Fix pymc3 to work with latest theano-pymc master

pymc3/sampling_jax.py

@@ -12,8 +12,8 @@
 import numpy as np
 import pandas as pd
 import theano
-import theano.sandbox.jax_linker
-import theano.sandbox.jaxify
+
+from theano.link.jax.jax_dispatch import jax_funcify
 
 import pymc3 as pm
 
@@ -46,7 +46,7 @@ def sample_tfp_nuts(
     seed = jax.random.PRNGKey(random_seed)
 
     fgraph = theano.gof.FunctionGraph(model.free_RVs, [model.logpt])
-    fns = theano.sandbox.jaxify.jax_funcify(fgraph)
+    fns = jax_funcify(fgraph)
     logp_fn_jax = fns[0]
 
     rv_names = [rv.name for rv in model.free_RVs]
@@ -131,7 +131,7 @@ def sample_numpyro_nuts(
     seed = jax.random.PRNGKey(random_seed)
 
     fgraph = theano.gof.FunctionGraph(model.free_RVs, [model.logpt])
-    fns = theano.sandbox.jaxify.jax_funcify(fgraph)
+    fns = jax_funcify(fgraph)
     logp_fn_jax = fns[0]
 
     rv_names = [rv.name for rv in model.free_RVs]

pymc3/tests/helpers.py

@@ -20,7 +20,7 @@
 import theano
 
 from theano.gradient import verify_grad as tt_verify_grad
-from theano.sandbox.rng_mrg import MRG_RandomStreams
+from theano.sandbox.rng_mrg import MRG_RandomStream as RandomStream
 
 from pymc3.theanof import set_tt_rng, tt_rng
 
@@ -35,7 +35,7 @@ def setup_class(cls):
     def setup_method(self):
         nr.seed(self.random_seed)
         self.old_tt_rng = tt_rng()
-        set_tt_rng(MRG_RandomStreams(self.random_seed))
+        set_tt_rng(RandomStream(self.random_seed))
 
     def teardown_method(self):
         set_tt_rng(self.old_tt_rng)

pymc3/tests/test_variational_inference.py

@@ -27,7 +27,7 @@
 
 from pymc3.tests import models
 from pymc3.tests.helpers import not_raises
-from pymc3.theanof import change_flags, intX
+from pymc3.theanof import intX
 from pymc3.variational import flows, opvi
 from pymc3.variational.approximations import (
     Empirical,
@@ -533,17 +533,20 @@ def test_scale_cost_to_minibatch_works(aux_total_size):
     sigma = 1.0
     y_obs = np.array([1.6, 1.4])
     beta = len(y_obs) / float(aux_total_size)
-    post_mu = np.array([1.88], dtype=theano.config.floatX)
-    post_sigma = np.array([1], dtype=theano.config.floatX)
 
     # TODO: theano_config
     # with pm.Model(theano_config=dict(floatX='float64')):
     # did not not work as expected
     # there were some numeric problems, so float64 is forced
-    with pm.theanof.change_flags(floatX="float64", warn_float64="ignore"):
+    with theano.config.change_flags(floatX="float64", warn_float64="ignore"):
+
+        assert theano.config.floatX == "float64"
+        assert theano.config.warn_float64 == "ignore"
+
+        post_mu = np.array([1.88], dtype=theano.config.floatX)
+        post_sigma = np.array([1], dtype=theano.config.floatX)
+
         with pm.Model():
-            assert theano.config.floatX == "float64"
-            assert theano.config.warn_float64 == "ignore"
             mu = pm.Normal("mu", mu=mu0, sigma=sigma)
             pm.Normal("y", mu=mu, sigma=1, observed=y_obs, total_size=aux_total_size)
             # Create variational gradient tensor
@@ -552,7 +555,7 @@ def test_scale_cost_to_minibatch_works(aux_total_size):
             mean_field_1.shared_params["mu"].set_value(post_mu)
             mean_field_1.shared_params["rho"].set_value(np.log(np.exp(post_sigma) - 1))
 
-            with pm.theanof.change_flags(compute_test_value="off"):
+            with theano.config.change_flags(compute_test_value="off"):
                 elbo_via_total_size_scaled = -pm.operators.KL(mean_field_1)()(10000)
 
         with pm.Model():
@@ -566,7 +569,7 @@ def test_scale_cost_to_minibatch_works(aux_total_size):
             mean_field_2.shared_params["mu"].set_value(post_mu)
             mean_field_2.shared_params["rho"].set_value(np.log(np.exp(post_sigma) - 1))
 
-        with pm.theanof.change_flags(compute_test_value="off"):
+        with theano.config.change_flags(compute_test_value="off"):
             elbo_via_total_size_unscaled = -pm.operators.KL(mean_field_2)()(10000)
 
         np.testing.assert_allclose(
@@ -583,9 +586,12 @@ def test_elbo_beta_kl(aux_total_size):
     sigma = 1.0
     y_obs = np.array([1.6, 1.4])
     beta = len(y_obs) / float(aux_total_size)
-    post_mu = np.array([1.88], dtype=theano.config.floatX)
-    post_sigma = np.array([1], dtype=theano.config.floatX)
-    with pm.theanof.change_flags(floatX="float64", warn_float64="ignore"):
+
+    with theano.config.change_flags(floatX="float64", warn_float64="ignore"):
+
+        post_mu = np.array([1.88], dtype=theano.config.floatX)
+        post_sigma = np.array([1], dtype=theano.config.floatX)
+
         with pm.Model():
             mu = pm.Normal("mu", mu=mu0, sigma=sigma)
             pm.Normal("y", mu=mu, sigma=1, observed=y_obs, total_size=aux_total_size)
@@ -595,7 +601,7 @@ def test_elbo_beta_kl(aux_total_size):
             mean_field_1.shared_params["mu"].set_value(post_mu)
             mean_field_1.shared_params["rho"].set_value(np.log(np.exp(post_sigma) - 1))
 
