orbit.utils.fitting.Solver_lib#

This is a collection of classes for a general optimization problem. The Scorer function should be minimized by changing the input parameters that are inside the trial point class instance. The user can define different search algorithms that are suitable for the particular problem. Also there are set of solver stoppers that will stop optimization process immediately (but stopper cannot interrupt processes inside the Scorer).

Functions

orbitFinalize([message])

Method.

Classes

ComboStopper()

The stopper implementation that will say True to "Should I Stop?" question if at least one of stoppers registered in it will say True.

MaxAccuracyStopper(max_accuracy)

The stopper implementation that will say True to "Should I Stop?" question if the difference between two last best scores is less than the max accuracy.

MaxIterationStopper(max_iteration)

The stopper implementation that will say True to "Should I Stop?" question if the number of Scorer evaluation is more then the maximal allowable iterations.

MaxTimeStopper(max_time)

The stopper implementation that will say True to "Should I Stop?" question if the time is up.

NamedObject([name])

Class.

ParamsDictObject()

Class.

RunForeverStopper()

The stopper implementation that always says False to "Should I Stop?" question.

Scoreboard(solver)

Scoreboard class keeps the trace of all best scores (as tuple (score,TrialPoint)) in the Scores History Stack.

ScoreboardActionListener()

This is an abstract class for actions inside Scoreboard.

Scorer()

This class calculates the score for the trial point instance.

SearchAgorithm()

The SearchAgorithm is an abstract class.

SolveStopper()

The SolveStopper defines if we should stop solver's optimization process because of some condition.

SolveStopperFactory()

The Factory for stoppers.

Solver()

The class is a main class of the general fitting package.

TrialPoint()

This a container class for VariableProxy instances.

VariableProxy(*arg, **kwargs)

This class represents the parameter for the score function in the fitting process.