Switch to theano-pymc. (#4138)

* Switch to theano-pymc.

* Mention in release-notes.

* Update RELEASE-NOTES.md

Co-authored-by: Alexandre ANDORRA <[email protected]>

Co-authored-by: Alexandre ANDORRA <[email protected]>

Author: twiecki

RELEASE-NOTES.md

@@ -3,6 +3,7 @@
 ## PyMC3 3.9.x (on deck)
 
 ### Maintenance
+- Switch the dependency of Theano to our own fork, [Theano-PyMC](https://github.com/pymc-devs/Theano-PyMC).
 - Mentioned the way to do any random walk with `theano.tensor.cumsum()` in `GaussianRandomWalk` docstrings (see [#4048](https://github.com/pymc-devs/pymc3/pull/4048)).
 - Fixed numerical instability in ExGaussian's logp by preventing `logpow` from returning `-inf` (see [#4050](https://github.com/pymc-devs/pymc3/pull/4050)).
 - Use dill to serialize user defined logp functions in `DensityDist`. The previous serialization code fails if it is used in notebooks on Windows and Mac. `dill` is now a required dependency. (see [#3844](https://github.com/pymc-devs/pymc3/issues/3844)).

environment-dev.yml

@@ -5,7 +5,7 @@ channels:
 dependencies:
   - python=3.6
   - arviz>=0.9
-  - theano>=1.0.5
+  - theano-pymc>=1.0.5
   - numpy>=1.13
   - scipy>=0.18
   - pandas >=0.18

requirements.txt

@@ -1,5 +1,5 @@
 arviz>=0.9.0
-theano>=1.0.5
+theano-pymc>=1.0.5
 numpy>=1.13.0
 scipy>=0.18.1
 pandas>=0.18.0