marcel ⊙

Orbits for a periodic 3-body system, showing the stability of a ML long-time integrator

If you are excited about 30x longer time steps in molecular dynamics using FlashMD, but are worried about it not being symplectic, Filippo has something new cooking that should make you even more excited. Head to the #arxiv for a preview arxiv.org/html/2508.01...

for the jax kids, i just made my personal training infra repo public: github.com/sirmarcel/ma... -- happy for some feedback/ideas/takes!

I benefited massively from www.ipam.ucla.edu/programs/lon.... I got into ML for science through that program. Now IPAM may be gone mathstodon.xyz/@tao/1149568...

This, if true, is incredible. A long program in 2011 at IPAM that I attended practically wrote the book on machine learning in materials science, which is now AI. Not funding IPAM would be deeply destructive to US leadership in critical future technologies! Hard to fathom.

Absolute insanity. IPAM is where I first heard about equivariant NNs! What a loss.

this is work at @labcosmo.bsky.social with Egor, Tulga, Philip & @micheleceriotti.bsky.social!

Learning Long-Range Representations with Equivariant Messages Machine learning interatomic potentials trained on first-principles reference data are quickly becoming indispensable for computational physics, biology, and chemistry. Equivariant message-passing neu...

... overall, it seems that we need more challenging benchmarks. or maybe LR behaviour is simply "not that complicated". we will see. the manuscript is here: arxiv.org/abs/2507.19382 & code & data are coming out soon-ish, once i run a few more ablations. looking forward to feedback!

... and show that it works pretty well on a few long-range benchmark datasets. but we also find that many benchmark tasks are easily solved even by short-range message passing, so we run a few more targeted experiments to see where it fails & confirm that LOREM works for these ...

✨preprint alert: "learning long-range representations w/ equivariant messages" in which we get into the fray of long-range MLIPs and propose to use equivariant charges w/ classical electrostatics as long-range MP mechanism. we design LOREM, an equivariant MLIP, around this ...

very proud of filippo

Very proud to send Filippo Bigi to Vancouver to give an oral presentation at @icmlconf.bsky.social about our investigation of the use of "dark-side forces" in atomistic simulations. The final version is here openreview.net/forum?id=OEl... and it's worth a read even if you already read the #preprint

So kudos to 🧑‍🚀 Filippo and @marceldotsci.bsky.social bsky.app/profile/labc..., thanks to funders @erc.europa.eu @snf-fns.ch @cscsch.bsky.social and @nccr-marvel.bsky.social, and go to Filippo's talk if you are at #icml25, to see this nice mix of #compchem and #machinelearning!

obvious in hindsight but news to me: accessing np.load-ed arrays by index is orders of magnitude slower than forcing them to RAM. notes.marcel.science/2025/numpy-l...

ok then thanks

very annoying when papers don‘t include raw data for plots

⚡️

🍾

Hm... yeah, it's okay, but my impression from talking to practitioners is that its speed is not particularly impressive. (Because of the tensor products.) That's why it's a bit concerning that the LLM wrote it into the table... 😇 "Hot take" indeed!

MACE is not generally considered fast, right? 🧐

cosmo has joined the universal FF club!

watch out cancer, mihail is coming for you 🔥

Two halves of a 3D printed ergonomic split keyboard surrounding pain relief gel.

Completed my post preprint project: a split keyboard to help with my wrist pain (pictured with the pain-relieving gel it is trying to replace).

sick!

anyone at the lausanne applied ML days?

@marceldotsci.bsky.social getting into very dangerous territory.

same down to last minute upload 🤝

Seminar by Jan Gerken: Emergent Equivariance in Deep Ensembles Join us for a seminar by Jan Gerken on Tuesday, February 4th, at 10:30 AM in MED 2 1124 (Coviz2). Jan is an Assistant Professor in the Division of Algebra and Geometry at Chalmers University of Techno...

we're having an interesting seminar next tuesday on equivariance in deep ensembles, for the EPFL squad: actu.epfl.ch/news/seminar...

Figure comparing automatic differentiation (AD) and automatic sparse differentiation (ASD). (a) Given a function f, AD backends return a function computing vector-Jacobian products (VJPs). (b) Standard AD computes Jacobians row-by-row by evaluating VJPs with all standard basis vectors. (c) ASD reduces the number of VJP evaluations by first detecting a sparsity pattern of non-zero values, coloring orthogonal rows in the pattern and simultaneously evaluating VJPs of orthogonal rows. The concepts shown in this figure directly translate to forward-mode, which computes Jacobians column-by-column instead of row-by-row.

You think Jacobian and Hessian matrices are prohibitively expensive to compute on your problem? Our latest preprint with @gdalle.bsky.social might change your mind!
arxiv.org/abs/2501.17737
🧵1/8

YouTube video by Orbital Big Data & Symmetry in Machine Learning Models, with Marcel Langer & Jonathan Schmidt

the recording of my talk at orbital materials, together with jonathan schmidt, is out: www.youtube.com/watch?v=0xpR...!

now on day ~2 of debugging bizarre multiprocessing issues with Google Grain, the alleged hot new data loading framework. has anyone here actually used it by any chance?