Multi-dimensional optimisation of non-convex functions.

• Uses LIPO algorithm described in [8].

Types

• using value_type = T
• using argument_type = Vector< T, N >
• using scalar_type = T

Methods

• template <class Func>
solve(argument_type lbound, argument_type ubound, Func func) const -> argument_type
• estimate_constant(bool rhs) -> void
• sample_probability(scalar_type rhs) -> void
• max_iterations(int rhs) -> void
• max_evaluations(int rhs) -> void
• function_precision(value_type rhs) -> void
• argument_precision(argument_type rhs) -> void
• constant(value_type rhs) -> void
• estimate_constant() const -> bool

  Estimate Lipschitz constant or not.

• sample_probability() const -> scalar_type

  Probability of choosing exploration step over exploitation step.

• max_iterations() const -> int

  Maximum number of iterations of the main loop.

• max_evaluations() const -> int

  Maximum number of function evaluations.

• function_precision() const -> value_type

  Function precision.

• argument_precision() const -> argument_type

  Argument precision.

• constant() const -> value_type

  Lipschitz constant (one for each dimension).