pymc.compute_deterministics

pymc.compute_deterministics(dataset, *, group=None, var_names=None, model=None, sample_dims=('chain', 'draw'), merge_dataset=False, extend_dataset=False, progressbar=True, backend=None, compile_kwargs=None)

Compute model deterministics given a dataset with values for model variables.

Parameters:

datasetDataset or DataTree

Dataset with values for model variables, such as idata.posterior. A whole InferenceData object can also be passed, in which case the group given by group is used.

groupstr, optional

Which group to use when dataset is a whole InferenceData object. If None, “posterior” is used, falling back to “prior” when there is no posterior group. Cannot be used when a single group is passed directly.

var_namessequence of str, optional

List of names of deterministic variable to compute. If None, compute all deterministics in the model.

modelBaseModel, optional

Model to use. If None, use context model.

sample_dimssequence of str, default (“chain”, “draw”)

Sample (batch) dimensions of the dataset over which to compute the deterministics.

merge_datasetbool, default False

Whether to include the values of the original dataset in the returned one.

Deprecated since version ``merge_dataset``: is deprecated and will be removed in a future release. Use extend_dataset instead.

extend_datasetbool, default False

Whether to add the deterministics to the original dataset in place, instead of returning a new one. The mutated input object is returned, so for an InferenceData the deterministics end up in the selected group. Cannot be combined with merge_dataset.

progressbarbool, default True

Whether to display a progress bar in the command line.

progressbar_themeTheme, optional

Custom theme for the progress bar.

backend: str, optional

Which computational backend to use. Recommended to be one of “numba”, “c”, and “jax”.

compile_kwargs: dict, optional

Additional arguments passed to model.compile_fn. compile_kwargs["mode"] cannot be combined with backend.

Returns:

Dataset or DataTree

Dataset with values for the deterministics. When merge_dataset is True, the values of the input dataset (or of the selected group) are included as well. When extend_dataset is True, the input object is returned instead, with the deterministics added to it.

Examples

import pymc as pm

with pm.Model(coords={"group": (0, 2, 4)}) as m:
    mu_raw = pm.Normal("mu_raw", 0, 1, dims="group")
    mu = pm.Deterministic("mu", mu_raw.cumsum(), dims="group")

    trace = pm.sample(var_names=["mu_raw"], chains=2, tune=5, draws=5)

assert "mu" not in trace.posterior

with m:
    pm.compute_deterministics(trace, extend_dataset=True)

assert "mu" in trace.posterior