["BayesicFitting"]

Global	Documentation
Home	Manual
Readme	Glossary
Notes	Design
Examples	Classes
	Trouble
	Style

class ChordEngine( Engine )	Source

Move a a walker in a random direction.

The ChordEngine draws a random line through the walker parameters in unit space, from unitMin (lowpoint) with lengths unitRange (highpoint).

A random point on the line is selected. If the corresponding parameter set has a likelihood < LowLhood, it is accepted. Otherwise either the highpoint is reset to the random point (if randompoint > walkerpoint) or the lowpoint is replaced by the randompoint (if walker < random). Then a new random point on the line is selected, until the point is accepted.

When the point is accepted, another random line is constructed through the new point and orthogonal to (all) previous ones. (The orthogonality is not implemented now. TBC).

This is an independent implementation inspired by the polychord engine described in "POLYCHORD: next-generation nested sampling", WJ Handley, MP Hobson and AN Lasenby. MNRAS (2015) Volume 453, Issue 4, p 4384–4398

Attributes

debug : bool
perform the step-out action too

Attributes from Engine

walkers, errdis, slow, maxtrials, nstep, rng, verbose, report, unitRange, unitMin

Author Do Kester.

ChordEngine( walkers, errdis, copy=None, **kwargs )

Constructor.

Parameters

walkers : WalkerList
walkers to be diffused
errdis : ErrorDistribution
error distribution to be used
copy : ChordEngine
to be copied
kwargs : for Engine
"slow", "seed", "verbose"

copy( )

Return copy of this.

execute( kw, lowLhood, append=False, iteration=0 )

Execute the engine by diffusing the parameters.

Parameters

kw : int
index of walker to diffuse
lowLhood : float
lower limit in logLikelihood
append : bool
set walker in place or append
iteration : int
iteration number

Returns

int : the number of successfull moves

stepOut( problem, ptry, usav, vel, t, tmax, lowLhood, fitIndex )

Check if endpoints are indeed outside the lowLhood domain.

Methods inherited from Engine