Source code for

#   Copyright 2020 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
#   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.

import collections
import io
import os
import pkgutil
import urllib.request
import warnings

from copy import copy
from typing import Any, Dict, List, Optional, Sequence, Tuple, Union, cast

import numpy as np
import pytensor
import pytensor.tensor as at

from pytensor.compile.sharedvalue import SharedVariable
from pytensor.graph.basic import Apply
from pytensor.tensor.type import TensorType
from pytensor.tensor.var import TensorConstant, TensorVariable

import pymc as pm

from pymc.pytensorf import convert_observed_data

__all__ = [
BASE_URL = "{filename}"

[docs]def get_data(filename): """Returns a BytesIO object for a package data file. Parameters ---------- filename: str file to load Returns ------- BytesIO of the data """ data_pkg = "pymc.tests" try: content = pkgutil.get_data(data_pkg, os.path.join("data", filename)) except FileNotFoundError: with urllib.request.urlopen(BASE_URL.format(filename=filename)) as handle: content = return io.BytesIO(content)
class GenTensorVariable(TensorVariable): def __init__(self, op, type, name=None): super().__init__(type=type, owner=None, name=name) self.op = op def set_gen(self, gen): self.op.set_gen(gen) def set_default(self, value): self.op.set_default(value) def clone(self): cp = self.__class__(self.op, self.type, cp.tag = copy(self.tag) return cp
[docs]class GeneratorAdapter: """ Helper class that helps to infer data type of generator with looking at the first item, preserving the order of the resulting generator """
[docs] def make_variable(self, gop, name=None): var = GenTensorVariable(gop, self.tensortype, name) var.tag.test_value = self.test_value return var
[docs] def __init__(self, generator): if not pm.vartypes.isgenerator(generator): raise TypeError("Object should be generator like") self.test_value = pm.smartfloatX(copy(next(generator))) # make pickling potentially possible self._yielded_test_value = False self.gen = generator self.tensortype = TensorType(self.test_value.dtype, ((False,) * self.test_value.ndim))
# python3 generator def __next__(self): if not self._yielded_test_value: self._yielded_test_value = True return self.test_value else: return pm.smartfloatX(copy(next(self.gen))) # python2 generator next = __next__ def __iter__(self): return self def __eq__(self, other): return id(self) == id(other) def __hash__(self): return hash(id(self))
[docs]class Minibatch(TensorVariable): """Multidimensional minibatch that is pure TensorVariable Parameters ---------- data: np.ndarray initial data batch_size: ``int`` or ``List[int|tuple(size, random_seed)]`` batch size for inference, random seed is needed for child random generators dtype: ``str`` cast data to specific type broadcastable: tuple[bool] change broadcastable pattern that defaults to ``(False, ) * ndim`` name: ``str`` name for tensor, defaults to "Minibatch" random_seed: ``int`` random seed that is used by default update_shared_f: ``callable`` returns :class:`ndarray` that will be carefully stored to underlying shared variable you can use it to change source of minibatches programmatically in_memory_size: ``int`` or ``List[int|slice|Ellipsis]`` data size for storing in ``pytensor.shared`` Attributes ---------- shared: shared tensor Used for storing data minibatch: minibatch tensor Used for training Notes ----- Below is a common use case of Minibatch with variational inference. Importantly, we need to make PyMC "aware" that a minibatch is being used in inference. Otherwise, we will get the wrong :math:`logp` for the model. the density of the model ``logp`` that is affected by Minibatch. See more in the examples below. To do so, we need to pass the ``total_size`` parameter to the observed node, which correctly scales the density of the model ``logp`` that is affected by Minibatch. See more in the examples below. Examples -------- Consider we have `data` as follows: >>> data = np.random.rand(100, 100) if we want a 1d slice of size 10 we do >>> x = Minibatch(data, batch_size=10) Note that your data is cast to ``floatX`` if it is not integer type But you still can add the ``dtype`` kwarg for :class:`Minibatch` if you need more control. If we want 10 sampled rows and columns ``[(size, seed), (size, seed)]`` we can use >>> x = Minibatch(data, batch_size=[(10, 42), (10, 42)], dtype='int32') >>> assert str(x.dtype) == 'int32' Or, more simply, we can use the default random seed = 42 ``[size, size]`` >>> x = Minibatch(data, batch_size=[10, 10]) In the above, `x` is a regular :class:`TensorVariable` that supports any math operations: >>> assert x.eval().shape == (10, 10) You can pass the Minibatch `x` to your desired model: >>> with pm.Model() as model: ... mu = pm.Flat('mu') ... sigma = pm.HalfNormal('sigma') ... lik = pm.Normal('lik', mu, sigma, observed=x, total_size=(100, 100)) Then you can perform regular Variational Inference out of the box >>> with model: ... approx = Important note: :class:``Minibatch`` has ``shared``, and ``minibatch`` attributes you can call later: >>> x.set_value(np.random.laplace(size=(100, 100))) and minibatches will be then from new storage it directly affects ``x.shared``. A less convenient convenient, but more explicit, way to achieve the same thing: >>> x.shared.set_value(pm.floatX(np.random.laplace(size=(100, 100)))) The programmatic way to change storage is as follows I import ``partial`` for simplicity >>> from functools import partial >>> datagen = partial(np.random.laplace, size=(100, 100)) >>> x = Minibatch(datagen(), batch_size=10, update_shared_f=datagen) >>> x.update_shared() To be more concrete about how we create a minibatch, here is a demo: 1. create a shared variable >>> shared = pytensor.shared(data) 2. take a random slice of size 10: >>> ridx = pm.at_rng().uniform(size=(10,), low=0, high=data.shape[0]-1e-10).astype('int64') 3) take the resulting slice: >>> minibatch = shared[ridx] That's done. Now you can use this minibatch somewhere else. You can see that the implementation does not require a fixed shape for the shared variable. Feel free to use that if needed. *FIXME: What is "that" which we can use here? A fixed shape? Should this say "but feel free to put a fixed shape on the shared variable, if appropriate?"* Suppose you need to make some replacements in the graph, e.g. change the minibatch to testdata >>> node = x ** 2 # arbitrary expressions on minibatch `x` >>> testdata = pm.floatX(np.random.laplace(size=(1000, 10))) Then you should create a `dict` with replacements: >>> replacements = {x: testdata} >>> rnode = pytensor.clone_replace(node, replacements) >>> assert (testdata ** 2 == rnode.eval()).all() *FIXME: In the following, what is the **reason** to replace the Minibatch variable with its shared variable? And in the following, the `rnode` is a **new** node, not a modification of a previously existing node, correct?* To replace a minibatch with its shared variable you should do the same things. The Minibatch variable is accessible through the `minibatch` attribute. For example >>> replacements = {x.minibatch: x.shared} >>> rnode = pytensor.clone_replace(node, replacements) For more complex slices some more code is needed that can seem not so clear >>> moredata = np.random.rand(10, 20, 30, 40, 50) The default ``total_size`` that can be passed to PyMC random node is then ``(10, 20, 30, 40, 50)`` but can be less verbose in some cases 1. Advanced indexing, ``total_size = (10, Ellipsis, 50)`` >>> x = Minibatch(moredata, [2, Ellipsis, 10]) We take the slice only for the first and last dimension >>> assert x.eval().shape == (2, 20, 30, 40, 10) 2. Skipping a particular dimension, ``total_size = (10, None, 30)``: >>> x = Minibatch(moredata, [2, None, 20]) >>> assert x.eval().shape == (2, 20, 20, 40, 50) 3. Mixing both of these together, ``total_size = (10, None, 30, Ellipsis, 50)``: >>> x = Minibatch(moredata, [2, None, 20, Ellipsis, 10]) >>> assert x.eval().shape == (2, 20, 20, 40, 10) """ RNG: Dict[str, List[Any]] = collections.defaultdict(list)
