Guillaume Fraux
@luthaf.bsky.social
Reposted by Guillaume Fraux
📢 chemiscope.org 1.0.0rc1 just dropped on pypi! We are making (a few) breaking changes to the interfaces, fixing a ton of bugs and introducing some exciting features (you can finally load datasets with > 100k points!). We'd be grateful if you test, break and report 🐛 github.com/lab-cosmo/ch...
January 5, 2026 at 2:42 PM
📢 chemiscope.org 1.0.0rc1 just dropped on pypi! We are making (a few) breaking changes to the interfaces, fixing a ton of bugs and introducing some exciting features (you can finally load datasets with > 100k points!). We'd be grateful if you test, break and report 🐛 github.com/lab-cosmo/ch...
There seems to be some initial work for webGPU with ggml (github.com/ggml-org/lla...), but it looks very recent and not yet usable. Looking forward to it!
ggml : add WebGPU backend · Issue #7773 · ggml-org/llama.cpp
I hope that this would be relatively easy to do since AFAIK WebGPU allows us to write kernels in a shader language, so we have experience how to create such backends. There has been some initial wo...
github.com
November 28, 2025 at 1:39 PM
There seems to be some initial work for webGPU with ggml (github.com/ggml-org/lla...), but it looks very recent and not yet usable. Looking forward to it!
Very nice! Is this able to use WebGPU when running the model?
Also, I guess this is limited to 4GiB of memory? Although it might not matter too much, since it would become very slow for systems that require that kind of memory anyway
Also, I guess this is limited to 4GiB of memory? Although it might not matter too much, since it would become very slow for systems that require that kind of memory anyway
November 28, 2025 at 11:26 AM
Very nice! Is this able to use WebGPU when running the model?
Also, I guess this is limited to 4GiB of memory? Although it might not matter too much, since it would become very slow for systems that require that kind of memory anyway
Also, I guess this is limited to 4GiB of memory? Although it might not matter too much, since it would become very slow for systems that require that kind of memory anyway
Reposted by Guillaume Fraux
📢 PET-MAD is here! 📢 It has been for a while for those who read the #arXiv, but now you get it preciously 💸 typeset by @natcomms.nature.com Take home: unconstrained architecture + good train set choices give you fast, accurate and stable universal MLIP that just works™️ www.nature.com/articles/s41...
PET-MAD as a lightweight universal interatomic potential for advanced materials modeling - Nature Communications
PET-MAD is a fast and lightweight universal machine-learning potential, trained on a small but diverse dataset, that delivers near-quantum accuracy in atomistic simulations for both organic and inorga...
www.nature.com
November 28, 2025 at 8:36 AM
📢 PET-MAD is here! 📢 It has been for a while for those who read the #arXiv, but now you get it preciously 💸 typeset by @natcomms.nature.com Take home: unconstrained architecture + good train set choices give you fast, accurate and stable universal MLIP that just works™️ www.nature.com/articles/s41...
Reposted by Guillaume Fraux
We're introducing ShiftML3, a new ShiftML model for chemical shielding predictions in organic solids.
* ShiftML3 predicts full chemical shielding tensors
* DFT accuracy for 1H, 13C, and 15N
* ASE integration
* GPU integration
Code: github.com/lab-cosmo/Sh...
Install from Pypi: pip install shiftml
* ShiftML3 predicts full chemical shielding tensors
* DFT accuracy for 1H, 13C, and 15N
* ASE integration
* GPU integration
Code: github.com/lab-cosmo/Sh...
Install from Pypi: pip install shiftml
GitHub - lab-cosmo/shiftml: A python package for the prediction of chemical shieldings of organic solids and beyond.
A python package for the prediction of chemical shieldings of organic solids and beyond. - lab-cosmo/shiftml
github.com
August 25, 2025 at 8:53 AM
We're introducing ShiftML3, a new ShiftML model for chemical shielding predictions in organic solids.
* ShiftML3 predicts full chemical shielding tensors
* DFT accuracy for 1H, 13C, and 15N
* ASE integration
* GPU integration
Code: github.com/lab-cosmo/Sh...
Install from Pypi: pip install shiftml
* ShiftML3 predicts full chemical shielding tensors
* DFT accuracy for 1H, 13C, and 15N
* ASE integration
* GPU integration
Code: github.com/lab-cosmo/Sh...
