capwitf/My-MathModeling-skills / math-model
math-model
VerifiedUse when 高教社杯/CUMCM 数学建模任务需要变量、假设、公式、约束、算法、验证设计、鲁棒性规划、灵敏度分析、结果 schema,或基于题面和本地证据的代码交接。
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 7.1 KB |
| agents/openai.yaml | config | 318 B |
| references/robustness-protocol.md | markdown | 3.1 KB |
references/robustness-protocol.md
View on GitHub# Robustness Protocol
Use this reference when a modeling route needs a validation tier or robustness plan. Keep the plan proportional: direct accounting should not receive a full sensitivity study, and fragile optimization or ranking claims should not pass with only a balance check.
## Tier Selection
| Situation | Minimum tier | Required evidence |
| --- | --- | --- |
| Direct formula accounting, deterministic unit conversion, or data transcription | `direct-check` | unit closure, formula back-substitution, balance residuals, boundary sanity |
| Required answer spans several given scenarios, regions, periods, or cases | `scenario-check` | all-case coverage, spread/ranking/pass-fail table, representative figure when useful |
| Main claim depends on uncertain parameters, weights, thresholds, calibrated values, or selected assumptions | `robustness-required` | sensitivity/stress table, tested range, stability criterion, downgrade rule |
| Solver, heuristic, stochastic, or tolerance-sensitive result supports a claim | `robustness-required` | tolerance or seed/restart evidence, solver gap/status, constraint violation summary |
| Exploratory probe used only to understand risk | `diagnostic-only` | diagnostic table/log, explicit ban from abstract and conclusion |
## Route Patterns
| Model route | Robustness checks to design |
| --- | --- |
| Linear/MILP/nonlinear optimization | constraint residuals, solver status/gap, tolerance stress, key cost/capacity/bound perturbation, baseline comparison |
| Heuristic or stochastic search | fixed seed record, multi-seed/restart distribution, convergence trace, best-vs-baseline comparison |
| Ranking, evaluation, TOPSIS, AHP, entropy weights | weight perturbation, indicator deletion, rank flip count, threshold margin |
| Prediction feeding a decision | train/validation separation, forecast error interval, error-to-scenario propagation, decision-impact table |
| Geometry, physics, or simulation | step-size or grid sensitivity, boundary/extreme cases, key size/angle/position perturbation |
| Monte Carlo convergence | pilot run variance, confidence half-width or relative half-width, seed-block or batch-means stability, replication increase rule |
| Calibrated simulation | parameter fit table, calibration error, holdout or back-test, decision impact of fitted-parameter uncertainty |
| Policy or system impact analysis | scale/penetration stress, synchronized vs diversified scenario, peak/ramp/shortage pressure metrics |
| Pure explanation or proof-style subquestion | boundary counterexample search, assumption-loss statement, formula back-substitution |
## Output Contract
Name robustness outputs semantically:
- `qX_sensitivity_summary.csv`
- `qX_robustness_summary.csv`
- `qX_stress_scenarios.csv`
- `qX_ranking_stability.csv`
- `qX_solver_diagnostics.csv`
Each table should include:
```csv
run_id,subquestion,check_id,baseline_case,stress_case,parameter_or_assumption,tested_range,metric,baseline_value,stress_value,unit,stability_criterion,status,validation_status,risk_note
```
Use `paper_role=robustness` only when the check is broad enough to support a paper claim. Narrow probes stay `diagnostic-only`.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack." | Stars: 197k | Updated | Claude CodeCodex | Verified | Install |
| ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |