Skip to content

Stochasticmodel API Reference

pyfastwell.stochastic.stochasticmodel.Stochasticmodel

Bases: object

Stochastic model for parameter sampling and forward modeling.

This class manages model parameters, samples them according to specified distributions, and applies them to a forward model for ensemble simulations.

Notes

  • Locked parameters remain constant across the ensemble.
  • Free parameters are sampled according to their specified distributions.
  • The class supports normal, uniform, and triangular distributions for sampling.

__init__ 

__init__(argdict: Optional[Dict[str, Any]] = None) -> None

Initialize the Stochasticmodel object.

Parameters:

argdict: dictionary with keys 'locked', 'free_param', and 'free dist' containing the model parameters

Notes

The model parameters are subdivided in 'locked' parameters, which typically include the time array for which simulation is performed and additional 'locked' parameters which are not variable for the ensemble. The 'free_parameters' contains a list of parameter names which are allowed to vary and are sampled for the Stochasticmodel corresponding to the 'free_dist' distribition specifications

get_from_dict 

get_from_dict(name: str, param: ndarray) -> Any

Get a parameter from the dictionary based on its name, either from 'locked' or 'free_param' If the name is not found in 'free_param', it will look in 'locked'. If the name is not found in either, it will raise an exception. This method is used to retrieve the parameter values for the ensemble realisations.

Parameters

name: name of the parameter to be retrieved from the dictionary (either 'locked' or 'free_param'), 'free_param' takes prevalence

param: ensembles realisation corresponding to 'free_param' sampled values

Returns

value of the parameter corresponding to the name, either from 'locked' or from 'free_param'

getmedianparams 

getmedianparams() -> np.ndarray

Get the median values of the free parameters

Returns

median values of the free parameters in one dimensional array

generate_ensemble 

generate_ensemble(nsamples: Optional[int] = None) -> List[np.ndarray]

Generate an ensemble of model parameters based on the distribution specifications in the dictionary

Parameters

nsamples: number of samples to generate

Returns

list of parameter arrays (dimension equal to number of free parameters), list size equal to nsamples

run_ensemble 

run_ensemble(m: List[ndarray], forward: Callable[[ndarray], Any]) -> List[np.ndarray]

Run the forward model for the ensemble of parameters

Parameters

m: ensemble of parameters, corresponding to the free parameters and created by generate_ensemble forward: forward model function

Returns

results of the forward model for the ensemble

plot_distribution 

plot_distribution(results: ndarray, name: str)

Plot the distribution of a parameter forward model results

Parameters:

results: results of the forward model for the ensemble ndarray

name: name of the plot

Returns

None

expectation_plot 

expectation_plot(results: ndarray, name: str, percent: bool = True, expectation: bool = True, pvals: Optional[List[float]] = None) -> None

Plot the expectation of a parameter forward model results

Parameters

results: results of the forward model for the ensemble ndarray

name: name of the plot

percent: if True, the y-axis is in percent, otherwise in fraction

expectation: if True, the y-axis is reversed interpreting the results as expectation

pvals: pvalues to be plotted (list)

Returns

None