Source code for pymc.sampling.mcmc

#   Copyright 2024 - present The PyMC Developers
#
#   Licensed under the Apache License, Version 2.0 (the "License");
#   you may not use this file except in compliance with the License.
#   You may obtain a copy of the License at
#
#       http://www.apache.org/licenses/LICENSE-2.0
#
#   Unless required by applicable law or agreed to in writing, software
#   distributed under the License is distributed on an "AS IS" BASIS,
#   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
#   See the License for the specific language governing permissions and
#   limitations under the License.

"""Functions for MCMC sampling."""

import contextlib
import logging
import pickle
import sys
import time
import warnings

from collections.abc import Callable, Iterator, Mapping, Sequence
from typing import (
    Any,
    Literal,
    TypeAlias,
    cast,
    overload,
)

import numpy as np
import pytensor.gradient as tg

from arviz import InferenceData, dict_to_dataset
from arviz.data.base import make_attrs
from pytensor.graph.basic import Variable
from rich.theme import Theme
from threadpoolctl import threadpool_limits
from typing_extensions import Protocol

import pymc as pm

from pymc.backends import RunType, TraceOrBackend, init_traces
from pymc.backends.arviz import (
    coords_and_dims_for_inferencedata,
    find_constants,
    find_observations,
)
from pymc.backends.base import IBaseTrace, MultiTrace, _choose_chains
from pymc.backends.zarr import ZarrChain, ZarrTrace
from pymc.blocking import DictToArrayBijection
from pymc.exceptions import SamplingError
from pymc.initial_point import PointType, StartDict, make_initial_point_fns_per_chain
from pymc.model import Model, modelcontext
from pymc.sampling.parallel import Draw, _cpu_count
from pymc.sampling.population import _sample_population
from pymc.stats.convergence import (
    log_warning_stats,
    log_warnings,
    run_convergence_checks,
)
from pymc.step_methods import NUTS, CompoundStep
from pymc.step_methods.arraystep import BlockedStep, PopulationArrayStepShared
from pymc.step_methods.hmc import quadpotential
from pymc.util import (
    ProgressBarManager,
    ProgressBarType,
    RandomSeed,
    RandomState,
    _get_seeds_per_chain,
    default_progress_theme,
    drop_warning_stat,
    get_random_generator,
    get_untransformed_name,
    is_transformed_name,
)
from pymc.vartypes import discrete_types

try:
    from zarr.storage import MemoryStore
except ImportError:
    MemoryStore = type("MemoryStore", (), {})

sys.setrecursionlimit(10000)

__all__ = [
    "init_nuts",
    "sample",
]

Step: TypeAlias = BlockedStep | CompoundStep


class SamplingIteratorCallback(Protocol):
    """Signature of the callable that may be passed to `pm.sample(callable=...)`."""

    def __call__(self, trace: IBaseTrace, draw: Draw):
        pass


_log = logging.getLogger(__name__)


def instantiate_steppers(
    model: Model,
    steps: list[Step],
    selected_steps: Mapping[type[BlockedStep], list[Any]],
    *,
    step_kwargs: dict[str, dict] | None = None,
    initial_point: PointType | None = None,
    compile_kwargs: dict | None = None,
) -> Step | list[Step]:
    """Instantiate steppers assigned to the model variables.

    This function is intended to be called automatically from ``sample()``, but
    may be called manually.

    Parameters
    ----------
    model : Model object
        A fully-specified model object.
    steps : list, array_like of shape (selected_steps, )
        A list of zero or more step function instances that have been assigned to some subset of
        the model's parameters.
    selected_steps : dict
        A dictionary that maps a step method class to a list of zero or more model variables.
    step_kwargs : dict, default=None
        Parameters for the samplers. Keys are the lower case names of
        the step method, values a dict of arguments. Defaults to None.

    Returns
    -------
    methods : list or step
        List of step methods associated with the model's variables, or step method
        if there is only one.
    """
    if step_kwargs is None:
        step_kwargs = {}

    used_keys = set()
    if selected_steps:
        if initial_point is None:
            initial_point = model.initial_point()

        for step_class, vars in selected_steps.items():
            if vars:
                name = getattr(step_class, "name")
                kwargs = step_kwargs.get(name, {})
                used_keys.add(name)
                step = step_class(
                    vars=vars,
                    model=model,
                    initial_point=initial_point,
                    compile_kwargs=compile_kwargs,
                    **kwargs,
                )
                steps.append(step)

    unused_args = set(step_kwargs).difference(used_keys)
    if unused_args:
        s = "s" if len(unused_args) > 1 else ""
        example_arg = sorted(unused_args)[0]
        example_step = (list(selected_steps.keys()) or pm.STEP_METHODS)[0]
        example_step_name = getattr(example_step, "name")
        raise ValueError(
            f"Invalid key{s} found in step_kwargs: {unused_args}. "
            "Keys must be step names and values valid kwargs for that stepper. "
            f'Did you mean {{"{example_step_name}": {{"{example_arg}": ...}}}}?'
        )

    if len(steps) == 1:
        return steps[0]

    return steps


def assign_step_methods(
    model: Model,
    step: Step | Sequence[Step] | None = None,
    methods: Sequence[type[BlockedStep]] | None = None,
) -> tuple[list[Step], dict[type[BlockedStep], list[Variable]]]:
    """Assign model variables to appropriate step methods.

    Passing a specified model will auto-assign its constituent value
    variables to step methods based on the characteristics of the respective
    random variables, and whether the logp can be differentiated with respect to it.

    This function is intended to be called automatically from ``sample()``, but
    may be called manually. Each step method passed should have a
    ``competence()`` method that returns an ordinal competence value
    corresponding to the variable passed to it. This value quantifies the
    appropriateness of the step method for sampling the variable.

    The outputs of this function can then be passed to `instantiate_steppers()`
    to initialize the assigned step samplers.

    Parameters
    ----------
    model : Model object
        A fully-specified model object.
    step : step function or iterable of step functions, optional
        One or more step functions that have been assigned to some subset of
        the model's parameters. Defaults to ``None`` (no assigned variables).
    methods : iterable of step method classes, optional
        The set of step methods from which the function may choose. Defaults
        to the main step methods provided by PyMC.

    Returns
    -------
    provided_steps: list of Step instances
        List of user provided instantiated step(s)
    assigned_steps: dict of Step class to Variable
        Dictionary with automatically selected step classes as keys and associated value variables as values
    """
    provided_steps: list[Step] = []
    assigned_vars: set[Variable] = set()

    if step is not None:
        if isinstance(step, BlockedStep | CompoundStep):
            provided_steps = [step]
        elif isinstance(step, Sequence):
            provided_steps = list(step)
        else:
            raise ValueError(f"Step should be a Step or a sequence of Steps, got {step}")

        for step in provided_steps:
            if not isinstance(step, BlockedStep | CompoundStep):
                if issubclass(step, BlockedStep | CompoundStep):
                    raise ValueError(f"Provided {step} was not initialized")
                else:
                    raise ValueError(f"{step} is not a Step instance")

            for var in step.vars:
                if var not in model.value_vars:
                    raise ValueError(
                        f"{var} assigned to {step} sampler is not a value variable in the model. "
                        "You can use `util.get_value_vars_from_user_vars` to parse user provided variables."
                    )
            assigned_vars = assigned_vars.union(set(step.vars))

    # Use competence classmethods to select step methods for remaining
    # variables
    methods_list: list[type[BlockedStep]] = list(methods or pm.STEP_METHODS)
    selected_steps: dict[type[BlockedStep], list] = {}
    model_logp = model.logp()

    for var in model.value_vars:
        if var not in assigned_vars:
            # determine if a gradient can be computed
            has_gradient = getattr(var, "dtype") not in discrete_types
            if has_gradient:
                try:
                    tg.grad(model_logp, var)  # type: ignore[arg-type]
                except (NotImplementedError, tg.NullTypeGradError):
                    has_gradient = False

            # select the best method
            rv_var = model.values_to_rvs[var]
            selected = max(
                methods_list,
                key=lambda method, var=rv_var, has_gradient=has_gradient: method._competence(  # type: ignore[misc]
                    var, has_gradient
                ),
            )
            selected_steps.setdefault(selected, []).append(var)

    return provided_steps, selected_steps


def _print_step_hierarchy(s: Step, level: int = 0) -> None:
    if isinstance(s, CompoundStep):
        _log.info(">" * level + "CompoundStep")
        for i in s.methods:
            _print_step_hierarchy(i, level + 1)
    else:
        varnames = ", ".join(
            [
                get_untransformed_name(v.name) if is_transformed_name(v.name) else v.name  # type: ignore[misc]
                for v in s.vars
            ]
        )
        _log.info(">" * level + f"{s.__class__.__name__}: [{varnames}]")


def all_continuous(vars):
    """Check that vars not include discrete variables."""
    if any((var.dtype in discrete_types) for var in vars):
        return False
    else:
        return True


def _sample_external_nuts(
    sampler: Literal["nutpie", "numpyro", "blackjax"],
    draws: int,
    tune: int,
    chains: int,
    target_accept: float,
    random_seed: RandomState | None,
    initvals: StartDict | Sequence[StartDict | None] | None,
    model: Model,
    var_names: Sequence[str] | None,
    progressbar: bool,
    idata_kwargs: dict | None,
    compute_convergence_checks: bool,
    nuts_sampler_kwargs: dict | None,
    **kwargs,
):
    if nuts_sampler_kwargs is None:
        nuts_sampler_kwargs = {}

    if sampler == "nutpie":
        try:
            import nutpie
        except ImportError as err:
            raise ImportError(
                "nutpie not found. Install it with conda install -c conda-forge nutpie"
            ) from err

        if initvals is not None:
            warnings.warn(
                "`initvals` are currently not passed to nutpie sampler. "
                "Use `init_mean` kwarg following nutpie specification instead.",
                UserWarning,
            )

        if idata_kwargs is not None:
            warnings.warn(
                "`idata_kwargs` are currently ignored by the nutpie sampler",
                UserWarning,
            )
        if var_names is not None:
            warnings.warn(
                "`var_names` are currently ignored by the nutpie sampler",
                UserWarning,
            )
        compile_kwargs = {}
        nuts_sampler_kwargs = nuts_sampler_kwargs.copy()
        for kwarg in ("backend", "gradient_backend"):
            if kwarg in nuts_sampler_kwargs:
                compile_kwargs[kwarg] = nuts_sampler_kwargs.pop(kwarg)
        compiled_model = nutpie.compile_pymc_model(
            model,
            **compile_kwargs,
        )
        t_start = time.time()
        idata = nutpie.sample(
            compiled_model,
            draws=draws,
            tune=tune,
            chains=chains,
            target_accept=target_accept,
            seed=_get_seeds_per_chain(random_seed, 1)[0],
            progress_bar=progressbar,
            **nuts_sampler_kwargs,
        )
        t_sample = time.time() - t_start
        # Temporary work-around. Revert once https://github.com/pymc-devs/nutpie/issues/74 is fixed
        # gather observed and constant data as nutpie.sample() has no access to the PyMC model
        coords, dims = coords_and_dims_for_inferencedata(model)
        constant_data = dict_to_dataset(
            find_constants(model),
            library=pm,
            coords=coords,
            dims=dims,
            default_dims=[],
        )
        observed_data = dict_to_dataset(
            find_observations(model),
            library=pm,
            coords=coords,
            dims=dims,
            default_dims=[],
        )
        attrs = make_attrs(
            {
                "sampling_time": t_sample,
                "tuning_steps": tune,
            },
            library=nutpie,
        )
        for k, v in attrs.items():
            idata.posterior.attrs[k] = v
        idata.add_groups(
            {"constant_data": constant_data, "observed_data": observed_data},
            coords=coords,
            dims=dims,
        )
        return idata

    elif sampler in ("numpyro", "blackjax"):
        import pymc.sampling.jax as pymc_jax

        idata = pymc_jax.sample_jax_nuts(
            draws=draws,
            tune=tune,
            chains=chains,
            target_accept=target_accept,
            random_seed=random_seed,
            initvals=initvals,
            model=model,
            var_names=var_names,
            progressbar=progressbar,
            nuts_sampler=sampler,
            idata_kwargs=idata_kwargs,
            compute_convergence_checks=compute_convergence_checks,
            **nuts_sampler_kwargs,
        )
        return idata

    else:
        raise ValueError(
            f"Sampler {sampler} not found. Choose one of ['nutpie', 'numpyro', 'blackjax', 'pymc']."
        )


@overload
def sample(
    draws: int = 1000,
    *,
    tune: int = 1000,
    chains: int | None = None,
    cores: int | None = None,
    random_seed: RandomState = None,
    progressbar: bool | ProgressBarType = True,
    progressbar_theme: Theme | None = default_progress_theme,
    step=None,
    var_names: Sequence[str] | None = None,
    nuts_sampler: Literal["pymc", "nutpie", "numpyro", "blackjax"] = "pymc",
    initvals: StartDict | Sequence[StartDict | None] | None = None,
    init: str = "auto",
    jitter_max_retries: int = 10,
    n_init: int = 200_000,
    trace: TraceOrBackend | None = None,
    discard_tuned_samples: bool = True,
    compute_convergence_checks: bool = True,
    keep_warning_stat: bool = False,
    return_inferencedata: Literal[True] = True,
    idata_kwargs: dict[str, Any] | None = None,
    nuts_sampler_kwargs: dict[str, Any] | None = None,
    callback=None,
    mp_ctx=None,
    blas_cores: int | None | Literal["auto"] = "auto",
    compile_kwargs: dict | None = None,
    **kwargs,
) -> InferenceData: ...


