Verification & Validation
Checking the code is right, and checking it matches reality.
- V&V
- Solved right + matches reality
Verification asks whether the models are solved correctly; validation asks whether they match experiment. Kronos does both — reproducible engines for verification, benchmarking against the multi-machine database for validation — and reports each honestly.
The distinction matters: a reproducible number is not automatically a validated one.
Common questions
Verification checks that a model solves its equations correctly; validation checks that those equations match reality. Both are needed before a computed number can be trusted for design.
By anchoring to the experimental database, benchmarking engines against known results, and publishing the engines openly so others can verify them — reproducibility plus benchmarking, not assertion.