| name | mathematical-modeling |
| description | Use when the problem frame is clear enough to introduce variables, assumptions, and candidate mathematical structures while documenting model tradeoffs and risks. |
Use when
Use this skill when the task has moved past basic clarification and now needs a mathematical representation that can support derivation, estimation, simulation, or proof planning.
Typical signals:
- Variables, symbols, states, or parameters need to be defined.
- Competing model families are possible and the tradeoffs matter.
- Assumptions must be made explicit before proceeding to calculations.
- The likely failure modes of a model need to be called out early.
Inputs
- The analyzed problem statement and success criteria.
- Explicit givens, unresolved uncertainties, and any accepted assumptions.
- Available data, constraints, physical interpretation, and intended use of the model.
Outputs
- A named candidate model or model family with justification.
- A variables and notation section with consistent symbols and meanings.
- A recorded assumptions section, including which assumptions are strong.
- A short tradeoff discussion covering why this model was chosen over alternatives.
- A
Likely failure regions section describing where the model may break down.
Hard rules
- Define variables and notation before using them in equations or prose.
- Record every modeling assumption explicitly.
- Mark strong assumptions as
strong assumption where they appear.
- State the main tradeoff behind the selected model.
- Name at least one likely failure region or invalid operating regime.
- Do not present a convenient model as ground truth.
Process
- State the modeling goal: prediction, explanation, optimization, approximation, control, or another target.
- Define variables, parameters, units, and notation with stable names.
- Propose one primary model and, when useful, one alternative model family.
- Record assumptions, distinguishing routine simplifications from strong assumptions.
- Explain the model selection tradeoff: realism vs tractability, interpretability vs fidelity, deterministic vs stochastic, continuous vs discrete, or similar.
- Mark likely failure regions, edge cases, and conditions where the model should not be trusted.
Variable and notation guidance
- Reuse standard notation only when it helps clarity in the stated domain.
- Prefer one symbol per concept and one concept per symbol.
- Include units or dimensions when they matter.
- Avoid introducing notation that is not used by the selected model.
Assumption recording rules
- Keep assumptions in their own section.
- If an assumption is convenient but weakly justified, say so directly.
- If an assumption materially narrows applicability, label it
strong assumption.
- If the model depends on unavailable data or calibration, say what is missing.
Model selection tradeoffs
- Favor the simplest model that can still answer the stated question.
- Prefer a richer model only when the extra complexity changes decisions or conclusions.
- If two models are viable, explain why one is the current default and what would trigger switching.
Failure regions
- Nonlinear effects ignored by a linear approximation.
- Boundary regimes, sparse data, extrapolation, or scale changes.
- Sensitivity to parameters that are uncertain or unobserved.
- Violations of independence, stationarity, equilibrium, or homogeneity assumptions.
Handoff guidance
- Move to planning or execution only after the notation, assumptions, and failure regions are explicit enough that downstream work can test the model instead of rediscovering it.