Skip to content

capwitf/My-MathModeling-skills / math-hub

math-hub

Verified

Use when 高教社杯/CUMCM 数学建模项目需要竞赛级范围控制、轻量 hub 状态、当前分问交付物锁定、文献启发式解题路线、结论/证据一致性、数学核验、结果一致性审查、官方规则合规、AI 使用披露跟踪、最终提交门禁或明确任务派发。

references/a-problem-pattern-library.md

View on GitHub
# A-Problem Pattern Library

## Purpose

Use this library during A-problem triage after reading the actual problem statement. It gives repeatable patterns, not final answers. Always derive deliverables, rules, data fields, and constraints from the current contest problem.

For 2024 bench-dragon examples and 2023 heliostat examples, read [a-problem-2024-casebank.md](a-problem-2024-casebank.md) as supporting evidence, not as a template to copy.

## No-Match / Novel Mechanism Fallback

If the current A problem does not clearly match one of the patterns below, do not force a nearest pattern. Use `new_mechanism_route` and mark the first route as `low-confidence` until the current problem creates its own mechanism map.

When using `new_mechanism_route`:

- 不得硬选主模式;不要为了让流程继续而强行归类。
- 从交付物反推: list the exact required tables, figures, decisions, parameters, or prose answers before naming a model.
- Identify the real-world mechanism in one sentence: conservation, geometry, motion, material response, signal extraction, economics, queueing, control, policy, or another problem-specific mechanism.
- Build a baseline from the simplest defensible accounting, physical relation, simulation, or direct calculation.
- Define the first validation action before advanced modeling: unit closure, boundary case, back-substitution, brute-force small case, visual reconstruction, or scenario stress.
- Only after this check may a new primary pattern be written into the project notes. Do not present a 2023/2024 pattern as a universal future-year rule.

## Pattern 1: Geometric Motion and Collision

| Item | Guidance |
| --- | --- |
| 题面信号 | moving linked bodies, curves, collision, turning space, path width, speed limits, rigid geometry |
| 第一版基准模型 | coordinate frame + path equation + recursive positions + simple distance or overlap check |
| 高分主模型 | geometry-aware recurrence with explicit frame transforms, candidate collision pruning, exact or high-precision threshold search, and complete state tables |
| 必要结果表 | time/entity position table, speed table, first-constraint-hit table, parameter-search table, final feasible solution table |
| 必要验证 | boundary cases, visual state reconstruction, threshold crossing evidence, smaller brute-force check, speed/position consistency |
| 常见扣分坑 | no coordinate frame, incomplete result table, collision drawn but not quantified, missing units, "first collision" asserted without proof |
| 可生成创新点 | collision candidate pruning, dynamic search, curve-segment state classifier, speed scaling relation, threshold-crossing ledger |
| 适合图表 | path reconstruction, local collision zoom, distance-to-threshold curve, speed heatmap, parameter search curve |

## Pattern 2: Physical Field or Efficiency Decomposition

| Item | Guidance |
| --- | --- |
| 题面信号 | heat, light, force, flow, diffusion, energy, efficiency, power, concentration, pressure |
| 第一版基准模型 | conservation/balance equation + component efficiency factors + direct numerical evaluation |
| 高分主模型 | mechanism decomposition, coordinate transforms, shadow/blocking or loss modeling, time/space aggregation, and output-power registry |
| 必要结果表 | component efficiency table, scenario/time summary, object-level metric table, annual/monthly or spatial aggregation table |
| 必要验证 | unit check, extreme sun/angle/boundary case, component contribution comparison, sensitivity to key size/position parameters |
| 常见扣分坑 | efficiency factors named but not derived, hidden geometry, no component contribution table, visual heatmap without units |
| 可生成创新点 | loss decomposition, projection or ray-sampling correction, symmetry-based variable grouping, component ablation |
| 适合图表 | coordinate-frame diagram, component loss stacked bars, spatial heatmap, monthly/seasonal trend, parameter sensitivity curve |

