pybop.optimisers.scipy_optimisers#

Classes#

BaseSciPyOptimiser

A base class for defining optimisation methods from the SciPy library.

SciPyDifferentialEvolution

Adapts SciPy's differential_evolution function for global optimisation.

SciPyMinimize

Adapts SciPy's minimize function for use as an optimisation strategy.

Module Contents#

class pybop.optimisers.scipy_optimisers.BaseSciPyOptimiser(cost, **optimiser_kwargs)[source]#

Bases: pybop.BaseOptimiser

A base class for defining optimisation methods from the SciPy library.

Parameters:

  • x0 (array_like) – Initial position from which optimisation will start.

  • bounds (dict, sequence or scipy.optimize.Bounds, optional) – Bounds for variables as supported by the selected method.

  • **optimiser_kwargs (optional) – Valid SciPy option keys and their values.

_run()[source]#

Internal method to run the optimisation using a PyBOP optimiser.

Returns:

result – The result of the optimisation including the optimised parameter values and cost.

Return type:

pybop.Result

_sanitise_inputs()[source]#

Check and remove any duplicate optimiser options.

key_mapping[source]#
num_resamples = 40[source]#
class pybop.optimisers.scipy_optimisers.SciPyDifferentialEvolution(cost, **optimiser_kwargs)[source]#

Bases: BaseSciPyOptimiser

Adapts SciPy’s differential_evolution function for global optimisation.

This class provides a global optimisation strategy based on differential evolution, useful for problems involving continuous parameters and potentially multiple local minima.

Parameters:

  • bounds (dict, sequence or scipy.optimize.Bounds) – Bounds for variables. Must be provided as it is essential for differential evolution. Each element is a tuple (min, max) for the corresponding variable.

  • **optimiser_kwargs (optional) –

    Valid SciPy differential_evolution options: strategy : str, optional

    The differential evolution strategy to use. Should be one of: - ‘best1bin’ - ‘best1exp’ - ‘rand1exp’ - ‘randtobest1exp’ - ‘currenttobest1exp’ - ‘best2exp’ - ‘rand2exp’ - ‘randtobest1bin’ - ‘currenttobest1bin’ - ‘best2bin’ - ‘rand2bin’ - ‘rand1bin’ Default is ‘best1bin’.

    maxiterint, optional

    Maximum number of generations. Default is 1000.

    popsizeint, optional

    Multiplier for setting the total population size. The population has popsize * len(x) individuals. Default is 15.

    tolfloat, optional

    Relative tolerance for convergence. Default is 0.01.

    mutationfloat or tuple(float, float), optional

    The mutation constant. If specified as a float, should be in [0, 2]. If specified as a tuple (min, max), dithering is used. Default is (0.5, 1.0).

    recombinationfloat, optional

    The recombination constant, should be in [0, 1]. Default is 0.7.

    seedint, optional

    Random seed for reproducibility.

    dispbool, optional

    Display status messages. Default is False.

    callbackcallable, optional

    Called after each iteration with the current result as argument.

    polishbool, optional

    If True, performs a local optimisation on the solution. Default is True.

    initstr or array-like, optional

    Specify initial population. Can be ‘latinhypercube’, ‘random’, or an array of shape (M, len(x)).

    atolfloat, optional

    Absolute tolerance for convergence. Default is 0.

    updating{‘immediate’, ‘deferred’}, optional

    If ‘immediate’, best solution vector is continuously updated within a single generation. Default is ‘immediate’.

    workersint or map-like callable, optional

    If workers is an int the population is subdivided into workers sections and evaluated in parallel. Default is 1.

    constraints{NonlinearConstraint, LinearConstraint, Bounds}, optional

    Constraints on the solver.

See also

scipy.optimize.differential_evolution

The SciPy method this class is based on.

Notes

Differential Evolution is a stochastic population based method that is useful for global optimisation problems. At each pass through the population the algorithm mutates each candidate solution by mixing with other candidate solutions to create a trial candidate. The fitness of all candidates is then evaluated and for each candidate if the trial candidate is an improvement, it takes its place in the population for the next iteration.

_run_optimiser()[source]#

Executes the optimization process using SciPy’s differential_evolution function.

Returns:

result – The result of the optimisation including the optimised parameter values and cost.

Return type:

scipy.optimize.OptimizeResult

_set_up_optimiser()[source]#

Parse optimiser options.

name()[source]#

Provides the name of the optimisation strategy.

class pybop.optimisers.scipy_optimisers.SciPyMinimize(cost, **optimiser_kwargs)[source]#

Bases: BaseSciPyOptimiser

Adapts SciPy’s minimize function for use as an optimisation strategy.

This class provides an interface to various scalar minimisation algorithms implemented in SciPy, allowing fine-tuning of the optimisation process through method selection and option configuration.

Parameters:

**optimiser_kwargs (optional) –

Valid SciPy Minimize option keys and their values: x0 : array_like

Initial position from which optimisation will start.

methodstr

The optimisation method, options include: ‘Nelder-Mead’, ‘Powell’, ‘CG’, ‘BFGS’, ‘Newton-CG’, ‘L-BFGS-B’, ‘TNC’, ‘COBYLA’, ‘SLSQP’, ‘trust-constr’, ‘dogleg’, ‘trust-ncg’, ‘trust-exact’, ‘trust-krylov’.

jac{callable, ‘2-point’, ‘3-point’, ‘cs’, bool}, optional

Method for computing the gradient vector.

hess{callable, ‘2-point’, ‘3-point’, ‘cs’, HessianUpdateStrategy}, optional

Method for computing the Hessian matrix.

hesspcallable, optional

Hessian of objective function times an arbitrary vector p.

boundssequence or scipy.optimize.Bounds, optional

Bounds on variables for L-BFGS-B, TNC, SLSQP, trust-constr methods.

constraints{Constraint, dict} or List of {Constraint, dict}, optional

Constraints definition for constrained optimisation.

tolfloat, optional

Tolerance for termination.

optionsdict, optional

Method-specific options. Common options include: maxiter : int

Maximum number of iterations.

dispbool

Set to True to print convergence messages.

ftolfloat

Function tolerance for termination.

gtolfloat

Gradient tolerance for termination.

epsfloat

Step size for finite difference approximation.

maxfevint

Maximum number of function evaluations.

maxcorint

Maximum number of variable metric corrections (L-BFGS-B).

See also

scipy.optimize.minimize

The SciPy method this class is based on.

Notes

Different optimisation methods may support different options. Consult SciPy’s documentation for method-specific options and constraints.

_run_optimiser()[source]#

Executes the optimisation process using SciPy’s minimize function.

Returns:

result – The result of the optimisation including the optimised parameter values and cost.

Return type:

scipy.optimize.OptimizeResult

_set_up_optimiser()[source]#

Parse optimiser options.

cost_wrapper(x)[source]#

Scale the cost function, preserving the sign convention, and eliminate nan values

name()[source]#

Provides the name of the optimisation strategy.

_cost0 = 1.0[source]#