Learn · Project evaluation
Axeyum for reviewers and research leads
Axeyum publishes countable claims about proof status, trust, evidence routes, comparison coverage, and automated production.
a103b3db3 Measured Source Axeyum fact ledger and implementation documentation After reading this guide, you should be able to
Learning objectives
- Identify the parts of Axeyum that are relevant to project evaluation.
- Distinguish implemented results from the current research frontier.
- Open a lesson or artifact that provides the underlying definitions and evidence.
Current state
Which claims can an external reviewer verify directly?
The ledger separates proved, computed, conjectured, open, and refuted facts. It records proof routes and computes axiom footprints for kernel facts. Benchmark pages report the committed populations, tool versions, limits, decisions, and disagreements used in solver comparisons.
The automated library process has completed a full select, produce, check, admit, and reschedule cycle on a recorded fact. Repeated autonomous production remains a current engineering target rather than a completed capability.
Why it matters
What the current work makes possible
- Every site-wide figure names a command or committed source file.
- Open and refuted facts remain visible beside proved facts.
- Imported proofs, kernel proofs, solver certificates, and CAS certificates have separate route labels.
- Individual artifact pages expose statements, commands, evidence notes, and dependency graphs.
Results to inspect
Open the evidence
What comes next
Extend the present base
The next evaluation work is longitudinal: measure how often automated runs retire open facts, how much human intervention each route needs, and whether later facts reuse the admitted result.
Independent replay and statement-quality checks can then grow alongside raw theorem counts, so increased production does not weaken the meaning of a proved fact.
Where to begin