Random Sampling Search Algorithm¶

monte_carlo_random¶

Module containing algorithm classes for Monte Carlo random sampling.

class MTfit.algorithms.monte_carlo.IterationSample(max_samples=600000, *args, **kwargs)[source]¶

Iteration based sampling algorithm

Algorithm that runs for a fixed number of samples.

Initialisation of IterationSample

Args: max_samples:[600000] Maximum time in seconds for the iteration to run for. number_samples:[10000] Number of samples to use per iteration. dc:[False] Boolean to select inversion constrained to double-couple space or over the full moment tensor space.
Keyword Arguments: basic_cdc:[False] Boolean to select whether to us the Basic CDC model (inversion constrained to double-couple space or over the full moment tensor space.

eigenvectors_mt_2_mt6(diag, a, b, c)¶

Converts eigenvectors and eigenvalues to moment tensor 6-vector.

Generates moment tensor 6-vector.

Args: diag: numpy matrix of eigenvalues. a: numpy array of eigenvectors corresponding to the first eigenvalue. b: numpy array of eigenvectors corresponding to the second eigenvalue. c: numpy array of eigenvectors corresponding to the third eigenvalue.
Returns: numpy matrix of moment tensor 6-vectors.

get_sampling_model(kwargs, file_sample, file_safe)¶: Get the sampling model from the entry points

initialise()¶

Initial task for ForwardTask

Returns: task,end=False task: the random samples starting the inversion.

iterate(result)¶

Iteration function for BaseMonteCarloRandomSample object

Carries out iteration of random sampling from previous result.

Args: result: dictionary result from MTfit.inversion.ForwardTask, added to the pdf
Returns: new_samples,end new_samples: New set of random samples to test. end: Boolean flag as to whether inversion has finished, if a MemoryError error is thrown, ends iteration.

output(normalise=True, convert=False, discard=10000)¶

Return the algorithm results for output.

Returns: Dictionary containing output.

random_clvd(*args)¶

Generate random CLVD moment tensors (size 6,number_samples)

Generates random CLVD moment tensors with random orientations.

Returns: numpy matrix of random CLVD moment tensors, size 6,number_samples

random_dc(*args)¶

Generate random double-couple moment tensors (size 6, number_samples)

Generates random double-couple moment tensors with random orientations.

Returns: numpy matrix of random double-couple moment tensors, size 6,number_samples

random_orthogonal_eigenvectors()¶

Generates random orthogonal eigenvectors.

Returns: list of numpy arrays of eigenvectors.

random_sample()¶

Return random sample

Extends _BaseAlgorithm.random_sample to account for multiple events, and checks the source type.

random_type(diag)¶

Generate random moment tensors from a given set of eigenvalues.

Generates random moment tensors with a given set of eigenvalues and random orientations.

Args: diag: numpy matrix of eigenvalues.
Returns: numpy matrix of random moment tensors, size 6,number_samples

members:

inherited-members:

The random sampling algorithms inherit from the BaseAlgorithm class:

class MTfit.algorithms.base.BaseAlgorithm(number_samples=10000, dc=False, quality_check=False, file_sample=False, file_safe=True, generate=False, *args, **kwargs)[source]¶

Base class for algorithms including the base methods required for forward modelling based inversions.

BaseAlgorithm initialisation

Args

number_samples:[10000] Number of samples to use per iteration. dc:[False] Boolean to select inversion constrained to double-couple

space or over the full moment tensor space.

quality_check:[False] Carry out quality check after 40,000 samples.: If there are too many non-zero samples, the inversion is stopped for that event.
file_sample:[False] Boolean to select whether to use file sampling: (saving the output to file).
file_safe:[False] Boolean to select whether to use file safe outputs: when using file sampling.
generate:[False] Boolean to select whether to generate the random: moment tensors in the forward task or not.

Keyword Arguments

basic_cdc:[False] Boolean to select whether to use the Basic CDC: model (inversion constrained to crack+double-couple space).
number_events:[1] Integer giving the number of events in this: inversion (for use with multiple event inversions).
sampling_prior:[6sphere_prior] string selector for the full moment: tensor sampling prior (MTfit.sampling_prior entry point)
sampling:[6sphere] string selector for the full moment tensor: sampling model (MTfit.sampling entry point)
model:[False] selector for alternate models using the: MTfit.sample_distribution entry_point

eigenvectors_mt_2_mt6(diag, a, b, c)[source]¶

Converts eigenvectors and eigenvalues to moment tensor 6-vector.

Generates moment tensor 6-vector.

Args: diag: numpy matrix of eigenvalues. a: numpy array of eigenvectors corresponding to the first eigenvalue. b: numpy array of eigenvectors corresponding to the second eigenvalue. c: numpy array of eigenvectors corresponding to the third eigenvalue.
Returns: numpy matrix of moment tensor 6-vectors.

get_sampling_model(kwargs, file_sample, file_safe)[source]¶: Get the sampling model from the entry points

initialise()[source]¶

Basic initialisation function

Return the first task

Returns: task, False task: New task for forward model

iterate(result)[source]¶

Basic iteration function

Args: result: Result from forward Task
Returns: task,End task: New task for forward model End: Boolean flag for whether the inversion is finished.

output(normalise=True, convert=False, discard=10000)[source]¶

Return the algorithm results for output.

Returns: Dictionary containing output.

random_clvd(*args)[source]¶

Generate random CLVD moment tensors (size 6,number_samples)

Generates random CLVD moment tensors with random orientations.

Returns: numpy matrix of random CLVD moment tensors, size 6,number_samples

random_dc(*args)[source]¶

Generate random double-couple moment tensors (size 6, number_samples)

Generates random double-couple moment tensors with random orientations.

Returns: numpy matrix of random double-couple moment tensors, size 6,number_samples

random_orthogonal_eigenvectors()[source]¶

Generates random orthogonal eigenvectors.

Returns: list of numpy arrays of eigenvectors.

random_sample()[source]¶

Return random samples

Returns:	results from generating the random samples.

random_type(diag)[source]¶

Generate random moment tensors from a given set of eigenvalues.

Generates random moment tensors with a given set of eigenvalues and random orientations.

Args: diag: numpy matrix of eigenvalues.
Returns: numpy matrix of random moment tensors, size 6,number_samples

They also inherit from the BaseMonteCarloRandomSample class:

class MTfit.algorithms.monte_carlo.BaseMonteCarloRandomSample(*args, **kwargs)[source]¶

Base class for Monte Carlo Random Sampling

Contains base functions for new samples, iteration and initialisation.

Initialisation of BaseMonteCarloRandomSample object

Args: number_samples:[10000] Number of samples to use per iteration. dc:[False] Boolean to select inversion constrained to double-couple space or over the full moment tensor space.
Keyword Arguments: basic_cdc:[False] Boolean to select whether to us the Basic CDC model (inversion constrained to double-couple space or over the full moment tensor space.