🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

When I came to US for graduate study, in 1986, the Harvard International Office made Ed and Jane my host parents. “My father likes math,” their son told me. “Me too,” I said. Years later, I learned that Ed invented chaos theory. en.wikipedia.org/wiki/Edward_No…

Check out my new pre-print on using differentiable molecular simulation to improve implicit solvent force fields for disordered proteins: biorxiv.org/content/10.110…. The resulting force field, GB99dms, is easy to use with OpenMM.

1/n: We are excited to share that our paper on Chroma, a general purpose diffusion model for proteins, is out today in nature! nature.com/articles/s4158… A couple of my favorite highlights in the 🧵below 👇

#Chroma has also generated a protein model in the shape of Japanese character！ アルファベット以外でもいけるようだ。 文字形状の点群への変換を経由しており、任意の文字でできるのかもしれない。 Colab Notebook: colab.research.google.com/github/generat…

The 62p of supplementary information of #AlphaFold2 vs the 123p of #Chroma 🤪 ... 🥲😭

Thrilled to share this work on materials discovery! We found that OOD generalization of GNNs improves predictably, with increasing data from quantum mechanical simulations. These GNNs allowed us to expand the number of known stable materials by an order of magnitude.

Oldies but goldies: R. Brockett, Dynamical systems that sort lists, diagonalize matrices, and solve linear programming problems, 1991. Brockett's flow progressively diagonalizes a symmetric matrix. hrl.harvard.edu/publications/b…

[1/N] Generative AI has revolutionized how we create text and images. How about designing novel materials? We at Microsoft Research #AI4Science are thrilled to announce MatterGen: our generative model that enables broad property-guided materials design. 👇 arxiv.org/abs/2312.03687

MatternGen can generate stable and functional materials using desired properties as prompts, validated with DFT. Congrats to dear friends and colleagues at Microsoft Research! Proud to be part of it.

Looking for diffusion model advancements at #NeurIPS2023? Come to check our oral work "Understanding Diffusion Objectives as the ELBO with Simple Data Augmentation" w/ Durk Kingma. New theoretical understanding, SOTA empirical results, and more! Arxiv: arxiv.org/abs/2303.00848

A foundation model for atomistic materials chemistry arxiv.org/abs/2401.00096 #compchem

Hey Tim Cook if you see a spike in M2 Ultra and M3 Max sales in Q4, please look at your Apple Machine Learning Research team. They did the magic: MLX 🪄 github.com/ml-explore/mlx

Our third year hosting this program. It has been incredibly rewarding for all of us in the team to interact with amazingly talented PhD students. Looking forward to another summer of fun science!

Making a commitment here that I’ll talk about symmetric protein design from a perspective you couldn’t have read about somewhere 😉

Announcing our new model for materials! FlowMM... - Generates stable & novel materials efficiently - Predicts crystal structure accurately - Generalizes Riemannian Flow Matching to point clouds w/ periodic boundaries arxiv.org/abs/2406.04713 Ricky T. Q. Chen Anuroop Sriram Brandon Wood

Protpardelle 🍝 came out in PNASNews last week! highlights of new results: - backbone-only model compares with RFdiffusion - flow-based likelihoods predict designability - better filtering on AFDB improves quality and diversity - measure scaffolding and repacking performance

You asked for it, so here it is. Visualizing CPU cache speeds relative to RAM. Cache optimization is important too!

CPU vs FPGA - what an easy and well explanation in 25 seconds

Introducing reversible molecular simulation: differentiable simulation but with constant memory. Useful for training classical and machine learning force fields to match experimental data. Feedback very welcome! arxiv.org/abs/2412.04374 (1/8)

Today we are releasing NeuralPLexer3 (NP3) technical report, along with opening-sourcing NPBench - our turnkey library for fair benchmarking of co-folding structure prediction models on diverse interactions - a task that has challenged this field. 👩‍💻: github.com/iambic-therape…