Module problem 

Declarative optimization problem builder.

Author: Moussa Leblouba Date: 5 March 2026 Modified: 2 May 2026

Structs

Constraint

A single constraint with optional gradient.

LinearConstraint

A linear constraint: a^T x {<=, =} b.

OptimProblem

Declarative optimization problem.

Enums

ConstraintKind

Kind of constraint: equality h(x)=0 or inequality g(x)<=0.

ObjectiveKind

Whether the objective is a scalar function or a least-squares residual.

ProblemHint

Hint about problem structure for closure-based objectives.

VarType

Variable type for mixed-integer problems.

Functions

finite_diff_gradient

Compute a finite-difference gradient using the canonical central FD step cbrt(S::EPSILON) * (1 + |x|), which scales with |x| to remain valid for large-magnitude parameters. The additive scaling prevents silent zero-gradient failures that occur when an unscaled step falls below the precision floor of x + h == x (manifesting for |x| > ~5e7 at f64 precision).

finite_diff_jacobian

Compute a finite-difference Jacobian of residual r at x using the canonical central FD step cbrt(S::EPSILON) * (1 + |x|), which scales with |x| to remain valid for large-magnitude parameters. The additive scaling prevents silent zero-Jacobian failures that occur when an unscaled step falls below the precision floor of x + h == x (manifesting for |x| > ~5e7 at f64 precision).

Type Aliases

ScalarFn

Boxed scalar function &[S] -> S.

VectorFn

Boxed vector function (&[S], &mut [S]).