Install from Pypi: pip install shiftml
Reposted by Guillaume Fraux
If you are using or you are considering using CP2K, check this out!
Our manuscript "The CP2K Program Package Made Simple", dedicated to the usage and applications of CP2K is available at: arxiv.org/abs/2508.15559
Contrary to the accompanying theory & code paper doi.org/10.1063/5.00..., the underlying theoretical concepts are minimized and only introduced as needed.
Contrary to the accompanying theory & code paper doi.org/10.1063/5.00..., the underlying theoretical concepts are minimized and only introduced as needed.
The CP2K Program Package Made Simple
CP2K is a versatile open-source software package for simulations across a wide range of atomistic systems, from isolated molecules in the gas phase to low-dimensional functional materials and interfac...
arxiv.org
August 25, 2025 at 10:15 AM
If you are using or you are considering using CP2K, check this out!
Reposted by Guillaume Fraux
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
The dark side of the forces: assessing non-conservative force...
The use of machine learning to estimate the energy of a group of atoms, and the forces that drive them to more stable configurations, have revolutionized the fields of computational chemistry and...
openreview.net
June 20, 2025 at 3:53 PM
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
Reposted by Guillaume Fraux
🎉 DFT-accurate, with built-in uncertainty quantification, providing chemical shielding anisotropy - ShiftML3.0 has it all! Building on a successful @nccr-marvel.bsky.social-funded collaboration with LRM🧲⚛️, it just landed on the arXiv arxiv.org/html/2506.13... and on pypi pypi.org/project/shif...
June 17, 2025 at 1:18 PM
🎉 DFT-accurate, with built-in uncertainty quantification, providing chemical shielding anisotropy - ShiftML3.0 has it all! Building on a successful @nccr-marvel.bsky.social-funded collaboration with LRM🧲⚛️, it just landed on the arXiv arxiv.org/html/2506.13... and on pypi pypi.org/project/shif...
The link above is missing a password, pleas use epfl.zoom.us/j/6836877674... instead!
June 13, 2025 at 7:42 AM
The link above is missing a password, pleas use epfl.zoom.us/j/6836877674... instead!
Reposted by Guillaume Fraux
We wouldn't be @labcosmo.bsky.social if we didn't want you to go break it, so head to atomistic-cookbook.org/examples/fla... for a crash-course 🧑🍳📖 recipe, but not before reading the warnings arxiv.org/html/2505.19.... Have fun! #compchem #machinelearning #md @nccr-marvel.bsky.social @erc.europa.eu
Long-stride trajectories with a universal FlashMD model - The Atomistic CookbookContentsMenuExpandLight modeDark modeAuto light/dark, in light modeAuto light/dark, in dark mode
atomistic-cookbook.org
May 27, 2025 at 7:03 AM
We wouldn't be @labcosmo.bsky.social if we didn't want you to go break it, so head to atomistic-cookbook.org/examples/fla... for a crash-course 🧑🍳📖 recipe, but not before reading the warnings arxiv.org/html/2505.19.... Have fun! #compchem #machinelearning #md @nccr-marvel.bsky.social @erc.europa.eu
Reposted by Guillaume Fraux
If you do, the rewards can be very impressive: you can run solvated alanine dipeptide and observe superionic behavior in LiPS with 16fs time step, and watch the Al(110) surface pre-melt in strides of 64fs. And all with the same universal model, no fine-tuning needed!
May 27, 2025 at 7:03 AM
If you do, the rewards can be very impressive: you can run solvated alanine dipeptide and observe superionic behavior in LiPS with 16fs time step, and watch the Al(110) surface pre-melt in strides of 64fs. And all with the same universal model, no fine-tuning needed!
Reposted by Guillaume Fraux
Much as for direct force prediction [ arxiv.org/html/2412.11... ] you better know what you are doing: you've no guarantee of energy conservation, or of equipartition, so you should know your thermostats VERY well. Caveat emptor.
May 27, 2025 at 7:03 AM
Much as for direct force prediction [ arxiv.org/html/2412.11... ] you better know what you are doing: you've no guarantee of energy conservation, or of equipartition, so you should know your thermostats VERY well. Caveat emptor.
