Peter Spackman
@crystalexplorer.net
Computers, crystals and chemistry. Research Fellow in the Computational Materials and Minerals Group at Curtin University. @[email protected]
Actually their webgpu support is better than I'd thought, I was wrong. I'll get it working if I can!
November 28, 2025 at 2:25 PM
Actually their webgpu support is better than I'd thought, I was wrong. I'll get it working if I can!
As far as I know ggml doesn’t have webgpu acceleration, but in principle it should absolutely be possible as there are plenty of other wasm and webgpu combinations! Yes I think given my current implementation is single core only and quite slow in wasm it’s not the memory that’ll bottleneck first!
November 28, 2025 at 1:10 PM
As far as I know ggml doesn’t have webgpu acceleration, but in principle it should absolutely be possible as there are plenty of other wasm and webgpu combinations! Yes I think given my current implementation is single core only and quite slow in wasm it’s not the memory that’ll bottleneck first!
Small enough to conveniently run directly in the browser via WASM… peterspackman.github.io/mlip.cpp/
mlip.js - ML Interatomic Potentials in the Browser
peterspackman.github.io
November 28, 2025 at 8:48 AM
Small enough to conveniently run directly in the browser via WASM… peterspackman.github.io/mlip.cpp/
This is all using the PET-MAD model (github.com/lab-cosmo/pe...). Wouldn't call my code production ready or anything but it's nice to see what's possible in a web browser.
November 28, 2025 at 5:09 AM
This is all using the PET-MAD model (github.com/lab-cosmo/pe...). Wouldn't call my code production ready or anything but it's nice to see what's possible in a web browser.
If that sounds like something you or someone you know would be interested in, feel free to reach out to [email protected] with any questions you might have. Or chat to me here!
November 13, 2025 at 11:47 PM
If that sounds like something you or someone you know would be interested in, feel free to reach out to [email protected] with any questions you might have. Or chat to me here!
Broadly, interactions of small hydrocarbons, nitriles etc. improved sampling methods for crystal structure prediction, and modelling to predict stabilities in growth conditions.
We've got a really nice group of people here at Curtin - and I think this is great opportunity to join us!
We've got a really nice group of people here at Curtin - and I think this is great opportunity to join us!
November 13, 2025 at 11:47 PM
Broadly, interactions of small hydrocarbons, nitriles etc. improved sampling methods for crystal structure prediction, and modelling to predict stabilities in growth conditions.
We've got a really nice group of people here at Curtin - and I think this is great opportunity to join us!
We've got a really nice group of people here at Curtin - and I think this is great opportunity to join us!
Given you’re doing NaCl and crystals already then maybe something with solvent? Ion pairing of NaCl in water? Always fun to show how such a “simple” problem is not at all simple… that or surface energies, kink site stuff or finite particle size effects maybe aren’t so intuitive for many
November 11, 2025 at 3:45 AM
Given you’re doing NaCl and crystals already then maybe something with solvent? Ion pairing of NaCl in water? Always fun to show how such a “simple” problem is not at all simple… that or surface energies, kink site stuff or finite particle size effects maybe aren’t so intuitive for many
It would be possible I think! Just a matter of putting in the right potential functions. It never ceases to amaze me how complex behaviour can get with simple models like this!
July 30, 2025 at 1:53 PM
It would be possible I think! Just a matter of putting in the right potential functions. It never ceases to amaze me how complex behaviour can get with simple models like this!