Ross ROSS = Recommend OSS · open-source software intelligence for agents

pdebench/PDEBench resource

PDEBench: An Extensive Benchmark for Scientific Machine Learning observed · 2026-08-28

github.com/pdebench/PDEBench · homepage · Python · NOASSERTION (other) observed · 2026-08-28

Health v2 · maintenance only

56/100

  • Activity 74
  • Release rhythm 8
  • Longevity 100

Flags: no_license

How is this computed?

round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10) — computed 2026-09-02. Adoption (stars, forks) is never an input.

  • gap_med: n/a
  • age_days: 1553
  • days_rel: n/a
  • days_push: 156
  • n_releases_24m: 0

Full methodology

Adoption not part of the score

1191 stars · 154 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

PDEBench is a benchmark suite for scientific machine learning consisting of code and large ready-to-use datasets of time-dependent partial differential equation (PDE) simulation tasks. It includes data generation tools, pretrained baselines (FNO, U-Net, PINN, gradient-based inverse methods), and an extensible API for evaluating new ML models against classical numerical solvers.

Use cases

  • benchmark my new neural operator model on PDE simulation tasks
  • download ready-to-use datasets of Navier-Stokes and other PDE solutions
  • train FNO or U-Net baselines on scientific ML problems
  • evaluate machine learning models for forward and inverse PDE problems
  • generate PDE simulation datasets with varied initial and boundary conditions
  • compare my physics-informed neural network against standard baselines
  • find a standardized benchmark for scientific machine learning research

When to choose

  • you need standardized, challenging PDE benchmarks with large precomputed datasets
  • you want to compare a new SciML model against established baselines like FNO, U-Net, or PINN
  • you need both forward and inverse problem tasks across a wide range of PDEs
  • you want an extensible benchmark with a user-friendly API for the SciML community

When to avoid

  • you need a general-purpose numerical PDE solver rather than an ML benchmark
  • you require a lightweight dataset for quick prototyping on limited compute
  • you need guaranteed compatibility with the very latest Python/JAX/PyTorch versions for all components
  • your domain is not physics simulation or PDE-based modeling

Facets

dataset · maturity active

benchmarking machine-learning simulation data-generation deep-learning machine-learning simulation artificial-intelligence python pde scientific-machine-learning neural-operators pinn navier-stokes fluid-dynamics benchmark-suite jax pytorch neurips-2022 physics scientific-computing linux gpu

7 sources

Member repositories

RepositoryRoleHealth v2
pdebench/PDEBenchmain56

For agents

markdown · JSON · MCP: product_card(name="pdebench/PDEBench")

Data as of 2026-08-30T08:39:29.467469+00:00 · Report a problem