# PolymathicAI/the_well

A 15TB Collection of Physics Simulation Datasets

Repository: https://github.com/PolymathicAI/the_well
Canonical: https://ross.abutalabs.com/products/the_well
Homepage: https://polymathic-ai.org/the_well/
Language: Jupyter Notebook
License: BSD-3-Clause
License Family: permissive
Last push: 2026-07-23T14:39:04+00:00

## Health v2 (maintenance only)
Score: 62/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 94, release rhythm 32, longevity 45
- inputs: {"age_days": 639, "days_push": 41, "days_rel": 295, "gap_med": 118, "n_releases_24m": 4}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 4389, forks 564 (observed 2026-08-28T04:08:47.211473+00:00)

## What it is
The Well is a 15TB collection of machine learning datasets containing numerical simulations of spatiotemporal physical systems, spanning 16 datasets across fluid dynamics, biology, acoustics, and astrophysics. It ships with a PyTorch-based Python package (WellDataset) for loading the data into training pipelines and serves as a benchmark suite for ML and computational science research.

## Use cases
- train neural surrogates for physics simulations
- benchmark machine learning models on spatiotemporal PDE data
- download large-scale fluid dynamics simulation datasets
- get datasets for magneto-hydrodynamics and astrophysics ML
- load physics simulation data into a PyTorch DataLoader
- find training data for learned weather or supernova simulators

## When to choose
- you need large, diverse, standardized physics simulation data for training or evaluating deep learning models
- you want a benchmark suite for spatiotemporal ML research
- you work in PyTorch and want a ready-made dataset API

## When to avoid
- you lack the storage or compute for 15TB-scale data
- you need small, lightweight toy datasets for quick experiments
- you need real-world sensor data rather than numerical simulations

## Facets
- artifact type: dataset
- maturity: active
- function: machine-learning, data-science, simulation, benchmarking
- domain: machine-learning, data-science, deep-learning
- platform: python, cross-platform
- tags: physics-simulation, spatiotemporal-data, pytorch, scientific-datasets, pde, fluid-dynamics, benchmark-suite, huggingface, physics, gpu

## Member repositories
- PolymathicAI/the_well (main) score 62

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:08:47.211473+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T18:21:13.017898+00:00, confidence not recorded.
  - readme: https://github.com/PolymathicAI/the_well (fetched 2026-08-28T04:08:47.211473+00:00, sha a333b6d268c3)
  - homepage: https://polymathic-ai.org/the_well/ (fetched 2026-08-29T09:09:16.331335+00:00, sha 192994d3cf6c)
  - registry_pypi: https://pypi.org/pypi/the_well/json (fetched 2026-08-29T09:09:16.340588+00:00, sha c92027735955)
- Data as of 2026-08-30T08:39:29.467469+00:00.