openmmslicer.resampling_methods module

This module contains different resampling algorithms, such as the basic MultinomialResampler and the more sophisticated SystematicResampler.

class openmmslicer.resampling_methods.MultinomialResampler

Bases: object

A basic multinomial resampler which resamples each sample independently of the others with replacement.

static resample(samples, weights=None, n_walkers=None, n_samples=1)

Resamples samples multinomially based on given weights.

Parameters

  • samples (list) – The input samples.

  • weights (list) – The weights associated with the samples. Default is equal weights.

  • n_walkers (int) – The number of resampled samples per batch. Default is the length of the samples.

  • n_samples (int) – The number of resample batches. Default is one.

Returns

resamples – A list of lists, containing the resampled samples.

Return type

[list]

class openmmslicer.resampling_methods.SystematicResampler

Bases: object

The most conservative resampler, which preserves as many samples as possible. Based on the method in: http://dx.doi.org/10.3150/12-BEJSP07. An additional review can be found in: http://people.isy.liu.se/rt/schon/Publications/HolSG2006.pdf.

static resample(samples, weights=None, n_walkers=None, n_samples=1)

Resamples samples systematically based on given weights.

Parameters

  • samples (list) – The input samples.

  • weights (list) – The weights associated with the samples. Default is equal weights.

  • n_walkers (int) – The number of resampled samples per batch. Default is the length of the samples.

  • n_samples (int) – The number of resample batches. Default is one.

Returns

resamples – A list of lists, containing the resampled samples.

Return type

[list]