|
1 | 1 | # Release Notes
|
2 | 2 |
|
3 |
| -## PyMC3 vNext (3.11.1) |
4 |
| - |
| 3 | +## PyMC3 vNext (TBD) |
5 | 4 | ### Breaking Changes
|
| 5 | ++ ... |
| 6 | + |
| 7 | +### New Features |
| 8 | ++ ... |
| 9 | + |
| 10 | +### Maintenance |
| 11 | ++ ... |
| 12 | + |
| 13 | +## PyMC3 3.11.1 (12 February 2021) |
6 | 14 |
|
7 | 15 | ### New Features
|
8 | 16 | - Automatic imputations now also work with `ndarray` data, not just `pd.Series` or `pd.DataFrame` (see[#4439](https://github.com/pymc-devs/pymc3/pull/4439)).
|
9 | 17 | - `pymc3.sampling_jax.sample_numpyro_nuts` now returns samples from transformed random variables, rather than from the unconstrained representation (see [#4427](https://github.com/pymc-devs/pymc3/pull/4427)).
|
10 | 18 |
|
11 | 19 | ### Maintenance
|
12 |
| -- We upgraded to `Theano-PyMC v1.1.2` which [includes bugfixes](https://github.com/pymc-devs/aesara/compare/rel-1.1.0...rel-1.1.2) for warning floods and compiledir locking (see [#4444](https://github.com/pymc-devs/pymc3/pull/4444)) |
13 |
| -- `Theano-PyMC v1.1.2` also fixed an important issue in `tt.switch` that affected the behavior of several PyMC distributions, including at least the `Bernoulli` and `TruncatedNormal` (see[#4448](https://github.com/pymc-devs/pymc3/pull/4448)) |
| 20 | +- We upgraded to `Theano-PyMC v1.1.2` which [includes bugfixes](https://github.com/pymc-devs/aesara/compare/rel-1.1.0...rel-1.1.2) for... |
| 21 | + - ⚠ a problem with `tt.switch` that affected the behavior of several distributions, including at least the following special cases (see [#4448](https://github.com/pymc-devs/pymc3/pull/4448)) |
| 22 | + 1. `Bernoulli` when all the observed values were the same (e.g., `[0, 0, 0, 0, 0]`). |
| 23 | + 2. `TruncatedNormal` when `sigma` was constant and `mu` was being automatically broadcasted to match the shape of observations. |
| 24 | + - Warning floods and compiledir locking (see [#4444](https://github.com/pymc-devs/pymc3/pull/4444)) |
14 | 25 | - `math.log1mexp_numpy` no longer raises RuntimeWarning when given very small inputs. These were commonly observed during NUTS sampling (see [#4428](https://github.com/pymc-devs/pymc3/pull/4428)).
|
15 | 26 | - `ScalarSharedVariable` can now be used as an input to other RVs directly (see [#4445](https://github.com/pymc-devs/pymc3/pull/4445)).
|
16 | 27 | - `pm.sample` and `pm.find_MAP` no longer change the `start` argument (see [#4458](https://github.com/pymc-devs/pymc3/pull/4458)).
|
17 | 28 | - Fixed `Dirichlet.logp` method to work with unit batch or event shapes (see [#4454](https://github.com/pymc-devs/pymc3/pull/4454)).
|
18 |
| -- Bugfix in logp and logcdf methods of `Triangular` distribution (see[#4470](https://github.com/pymc-devs/pymc3/pull/4470)). |
| 29 | +- Bugfix in logp and logcdf methods of `Triangular` distribution (see [#4470](https://github.com/pymc-devs/pymc3/pull/4470)). |
| 30 | + |
| 31 | +**Release manager** for 3.11.1: Michael Osthege ([@michaelosthege](https://github.com/michaelosthege)) |
19 | 32 |
|
20 | 33 | ## PyMC3 3.11.0 (21 January 2021)
|
21 | 34 |
|
|
0 commit comments