computational-math-domain-skill

name	computational-math-domain-skill
description	Use when classifying a computational math repository or task into a broad domain before choosing specialist Skills, references, runtimes, and validation evidence.

Computational Math Domain Skill

This Skill routes a user goal, paper code repository, or local source tree into a broad computational math domain. It keeps the project from creating one Skill per domain too early: most domains begin as reference cards here, and only become standalone specialist Skills after they need their own workflow, scripts, failure modes, and evals.

For end-to-end reproduction workflows, this Skill is selected by computational_math_reproduction_workflow_skill before deeper specialist Skills.

When To Use

• A repository may be computational math, but the specific domain is unclear.
• The user asks for reproduction beyond continuous optimization.
• The source includes MATLAB, Python, Julia, C++, or mixed-language numerical code.
• A run plan needs domain-specific evidence or validation criteria before execution.

When Not To Use

• The domain is already known and a mature specialist Skill exists.
• The task is only runtime setup, such as MATLAB MCP configuration. Use matlab_runtime_skill for MATLAB execution boundaries.

Domain Routing

Read references/INDEX.md first. It routes observable signals to compact domain cards:

• continuous optimization;
• numerical linear algebra;
• differential equations;
• PDE/FEM;
• stochastic simulation;
• inverse problems.

Continuous optimization is currently the first mature specialist domain. If the index routes to continuous optimization with enough evidence, continue with continuous_optimization_skill.

Workflow

1. Inspect README, paper notes, scripts, dependency files, example commands, and output names.
2. Use references/INDEX.md to identify candidate domains.
3. Read only the relevant domain cards.
4. Record evidence, uncertainty, expected validation signals, and likely runtime requirements.
5. Route to mature specialist Skills when available; otherwise keep the domain card as guidance for repo_reproduction_skill, failure_diagnosis_skill, visualization_skill, and report_generation_skill.

Output Contract

In conversation or in plan.md, summarize:

• candidate domain;
• source evidence;
• likely runtime backend;
• expected evidence for reproduction;
• domain-specific failure risks;
• whether a mature specialist Skill exists.

If evidence is ambiguous, ask for human review before treating the classification as settled.