@overload
def sample(
    draws: int = 1000,
    *,
    tune: int = 1000,
    chains: int | None = None,
    cores: int | None = None,
    random_seed: RandomState = None,
    progressbar: bool | ProgressBarType = True,
    progressbar_theme: Theme | None = default_progress_theme,
    step=None,
    var_names: Sequence[str] | None = None,
    nuts_sampler: Literal["pymc", "nutpie", "numpyro", "blackjax"] = "pymc",
    initvals: StartDict | Sequence[StartDict | None] | None = None,
    init: str = "auto",
    jitter_max_retries: int = 10,
    n_init: int = 200_000,
    trace: TraceOrBackend | None = None,
    discard_tuned_samples: bool = True,
    compute_convergence_checks: bool = True,
    keep_warning_stat: bool = False,
    return_inferencedata: Literal[False],
    idata_kwargs: dict[str, Any] | None = None,
    nuts_sampler_kwargs: dict[str, Any] | None = None,
    callback=None,
    mp_ctx=None,
    model: Model | None = None,
    blas_cores: int | None | Literal["auto"] = "auto",
    compile_kwargs: dict | None = None,
    **kwargs,
) -> MultiTrace: ...


[docs] def sample( draws: int = 1000, *, tune: int = 1000, chains: int | None = None, cores: int | None = None, random_seed: RandomState = None, progressbar: bool | ProgressBarType = True, progressbar_theme: Theme | None = None, step=None, var_names: Sequence[str] | None = None, nuts_sampler: Literal["pymc", "nutpie", "numpyro", "blackjax"] = "pymc", initvals: StartDict | Sequence[StartDict | None] | None = None, init: str = "auto", jitter_max_retries: int = 10, n_init: int = 200_000, trace: TraceOrBackend | None = None, discard_tuned_samples: bool = True, compute_convergence_checks: bool = True, keep_warning_stat: bool = False, return_inferencedata: bool = True, idata_kwargs: dict[str, Any] | None = None, nuts_sampler_kwargs: dict[str, Any] | None = None, callback=None, mp_ctx=None, blas_cores: int | None | Literal["auto"] = "auto", model: Model | None = None, compile_kwargs: dict | None = None, **kwargs, ) -> InferenceData | MultiTrace | ZarrTrace: r"""Draw samples from the posterior using the given step methods. Multiple step methods are supported via compound step methods. Parameters ---------- draws : int The number of samples to draw. Defaults to 1000. The number of tuned samples are discarded by default. See ``discard_tuned_samples``. tune : int Number of iterations to tune, defaults to 1000. Samplers adjust the step sizes, scalings or similar during tuning. Tuning samples will be drawn in addition to the number specified in the ``draws`` argument, and will be discarded unless ``discard_tuned_samples`` is set to False. chains : int The number of chains to sample. Running independent chains is important for some convergence statistics and can also reveal multiple modes in the posterior. If ``None``, then set to either ``cores`` or 2, whichever is larger. cores : int The number of chains to run in parallel. If ``None``, set to the number of CPUs in the system, but at most 4. random_seed : int, array-like of int, or Generator, optional Random seed(s) used by the sampling steps. Each step will create its own :py:class:`~numpy.random.Generator` object to make its random draws in a way that is indepedent from all other steppers and all other chains. A ``TypeError`` will be raised if a legacy :py:class:`~numpy.random.RandomState` object is passed. We no longer support ``RandomState`` objects because their seeding mechanism does not allow easy spawning of new independent random streams that are needed by the step methods. progressbar: bool or ProgressType, optional How and whether to display the progress bar. If False, no progress bar is displayed. Otherwise, you can ask for one of the following: - "combined": A single progress bar that displays the total progress across all chains. Only timing information is shown. - "split": A separate progress bar for each chain. Only timing information is shown. - "combined+stats" or "stats+combined": A single progress bar displaying the total progress across all chains. Aggregate sample statistics are also displayed. - "split+stats" or "stats+split": A separate progress bar for each chain. Sample statistics for each chain are also displayed. If True, the default is "split+stats" is used. step : function or iterable of functions A step function or collection of functions. If there are variables without step methods, step methods for those variables will be assigned automatically. By default the NUTS step method will be used, if appropriate to the model. var_names : list of str, optional Names of variables to be stored in the trace. Defaults to all free variables and deterministics. nuts_sampler : str Which NUTS implementation to run. One of ["pymc", "nutpie", "blackjax", "numpyro"]. This requires the chosen sampler to be installed. All samplers, except "pymc", require the full model to be continuous. blas_cores: int or "auto" or None, default = "auto" The total number of threads blas and openmp functions should use during sampling. Setting it to "auto" will ensure that the total number of active blas threads is the same as the `cores` argument. If set to an integer, the sampler will try to use that total number of blas threads. If `blas_cores` is not divisible by `cores`, it might get rounded down. If set to None, this will keep the default behavior of whatever blas implementation is used at runtime. initvals : optional, dict, array of dict Dict or list of dicts with initial value strategies to use instead of the defaults from `Model.initial_values`. The keys should be names of transformed random variables. Initialization methods for NUTS (see ``init`` keyword) can overwrite the default. init : str Initialization method to use for auto-assigned NUTS samplers. See `pm.init_nuts` for a list of all options. This argument is ignored when manually passing the NUTS step method. Only applicable to the pymc nuts sampler. jitter_max_retries : int Maximum number of repeated attempts (per chain) at creating an initial matrix with uniform jitter that yields a finite probability. This applies to ``jitter+adapt_diag`` and ``jitter+adapt_full`` init methods. n_init : int Number of iterations of initializer. Only works for 'ADVI' init methods. trace : backend, optional A backend instance or None. If ``None``, a ``MultiTrace`` object with underlying ``NDArray`` trace objects is used. If ``trace`` is a :class:`~pymc.backends.zarr.ZarrTrace` instance, the drawn samples will be written onto the desired storage while sampling is on-going. This means sampling runs that, for whatever reason, die in the middle of their execution will write the partial results onto the storage. If the storage persist on disk, these results should be available even after a server crash. See :class:`~pymc.backends.zarr.ZarrTrace` for