README.md

metadata

license: cc-by-nc-nd-4.0

OpenOAT: A Massive Large Minimum Compliance Topology Optimization Dataset

OpenOAT is a dataset of random FEA problems solved using a SIMP optimizer and includes fully random resolution and aspect ratio problems. The dataset has three parts:

1. Pre-Training Data: 2M sample topologies without labeled boundary condition and loads.
2. Labeled Data: 700K samples(also in the pre-training) with labeled boundary conditions and loads and volume fraction:
3. Test Data: 15K labeled samples not present in either prior datasets for testing.