Deep learning accelerates discovery of complex nanomaterials Nature Computational Science - A physics-infused heterogeneous graph neural network has been developed to address challenges in designing complex nanomaterials with spatially varying compositions....

Emory and I also wrote a higher level summary of the work available here: rdcu.be/eUVIw

Gradient-based optimization of complex nanoparticle heterostructures enabled by deep learning on heterogeneous graphs - Nature Computational Science Graph neural networks built on physically motivated representations enable gradient-based optimization of complex upconverting nanoparticle heterostructures, revealing photophysical design rules and a...

Out in @natcomputsci.nature.com: A roadmap for inverse design of #nanomaterials heterostructures via HT data gen -> representation dev -> heteroGNN training -> gradient-based global opt! w/ @emorychannano.bsky.social @ewcspottesmith.bsky.social
www.nature.com/articles/s43...
Free link rdcu.be/eTH72

@berkeleylab.lbl.gov
@molecularfoundry.lbl.gov @mitcheme.bsky.social

For a more digestible summary of our recent @natcomputsci.nature.com article, here's the research brief that @samblau.bsky.social and I wrote... complete with a #BehindThePaper tidbit that thankfully has less drama than a VH1 Behind the Music episode (free link: rdcu.be/eUVIw)

Our manuscript on deep learning of upconverting nanoparticle heterostructures w/ @samblau.bsky.social is finally out in @natcomputsci.nature.com!

See 🧵for links to free access via Readcube and chemrxiv.

@berkeleylab.lbl.gov @molecularfoundry.lbl.gov @mitcheme.bsky.social

I'm hiring postdocs @berkeleylab.lbl.gov to drive cutting-edge research involving MLIPs, high-throughput workflows, chemical reaction networks, generative models, and open-source software dev. Full position description + application here: forms.gle/zePBZDmciXez... #Chempostdoc #AI4Science

Come work with @emorychannano.bsky.social and me!
#robotics #nanochemistry #machinelearning #UCNPs

Interested in learning more about our recently published OMol25 dataset and the advances that it's bringing to atomistic machine learning? Check out this talk that my boy @samblau.bsky.social gave as part of the "Modeling Talk Series".

#CompChem ⚗️ 🧪 #SciML

I'm presenting OMol25 tomorrow 7/29 at 9 AM PST as part of a talk series at Google. Learn how we built the dataset + how MLIPs trained on OMol are revolutionizing comp chem!
Meet: lnkd.in/g4AAWkcK
YouTube Stream: lnkd.in/ggmtMtTR
Join group: lnkd.in/g5ciuNuX

OMol25 was calculated with ORCA. I want to acknowledge the work of the ORCA team to improve the quality of the gradient + the robustness of SCF convergence for complicated systems as part of the OMol effort - it was much appreciated and critical to ensuring that we're releasing high quality data!

Computational Chemistry Unlocked: A Record-Breaking Dataset to Train AI Models has Launched - Berkeley Lab Scientists will finally be able to simulate the chemistry that drives our bodies, our environment, and our technologies.

🚨 Just dropped: Open Molecules 2025 — a record-breaking dataset co-led by Berkeley Lab + Meta FAIR.

100M+ DFT snapshots. Built to train #AI for real-world chemistry 🧪.

Could reshape discovery in batteries, drug discovery & much more! @cs.lbl.gov ⬇️

We can't wait to see what the community does with OMol! Don't hesitate to reach out with feedback on the data, models, or paper - we aren't going to submit to a journal until the leaderboard goes up, which means we have time to incorporate community feedback (within reason) 10/10

A special shout out to co-first authors Daniel Levine and Muhammed Shuaibi who moved mountains making OMol a reality. I also want to recognize the substantial and critical contributions of @ewcspottesmith.bsky.social, Michael Taylor, Muhammad Hasyim, and Kyle Michel 9/N

Co-leading OMol with Brandon and Larry was a joy and an honor - as was assembling a world-leading team of scientists from 2 companies, 2 national labs, and 6 universities who were excited to help build an open-source, revolutionary molecular DFT dataset to push science forward 8/N

