Agent objective
- Interact with the supplied stateful environment.
- Produce verifier-checkable actions or artifacts.
- Maximise scalar reward under the package contract.
A five-environment research suite where advanced mathematics becomes a sequence of consequential decisions. Agents navigate higher coherence, Stokes geometry, singular learning, generalized symmetry, and mean-field games while managing noisy evidence, nuisance variables, scarce resources, and irreversible interventions; success requires calibrated beliefs and sealed prediction, not merely naming the right concept.
A five-environment research suite where advanced mathematics becomes a sequence of consequential decisions. Agents navigate higher coherence, Stokes geometry, singular learning, generalized symmetry, and mean-field games while managing noisy evidence, nuisance variables, scarce resources, and irreversible interventions; success requires calibrated beliefs and sealed prediction, not merely naming the right concept.
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 |
|---|---|---|
| HigherCoherenceLab | Weak 2-categories and higher coherence | Reconstruct interacting coherence evidence while deciding when localization and strictification claims are justified. |
| RiemannHilbertStokesLab | Irregular singularities and Stokes geometry | Choose contour coordinates and deformations while avoiding irreversible crossings that destroy later diagnostics. |
| SingularLearningGeometryLab | Singular statistics and resolution geometry | Select a chart, separate non-identifiability from singularity, and judge whether symmetry breaking clarifies or biases inference. |
| GeneralizedSymmetryAnomalyLab | Higher symmetries, anomalies, topological phases | Allocate manifold and defect probes before irreversible gauging or condensation and certify the anomaly mechanism. |
| MeanFieldMasterEquationLab | Mean-field games and master equations | Coordinate finite-population and measure-space probes, adaptive controls, and principled identification or abstention. |
The supplied ten-episode run averaged 0.6627 with no formal passes, despite mean terminal-decision and path scores near 0.80. One episode predicted sealed behavior perfectly but failed confidence and path gates; another had near-perfect posterior and decision scores but poor prediction. These traces show why identification, forecasting, and sequential intervention are independently necessary. The release reports 132/132 tests.
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.
5 reported runs.
ulam_my_scores_and_log.zip:ulam_my_scores_summary.json
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
f3683756e9f51ce96c396c4fb699e42835faaf06d4d705a9e1779b891776c051One 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.