Physics
Run parameter sweeps of thousands of settings at once.
Antinuous runs parameter sweeps of thousands of settings in parallel and fits models to your data with uncertainties. Every run is code you can read and rerun, and it flags new arXiv preprints each week.
Use cases
Build fast stand-ins for slow simulations
Slow runs limit how many designs or parameters you can explore.
You ask
Train a surrogate on our 500 plasma simulation outputs, then use it to infer the input parameters that best match this measured spectrum.
What it does
- Antinuous reads your existing simulation outputs.
- It starts from a workspace with physics machine learning libraries already installed.
- It trains a stand-in model and tests it against held-out runs.
- It saves runs and predictions in a viewer you can open and compare.
You get
A notebook with the trained model, its test error and the inferred parameters.
Past simulation runs train a faster model for design scans and inference.
Who it's for
Computational physicistsRun parameter sweeps and chart every run in one notebook.
Experimental physicistsFit models to measured data with uncertainties.
Theoretical physicistsCheck derivations against known limits in code.
Graduate studentsTrack new arXiv preprints on your questions each week.
Works with
arXiv
GitHub
Slack
Gmail
Google Calendar
Google Slides
PowerPoint