-            with pm.theanof.change_flags(compute_test_value="off"):
+            with theano.config.change_flags(compute_test_value="off"):
                 elbo_via_total_size_scaled = -pm.operators.KL(mean_field_1)()(10000)
 
         with pm.Model():
@@ -606,7 +612,7 @@ def test_elbo_beta_kl(aux_total_size):
             mean_field_3.shared_params["mu"].set_value(post_mu)
             mean_field_3.shared_params["rho"].set_value(np.log(np.exp(post_sigma) - 1))
 
-            with pm.theanof.change_flags(compute_test_value="off"):
+            with theano.config.change_flags(compute_test_value="off"):
                 elbo_via_beta_kl = -pm.operators.KL(mean_field_3, beta=beta)()(10000)
 
         np.testing.assert_allclose(
@@ -1014,7 +1020,7 @@ def init_(**kw):
 def test_flow_det(flow_spec):
     z0 = tt.arange(0, 20).astype("float32")
     flow = flow_spec(dim=20, z0=z0.dimshuffle("x", 0))
-    with change_flags(compute_test_value="off"):
+    with theano.config.change_flags(compute_test_value="off"):
         z1 = flow.forward.flatten()
         J = tt.jacobian(z1, z0)
         logJdet = tt.log(tt.abs_(tt.nlinalg.det(J)))
@@ -1030,7 +1036,7 @@ def test_flow_det_local(flow_spec):
         params[k] = np.random.randn(1, *shp).astype("float32")
     flow = flow_spec(dim=12, z0=z0.reshape((1, 1, 12)), **params)
     assert flow.batched
-    with change_flags(compute_test_value="off"):
+    with theano.config.change_flags(compute_test_value="off"):
         z1 = flow.forward.flatten()
         J = tt.jacobian(z1, z0)
         logJdet = tt.log(tt.abs_(tt.nlinalg.det(J)))

pymc3/theanof.py

@@ -19,7 +19,7 @@
 from theano import tensor as tt
 from theano.gof import Op
 from theano.gof.graph import inputs
-from theano.sandbox.rng_mrg import MRG_RandomStreams
+from theano.sandbox.rng_mrg import MRG_RandomStream as RandomStream
 
 from pymc3.blocking import ArrayOrdering
 from pymc3.data import GeneratorAdapter
@@ -394,7 +394,7 @@ def generator(gen, default=None):
     return GeneratorOp(gen, default)()
 
 
-_tt_rng = MRG_RandomStreams()
+_tt_rng = RandomStream()
 
 
 def tt_rng(random_seed=None):
@@ -409,14 +409,14 @@ def tt_rng(random_seed=None):
 
     Returns
     -------
-    `theano.sandbox.rng_mrg.MRG_RandomStreams` instance
-        `theano.sandbox.rng_mrg.MRG_RandomStreams`
+    `theano.tensor.random.utils.RandomStream` instance
+        `theano.tensor.random.utils.RandomStream`
         instance passed to the most recent call of `set_tt_rng`
     """
     if random_seed is None:
         return _tt_rng
     else:
-        ret = MRG_RandomStreams(random_seed)
+        ret = RandomStream(random_seed)
         return ret
 
 
@@ -426,14 +426,14 @@ def set_tt_rng(new_rng):
 
     Parameters
     ----------
-    new_rng: `theano.sandbox.rng_mrg.MRG_RandomStreams` instance
+    new_rng: `theano.tensor.random.utils.RandomStream` instance
         The random number generator to use.
     """
     # pylint: disable=global-statement
     global _tt_rng
     # pylint: enable=global-statement
     if isinstance(new_rng, int):
-        new_rng = MRG_RandomStreams(new_rng)
+        new_rng = RandomStream(new_rng)
     _tt_rng = new_rng
 
 

pymc3/variational/opvi.py

@@ -1078,9 +1078,9 @@ def _new_initial(self, size, deterministic, more_replacements=None):
             if deterministic:
                 return tt.ones(shape, dtype) * dist_map
             else:
-                return getattr(self._rng, dist_name)(shape)
+                return getattr(self._rng, dist_name)(size=shape)
         else:
-            sample = getattr(self._rng, dist_name)(shape)
+            sample = getattr(self._rng, dist_name)(size=shape)
             initial = tt.switch(deterministic, tt.ones(shape, dtype) * dist_map, sample)
             return initial
 

requirements.txt

@@ -5,5 +5,5 @@ numpy>=1.13.0
 pandas>=0.18.0
 patsy>=0.5.1
 scipy>=0.18.1
-theano-pymc==1.0.12
+theano-pymc==1.0.14
 typing-extensions>=3.7.4