simpeg.directives.UpdateIRLS#

class simpeg.directives.UpdateIRLS(cooling_rate=1, cooling_factor=2.0, chifact_start=1.0, chifact_target=1.0, irls_cooling_factor=1.2, f_min_change=0.01, max_irls_iterations=20, misfit_tolerance=0.1, percentile=100.0, verbose=True, **kwargs)[source]#

Bases: InversionDirective

Directive to control the IRLS iterations for Sparse.

Parameters:

cooling_rate: int: Number of iterations to cool beta.
cooling_factor: float: Factor to cool beta.
chifact_start: float: Starting chi factor for the IRLS iterations.
chifact_target: float: Target chi factor for the IRLS iterations.
irls_cooling_factor: float: Factor to cool the IRLS threshold epsilon.
f_min_change: float: Minimum change in the regularization function to continue the IRLS iterations.
max_irls_iterations: int: Maximum number of IRLS iterations.
misfit_tolerance: float: Tolerance for the target misfit.
percentile: float: Percentile of the function values used to determine the initial IRLS threshold.
verbose: bool: Print information to the screen.

Attributes

`chifact_start`	Target chi factor to start the IRLS process.
`chifact_target`	Targer chi factor to maintain during the IRLS process.
`cooling_factor`	Beta is divided by this value every `cooling_rate` iterations.
`cooling_rate`	Cool beta after this number of iterations.
`dmisfit`	Data misfit associated with the directive.
`f_min_change`	Target chi factor to start the IRLS process.
`invProb`	Inverse problem associated with the directive.
`inversion`	Inversion object associated with the directive.
`irls_cooling_factor`	IRLS threshold parameter (epsilon) is divided by this value every iteration.
`max_irls_iterations`	Maximum irls iterations.
`metrics`	Various metrics used by the IRLS algorithm.
`misfit_tolerance`	Tolerance on deviation from the target chi factor, as a fractional percent.
`opt`	Optimization algorithm associated with the directive.
`percentile`	Tolerance on deviation from the target chi factor, as a fractional percent.
`reg`	Regularization associated with the directive.
`simulation`	Return simulation for all data misfits.
`survey`	Return survey for all data misfits
`verbose`	Whether or not to print debugging information.

Methods

`adjust_cooling_schedule`()	Adjust the cooling schedule based on the misfit.
`endIter`()	Check on progress of the inversion and start/update the IRLS process.
`finish`()	Update inversion parameter(s) according to directive at end of inversion.
`initialize`()	Initialize the IRLS iterations with l2-norm regularization (mode:1).
`misfit_from_chi_factor`(chi_factor)	Compute the target misfit from the chi factor.
`start_irls`()	Start the IRLS iterations by computing the initial threshold values.
`stopping_criteria`()	Check for stopping criteria of max_irls_iteration or minimum change.
`validate`([directiveList])	Validate directive.