Xiang Fu (@xiangfu_ml) / X

Xiang Fu

Periodic Labs

423 posts

Xiang Fu

Periodic Labs

@xiangfu_ml

mts @periodiclabs. prev @AIatMeta @MIT_CSAIL

San Francisco, CA

Joined July 2019

New to X?

Sign up now to get your own personalized timeline!

By signing up, you agree to the Terms of Service and Privacy Policy, including Cookie Use.

Terms·Privacy·Cookies·Accessibility·Ads Info·© 2026 X Corp.

Don't miss what's happening

People on X are the first to know.

Pinned
Xiang Fu
@xiangfu_ml
Sep 30, 2025
I joined @periodiclabs in May. We’re building AI scientists + autonomous labs to create breakthroughs you can hold. Longer take: xiangfu.co/science More on Periodic: periodic.com
Liam Fedus
@LiamFedus
Sep 30, 2025
Today, @ekindogus and I are excited to introduce @periodiclabs. Our goal is to create an AI scientist. Science works by conjecturing how the world might be, running experiments, and learning from the results. Intelligence is necessary, but not sufficient. New knowledge is
54K
Xiang Fu
@xiangfu_ml
Apr 25, 2022
New paper: an end-to-end approach for learning to simulate coarse-grained molecular dynamics! - Multi-scale GNN - complex systems: polymers and batteries - very long stable rollouts - orders of magnitude faster than classical force fields! arxiv.org/abs/2204.10348 🧵1/n
GIF
Xiang Fu
@xiangfu_ml
Oct 14, 2022
Are SOTA ML Force Fields capable of simulating a variety of molecular dynamics (MD) systems? What causes a model to fail in simulations? We explore these questions with a comprehensive benchmark study: Paper: arxiv.org/abs/2210.07237 Code/Data: github.com/kyonofx/MDsim 🧵1/n
GIF
Xiang Fu
@xiangfu_ml
May 30, 2024
Charge density is the core attribute of atomic systems in DFT. When representing and predicting charge density with ML, it is challenging to balance accuracy and efficiency. We propose a recipe that achieves SOTA on both: arxiv.org/abs/2405.19276 1/5
45K
Xiang Fu
@xiangfu_ml
Sep 28, 2024
Now accepted to NeurIPS 2024 with code and pretrained models available: github.com/kyonofx/scdp
Xiang Fu
@xiangfu_ml
May 30, 2024
Charge density is the core attribute of atomic systems in DFT. When representing and predicting charge density with ML, it is challenging to balance accuracy and efficiency. We propose a recipe that achieves SOTA on both: arxiv.org/abs/2405.19276 1/5
16K
Xiang Fu
@xiangfu_ml
Apr 25, 2022
Very excited to share our latest work in an attempt to bring MD simulation to large scale! Thanks @ak92501 for tweeting!
AK
@_akhaliq
Apr 25, 2022
Simulate Time-integrated Coarse-grained Molecular Dynamics with Geometric Machine Learning abs: arxiv.org/abs/2204.10348 project page: xiangfu.co/mlcgmd
00:00
Xiang Fu
@xiangfu_ml
Oct 18, 2024
One of the largest open dataset for materials quantum mechanical calculations and state-of-the-art ML potentials, open-sourced for both commercial and non-commercial use. We are eager to hear your feedback!
FAIR Chemistry
@OpenCatalyst
Oct 18, 2024
Introducing Meta’s Open Materials 2024 (OMat24) Dataset and Models! All under permissive open licenses for commercial and non-commercial use! Paper: arxiv.org/abs/2410.12771 Dataset: huggingface.co/datasets/fairc… Models: huggingface.co/fairchem/OMAT24 🧵1/x
8.4K
Xiang Fu
@xiangfu_ml
Oct 18, 2023
Metal-organic frameworks (MOFs) are emerging as a promising class of materials for gas separation and storage to address climate change 🌍. MOFDiff blends coarse-graining and a diffusion generative model to design MOFs for carbon capture. arxiv.org/abs/2310.10732 1/6
GIF
27K
Xiang Fu
@xiangfu_ml
Feb 18, 2025
For existing MLIPs, lower test errors do not always translate to better performance in downstream tasks. We bridge this gap by proposing eSEN -- SOTA performance on compliant Matbench-Discovery (F1 0.831, κSRME 0.321) and phonon prediction. arxiv.org/abs/2502.12147 1/6
arxiv.org
Learning Smooth and Expressive Interatomic Potentials for Physical...
Machine learning interatomic potentials (MLIPs) have become increasingly effective at approximating quantum mechanical calculations at a fraction of the computational cost. However, lower errors...
13K
Xiang Fu
@xiangfu_ml
Jun 29, 2022
The code and single-chain polymer dataset are now available: Code: github.com/kyonofx/mlcgmd Data: zenodo.org/record/6764836… Join us next week on July 5 at 11 am EST if you are interested in learning to simulate MD: hannes-stark.com/logag-reading-…
Xiang Fu
@xiangfu_ml
Apr 25, 2022
New paper: an end-to-end approach for learning to simulate coarse-grained molecular dynamics! - Multi-scale GNN - complex systems: polymers and batteries - very long stable rollouts - orders of magnitude faster than classical force fields! arxiv.org/abs/2204.10348 🧵1/n
GIF
Xiang Fu
@xiangfu_ml
Jun 5, 2023
This summer I am interning at Microsoft Research AI4Science in Cambridge, UK 😄 please feel free to reach out if you’d like to catch up on ML for MD simulation, coarse graining, or anything else!
9.3K
Xiang Fu
@xiangfu_ml
Oct 24, 2023
Running MD simulations with ML force fields? Consider learning the scale separation for a potential ~2-4x speed boost using Multi-scale integration: working paper: arxiv.org/abs/2310.13756 Great collaborating with Alby Musaelian, @ndrsjhnssn, Tommi Jaakkola, and @bkoz37
7.4K
Xiang Fu
@xiangfu_ml
Mar 11, 2024
The code for MOFDiff (accepted to ICLR 2024) is now open-sourced: github.com/microsoft/MOFD…
Xiang Fu
@xiangfu_ml
Oct 18, 2023
Metal-organic frameworks (MOFs) are emerging as a promising class of materials for gas separation and storage to address climate change 🌍. MOFDiff blends coarse-graining and a diffusion generative model to design MOFs for carbon capture. arxiv.org/abs/2310.10732 1/6
GIF
11K
Xiang Fu
@xiangfu_ml
Dec 11, 2024
Presenting this work at NeurIPS tomorrow morning! I will be at NeurIPS from 12/11 to 12/14, let me know if you’d like to chat about AI for electronic structures, molecular dynamics, materials design, or the FAIR chemistry team!
Xiang Fu
@xiangfu_ml
May 30, 2024
Charge density is the core attribute of atomic systems in DFT. When representing and predicting charge density with ML, it is challenging to balance accuracy and efficiency. We propose a recipe that achieves SOTA on both: arxiv.org/abs/2405.19276 1/5
4.6K