Reposted by Guillaume Fraux
Filippo's idea was to use the heavy-duty PET-MAD model [ arxiv.org/html/2503.14... ] to generate a bunch of trajectories of wildly different compounds and use a PET-like architecture to learn (q',p') from (q,p), and and to think A LOT about the many things that could possibly go wrong.
PET-MAD, a universal interatomic potential for advanced materials modeling
arxiv.org
May 27, 2025 at 7:03 AM
Filippo's idea was to use the heavy-duty PET-MAD model [ arxiv.org/html/2503.14... ] to generate a bunch of trajectories of wildly different compounds and use a PET-like architecture to learn (q',p') from (q,p), and and to think A LOT about the many things that could possibly go wrong.
Reposted by Guillaume Fraux
There are reports as early as 2021 [cf. arxiv.org/abs/2111.15176 ] of using neural nets to predict a MD trajectory in large strides, but these were usually limited to a single system in a given thermodynamic state point. Nice, but not life-changing.
Learning Large-Time-Step Molecular Dynamics with Graph Neural Networks
Molecular dynamics (MD) simulation predicts the trajectory of atoms by solving Newton's equation of motion with a numeric integrator. Due to physical constraints, the time step of the integrator need ...
arxiv.org
May 27, 2025 at 7:03 AM
There are reports as early as 2021 [cf. arxiv.org/abs/2111.15176 ] of using neural nets to predict a MD trajectory in large strides, but these were usually limited to a single system in a given thermodynamic state point. Nice, but not life-changing.
Reposted by Guillaume Fraux
📢 Running molecular dynamics with time steps up to 64fs for any atomistic system, from Al(110) to Ala2? Thanks to 🧑🚀 Filippo Bigi and Sanggyu Chong, with some help from Agustinus Kristiadis, this is not as crazy as it sounds. Let us briefly introduce FlashMD⚡ arxiv.org/html/2505.19...
May 27, 2025 at 7:03 AM
📢 Running molecular dynamics with time steps up to 64fs for any atomistic system, from Al(110) to Ala2? Thanks to 🧑🚀 Filippo Bigi and Sanggyu Chong, with some help from Agustinus Kristiadis, this is not as crazy as it sounds. Let us briefly introduce FlashMD⚡ arxiv.org/html/2505.19...
Reposted by Guillaume Fraux
If you don't want to read, but want to cook, guess what? The 🧑🍳📖 #atomistic-cookbook has you covered. Head to atomistic-cookbook.org/examples/lea... to see how to use this, as simple as `mtt train mymodel.yaml`.
Equivariant model for tensorial properties based on scalar features - The Atomistic CookbookContentsMenuExpandLight modeDark modeAuto light/dark, in light modeAuto light/dark, in dark mode
atomistic-cookbook.org
May 9, 2025 at 6:50 AM
If you don't want to read, but want to cook, guess what? The 🧑🍳📖 #atomistic-cookbook has you covered. Head to atomistic-cookbook.org/examples/lea... to see how to use this, as simple as `mtt train mymodel.yaml`.
Reposted by Guillaume Fraux
If you don't have time for the 20-pages appendix, the TL;DR is that approximating tensors is harder than the vector case, but can be made as simple as possible using angular momentum theory. The practical implementation we propose is not as rigorous, but works well and is very fast in practice.
May 9, 2025 at 6:50 AM
If you don't have time for the 20-pages appendix, the TL;DR is that approximating tensors is harder than the vector case, but can be made as simple as possible using angular momentum theory. The practical implementation we propose is not as rigorous, but works well and is very fast in practice.
Reposted by Guillaume Fraux
First #preprint from recent 🧑🚀 Michelangelo Domina (+ @ppegolo.bsky.social and Filippo) provides theoretical foundations and practical architectures to build scalar-function-based approximations of tensors, in the spirit of the "scalars are universals" paper #compchem arxiv.org/html/2505.05...
May 9, 2025 at 6:50 AM
First #preprint from recent 🧑🚀 Michelangelo Domina (+ @ppegolo.bsky.social and Filippo) provides theoretical foundations and practical architectures to build scalar-function-based approximations of tensors, in the spirit of the "scalars are universals" paper #compchem arxiv.org/html/2505.05...
