bluemira.optimisation.problem

Interface for defining an optimisation problem.

Classes

`OptimisationProblemBase`	Common base class for OptimisationProblem classes.
`OptimisationProblem`	Interface for an optimisation problem.

Module Contents

class bluemira.optimisation.problem.OptimisationProblemBase

Common base class for OptimisationProblem classes.

__MethodT

__AnyT

_overridden_or_default(f: __MethodT, cls: type[Any], default: __AnyT) → __MethodT | __AnyT

If the given object is not a member of this class return a default.

This can be used to decide whether a function has been overridden or not. Which is useful in this class for the df_objective case, where overriding the method is possible, but not necessary. We want it to appear in the class interface, but we want to be able to tell if it’s been overridden so we can use an approximate gradient if it has not been.

Returns:

overridden or default object

Parameters:

f (__MethodT)
cls (type[Any])
default (__AnyT)

Return type:

__MethodT | __AnyT

static __is_method(f: __MethodT, cls: type[Any]) → bool

Determine if the given method is a member of this base class or not.

Note that f must be a bound method, i.e., it needs the __func__ dunder method.

Returns:

true if the given method is a member of this base class else false

Parameters:

f (__MethodT)
cls (type[Any])

Return type:

bool

class bluemira.optimisation.problem.OptimisationProblem

Bases: abc.ABC, OptimisationProblemBase

Interface for an optimisation problem.

This is an alternative to running an optimisation using the optimise() function.

Using this interface to define an optimisation can provide a few benefits, including:

Shared state between optimisation functions and constraints. This can enable things like shared parameters and dynamic constraints.

Switch out optimisation problems using Liskov Substitution.

Logical grouping of related functions.

abstract objective(x: numpy.ndarray) → float

The objective function to minimise.

Parameters:: x (numpy.ndarray)
Return type:: float

df_objective(x: numpy.ndarray) → numpy.ndarray

The gradient of the objective function at x.

Parameters:: x (numpy.ndarray)
Return type:: numpy.ndarray

eq_constraints() → list[bluemira.optimisation.typed.ConstraintT]

Returns:: The equality constraints on the optimisation.
Return type:: list[bluemira.optimisation.typed.ConstraintT]

ineq_constraints() → list[bluemira.optimisation.typed.ConstraintT]

Returns:: The inequality constraints on the optimisation.
Return type:: list[bluemira.optimisation.typed.ConstraintT]

bounds() → tuple[numpy.typing.ArrayLike, numpy.typing.ArrayLike]

Returns:: The lower and upper bounds of the optimisation parameters.
Return type:: tuple[numpy.typing.ArrayLike, numpy.typing.ArrayLike]

Notes

Each set of bounds must be convertible to a numpy array of floats. If the lower or upper bound is a scalar value, that value is set as the bound for each of the optimisation parameters.

optimise(x0: numpy.ndarray, *, algorithm: bluemira.optimisation._algorithm.AlgorithmType = Algorithm.SLSQP, opt_conditions: collections.abc.Mapping[str, int | float] | None = None, opt_parameters: collections.abc.Mapping[str, Any] | None = None, keep_history: bool = False, check_constraints: bool = True, check_constraints_warn: bool = True) → bluemira.optimisation._optimise.OptimiserResult

Perform the optimisation.

See optimise() for more function parameter details.

Returns:

the result of optimisation

Parameters:

x0 (numpy.ndarray)
algorithm (bluemira.optimisation._algorithm.AlgorithmType)
opt_conditions (collections.abc.Mapping[str, int | float] | None)
opt_parameters (collections.abc.Mapping[str, Any] | None)
keep_history (bool)
check_constraints (bool)
check_constraints_warn (bool)

Return type:

bluemira.optimisation._optimise.OptimiserResult

check_constraints(x: numpy.ndarray, *, warn: bool = True) → bool

Check if the given parameterisation violates this optimiser’s constraints.

Parameters:

x (numpy.ndarray) – The parameterisation to check the constraints against.
warn (bool) – If True print a warning that lists the violated constraints.

Return type:

True if any constraints are violated by the parameterisation.