COSMO Lab

@labcosmo.bsky.social

1324 Followers 195 Following

Computational Science and Modelling of materials and molecules at the atomic-scale, with machine learning.

chemiscope 1.0 can be used online, in jubyter notebooks, in documentation pages and in streamlit apps

Snapshot of a ring-polymer instanton calculation for the reaction of methane with a H radical

Interpolative and extrapolative performance of different architectures for intermediate H cluster densities, with and without fragments included into training.

$Changes in the learned expansion as the fraction of fragments added to the training set is increased$

Body-ordered energy decomposition for low and high-density hydrogen clusters

The body ordered expansion, equations

Sergey collects the edmx doctoral distinction prize

Cluster highlights in chemiscope 1.0 RC3

Pareto front for PET-OMAT models

Table showing results of a few representative universal model on the matbench leaderboard

Screenshot of the matbench discovery leaderboard as of 14.01.2026, showing a PET based model in the top position

Zooming in on a large-scale dataset to showcase the new adaptive resolution features in chemiscope

A cheeshire cat sitting on a tree overlooking a winter landscape, with snapshots of the many classes of materials that the PET-MAD universal interatomic potential can be used for

example of a streamlit app integrating a chemiscope viewer

Features reconstruction errors between the latent spaces of several universal MLIPs

error plots for the PET-MAD-DOS model on different datasets

Reaction energies of pristine and reconstructed surfaces with water

Surface energy diagram of LPS before and after reconstruction

A cartoon explaining how mild finite-temperature conditions induce disorder and dynamical reconstruction on the surfaces of lithium thiophosphates

A metatensor.org sticker on top of the mattehorn

Long-time relaxation of the potential energy in glassy GeTe

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

Training curves for "conservative fine tuning" a PET model

code snippet for a PLUMED input reading
```
histo: METATOMIC ...
MODEL=histo-cv.pt
EXTENSIONS_DIRECTORY=./extensions/
SPECIES1=1-38
SPECIES_TO_TYPES=18
...

cv1: SELECT_COMPONENTS ARG=histo COMPONENTS=1
cv2: SELECT_COMPONENTS ARG=histo COMPONENTS=2
```

snapshot of a trajectory of a Ar38 cluster undergoing a transition from the fcc to the icosahedral minima

Demonstration of multiple time stepping with non-conservative forces

A plot showing the speedup achieved by conservative fine tuning

A scheme explaining why the extended receptive radius afforded for the derivatives of an interatomic potential don't happen when doing direct force prediction.

A plot of energy conservation for conservative models (that work) and non-conservative models with direct forces (that blow up)

Darth Vader dressed as Santa presents a non-conservative forcefield

Scheme of the GNN architecture of the FlashMD method.

Schematic representation of the lambda-MCoV architecture for predicting tensorial quantities using scalar approximators

Plot of the potential and total energy trends along a MD trajectory, showing drift for non-conservative forces, and how that is fixed using multiple time stepping.

A schematic of the functioning of a ML/QM hybrid framework

Nuclear quantum effects in water

Group picture of the Laboratory of Computational Science and Modeling in 2025

Header for the atomistic-cookbook.org recipe for the PET-MAD universal potential