Reposted by Guillaume Fraux
You can use these safely in MD, by multiple time stepping (a '90s classic: doi.org/10.1063/1.46...) so you get reliable conservative trajectories, for any material, twice as fast! Let us know if it works for you, and even more importantly, if it doesn't! 👉 atomistic-cookbook.org/examples/pet...
Reversible multiple time scale molecular dynamics
The Trotter factorization of the Liouville propagator is used to generate new reversible molecular dynamics integrators. This strategy is applied to derive reve
doi.org
May 7, 2025 at 5:24 AM
You can use these safely in MD, by multiple time stepping (a '90s classic: doi.org/10.1063/1.46...) so you get reliable conservative trajectories, for any material, twice as fast! Let us know if it works for you, and even more importantly, if it doesn't! 👉 atomistic-cookbook.org/examples/pet...
Reposted by Guillaume Fraux
The PET-MAD universal forcefield mingled with the dark side, and got twice as fast 🚀. Read on, or head to the 🧑🍳📖 atomistic-cookbook.org/examples/pet..., if you are curious of what this is all about. #atomistic-cookbook #compchem #machinelearning #mlip🧵
May 7, 2025 at 5:24 AM
The PET-MAD universal forcefield mingled with the dark side, and got twice as fast 🚀. Read on, or head to the 🧑🍳📖 atomistic-cookbook.org/examples/pet..., if you are curious of what this is all about. #atomistic-cookbook #compchem #machinelearning #mlip🧵
Reposted by Guillaume Fraux
📢 For those who missed the #preprint, torch-pme is now published in @aip.bsky.social #JChemPhys. #compchem classics like Ewald and P3M meet #machine learning. Read all about it here pubs.aip.org/aip/jcp/arti...
Fast and flexible long-range models for atomistic machine learning
Most atomistic machine learning (ML) models rely on a locality ansatz and decompose the energy into a sum of short-ranged, atom-centered contributions. This lea
pubs.aip.org
April 8, 2025 at 1:31 PM
📢 For those who missed the #preprint, torch-pme is now published in @aip.bsky.social #JChemPhys. #compchem classics like Ewald and P3M meet #machine learning. Read all about it here pubs.aip.org/aip/jcp/arti...
Reposted by Guillaume Fraux
You can read all about it here, arxiv.org/abs/2503.14118, see it in action as a #cookbook recipe 🧑🍳 atomistic-cookbook.org/examples/pet... or rush to install it following the instructions here github.com/lab-cosmo/pe...
March 19, 2025 at 7:23 AM
You can read all about it here, arxiv.org/abs/2503.14118, see it in action as a #cookbook recipe 🧑🍳 atomistic-cookbook.org/examples/pet... or rush to install it following the instructions here github.com/lab-cosmo/pe...
Reposted by Guillaume Fraux
📢 PET-MAD has just landed! 📢 What if I told you that you can match & improve the accuracy of other "universal" #machinelearning potentials training on fewer than 100k atomic structures? And be *faster* with an unconstrained architecture that is conservative with tiny symmetry breaking? Sounds like 🧑🚀
March 19, 2025 at 7:23 AM
📢 PET-MAD has just landed! 📢 What if I told you that you can match & improve the accuracy of other "universal" #machinelearning potentials training on fewer than 100k atomic structures? And be *faster* with an unconstrained architecture that is conservative with tiny symmetry breaking? Sounds like 🧑🚀
Reposted by Guillaume Fraux
Should be super-easy to adapt to whatever tensor you're trying to learn (and to extend to more sophisticated models than symmetry-adapted regression). Let us know what you cook with it 😋!
March 13, 2025 at 5:30 PM
Should be super-easy to adapt to whatever tensor you're trying to learn (and to extend to more sophisticated models than symmetry-adapted regression). Let us know what you cook with it 😋!
Reposted by Guillaume Fraux
In particular, this recipe relies on metatensor.github.io/featomic to compute the features, and the docs.metatensor.org/latest/torch... backend of #metatensor to export a self-contained ASE-compatible calculator. Easy to use, fast, and accurate.
metatensor.github.io
March 13, 2025 at 5:30 PM
In particular, this recipe relies on metatensor.github.io/featomic to compute the features, and the docs.metatensor.org/latest/torch... backend of #metatensor to export a self-contained ASE-compatible calculator. Easy to use, fast, and accurate.