Agent objective
- Interact with the supplied stateful environment.
- Produce verifier-checkable actions or artifacts.
- Maximise scalar reward under the package contract.
Five research mathematics environments built around exact nuisance equivalences: familiar summaries are identical across candidate systems, so agents must discover which advanced invariant actually carries information. The suite spans cellular sheaves, Schrödinger bridges, inverse operators, algebraic varieties, and random-matrix spectra with strict budgets and sealed generalization.
Five research mathematics environments built around exact nuisance equivalences: familiar summaries are identical across candidate systems, so agents must discover which advanced invariant actually carries information. The suite spans cellular sheaves, Schrödinger bridges, inverse operators, algebraic varieties, and random-matrix spectra with strict budgets and sealed generalization.
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 |
|---|---|---|
| SheafHolonomy | Cellular sheaves and connection geometry | Combine invariant and anchored probes to determine hidden cycle geometry when local edge spectra are exact decoys. |
| SchrodingerBridge | Entropic transport and reciprocal processes | Select coupling, potential, barycentric, and bridge measurements across scales to recover geometry hidden by matching marginals. |
| OperatorTomography | Inverse PDEs and spectral geometry | Infer eigenspace alignment with forces, sensors, and boundaries although every candidate shares the full eigenvalue spectrum. |
| VarietyChow | Computational algebraic geometry | Use mixed algebraic probes to uncover coordinate relationships hidden by identical degree, Hilbert, marginal, and single-coordinate data. |
| FreeSpectrum | Free probability and random-matrix theory | Combine analytic transforms, edge measurements, and perturbation response to distinguish laws matched on mass and low moments. |
The public-contract evaluation reached 0.7206 mean reward and 24/40 passes, solving all evaluated SheafHolonomy, OperatorTomography, and VarietyChow episodes but none in two other families. The report openly traces that split to missing public likelihood parameters, while an evaluator-side control passed 40/40. This makes the package unusually transparent about public-agent identifiability versus simulator solvability. Validation passed 77/77 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.
40 reported runs.
ulam_rlvr_score_report.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
4846fa16b7cd86e882bb06c9f6801f0e48330df760aa815f70f246754d76c9a0One 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.