[docs] @pytensor.config.change_flags(compute_test_value="raise") def __init__( self, data, batch_size=128, dtype=None, broadcastable=None, shape=None, name="Minibatch", random_seed=42, update_shared_f=None, in_memory_size=None, ): if dtype is None: data = pm.smartfloatX(np.asarray(data)) else: data = np.asarray(data, dtype) in_memory_slc = self.make_static_slices(in_memory_size) self.shared = pytensor.shared(data[tuple(in_memory_slc)]) self.update_shared_f = update_shared_f self.random_slc = self.make_random_slices(self.shared.shape, batch_size, random_seed) minibatch = self.shared[self.random_slc] if broadcastable is not None: warnings.warn( "Minibatch `broadcastable` argument is deprecated. Use `shape` instead", FutureWarning, ) assert shape is None shape = [1 if b else None for b in broadcastable] if shape is not None: minibatch = at.specify_shape(minibatch, shape) self.minibatch = minibatch super().__init__(self.minibatch.type, None, None, name=name) Apply(pytensor.compile.view_op, inputs=[self.minibatch], outputs=[self]) self.tag.test_value = copy(self.minibatch.tag.test_value)
[docs] def rslice(self, total, size, seed): if size is None: return slice(None) elif isinstance(size, int): rng = pm.at_rng(seed) Minibatch.RNG[id(self)].append(rng) return rng.uniform(size=(size,), low=0.0, high=pm.floatX(total) - 1e-16).astype("int64") else: raise TypeError("Unrecognized size type, %r" % size)
def __del__(self): del Minibatch.RNG[id(self)]
[docs] @staticmethod def make_static_slices(user_size): if user_size is None: return [Ellipsis] elif isinstance(user_size, int): return slice(None, user_size) elif isinstance(user_size, (list, tuple)): slc = list() for i in user_size: if isinstance(i, int): slc.append(i) elif i is None: slc.append(slice(None)) elif i is Ellipsis: slc.append(Ellipsis) elif isinstance(i, slice): slc.append(i) else: raise TypeError("Unrecognized size type, %r" % user_size) return slc else: raise TypeError("Unrecognized size type, %r" % user_size)
[docs] def make_random_slices(self, in_memory_shape, batch_size, default_random_seed): if batch_size is None: return [Ellipsis] elif isinstance(batch_size, int): slc = [self.rslice(in_memory_shape[0], batch_size, default_random_seed)] elif isinstance(batch_size, (list, tuple)): def check(t): if t is Ellipsis or t is None: return True else: if isinstance(t, (tuple, list)): if not len(t) == 2: return False else: return isinstance(t[0], int) and isinstance(t[1], int) elif isinstance(t, int): return True else: return False # end check definition if not all(check(t) for t in batch_size): raise TypeError( "Unrecognized `batch_size` type, expected " "int or List[int|tuple(size, random_seed)] where " "size and random seed are both ints, got %r" % batch_size ) batch_size = [(i, default_random_seed) if isinstance(i, int) else i for i in batch_size] shape = in_memory_shape if Ellipsis in batch_size: sep = batch_size.index(Ellipsis) begin = batch_size[:sep] end = batch_size[sep + 1 :] if Ellipsis in end: raise ValueError( "Double Ellipsis in `batch_size` is restricted, got %r" % batch_size ) if len(end) > 0: shp_mid = shape[sep : -len(end)] mid = [at.arange(s) for s in shp_mid] else: mid = [] else: begin = batch_size end = [] mid = [] if (len(begin) + len(end)) > len(in_memory_shape.eval()): raise ValueError( "Length of `batch_size` is too big, " "number of ints is bigger that ndim, got %r" % batch_size ) if len(end) > 0: shp_end = shape[-len(end) :] else: shp_end = np.asarray([]) shp_begin = shape[: len(begin)] slc_begin = [ self.rslice(shp_begin[i], t[0], t[1]) if t is not None else at.arange(shp_begin[i]) for i, t in enumerate(begin) ] slc_end = [ self.rslice(shp_end[i], t[0], t[1]) if t is not None else at.arange(shp_end[i]) for i, t in enumerate(end) ] slc = slc_begin + mid + slc_end else: raise TypeError("Unrecognized size type, %r" % batch_size) return pm.pytensorf.ix_(*slc)