more information. discard_tuned_samples : bool Whether to discard posterior samples of the tune interval. compute_convergence_checks : bool, default=True Whether to compute sampler statistics like Gelman-Rubin and ``effective_n``. keep_warning_stat : bool If ``True`` the "warning" stat emitted by, for example, HMC samplers will be kept in the returned ``idata.sample_stats`` group. This leads to the ``idata`` not supporting ``.to_netcdf()`` or ``.to_zarr()`` and should only be set to ``True`` if you intend to use the "warning" objects right away. Defaults to ``False`` such that ``pm.drop_warning_stat`` is applied automatically, making the ``InferenceData`` compatible with saving. return_inferencedata : bool Whether to return the trace as an :class:`arviz:arviz.InferenceData` (True) object or a `MultiTrace` (False). Defaults to `True`. idata_kwargs : dict, optional Keyword arguments for :func:`pymc.to_inference_data` nuts_sampler_kwargs : dict, optional Keyword arguments for the sampling library that implements nuts. Only used when an external sampler is specified via the `nuts_sampler` kwarg. callback : function, default=None A function which gets called for every sample from the trace of a chain. The function is called with the trace and the current draw and will contain all samples for a single trace. the ``draw.chain`` argument can be used to determine which of the active chains the sample is drawn from. Sampling can be interrupted by throwing a ``KeyboardInterrupt`` in the callback. mp_ctx : multiprocessing.context.BaseContent A multiprocessing context for parallel sampling. See multiprocessing documentation for details. model : Model (optional if in ``with`` context) Model to sample from. The model needs to have free random variables. compile_kwargs: dict, optional Dictionary with keyword argument to pass to the functions compiled by the step methods. Returns ------- trace : pymc.backends.base.MultiTrace | pymc.backends.zarr.ZarrTrace | arviz.InferenceData A ``MultiTrace``, :class:`~arviz.InferenceData` or :class:`~pymc.backends.zarr.ZarrTrace` object that contains the samples. A ``ZarrTrace`` is only returned if the supplied ``trace`` argument is a ``ZarrTrace`` instance. Refer to :class:`~pymc.backends.zarr.ZarrTrace` for the benefits this backend provides. Notes ----- Optional keyword arguments can be passed to ``sample`` to be delivered to the ``step_method``\ s used during sampling. For example: 1. ``target_accept`` to NUTS: nuts={'target_accept':0.9} 2. ``transit_p`` to BinaryGibbsMetropolis: binary_gibbs_metropolis={'transit_p':.7} Note that available step names are: ``nuts``, ``hmc``, ``metropolis``, ``binary_metropolis``, ``binary_gibbs_metropolis``, ``categorical_gibbs_metropolis``, ``DEMetropolis``, ``DEMetropolisZ``, ``slice`` The NUTS step method has several options including: * target_accept : float in [0, 1]. The step size is tuned such that we approximate this acceptance rate. Higher values like 0.9 or 0.95 often work better for problematic posteriors. This argument can be passed directly to sample. * max_treedepth : The maximum depth of the trajectory tree * step_scale : float, default 0.25 The initial guess for the step size scaled down by :math:`1/n**(1/4)`, where n is the dimensionality of the parameter space Alternatively, if you manually declare the ``step_method``\ s, within the ``step`` kwarg, then you can address the ``step_method`` kwargs directly. e.g. for a CompoundStep comprising NUTS and BinaryGibbsMetropolis, you could send :: step = [ pm.NUTS([freeRV1, freeRV2], target_accept=0.9), pm.BinaryGibbsMetropolis([freeRV3], transit_p=0.7), ] You can find a full list of arguments in the docstring of the step methods. Examples -------- .. code:: ipython In [1]: import pymc as pm ...: n = 100 ...: h = 61 ...: alpha = 2 ...: beta = 2 In [2]: with pm.Model() as model: # context management ...: p = pm.Beta("p", alpha=alpha, beta=beta) ...: y = pm.Binomial("y", n=n, p=p, observed=h) ...: idata = pm.sample() In [3]: az.summary(idata, kind="stats") Out[3]: mean sd hdi_3% hdi_97% p 0.609 0.047 0.528 0.699 """ if "start" in kwargs: if initvals is not None: raise ValueError("Passing both `start` and `initvals` is not supported.") warnings.warn( "The `start` kwarg was renamed to `initvals` and can now do more. Please check the docstring.", FutureWarning, stacklevel=2, ) initvals = kwargs.pop("start") if nuts_sampler_kwargs is None: nuts_sampler_kwargs = {} if "target_accept" in kwargs: if "nuts" in kwargs and "target_accept" in kwargs["nuts"]: raise ValueError( "`target_accept` was defined twice. Please specify it either as a direct keyword argument or in the `nuts` kwarg." ) if "nuts" in kwargs: kwargs["nuts"]["target_accept"] = kwargs.pop("target_accept") else: kwargs["nuts"] = {"target_accept": kwargs.pop("target_accept")} if isinstance(trace, list): raise ValueError("Please use `var_names` keyword argument for partial traces.") # progressbar might be a string, which is used by the ProgressManager in the pymc samplers. External samplers and # ADVI initialization expect just a bool. progress_bool = bool(progressbar) model = modelcontext(model) if not model.free_RVs: raise SamplingError( "Cannot sample from the model, since the model does not contain any free variables." ) if cores is None: cores = min(4, _cpu_count()) if chains is None: chains = max(2, cores) if blas_cores == "auto": blas_cores = cores cores = min(cores, chains) num_blas_cores_per_chain: int | None joined_blas_limiter: Callable[[], Any] if blas_cores is None: joined_blas_limiter = contextlib.nullcontext num_blas_cores_per_chain = None elif isinstance(blas_cores, int): def joined_blas_limiter(): return threadpool_limits(limits=blas_cores) num_blas_cores_per_chain = blas_cores // cores else: raise ValueError( f"Invalid argument `blas_cores`, must be int, 'auto' or None: {blas_cores}" ) if random_seed == -1: warnings.warn( "Setting random_seed = -1 is deprecated. Pass `None` to not specify a seed.", FutureWarning, ) random_seed = None elif isinstance(random_seed, tuple | list): warnings.warn( "A list or tuple of random_seed no longer specifies the specific random_seed of each chain. " "Use a single seed instead.", UserWarning, ) rngs = get_random_generator(random_seed).spawn(chains) random_seed_list = [rng.integers(2**30) for rng in rngs] if not discard_tuned_samples and not return_inferencedata and not isinstance(trace, ZarrTrace): warnings.warn( "Tuning samples will be included in the returned `MultiTrace` object, which can lead to" " complications in your downstream analysis. Please consider to switch to `InferenceData`:\n" "`pm.sample(..., return_inferencedata=True)`", UserWarning, stacklevel=2, ) # small trace warning if draws == 0: msg = "Tuning was enabled