gwin.models package

Submodules

• gwin.models.analytic module
• gwin.models.base module
• gwin.models.base_data module
• gwin.models.gaussian_noise module
• gwin.models.marginalized_gaussian_noise module

Module contents

This package provides classes and functions for evaluating Bayesian statistics assuming various noise models.

class gwin.models.CallModel(model, callstat, return_all_stats=True)

Bases: object

Wrapper class for calling models from a sampler.

This class can be called like a function, with the parameter values to evaluate provided as a list in the same order as the model’s variable_params. In that case, the model is updated with the provided parameters and then the callstat retrieved. If return_all_stats is set to True, then all of the stats specified by the model’s default_stats will be returned as a tuple, in addition to the stat value.

The model’s attributes are promoted to this class’s namespace, so that any attribute and method of model may be called directly from this class.

This class must be initalized prior to the creation of a Pool object.

Parameters:

• model (Model instance) – The model to call.
• callstat (str) – The statistic to call.
• return_all_stats (bool, optional) – Whether or not to return all of the other statistics along with the callstat value.

Examples

Create a wrapper around an instance of the TestNormal model, with the callstat set to logposterior:

>>> from gwin.models import TestNormal, CallModel
>>> model = TestNormal(['x', 'y'])
>>> call_model = CallModel(model, 'logposterior')

Now call on a set of parameter values:

>>> call_model([0.1, -0.2])
(-1.8628770664093453, (0.0, 0.0, -1.8628770664093453))

Note that a tuple of all of the model’s default_stats were returned in addition to the logposterior value. We can shut this off by toggling return_all_stats:

>>> call_model.return_all_stats = False
>>> call_model([0.1, -0.2])
-1.8628770664093453

Attributes of the model can be called from the call model. For example:

>>> call_model.variable_params
('x', 'y')

gwin.models.read_from_config(cp, **kwargs)

Initializes a model from the given config file.

The section must have a name argument. The name argument corresponds to the name of the class to initialize.

Parameters:

• cp (WorkflowConfigParser) – Config file parser to read.
• **kwargs – All other keyword arguments are passed to the from_config method of the class specified by the name argument.

Returns:

The initialized model.

Return type:

cls