Skip to content

Dataset · AI4Science-WestlakeU

RealPDEBench: Real Measurements and Physics Simulations

AI4Science-WestlakeU/RealPDEBench

Paired real-world physical measurements and numerical simulations of complex physical systems for forecasting and sim-to-real transfer.

What's inside

A repository release of the RealPDEBench benchmark by Westlake University for time-series forecasting tasks. It contains paired data of actual physical measurements and corresponding numerical simulations (including distribution in HDF5 format). The dataset is designed to evaluate spatio-temporal forecasting and sim-to-real model transfer.

What you can do with it

  • Benchmark neural networks for solving partial differential equations (PDEs)
  • Test methods for transferring physics models from simulation to real-world data
Tasks
Time series forecasting
Downloads this month
32.6K
Likes
10

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("AI4Science-WestlakeU/RealPDEBench")