Right now, OMol data has energy, forces, partial charges, partial spins, and HOMO/LUMO. But we have far more info that we still need to parse and hope to do a battery of GBW postprocessing. Plus we have 10 petabytes of electron densities. Lots more to come! 7/N

And check out the UMA demo (facebook-fairchem-uma-demo.hf.space UMA is trained on OMol + other FAIR Chemistry datasets) - metal complexes at +1 vs +2 correctly optimize to tetrahedral/planar and reduced ethylene carbonate correctly ring-opens while a neutral EC remains stable 6/N

The Open Molecules 2025 (OMol25) Dataset, Evaluations, and Models Machine learning (ML) models hold the promise of transforming atomic simulations by delivering quantum chemical accuracy at a fraction of the computational cost. Realization of this potential would en...

Data, models, & paper are available to download now! 5/N
Paper: arxiv.org/abs/2505.08762
Data + models: huggingface.co/facebook/OMo...

We're also releasing baseline models trained on OMol. To guide future MLIP development, we built novel evaluations on intermolecular interactions, conformers, and charge/spin. We hope to include frequency, ΔG, and TSopt tasks when we put up a public leaderboard in the summer 4/N

OMol was constructed via an unprecedented diversity of methods: MD, ML-MD, RPMD, rattling, Architector, rxn path interpolation, AFIR, optimization, and scaled separation. We also recalculated some previous datasets and did additional sampling/structure generation atop others 3/N

OMol covers 83 elements, a wide range of intra and intermolecular interactions, explicit solvation, reactive structures, conformers, charges -10 to 10, 0-10 unpaired electrons, and 2-350 atoms per snapshot. It required >6B CPU hrs, 10x more than any prev MLIP training dataset 2/N

The Open Molecules 2025 dataset is out! With >100M gold-standard ωB97M-V/def2-TZVPD calcs of biomolecules, electrolytes, metal complexes, and small molecules, OMol is by far the largest, most diverse, and highest quality molecular DFT dataset for training MLIPs ever made 1/N

IIDAI Seminar, 5/1/2025, Samuel M. Blau (Berkeley Lab) YouTube video by Coordinated Science Laboratory

It was a pleasure to give an IIDAI seminar on nanoparticle ML for gradient-based heterostructure optimization (w/ @emorychannano.bsky.social ) and neural network path opt for finding reaction transition states on MLIPs (w/ @thglab.bsky.social) - find the talk here: www.youtube.com/watch?v=-4jB...

🧠 New postdoctoral researcher position at Princeton for those interested in data science and machine learning! Specify my group if you are interested in working together. Deadline is May 31. Details: puwebp.princeton.edu/AcadHire/app...

Final day to submit abstracts for ACS Fall 2025! Reminder that @ewcspottesmith.bsky.social , Brett Savoie (Notre Dame), and I are organizing a symposium on "Chemical Reaction Networks, Retrosynthesis, and Reaction Prediction". Will be a mix of invited and contributed talks - please submit! #CompChem

the @gpggrp.bsky.social is at the ACS Spring 2025! come check out the works of Daniil Boiko and Rob MacKnight at the "ML + AI in Organic Chemistry" Symposium (Hall B-1, Room 4) today! extreme scaling of experimental chemical reactions via MS and an OS for autonomous comp chem!

Looking forward to speaking at ACS on Sunday at 5:30! Come learn about "Popcornn" - a new method for double-ended transition state optimization atop machine learned interatomic potentials that is substantially better than NEB or GSM.

Fantastic new work from Aditi & co that shows how to leverage the expressivity + accuracy of massive pre-trained MLIPs to distill smaller, much faster models that are still extremely accurate to drive downstream simulations - no need to compromise on speed vs accuracy!

Applications closing in one week! If you’re interested in a prestigious postdoc at the intersection of AI/ML and nuclear nonproliferation, don’t hesitate to apply - come work with me on fascinating f-block chemistry and computational/ML methods! (Must be a US citizen)

@samblau.bsky.social, Brett Savoie (Notre Dame), and I are organizing a symposium for @amerchemsociety.bsky.social Fall 2025 called "Chemical Reaction Networks, Retrosynthesis, and Reaction Prediction" under @acscomp.bsky.social.

#reactionnetwork #CRN #retrosynthesis 🧪 ⚗️ #CompChem