Model

digraph inheritance456558ffd0 { bgcolor=transparent; rankdir=LR; size="8.0, 12.0"; "ABC" [fillcolor=white,fontname="Vera Sans, DejaVu Sans, Liberation Sans, Arial, Helvetica, sans",fontsize=10,height=0.25,shape=box,style="setlinewidth(0.5),filled",tooltip="Helper class that provides a standard way to create an ABC using"]; "Model" [URL="#geobipy.src.classes.model.Model.Model",fillcolor=white,fontname="Vera Sans, DejaVu Sans, Liberation Sans, Arial, Helvetica, sans",fontsize=10,height=0.25,shape=box,style="setlinewidth(0.5),filled",target="_top",tooltip="Generic model class with an attached mesh."]; "myObject" -> "Model" [arrowsize=0.5,style="setlinewidth(0.5)"]; "myObject" [URL="../core/myObject.html#geobipy.src.classes.core.myObject.myObject",fillcolor=white,fontname="Vera Sans, DejaVu Sans, Liberation Sans, Arial, Helvetica, sans",fontsize=10,height=0.25,shape=box,style="setlinewidth(0.5),filled",target="_top"]; "ABC" -> "myObject" [arrowsize=0.5,style="setlinewidth(0.5)"]; }

@Model_Class Module describing a Model

class geobipy.src.classes.model.Model.Model(mesh=None, values=None)

Generic model class with an attached mesh.

classmethod fromHdf(grp, index=None, skip_posterior=False): Reads in the object from a HDF file

property gradient

Compute the gradient

Parameter gradient :math:”nabla_{z}sigma” at the ith layer is computed via

gradient_probability(log=True)

Evaluate the prior for the gradient of the parameter with depth

Parameters:: hmin (float64) – The minimum thickness of any layer.
Returns:: out – The probability given the prior on the gradient of the parameters with depth.
Return type:: numpy.float64

local_inverse_hessian(observation=None)

Generate a localized Hessian matrix using a dataPoint and the current realization of the Model1D.

Parameters:: observation (geobipy.DataPoint, geobipy.Dataset, optional) – The observed data to use when computing the local estimate of the variance.
Returns:: out – Hessian matrix
Return type:: array_like

local_precision(observation=None)

Generate a localized inverse Hessian matrix using a dataPoint and the current realization of the Model1D.

Parameters:: datapoint (geobipy.DataPoint, optional) – The data point to use when computing the local estimate of the variance. If None, only the prior derivative is used.
Returns:: out – Inverse Hessian matrix
Return type:: array_like

local_variance(observation=None)

Generate a localized inverse Hessian matrix using a dataPoint and the current realization of the Model1D.

Parameters:: datapoint (geobipy.DataPoint, optional) – The data point to use when computing the local estimate of the variance. If None, only the prior derivative is used.
Returns:: out – Inverse Hessian matrix
Return type:: array_like

pad(shape)

Copies the properties of a model including all priors or proposals, but pads memory to the given size

Parameters:: size (int, tuple) – Create memory upto this size.
Returns:: out – Padded model
Return type:: geobipy.Model1D

pcolor(**kwargs)

Plot like an image

Parameters:

alpha (scalar or array_like, optional) – If alpha is scalar, behaves like standard matplotlib alpha and opacity is applied to entire plot If array_like, each pixel is given an individual alpha value.
log ('e' or float, optional) – Take the log of the colour to a base. ‘e’ if log = ‘e’, and a number e.g. log = 10. Values in c that are <= 0 are masked.
equalize (bool, optional) – Equalize the histogram of the colourmap so that all colours have an equal amount.
nbins (int, optional) – Number of bins to use for histogram equalization.
xscale (str, optional) – Scale the x axis? e.g. xscale = ‘linear’ or ‘log’
yscale (str, optional) – Scale the y axis? e.g. yscale = ‘linear’ or ‘log’.
flipX (bool, optional) – Flip the X axis
flipY (bool, optional) – Flip the Y axis
grid (bool, optional) – Plot the grid
noColorbar (bool, optional) – Turn off the colour bar, useful if multiple plotting plotting routines are used on the same figure.
trim (bool, optional) – Set the x and y limits to the first and last non zero values along each axis.

perturb(*args, **kwargs)

Perturb a model’s structure and parameter values.

Uses a stochastic newtown approach if a datapoint is provided. Otherwise, uses the existing proposal distribution attached to self.par to generate new values.

Parameters:

observation (geobipy.DataPoint, optional) – The datapoint to use to perturb using a stochastic Newton approach.

Returns:

remappedModel (geobipy.Model) – The current model remapped onto the perturbed dimension.
perturbedModel (geobipy.Model) – The model with perturbed structure and parameter values.

probability(solve_value, solve_gradient)

Evaluate the prior probability for the 1D Model.

Parameters:

sPar (bool) – Evaluate the prior on the parameters in the final probability
sGradient (bool) – Evaluate the prior on the parameter gradient in the final probability
components (bool, optional) – Return all components used in the final probability as well as the final probability

Returns:

probability (numpy.float64) – The probability
components (array_like, optional) – Return the components of the probability, i.e. the individually evaluated priors as a second return argument if comonents=True on input.

proposal_probabilities(remapped_model, observation=None, structure_only=False, alpha=1.0)

Return the forward and reverse proposal probabilities for the model

Returns the denominator and numerator for the model’s components of the proposal ratio.

(1)\[q(k, \boldsymbol{z} | \boldsymbol{m}^{'})\]

and

(2)\[q(k^{'}, \boldsymbol{z}^{'} | \boldsymbol{m})\]

Each component is dependent on the event that was chosen during perturbation.

Parameters:

remappedModel (geobipy.Model1D) – The current model, remapped onto the dimension of self.
observation (geobipy.DataPoint) – The perturbed datapoint that was used to generate self.

Returns:

forward (float) – The forward proposal probability
reverse (float) – The reverse proposal probability

set_priors(values_prior=None, gradient_prior=None, **kwargs)

Setup the priors of a 1D model.

Parameters:

halfSpaceValue (float) – Value of the parameter for the halfspace.
min_edge (float64) – Minimum depth possible for the model
max_edge (float64) – Maximum depth possible for the model
max_cells (int) – Maximum number of layers allowable in the model
parameterPrior (bool) – Sets a prior on the parameter values
gradientPrior (bool) – Sets a prior on the gradient of the parameter values
parameterLimits (array_like, optional) – Length 2 array with the bounds on the parameter values to impose.
min_width (float64, optional) – Minimum thickness of any layer. If min_width = None, min_width is computed from min_edge, max_edge, and max_cells (recommended).
factor (float, optional) – Tuning parameter used in the std of the parameter prior.
prng (numpy.random.RandomState(), optional) – Random number generator, if none is given, will use numpy’s global generator.