Quantum-accurate chemistry, right in your browser.
SimpleAtom is a free, open interface to MACE machine-learning force fields. Compute energies, relax geometries, and run molecular dynamics on your molecules and materials — with no installation, no command line, and no supercomputer account.
Built on MACE, the open-source machine-learning potential developed at the University of Cambridge.
- Free
- Open source (MIT)
- No account required
- Runs entirely in your browser
Live in your browser · ethanol, C₂H₅OH
From structure to insight in four steps
A guided workflow that mirrors how computational chemists actually work — without the setup that usually stands in the way.
- 01
Provide a structure
Upload XYZ, CIF, POSCAR, or PDB files, paste a SMILES string, sketch a molecule, or pick from a curated benchmark catalog.
- 02
Choose a model & calculation
Select a MACE foundation model (or your own fine-tuned checkpoint), then a single-point energy, geometry optimization, or molecular-dynamics run.
- 03
Run the calculation
The structure is evaluated with ASE and mace-torch on managed compute — no local install, no queue script, no cluster account.
- 04
Explore & share results
Inspect energies, forces, and trajectories in interactive 3D and publication-quality charts, then export or share a permanent link.
Built for real computational chemistry
The methods practitioners rely on, wrapped in an interface that anyone can pick up — from experimentalists to students.
Foundation models
Run MACE-MP-0 and MACE-OFF out of the box, or upload a fine-tuned .model checkpoint for a custom potential.
Geometry optimization
Relax structures to a local minimum with BFGS, converging forces to a target threshold in eV/Å.
Molecular dynamics
Propagate NVE, NVT (Langevin), or NPT trajectories with Maxwell–Boltzmann initial velocities and femtosecond timesteps.
Interactive 3D structures
Visualize geometries and per-atom force vectors, and animate MD trajectories with dual rendering engines.
Model benchmarking
Compare foundation models and your own checkpoints across reference structures on energy and force error metrics.
Reproducible sharing
Every result can be exported (CSV, JSON, PDF) or published as a permanent, read-only link for collaborators.
Two foundation models, one interface
MACE ships pre-trained potentials for distinct chemical domains. SimpleAtom picks sensible defaults, and explains the trade-offs so you choose deliberately.
MACE-MP-0
Materials & inorganic chemistry
Trained on PBE+U DFT (Materials Project)
- Crystals, surfaces, and bulk systems across 89 elements
- Periodic boundary conditions and cell-aware dynamics
- Optional D3 dispersion correction
MACE-OFF
Organic molecules
Trained on ωB97M-D3BJ (transferable organic chemistry)
- Neutral organic molecules (H, C, N, O, F, P, S, Cl, Br, I)
- Dispersion already included — no separate D3 term
- Ideal for conformers, drug-like molecules, and reactions
“Computational chemistry has long lived on remote clusters, behind the command line. SimpleAtom moves the first step into the browser — so a good idea doesn’t have to wait on an installation.”
Designed for accessibility, reproducibility, and teaching
Run your first calculation
Start with a demo molecule or bring your own structure. No account, no setup — results in seconds.