throughout the whole trace." _log.warning(msg) elif draws < 100: msg = f"Only {draws} samples per chain. Reliable r-hat and ESS diagnostics require longer chains for accurate estimate." _log.warning(msg) provided_steps, selected_steps = assign_step_methods(model, step, methods=pm.STEP_METHODS) exclusive_nuts = ( # User provided an instantiated NUTS step, and nothing else is needed (not selected_steps and len(provided_steps) == 1 and isinstance(provided_steps[0], NUTS)) or # Only automatically selected NUTS step is needed ( not provided_steps and len(selected_steps) == 1 and issubclass(next(iter(selected_steps)), NUTS) ) ) if nuts_sampler != "pymc": if not exclusive_nuts: raise ValueError( "Model can not be sampled with NUTS alone. It either has discrete variables or a non-differentiable log-probability." ) with joined_blas_limiter(): return _sample_external_nuts( sampler=nuts_sampler, draws=draws, tune=tune, chains=chains, target_accept=kwargs.pop("nuts", {}).get("target_accept", 0.8), random_seed=random_seed, initvals=initvals, model=model, var_names=var_names, progressbar=progress_bool, idata_kwargs=idata_kwargs, compute_convergence_checks=compute_convergence_checks, nuts_sampler_kwargs=nuts_sampler_kwargs, **kwargs, ) if exclusive_nuts and not provided_steps: # Special path for NUTS initialization if "nuts" in kwargs: nuts_kwargs = kwargs.pop("nuts") [kwargs.setdefault(k, v) for k, v in nuts_kwargs.items()] with joined_blas_limiter(): initial_points, step = init_nuts( init=init, chains=chains, n_init=n_init, model=model, random_seed=random_seed_list, progressbar=progress_bool, jitter_max_retries=jitter_max_retries, tune=tune, initvals=initvals, compile_kwargs=compile_kwargs, **kwargs, ) else: # Get initial points ipfns = make_initial_point_fns_per_chain( model=model, overrides=initvals, jitter_rvs=set(), chains=chains, ) initial_points = [ipfn(seed) for ipfn, seed in zip(ipfns, random_seed_list)] # Instantiate automatically selected steps step = instantiate_steppers( model, steps=provided_steps, selected_steps=selected_steps, step_kwargs=kwargs, initial_point=initial_points[0], compile_kwargs=compile_kwargs, ) if isinstance(step, list): step = CompoundStep(step) if var_names is not None: trace_vars = [v for v in model.unobserved_RVs if v.name in var_names] trace_vars = model.replace_rvs_by_values(trace_vars) assert len(trace_vars) == len(var_names), "Not all var_names were found in the model" else: trace_vars = None # Create trace backends for each chain run, traces = init_traces( backend=trace, chains=chains, expected_length=draws + tune, step=step, trace_vars=trace_vars, initial_point=initial_points[0], model=model, tune=tune, ) sample_args = { # draws is now the total number of draws, including tuning "draws": draws + tune, "step": step, "start": initial_points, "traces": traces, "chains": chains, "tune": tune, "var_names": var_names, "progressbar": progressbar, "progressbar_theme": progressbar_theme, "model": model, "cores": cores, "callback": callback, "discard_tuned_samples": discard_tuned_samples, } parallel_args = { "mp_ctx": mp_ctx, "blas_cores": num_blas_cores_per_chain, } sample_args.update(kwargs) has_population_samplers = np.any( [ isinstance(m, PopulationArrayStepShared) for m in (step.methods if isinstance(step, CompoundStep) else [step]) ] ) parallel = cores > 1 and chains > 1 and not has_population_samplers # At some point it was decided that PyMC should not set a global seed by default, # unless the user specified a seed. This is a symptom of the fact that PyMC samplers # are built around global seeding. This branch makes sure we maintain this unspoken # rule. See https://github.com/pymc-devs/pymc/pull/1395. if parallel: # For parallel sampling we can pass the list of random seeds directly, as # global seeding will only be called inside each process sample_args["rngs"] = rngs else: # We pass None if the original random seed was None. The single core sampler # methods will only set a global seed when it is not None. sample_args["rngs"] = rngs t_start = time.time() if parallel: _log.info(f"Multiprocess sampling ({chains} chains in {cores} jobs)") _print_step_hierarchy(step) try: _mp_sample(**sample_args, **parallel_args) except pickle.PickleError: _log.warning("Could not pickle model, sampling singlethreaded.") _log.debug("Pickling error:", exc_info=True) parallel = False except AttributeError as e: if not str(e).startswith("AttributeError: Can't pickle"): raise _log.warning("Could not pickle model, sampling singlethreaded.") _log.debug("Pickling error:", exc_info=True) parallel = False if not parallel: if has_population_samplers: _log.info(f"Population sampling ({chains} chains)") _print_step_hierarchy(step) with joined_blas_limiter(): _sample_population( initial_points=initial_points, parallelize=cores > 1, **sample_args ) else: _log.info(f"Sequential sampling ({chains} chains in 1 job)") _print_step_hierarchy(step) with joined_blas_limiter(): _sample_many(**sample_args) t_sampling = time.time() - t_start # Packaging, validating and returning the result was extracted # into a function to make it easier to test and refactor. return _sample_return( run=run, traces=trace if isinstance(trace, ZarrTrace) else traces, tune=tune, t_sampling=t_sampling, discard_tuned_samples=discard_tuned_samples, compute_convergence_checks=compute_convergence_checks, return_inferencedata=return_inferencedata, keep_warning_stat=keep_warning_stat, idata_kwargs=idata_kwargs or {}, model=model, )
def _sample_return( *, run: RunType | None, traces: Sequence[IBaseTrace] | ZarrTrace, tune: int, t_sampling: float, discard_tuned_samples: bool, compute_convergence_checks: bool, return_inferencedata: bool, keep_warning_stat: bool, idata_kwargs: dict[str, Any], model: Model, ) -> InferenceData | MultiTrace | ZarrTrace: """Pick/slice chains, run diagnostics and convert to the desired return type. Final step of `pm.sampler`. """ if isinstance(traces, ZarrTrace): # Split warmup from posterior samples traces.split_warmup_groups() # Set sampling time traces.sampling_time = t_sampling # Compute number of actual draws per chain total_draws_per_chain = traces._sampling_state.draw_idx[:] n_chains = len(traces.straces) desired_tune = traces.tuning_steps desired_draw = len(traces.posterior.draw) tuning_steps_per_chain = np.clip(total_draws_per_chain, 0, desired_tune) draws_per_chain = total_draws_per_chain - tuning_steps_per_chain total_n_tune = tuning_steps_per_chain.sum() total_draws = draws_per_chain.sum() _log.info( f"Sampling {n_chains} chain{'s' if n_chains > 1 else ''} for {desired_tune:_d} desired tune and {desired_draw:_d} desired draw iterations " f"(Actually sampled {total_n_tune:_d} tune and {total_draws:_d} draws total) " f"took {t_sampling:.0f} seconds." ) if compute_convergence_checks