Class adjuster

source code

The adjuster class controls the step size.

Instance Methods

__init__(self, F, vscale, vse=0.0, vsmax=1e+30)
x.__init__(...) initializes x; see help(type(x)) for signature

source code

reset(self)
Called when old habits should be forgotten.

source code

f(self)
This allows the desired fraction of accepted steps to depend on self.vscale.

source code

inctry(self, accepted)

source code

float

vs(self)
We stick in the factor of random.lognormvariate() so that all sizes of move are possible and thus we can prove that we can random-walk to any point in a connected region.

source code

status(self)
Returns: misc.

source code

Inherited from object: __delattr__, __format__, __getattribute__, __hash__, __new__, __reduce__, __reduce_ex__, __repr__, __setattr__, __sizeof__, __str__, __subclasshook__

Class Variables
	Z0 = `1.5`
	DZ = `0.3`
	TOL = `0.2`
	STABEXP = `1.0`

Instance Variables
	vsmax Used when the acceptance probability is larger than 25%.

Properties
Inherited from `object`: `__class__`

Method Details

init(self, F, vscale, vse=0.0, vsmax=1e+30)
(Constructor)

source code

x.__init__(...) initializes x; see help(type(x)) for signature

Overrides: object.__init__: (inherited documentation)

f(self)

source code

This allows the desired fraction of accepted steps to depend on self.vscale.

Requires: This method should only be called with self.vse != 0 where self.vscale can normally be expected to be fairly close to unity, and where small values of self.vse indicate trouble. In the mcmc.Bootstepper.step_boot case this is true.

vs(self)

source code

We stick in the factor of random.lognormvariate() so that all sizes of move are possible and thus we can prove that we can random-walk to any point in a connected region. This makes the proof of ergodicity simpler.

Returns: float: a scale factor for the step size.

status(self)

source code

Returns:: misc. status information for debugging purposes.

Instance Variable Details

vsmax

Used when the acceptance probability is larger than 25%. Large acceptance probabilities can happen if the probability is everywhere about equal. (E.g. a data fitting problem with almost no data)

Class adjuster

__init__(self, F, vscale, vse=0.0, vsmax=1e+30) (Constructor)

f(self)

vs(self)

status(self)

vsmax

init(self, F, vscale, vse=0.0, vsmax=1e+30)
(Constructor)