Testing a solver prototype before further development
A completed benchmark compared a combined-solver prototype with two alternatives, checked accepted solutions, and preserved the execution evidence. The result supported narrowing the research claim or redesigning the prototype.
Read the example →Checking a quantum optimization across compiler targets
A preregistered study tested a quantum-circuit optimization against an existing alternative across two compiler toolchains. One resource estimate favored the prototype, while the overall result remained mixed.
Read the example →Reproducing a research result with a separate implementation
A research-assurance pilot fixed an experiment plan before execution, preserved the run history, and checked the result with a separate implementation. Its evidence package met the pilot’s recipient policy and passed the recorded reproduction, mutation, and clean-extraction checks.
Read the example →