## Pattern 3: High-Dimensional Engineering Optimization

| Item | Guidance |
| --- | --- |
| 题面信号 | many design variables, layout, allocation, sizing, siting, cost/benefit objective, hard feasibility constraints |
| 第一版基准模型 | fixed layout or uniform parameters + feasibility filter + objective evaluation |
| 高分主模型 | decomposition or dimension reduction, constrained search/solver, run record, baseline comparison, and feasibility diagnostics |
| 必要结果表 | design-variable table, objective and constraint table, solver/run record, baseline-vs-final comparison, parameter scan |
| 必要验证 | feasibility check, repeated seeds or deterministic restart, sensitivity to weights/bounds, local perturbation around final design |
| 常见扣分坑 | final design not completely tabulated, no solver status, no baseline denominator, black-box heuristic with no diagnostics |
| 可生成创新点 | symmetry/ring/layer grouping, coarse-to-fine search, adaptive feasible-region pruning, hybrid mechanism + heuristic loop |
| 适合图表 | design layout map, convergence curve, constraint violation curve, objective-vs-parameter curve, feasible-region plot |

## Pattern 4: Prediction to Decision

| Item | Guidance |
| --- | --- |
| 题面信号 | demand/risk/traffic/fault/price forecast followed by scheduling, allocation, routing, or policy |
| 第一版基准模型 | simple forecast + deterministic decision using predicted mean |
| 高分主模型 | uncertainty intervals or scenarios feed into robust/stochastic optimization with decision-impact evaluation |
| 必要结果表 | prediction accuracy table, scenario table, decision table, robustness comparison table |
| 必要验证 | leakage check, train/test split, forecast error propagation, worst-case or percentile performance |
| 常见扣分坑 | forecast and decision are separate chapters, only RMSE reported, no decision impact |
| 可生成创新点 | error-to-scenario conversion, robust constraint, risk-weighted objective, decision-aware metric |
| 适合图表 | prediction interval plot, scenario fan chart, decision sensitivity curve, robust-vs-deterministic comparison |

## Pattern 5: Evaluation, Ranking, and Policy Selection

| Item | Guidance |
| --- | --- |
| 题面信号 | ranking, risk level, comprehensive index, scheme comparison, priority order, policy recommendation |
| 第一版基准模型 | normalized indicators + equal weights + ranking |
| 高分主模型 | indicator provenance, redundancy check, weight sensitivity, rank stability, and actionable recommendation |
| 必要结果表 | indicator dictionary, standardized data table, weight table, ranking table, rank-stability table |
| 必要验证 | direction check, collinearity/redundancy, weight perturbation, top-k stability, expert or baseline comparison |
| 常见扣分坑 | AHP/TOPSIS name-dropping, indicator direction mismatch, ranking changes under tiny weight changes |
| 可生成创新点 | task-specific indicator, stability-weighted ranking, constraint-aware score, explainable top-k policy |
| 适合图表 | ranking bar chart, weight tornado chart, stability heatmap, radar chart only when dimensions are few and units are clear |

## Pattern 6: Discrete Event, Queue, and Reliability Simulation

| Item | Guidance |
| --- | --- |
| 题面信号 | arrival, waiting, service, congestion, repair, emergency response, failure, inventory state |
| 第一版基准模型 | state machine + baseline analytic or simple simulation |
| 高分主模型 | event logic, capacity/mutual-exclusion constraints, Monte Carlo or scenario simulation, optimization or policy comparison |
| 必要结果表 | event table, state-transition table, waiting/utilization table, scenario comparison, seed/run record |
| 必要验证 | multiple seeds, confidence interval, load stress test, conservation or final-state check |
| 常见扣分坑 | state transitions implied by prose, one simulation run, no seed, no steady-state or extreme-load check |
| 可生成创新点 | event-priority rule, adaptive dispatch policy, stress-test scenario design, state compression |
| 适合图表 | event timeline, queue length curve, utilization heatmap, confidence interval plot, scenario comparison table |

## Quick Pattern Selection

Use this output after reading the problem statement:

```text
match_confidence:
主模式:
辅模式:
if no clear match: new_mechanism_route
题面信号:
第一版基准:
高分主模型:
必须结果表:
必须验证:
潜在创新:
推荐图表:
当前阻塞:
```

If a pattern does not force a complete result table and validation action, the pattern was selected too vaguely.
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: 198kUpdated Claude CodeCodexVerifiedInstall
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: 194kUpdated Claude CodeCodexVerifiedInstall
ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report.Stars: 194kUpdated Claude CodeCodexVerifiedInstall
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: 194kUpdated Claude CodeCodexVerifiedInstall