Works with the AI Agents you already use
mcm math modeling
An end-to-end mathematical modeling competition pipeline for MCM/ICM and CUMCM with checkpoint-enforced quality gates.
Free
mcm math modeling
Example session with this skill installed
I have the 2025 CUMCM Problem A PDF. I want to use LaTeX for the paper. Please analyze the problem, suggest models for each sub-problem, and start the workflow.
- Read your context and instructions
- Compiled the mcm math modeling
I have initialized the workflow for CUMCM Problem A. I have read the problem statement, identified 3 sub-problems, and verified the checkpoint script verify_checkpoint.py.
Modeling Analysis and Preferences
The project directory has been structured with plan.md and todo.md. I have performed a deep reading to identify distinctive constraints regarding the geometric parameters and physical intuition required for this dynamics problem.
| Category | Selection |
|---|---|
| Contest | CUMCM (National Undergraduate Mathematical Contest in Modeling) |
| Typesetting | LaTeX (using xelatex and cumcm template) |
| Language | Chinese (Standard for CUMCM) |
| Sub-problems | 3 (identified from PDF pages 1-4) |
Constraint Traceability Matrix (Partial)
| # | Requirement | Type | Mathematical Expression | Implementation |
|---|---|---|---|---|
| C1 | 掉头路径为两段相切圆弧 | Given Method | $R_1, R_2$ tangency | geometry.py: check_tangent() |
| C2 | 前弧半径:后弧半径=2:1 | Parameter | $R_1 = 2R_2$ | solver.py: radius_ratio |
| C3 | 边界最大偏离距离 < 0.5m | Constraint | $|P_{track} - P_{bound}| < 0.5$ | problem.py: bound_penalty |
Proposed Model Comparison
Sub-problem 1 (Trajectory): Analytic Geometry vs. B-Spline. Recommendation: Analytic Geometry (⭐⭐⭐) due to strict circular arc requirements.
2.
Sub-problem 2 (Optimization): Genetic Algorithm vs. Particle Swarm vs. Sequential Quadratic Programming. Recommendation: SQP (⭐⭐⭐) for high-precision local convergence.
Next steps
- Review the Constraint Traceability Matrix in
reports/ANALYSIS_MODELING_REPORT.md. - Confirm the Model Selection for Sub-problem 1 to trigger the solver implementation.
- Run
python scripts/verify_checkpoint.py cp1once selections are finalized.
Connects securely to your tools. The creator never sees your data.
About this skill
Winning a mathematical modeling competition requires more than just functional code. It demands rigorous model selection, constraint traceability, and high-density documentation that meets contest standards. This skill provides an end-to-end workflow for contests like MCM/ICM (COMAP), CUMCM (国赛), and other regional events, transforming a problem statement into a submission-ready paper.
What it does
- Problem analysis lists every sub-problem and constraint to prevent the 15-20 point penalties common in contest failures.
- Model selection compares multiple candidate methods (e.g., optimization, statistics, mechanics) before implementation begins.
- Authentic solving executes real numerical solvers via NumPy and SciPy, generating logs that prove mathematical validity.
- Figure generation produces a diverse set of at least 12-15 data-driven and conceptual diagrams using Matplotlib and Draw.io.
- Paper typesetting handles professional formatting in LaTeX or Typst using 34 specific contest templates.
- Checkpoint enforcement runs executable Python scripts to verify modeling integrity, solver consistency, and paper density before proceeding.
How it works
- Stage 0 initializes the project, sets preferences for typesetting engines, and establishes the specific contest rules.
- Stage 1 extracts the problem statement, builds a constraint traceability matrix, and selects the mathematical models.
- Stage 2 implements solvers, performs sensitivity analysis, and generates primary data visualizations.
- Stage 3 compiles the paper, populating sections with technical roadmaps and abstract versions.
- Stage 4 executes final verification scripts to catch layout errors, numerical inconsistencies, or AI-residue leaks.
Frameworks & tools
The skill utilizes Python (NumPy, SciPy, Matplotlib), LaTeX (XeLaTeX), Typst, and Draw.io. It includes scripts for checkpoint verification, solver templates (GA/PSO/SA), and paper auditing.
Why this beats prompting it yourself
Generic prompts often lead to "thin" papers with insufficient figure density or hallucinated citations that reviewers immediately flag. This skill enforces domain-specific norms, such as ensuring 65% figure coverage and mandatory sensitivity analysis, which are the hallmarks of O-award and F-award submissions.
Use cases
- Drafting a full MCM/ICM paper including the summary sheet and technical roadmap.
- Validating the feasibility of an optimization model against a complex CUMCM constraint set.
- Generating a complete set of Nature-style figures and sensitivity plots for a dynamics problem.
- Auditing a draft paper for compliance with COMAP formatting and AI disclosure rules.
Known limitations
Does not support general data analysis without a competition framing. Requires local installation of TeX Live or Typst for paper compilation. Visual inspection of PDF layout requires an agent with vision capabilities.
Requires Python 3.10+, LaTeX/Typst for compilation, and basic shell access. Best results for MCM/ICM and CUMCM contests. Cannot verify PDF layout without agent vision capabilities.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 13 days ago
- Free with an Agensi account
Requires