearning enable efficient identification of an experimentally-validated broad coronavirus inhibitor.
nature.com/articles/s41467-0โฆ
Leon Klein retweeted
Really excited to (finally) share the updated JAMUN preprint and codebase! We perform Langevin molecular dynamics in a smoothed space which allows us to take larger integrator steps. This requires learning a score function only at a single noise level, unlike diffusion models.
Leon Klein retweeted
Our development of machine-learned transferable coarse-grained models in now on Nat Chem! doi.org/10.1038/s41557-025-0โฆ
I am so proud of my group for this work! Particularly first authors Nick Charron, Klara Bonneau, @sayeg84, Andrea Guljas.
Leon Klein retweeted
Come check out SBG happening now! W-115 11-1:30 with
@charliebtan
@bose_joey
Chen Lin
@leonklein26
@mmbronstein
Leon Klein retweeted
Main Conference:
๐ Title: Scalable Equilibrium Sampling with Sequential Boltzmann Generators
๐ When: Wed 16 Jul 11 a.m. PDT โ 1:30 p.m. PDT
๐บ๏ธ Where: West Exhibition Hall B2-B3 W-115
๐ arXiv: arxiv.org/abs/2502.18462
w/ @charliebtan @WillLin1028 @leonklein26 @mmbronstein @AlexanderTong7
Leon Klein retweeted
๐Personal update: I'm thrilled to announce that I'm joining Imperial College London @imperialcollege as an Assistant Professor of Computing @ICComputing starting January 2026.
My future lab and I will continue to work on building better Generative Models ๐ค, the hardest AI4Science applications in computational biology ๐งฌand chemistry ๐งช, and also a sprinkling of Deep Learning theory ๐ that supports these goals.
