Research

replay measurements, gated models, and their documented limits

The measured work behind one pipeline: replay existing software trajectories, recover the state they produce, and compile demonstrated outcomes into objective tasks and checks. Negative results and exclusions are published beside the positive measurements.

• The state of the art in fluid-dynamics foundation models, in common unitsAug 2026 · 46 published rows · 7 conversions, 39 MISSING
• What CFD-surrogate benchmarks actually measureAug 2026 · convention, spread, offset, mesh
• Watch it compilelive demonstration · sealed model
• The oracle: a verifier compiled from one recorded steplive artifact
• What a command does: four pre-registered resultsAug 2026 · internal
• Recovering next state from public software runsJul 2026
• Reading Frozen World Models Through Action Effects (PDF)Jul 2026