[docs] def update_shared(self): if self.update_shared_f is None: raise NotImplementedError("No `update_shared_f` was provided to `__init__`") self.set_value(np.asarray(self.update_shared_f(), self.dtype))
[docs] def set_value(self, value): self.shared.set_value(np.asarray(value, self.dtype))
[docs] def clone(self): ret = self.type() = ret.tag = copy(self.tag) return ret
[docs]def align_minibatches(batches=None): if batches is None: for rngs in Minibatch.RNG.values(): for rng in rngs: rng.seed() else: for b in batches: if not isinstance(b, Minibatch): raise TypeError(f"{b} is not a Minibatch") for rng in Minibatch.RNG[id(b)]: rng.seed()
def determine_coords( model, value, dims: Optional[Sequence[Optional[str]]] = None, coords: Optional[Dict[str, Sequence]] = None, ) -> Tuple[Dict[str, Sequence], Sequence[Optional[str]]]: """Determines coordinate values from data or the model (via ``dims``).""" if coords is None: coords = {} # If value is a df or a series, we interpret the index as coords: if hasattr(value, "index"): dim_name = None if dims is not None: dim_name = dims[0] if dim_name is None and is not None: dim_name = if dim_name is not None: coords[dim_name] = value.index # If value is a df, we also interpret the columns as coords: if hasattr(value, "columns"): dim_name = None if dims is not None: dim_name = dims[1] if dim_name is None and is not None: dim_name = if dim_name is not None: coords[dim_name] = value.columns if isinstance(value, np.ndarray) and dims is not None: if len(dims) != value.ndim: raise pm.exceptions.ShapeError( "Invalid data shape. The rank of the dataset must match the " "length of `dims`.", actual=value.shape, expected=value.ndim, ) for size, dim in zip(value.shape, dims): coord = model.coords.get(dim, None) if coord is None and dim is not None: coords[dim] = range(size) if dims is None: # TODO: Also determine dim names from the index dims = [None] * np.ndim(value) return coords, dims
[docs]def ConstantData( name: str, value, *, dims: Optional[Sequence[str]] = None, coords: Optional[Dict[str, Sequence]] = None, export_index_as_coords=False, **kwargs, ) -> TensorConstant: """Alias for ``pm.Data(..., mutable=False)``. Registers the ``value`` as a :class:`~pytensor.tensor.TensorConstant` with the model. For more information, please reference :class:`pymc.Data`. """ var = Data( name, value, dims=dims, coords=coords, export_index_as_coords=export_index_as_coords, mutable=False, **kwargs, ) return cast(TensorConstant, var)
[docs]def MutableData( name: str, value, *, dims: Optional[Sequence[str]] = None, coords: Optional[Dict[str, Sequence]] = None, export_index_as_coords=False, **kwargs, ) -> SharedVariable: """Alias for ``pm.Data(..., mutable=True)``. Registers the ``value`` as a :class:`~pytensor.compile.sharedvalue.SharedVariable` with the model. For more information, please reference :class:`pymc.Data`. """ var = Data( name, value, dims=dims, coords=coords, export_index_as_coords=export_index_as_coords, mutable=True, **kwargs, ) return cast(SharedVariable, var)
[docs]def Data( name: str, value, *, dims: Optional[Sequence[str]] = None, coords: Optional[Dict[str, Sequence]] = None, export_index_as_coords=False, mutable: Optional[bool] = None, **kwargs, ) -> Union[SharedVariable, TensorConstant]: """Data container that registers a data variable with the model. Depending on the ``mutable`` setting (default: True), the variable is registered as a :class:`~pytensor.compile.sharedvalue.SharedVariable`, enabling it to be altered in value and shape, but NOT in dimensionality using :func:`pymc.set_data`. To set the value of the data container variable, check out :meth:`pymc.Model.set_data`. When making predictions or doing posterior predictive sampling, the shape of the registered data variable will most likely need to be changed. If you encounter an PyTensor shape mismatch error, refer to the documentation for :meth:`pymc.model.set_data`. For more information, read the notebook :ref:`nb:data_container`. Parameters ---------- name : str The name for this