or return_inferencedata: idata = traces.to_inferencedata(save_warmup=not discard_tuned_samples) log_likelihood = idata_kwargs.pop("log_likelihood", False) if log_likelihood: from pymc.stats.log_density import compute_log_likelihood idata = compute_log_likelihood( idata, var_names=None if log_likelihood is True else log_likelihood, extend_inferencedata=True, model=model, sample_dims=["chain", "draw"], progressbar=False, ) if compute_convergence_checks: warns = run_convergence_checks(idata, model) for warn in warns: traces._sampling_state.global_warnings.append(np.array([warn])) log_warnings(warns) if return_inferencedata: # By default we drop the "warning" stat which contains `SamplerWarning` # objects that can not be stored with `.to_netcdf()`. if not keep_warning_stat: return drop_warning_stat(idata) return idata return traces # Pick and slice chains to keep the maximum number of samples if discard_tuned_samples: traces, length = _choose_chains(traces, tune) else: traces, length = _choose_chains(traces, 0) mtrace = MultiTrace(traces)[:length] # count the number of tune/draw iterations that happened # ideally via the "tune" statistic, but not all samplers record it! if "tune" in mtrace.stat_names: # Get the tune stat directly from chain 0, sampler 0 stat = mtrace._straces[0].get_sampler_stats("tune", sampler_idx=0) stat = tuple(stat) n_tune = stat.count(True) n_draws = stat.count(False) else: # these may be wrong when KeyboardInterrupt happened, but they're better than nothing n_tune = min(tune, len(mtrace)) n_draws = max(0, len(mtrace) - n_tune) if discard_tuned_samples: mtrace = mtrace[n_tune:] # save metadata in SamplerReport mtrace.report._n_tune = n_tune mtrace.report._n_draws = n_draws mtrace.report._t_sampling = t_sampling n_chains = len(mtrace.chains) _log.info( f"Sampling {n_chains} chain{'s' if n_chains > 1 else ''} for {n_tune:_d} tune and {n_draws:_d} draw iterations " f"({n_tune * n_chains:_d} + {n_draws * n_chains:_d} draws total) " f"took {t_sampling:.0f} seconds." ) if compute_convergence_checks or return_inferencedata: ikwargs: dict[str, Any] = {"model": model, "save_warmup": not discard_tuned_samples} ikwargs.update(idata_kwargs) idata = pm.to_inference_data(mtrace, **ikwargs) if compute_convergence_checks: warns = run_convergence_checks(idata, model) mtrace.report._add_warnings(warns) log_warnings(warns) if return_inferencedata: # By default we drop the "warning" stat which contains `SamplerWarning` # objects that can not be stored with `.to_netcdf()`. if not keep_warning_stat: return drop_warning_stat(idata) return idata return mtrace def _check_start_shape(model, start: PointType): """Check that the prior evaluations and initial points have identical shapes. Parameters ---------- model : pm.Model The current model on context. start : dict The complete dictionary mapping (transformed) variable names to numeric initial values. """ e = "" try: actual_shapes = model.eval_rv_shapes() except NotImplementedError as ex: warnings.warn(f"Unable to validate shapes: {ex.args[0]}", UserWarning) return for name, sval in start.items(): ashape = actual_shapes.get(name) sshape = np.shape(sval) if ashape != tuple(sshape): e += f"\nExpected shape {ashape} for var '{name}', got: {sshape}" if e != "": raise ValueError(f"Bad shape in start point:{e}") def _sample_many( *, draws: int, chains: int, traces: Sequence[IBaseTrace], start: Sequence[PointType], rngs: Sequence[np.random.Generator], step: Step, callback: SamplingIteratorCallback | None = None, **kwargs, ): """Sample all chains sequentially. Parameters ---------- draws: int The number of samples to draw chains: int Total number of chains to sample. start: list Starting points for each chain rngs: list of random Generators A list of :py:class:`~numpy.random.Generator` objects, one for each chain step: function Step function """ initial_step_state = step.sampling_state progress_manager = ProgressBarManager( step_method=step, chains=chains, draws=draws - kwargs.get("tune", 0), tune=kwargs.get("tune", 0), progressbar=kwargs.get("progressbar", True), progressbar_theme=kwargs.get("progressbar_theme", default_progress_theme), ) with progress_manager: for i in range(chains): step.sampling_state = initial_step_state _sample( draws=draws, chain=i, start=start[i], step=step, trace=traces[i], rng=rngs[i], callback=callback, progress_manager=progress_manager, **kwargs, ) return def _sample( *, chain: int, rng: np.random.Generator, start: PointType, draws: int, step: Step, trace: IBaseTrace, tune: int, model: Model | None = None, callback=None, progress_manager: ProgressBarManager, **kwargs, ) -> None: """Sample one chain (singleprocess). Multiple step methods are supported via compound step methods. Parameters ---------- chain : int Number of the chain that the samples will belong to. random_seed : Generator Single random seed start : dict Starting point in parameter space (or partial point) draws : int The number of samples to draw step : Step Step class instance used to generate samples. trace A chain backend to record draws and stats. tune : int Number of iterations to tune. model : Model, optional PyMC model. If None, the model is taken from the current context. progress_manager: ProgressBarManager Helper class used to handle progress bar styling and updates """ sampling_gen = _iter_sample( draws=draws, step=step, start=start, trace=trace, chain=chain, tune=tune, model=model, rng=rng, callback=callback, ) try: for it, stats in enumerate(sampling_gen): progress_manager.update( chain_idx=chain, is_last=False, draw=it, stats=stats, tuning=it > tune ) if not progress_manager.combined_progress or chain == progress_manager.chains - 1: progress_manager.update( chain_idx=chain, is_last=True, draw=it, stats=stats, tuning=False ) except KeyboardInterrupt: pass def _iter_sample( *, draws: int, step: Step, start: PointType, trace: IBaseTrace, chain: int = 0, tune: int = 0, rng: np.random.Generator, model: Model | None = None, callback: SamplingIteratorCallback | None = None, ) -> Iterator[list[dict[str, Any]]]: """Sample one chain with a generator (singleprocess). Parameters ---------- draws : int The number of samples to draw step : function Step function start : dict Starting point in parameter space (or partial point). Must contain numeric (transformed) initial values for all (transformed) free variables. trace A chain backend to record draws and stats. chain : int, optional Chain number used to store sample in backend. tune : int, optional Number of iterations to tune (defaults to 0). model : Model (optional if in ``with`` context) random_seed : single random seed, optional Yields ------ stats : list of dict Dictionary of statistics returned by step sampler """ draws = int(draws) if draws < 1: raise ValueError("Argument `draws` must be greater than 0.") step.set_rng(rng) point = start if isinstance(trace, ZarrChain): trace.link_stepper(step) try: step.tune = bool(tune) if hasattr(step, "reset_tuning"): step.reset_tuning() for i in range(draws): if i == 0 and hasattr(step, "iter_count"): step.iter_count = 0 if i == tune: step.stop_tuning() point, stats = step.step(point) trace.record(point, stats) log_warning_stats(stats) if callback is not None: callback( trace=trace, draw=Draw(chain, i == draws, i, i < tune, stats, point), ) yield stats except (KeyboardInterrupt, BaseException): if isinstance(trace, ZarrChain): trace.record_sampling_state(step=step) trace.close() raise else: if isinstance(trace, ZarrChain): trace.record_sampling_state(step=step) trace.close() def _mp_sample( *, draws: int, tune: int, step, chains: int, cores: int, rngs: Sequence[np.random.Generator], start: Sequence[PointType], progressbar: bool = True, progressbar_theme: Theme | None = default_progress_theme, traces: Sequence[IBaseTrace], model: Model | None = None, callback: SamplingIteratorCallback | None = None, blas_cores: int | None = None, mp_ctx=None, **kwargs, ) -> None: """Sample all chains (multiprocess). Parameters ---------- draws : int The number of samples to draw tune : int Number of iterations to tune. step : function Step function chains : int The number of chains to sample. cores : int The number of chains to run in parallel. rngs: list of random Generators A list of :py:class:`~numpy.random.Generator` objects, one for each chain start : list Starting points for each chain. Dicts must contain numeric (transformed) initial values for all (transformed) free variables. progressbar : bool Whether or not to display a progress bar in the command line. progressbar_theme : Theme Optional custom theme for the progress bar. traces Recording backends for each chain. model : Model (optional if in ``with`` context) callback A function which gets called for every sample from the trace of a chain. The function is called with the trace and the current draw and will contain all samples for a single trace. the ``draw.chain`` argument can be used to determine which of the active chains the sample is drawn from. Sampling can be interrupted by throwing a ``KeyboardInterrupt`` in the callback. """ import pymc.sampling.parallel as ps # We did draws += tune in pm.sample draws -= tune zarr_chains: list[ZarrChain] | None = None zarr_recording = False if all(isinstance(trace, ZarrChain) for trace in traces): if isinstance(cast(ZarrChain, traces[0])._posterior.store, MemoryStore): warnings.warn( "Parallel sampling with MemoryStore zarr store wont write the processes " "step method sampling state. If you wish to be able to access the step " "method sampling state, please use a different storage backend, e.g. " "DirectoryStore or ZipStore" ) else: zarr_chains = cast(list[ZarrChain], traces) zarr_recording = True sampler = ps.ParallelSampler( draws=draws, tune=tune, chains=chains, cores=cores, rngs=rngs, start_points=start, step_method=step, progressbar=progressbar, progressbar_theme=progressbar_theme, blas_cores=blas_cores, mp_ctx=mp_ctx, zarr_chains=zarr_chains, ) try: try: with sampler: for draw in sampler: strace = traces[draw.chain] if not zarr_recording: # Zarr recording happens in each process strace.record(draw.point, draw.stats) log_warning_stats(draw.stats) if callback is not None: callback(trace=strace, draw=draw) except ps.ParallelSamplingError as error: for strace in traces: strace.close() raise except KeyboardInterrupt: pass finally: for strace in traces: strace.close() def _init_jitter( model: Model, initvals: StartDict | Sequence[StartDict | None] | None, seeds: Sequence[int] | np.ndarray, jitter: bool, jitter_max_retries: int, logp_fn: Callable[[PointType], np.ndarray] | None = None, ) -> list[PointType]: """Apply a uniform jitter in [-1, 1] to the test value as starting point in each chain. ``model.check_start_vals`` is used to test whether the jittered starting values produce a finite log probability. Invalid values are resampled unless `jitter_max_retries` is achieved, in which case the last sampled values are returned. Parameters ---------- jitter: bool Whether to apply jitter or not. jitter_max_retries : int Maximum number of repeated attempts at initializing values (per chain). logp_fn: Callable[[dict[str, np.ndarray]], np.ndarray | jax.Array] | None logp function that takes the output of initial point functions as input. If None, will use the results of model.compile_logp(). Returns ------- initial_points : list[dict[str, np.ndarray]] List of starting points for the sampler """ ipfns = make_initial_point_fns_per_chain( model=model, overrides=initvals, jitter_rvs=set(model.free_RVs) if jitter else set(), chains=len(seeds), ) if not jitter: return [ipfn(seed) for ipfn, seed in zip(ipfns, seeds)] if logp_fn is None: model_logp_fn: Callable[[PointType], np.ndarray] = model.compile_logp() else: model_logp_fn = logp_fn initial_points = [] for ipfn, seed in zip(ipfns, seeds): rng = np.random.default_rng(seed) for i in range(jitter_max_retries + 1): point = ipfn(seed) point_logp = model_logp_fn(point) if not np.isfinite(point_logp): if i == jitter_max_retries: # Print informative message on last attempted point model.check_start_vals(point) # Retry with a new seed seed = rng.integers(2**30, dtype=np.int64) else: break initial_points.append(point) return initial_points
[docs] def init_nuts( *, init: str = "auto", chains: int = 1, n_init: int = 500_000, model: Model | None = None, random_seed: RandomSeed = None, progressbar=True, jitter_max_retries: int = 10, tune: int | None = None, initvals: StartDict | Sequence[StartDict | None] | None = None, compile_kwargs: dict | None = None, **kwargs, ) -> tuple[Sequence[PointType], NUTS]: """Set up the mass matrix initialization for NUTS. NUTS convergence and sampling speed is extremely dependent on the choice of mass/scaling matrix. This function implements different methods for choosing or adapting the mass matrix. Parameters ---------- init : str Initialization method to use. * auto: Choose a default initialization method automatically. Currently, this is ``jitter+adapt_diag``, but this can change in the future. If you depend on the exact behaviour, choose an initialization method explicitly. * adapt_diag: Start with a identity mass matrix and then adapt a diagonal based on the variance of the tuning samples. All chains use the test value (usually the prior mean) as starting point. * jitter+adapt_diag: Same as ``adapt_diag``, but use test value plus a uniform jitter in [-1, 1] as starting point in each chain. * jitter+adapt_diag_grad: An experimental initialization method that uses information from gradients and samples during tuning. * advi+adapt_diag: Run ADVI