Design Problem Authoring
This section covers problem authoring for the design command. Design problems extend the base problem dictionary with DesignProblem keys and can follow two distinct workflows:
- Inverse FORM: compute partial factors/design point values that satisfy a target reliability (
TargetBetaorTargetPf). - Objective optimization: compute design variable values that minimize an objective while satisfying reliability constraints.
When To Use Which Design Type
- Use Inverse FORM when code calibration, load/resistance factor calibration, or characteristic-value back-calculation is the goal.
- Use objective optimization when cost/weight/performance minimization is the goal and reliability is one of the constraints.
Design Documentation Layout
problem-authoring/design/design-keys.md: fullDesignProblemdictionary reference and validation rules.problem-authoring/design/inverse-form.md: worked Inverse FORM examples (AT625single-case andAT624multi-case).problem-authoring/design/optimization.md: worked objective optimization example (Sorensen81Problem.py) including additional optimization keys.
Notes On Cases
- A design problem can define a single default case or multiple named cases.
- This applies to both workflows: Inverse FORM and objective optimization.
- If
DesignProblem.casesis omitted, Reliafy creates adefaultcase from the top-levelStochasticVariablesdefinition. - Multi-case definitions are useful when target reliability remains fixed but statistical definitions or design context vary per scenario.