variable. value : array_like or pandas.Series, pandas.Dataframe A value to associate with this variable. dims : str or tuple of str, optional Dimension names of the random variables (as opposed to the shapes of these random variables). Use this when ``value`` is a pandas Series or DataFrame. The ``dims`` will then be the name of the Series / DataFrame's columns. See ArviZ documentation for more information about dimensions and coordinates: :ref:`arviz:quickstart`. If this parameter is not specified, the random variables will not have dimension names. coords : dict, optional Coordinate values to set for new dimensions introduced by this ``Data`` variable. export_index_as_coords : bool, default=False If True, the ``Data`` container will try to infer what the coordinates and dimension names should be if there is an index in ``value``. mutable : bool, optional Switches between creating a :class:`~pytensor.compile.sharedvalue.SharedVariable` (``mutable=True``) vs. creating a :class:`~pytensor.tensor.TensorConstant` (``mutable=False``). Consider using :class:`pymc.ConstantData` or :class:`pymc.MutableData` as less verbose alternatives to ``pm.Data(..., mutable=...)``. If this parameter is not specified, the value it takes will depend on the version of the package. Since ``v4.1.0`` the default value is ``mutable=False``, with previous versions having ``mutable=True``. **kwargs : dict, optional Extra arguments passed to :func:`pytensor.shared`. Examples -------- >>> import pymc as pm >>> import numpy as np >>> # We generate 10 datasets >>> true_mu = [np.random.randn() for _ in range(10)] >>> observed_data = [mu + np.random.randn(20) for mu in true_mu] >>> with pm.Model() as model: ... data = pm.MutableData('data', observed_data[0]) ... mu = pm.Normal('mu', 0, 10) ... pm.Normal('y', mu=mu, sigma=1, observed=data) >>> # Generate one trace for each dataset >>> idatas = [] >>> for data_vals in observed_data: ... with model: ... # Switch out the observed dataset ... model.set_data('data', data_vals) ... idatas.append(pm.sample()) """ if coords is None: coords = {} if isinstance(value, list): value = np.array(value) # Add data container to the named variables of the model. model = pm.Model.get_context(error_if_none=False) if model is None: raise TypeError( "No model on context stack, which is needed to instantiate a data container. " "Add variable inside a 'with model:' block." ) name = model.name_for(name) # `convert_observed_data` takes care of parameter `value` and # transforms it to something digestible for PyTensor. arr = convert_observed_data(value) if mutable is None: warnings.warn( "The `mutable` kwarg was not specified. Before v4.1.0 it defaulted to `pm.Data(mutable=True)`," " which is equivalent to using `pm.MutableData()`." " In v4.1.0 the default changed to `pm.Data(mutable=False)`, equivalent to `pm.ConstantData`." " Use `pm.ConstantData`/`pm.MutableData` or pass `pm.Data(..., mutable=False/True)` to avoid this warning.", UserWarning, ) mutable = False if mutable: x = pytensor.shared(arr, name, **kwargs) else: x = at.as_tensor_variable(arr, name, **kwargs) if isinstance(dims, str): dims = (dims,) if not (dims is None or len(dims) == x.ndim): raise pm.exceptions.ShapeError( "Length of `dims` must match the dimensions of the dataset.", actual=len(dims), expected=x.ndim, ) # Optionally infer coords and dims from the input value. if export_index_as_coords: coords, dims = determine_coords(model, value, dims) if dims: if not mutable: # Use the dimension lengths from the before it was tensorified. # These can still be tensors, but in many cases they are numeric. xshape = np.shape(arr) else: xshape = x.shape # Register new dimension lengths for d, dname in enumerate(dims): if dname not in model.dim_lengths: model.add_coord( name=dname, # Note: Coordinate values can't be taken from # the value, because it could be N-dimensional. values=coords.get(dname, None), mutable=mutable, length=xshape[d], ) model.add_named_variable(x, dims=dims) return x