Agent objective
- Interact with the supplied stateful environment.
- Produce verifier-checkable actions or artifacts.
- Maximise scalar reward under the package contract.
Five omega-tier mathematics environments with 512 seed-shuffled candidates and only two noisy measurements. Across Stokes phenomena, decoder pseudocodewords, singular learning theory, higher associativity, and flag geometry, agents must escape exact seed equivalence, reason through nonlinear interactions, and submit calibrated posteriors that transfer to unseen mathematical functionals.
Five omega-tier mathematics environments with 512 seed-shuffled candidates and only two noisy measurements. Across Stokes phenomena, decoder pseudocodewords, singular learning theory, higher associativity, and flag geometry, agents must escape exact seed equivalence, reason through nonlinear interactions, and submit calibrated posteriors that transfer to unseen mathematical functionals.
Enough detail to understand the intellectual terrain; generated instances, hidden mechanisms, and solution paths remain inside the private package.
| Environment | Mathematical or technical frontier | Adaptive research problem |
|---|---|---|
| StokesLabyrinth | Irregular equations and resurgence | Choose continuation measurements with complementary sector and vector nullspaces beyond the shared formal normal form. |
| DecoderPseudocodeword | Iterative decoding and trapping-set geometry | Probe soft beliefs, syndrome landscapes, and Jacobians when all candidates decode clean and single-bit cases identically. |
| RLCTScanner | Singular learning and resolution geometry | Infer singular risk structure through finite-temperature evidence, moments, sectors, and susceptibility beyond identical local derivatives. |
| AssociahedronAudit | Operads and higher associativity | Compare parenthesizations and coherence networks away from a shared unit and tangent algebra. |
| FlagManifoldXRay | Grassmannians, flags, Plücker geometry | Use externally marked orientation probes when dimensions, projector spectra, and intrinsic nesting all agree. |
An isolated tool-using copy policy ran 60 fresh expert episodes and averaged 0.7047, with only 3 strict passes despite 0.8706 experiment design and 0.8532 held-out prediction. DecoderPseudocodeword was especially hard, while FlagManifoldXRay showed stronger identification. The result demonstrates that two good probes do not automatically produce a calibrated 512-way posterior. This is a tool policy, not a separately hosted model rollout.
We publish aggregate behavior and task structure, while withholding generated instances, hidden labels, exact successful probes, private checks, and solution trajectories.
Shown with its provenance and limitations; it is not a performance guarantee.
Reported result from the evaluation artifact supplied with this package.
As identified by the supplied artifact.
60 reported runs.
independent_copy_agent_report.md
Machine-readable provenance and the exact displayed metric are available in results.json.
The paid ZIP will live in a private R2 bucket. Vercel authorizes the buyer and issues a 2–5 minute object URL; R2 serves the bytes directly.
Authenticated buyer + entitlement check
+ private R2 object + 2–5 minute signed URL
= direct, auditable download
Package SHA-256
3f84ba8779a6430d7b4f70139d301774f6d9d099a8dc35fac5813373b6c3d03fOne purchase licenses this identified item to one legal organisation for worldwide, perpetual commercial model training, evaluation, research and development. Redistribution and resale of the package are not permitted.