paref.express.info#

Classes

`GprBbf`(**kwargs)	Approximate a blackbox function with a Gaussian process regression (GPR) in the BlackboxFunction interface
`Info`(blackbox_function[, training_iter, ...])	Obtain relevant information about the Pareto front of a blackbox function and model fitness

class paref.express.info.GprBbf(**kwargs)[source]#

Bases: BlackboxFunction

Approximate a blackbox function with a Gaussian process regression (GPR) in the BlackboxFunction interface

Initialize storage for evaluations of the blackbox function

__call__(**kwargs)#: Store evaluation of the blackbox function

property design_space#

property dimension_design_space: int#

property dimension_target_space: int#

class paref.express.info.Info(blackbox_function: BlackboxFunction, training_iter=2000, learning_rate=0.05)[source]#

Bases: object

Obtain relevant information about the Pareto front of a blackbox function and model fitness

topology:
the shape of your Pareto front (Info.topology)

suggestion_pareto_points:
suggestions for Pareto points to evaluate, how and why (Info.suggestion_pareto_points)

minima:
the estimated minima of each component (Info.minima)

model fitness:
how well the model approximates the bbf, how to improve it and how certain its estimation is (Info.model_fitness)

Warning

Paref’s Info class is still under development. If you run into any problems, errors or have suggestions how to make it user-friendlier, please contact me or open an issue on GitHub. Many thanks!

Parameters:

blackbox_function (BlackboxFunction) – underlying blackbox function
training_iter (int default 2000) – number of training iterations for the underlying approximate of the bbf (GPR)
learning_rate (float default 0.01) – learning rate for the underlying approximate of the bbf (GPR)

update()[source]#

Update the information about the Pareto front

Call this method whenever you have evaluated the blackbox function at some design.

property minima#: Estimated minimum of each component

property model: GPR#

The underlying model of the blackbox function

Returns:: underlying surrogate (GPR) of the blackbox function
Return type:: GPR

property model_fitness#

Information about the model fitness

How well does the model approximate the blackbox function? Are the training iterations sufficient, i.e. did the training convergence? How uncertain is the model? How many evaluations are needed to obtain good results?

property suggestion_pareto_points#

Suggestions for Pareto points to evaluate, how and why

The suggestion mainly focuses on the approximate shape of the Pareto front and its implication for the trade-offs between the components.

property topology#

Obtain topological information about the Pareto front

Is there are global optimum? Is the Pareto front rather convex or concave or linear? Of which dimension is the Pareto front?