and then adapt the resulting diagonal mass matrix based on the sample variance of the tuning samples. * advi: Run ADVI to estimate posterior mean and diagonal mass matrix. * advi_map: Initialize ADVI with MAP and use MAP as starting point. * map: Use the MAP as starting point. This is discouraged. * adapt_full: Adapt a dense mass matrix using the sample covariances. All chains use the test value (usually the prior mean) as starting point. * jitter+adapt_full: Same as ``adapt_full``, but use test value plus a uniform jitter in [-1, 1] as starting point in each chain. chains : int Number of jobs to start. initvals : optional, dict or list of dicts Dict or list of dicts with initial values to use instead of the defaults from `Model.initial_values`. The keys should be names of transformed random variables. n_init : int Number of iterations of initializer. Only works for 'ADVI' init methods. model : Model (optional if in ``with`` context) random_seed : int, array-like of int, RandomState or Generator, optional Seed for the random number generator. progressbar : bool Whether or not to display a progressbar for advi sampling. jitter_max_retries : int Maximum number of repeated attempts (per chain) at creating an initial matrix with uniform jitter that yields a finite probability. This applies to ``jitter+adapt_diag`` and ``jitter+adapt_full`` init methods. **kwargs : keyword arguments Extra keyword arguments are forwarded to pymc.NUTS. Returns ------- initial_points : list Starting points for each chain. nuts_sampler : ``pymc.step_methods.NUTS`` Instantiated and initialized NUTS sampler object """ model = modelcontext(model) vars = kwargs.get("vars", model.value_vars) if set(vars) != set(model.value_vars): raise ValueError("Must use init_nuts on all variables of a model.") if not all_continuous(vars): raise ValueError("init_nuts can only be used for models with continuous variables.") if not isinstance(init, str): raise TypeError("init must be a string.") init = init.lower() if init == "auto": init = "jitter+adapt_diag" if compile_kwargs is None: compile_kwargs = {} random_seed_list = _get_seeds_per_chain(random_seed, chains) _log.info(f"Initializing NUTS using {init}...") cb = [] if "advi" in init: cb = [ pm.callbacks.CheckParametersConvergence(tolerance=1e-2, diff="absolute"), pm.callbacks.CheckParametersConvergence(tolerance=1e-2, diff="relative"), ] logp_dlogp_func = model.logp_dlogp_function(ravel_inputs=True, **compile_kwargs) logp_dlogp_func.trust_input = True def model_logp_fn(ip: PointType) -> np.ndarray: q, _ = DictToArrayBijection.map(ip) return logp_dlogp_func([q], extra_vars={})[0] initial_points = _init_jitter( model, initvals, seeds=random_seed_list, jitter="jitter" in init, jitter_max_retries=jitter_max_retries, logp_fn=model_logp_fn, ) apoints = [DictToArrayBijection.map(point) for point in initial_points] apoints_data = [apoint.data for apoint in apoints] potential: quadpotential.QuadPotential if init == "adapt_diag": mean = np.mean(apoints_data, axis=0) var = np.ones_like(mean) n = len(var) potential = quadpotential.QuadPotentialDiagAdapt(n, mean, var, 10, rng=random_seed_list[0]) elif init == "jitter+adapt_diag": mean = np.mean(apoints_data, axis=0) var = np.ones_like(mean) n = len(var) potential = quadpotential.QuadPotentialDiagAdapt(n, mean, var, 10, rng=random_seed_list[0]) elif init == "jitter+adapt_diag_grad": mean = np.mean(apoints_data, axis=0) var = np.ones_like(mean) n = len(var) if tune is not None and tune > 250: stop_adaptation = tune - 50 else: stop_adaptation = None potential = quadpotential.QuadPotentialDiagAdaptExp( n, mean, alpha=0.02, use_grads=True, stop_adaptation=stop_adaptation, rng=random_seed_list[0], ) elif init == "advi+adapt_diag": approx = pm.fit( random_seed=random_seed_list[0], n=n_init, method="advi", model=model, callbacks=cb, progressbar=progressbar, obj_optimizer=pm.adagrad_window, compile_kwargs=compile_kwargs, ) approx_sample = approx.sample( draws=chains, random_seed=random_seed_list[0], return_inferencedata=False ) initial_points = [approx_sample[i] for i in range(chains)] std_apoint = approx.std.eval() cov = std_apoint**2 mean = approx.mean.get_value() weight = 50 n = len(cov) potential = quadpotential.QuadPotentialDiagAdapt( n, mean, cov, weight, rng=random_seed_list[0] ) elif init == "advi": approx = pm.fit( random_seed=random_seed_list[0], n=n_init, method="advi", model=model, callbacks=cb, progressbar=progressbar, obj_optimizer=pm.adagrad_window, compile_kwargs=compile_kwargs, ) approx_sample = approx.sample( draws=chains, random_seed=random_seed_list[0], return_inferencedata=False ) initial_points = [approx_sample[i] for i in range(chains)] cov = approx.std.eval() ** 2 potential = quadpotential.QuadPotentialDiag(cov, rng=random_seed_list[0]) elif init == "advi_map": start = pm.find_MAP(include_transformed=True, seed=random_seed_list[0]) approx = pm.MeanField(model=model, start=start) pm.fit( random_seed=random_seed_list[0], n=n_init, method=pm.KLqp(approx), callbacks=cb, progressbar=progressbar, obj_optimizer=pm.adagrad_window, compile_kwargs=compile_kwargs, ) approx_sample = approx.sample( draws=chains, random_seed=random_seed_list[0], return_inferencedata=False ) initial_points = [approx_sample[i] for i in range(chains)] cov = approx.std.eval() ** 2 potential = quadpotential.QuadPotentialDiag(cov, rng=random_seed_list[0]) elif init == "map": start = pm.find_MAP(include_transformed=True, seed=random_seed_list[0]) cov = -pm.find_hessian(point=start, negate_output=False) initial_points = [start] * chains potential = quadpotential.QuadPotentialFull(cov, rng=random_seed_list[0]) elif init == "adapt_full": mean = np.mean(apoints_data * chains, axis=0) initial_point = initial_points[0] initial_point_model_size = sum(initial_point[n.name].size for n in model.value_vars) cov = np.eye(initial_point_model_size) potential = quadpotential.QuadPotentialFullAdapt( initial_point_model_size, mean, cov, 10, rng=random_seed_list[0] ) elif init == "jitter+adapt_full": mean = np.mean(apoints_data, axis=0) initial_point = initial_points[0] initial_point_model_size = sum(initial_point[n.name].size for n in model.value_vars) cov = np.eye(initial_point_model_size) potential = quadpotential.QuadPotentialFullAdapt( initial_point_model_size, mean, cov, 10, rng=random_seed_list[0] ) else: raise ValueError(f"Unknown initializer: {init}.") step = pm.NUTS( potential=potential, model=model, rng=random_seed_list[0], initial_point=initial_points[0], logp_dlogp_func=logp_dlogp_func, **kwargs, ) # Filter deterministics from initial_points value_var_names = [var.name for var in model.value_vars] initial_points = [ {k: v for k, v in initial_point.items() if k in value_var_names} for initial_point in initial_points ] return initial_points, step