pylfi.priors.Prior

class pylfi.priors.Prior(distr_name, *params, name=None, tex=None, **kwargs)

Initialize a prior.

In the Bayesian paradigm, all available information about an unknown parameter is incorporated in a prior probability distribution, which describes the range of possible parameter values.

Parameters

distr_namestr

Any distribution from scipy.stats as a string.

params:

Parameters of the prior distribution. Typically these would be shape parameters or loc and scale passed as positional arguments.

namestr

Name of the unknown parameter, which is used to keep track and access the parameter in the sampling algorithms. Default: None.

texstr, optional

LaTeX typesetting for the parameter name. pyLFI includes procedures for automatically plotting priors and posteriors, and will use the tex name of the parameter as axis labels if provided. Default: None.

kwargs:

kwargs are passed to the scipy distribution methods. Typically these would be loc and scale.

Attributes

distr_name	Name of the scipy.stats distribution.
name	Parameter name.
tex	Parameter name with LaTeX typesetting.

Methods

logpdf(x)	Evaluate the log of the probability density function (pdf).
pdf(x)	Evaluate the probability density function (pdf).
plot_prior(x[, color, facecolor, alpha, ax])	Plot prior pdf or pmf evaluated at x.
pmf(x)	Evaluate the probability mass function (pmf).
rvs([size, seed])	Draw random variate.

rvs(size=None, seed=None)

Draw random variate.

Parameters

size{int, tuple}, optional

Output size of a single random draw. Default: None.

seedint, optional

Seed for reproducibility.

Returns

rvsnumpy.ndarray

Random variables.

pdf(x)

Evaluate the probability density function (pdf).

Method for continuous distributions.

Parameters

xarray_like

Evaluation points.

Returns

pdfnumpy.ndarray

pdf evaluated at x.

logpdf(x)

Evaluate the log of the probability density function (pdf).

Method for continuous distributions.

Parameters

xarray_like

Evaluation points.

Returns

logpdfnumpy.ndarray

Log of pdf evaluated at x.

pmf(x)

Evaluate the probability mass function (pmf).

Method for discrete distributions.

Parameters

xarray_like

Evaluation points.

Returns

pmfnumpy.ndarray

pmf evaluated at x.

property name

Parameter name.

Returns

str

property distr_name

Name of the scipy.stats distribution.

Returns

str

property tex

Parameter name with LaTeX typesetting.

Returns

str

plot_prior(x, color='C0', facecolor='lightblue', alpha=0.5, ax=None, **kwargs)

Plot prior pdf or pmf evaluated at x.

Parameters

xarray_like

Evaluation points.

colorstr, optional

Set the color of the line. Default: C0.

facecolorstr, optional

Set the face color of area under the curve. Default: lightblue.

alphafloat, optional

Set the alpha value used for blending the face color. Must be within the 0-1 range. Default: 0.5.

axmatplotlib.axes.Axes, optional

Pre-existing axes for the plot. Otherwise, call matplotlib.pyplot.gca internally.

kwargs:

kwargs are passed to matplotlib.pyplot.plot.

previous

pylfi.inferences.MCMCABC

next

pylfi.utils.check_and_set_jobs