|
2 | 2 |
|
3 | 3 | The release log for BoTorch. |
4 | 4 |
|
| 5 | +## [0.8.1] - Jan 5, 2023 |
| 6 | + |
| 7 | +### Highlights |
| 8 | +* This release includes changes for compatibility with the newest versions of linear_operator and gpytorch. |
| 9 | +* Several acquisition functions now have "Log" counterparts, which provide better |
| 10 | +numerical behavior for improvement-based acquisition functions in areas where the probability of |
| 11 | +improvement is low. For example, `LogExpectedImprovement` (#1565) should behave better than |
| 12 | +`ExpectedImprovement`. These new acquisition functions are |
| 13 | + * `LogExpectedImprovement` (#1565). |
| 14 | + * `LogNoisyExpectedImprovement` (#1577). |
| 15 | + * `LogProbabilityOfImprovement` (#1594). |
| 16 | + * `LogConstrainedExpectedImprovement` (#1594). |
| 17 | +* Bug fix: Stop `ModelListGP.posterior` from quietly ignoring `Log`, `Power`, and `Bilog` outcome transforms (#1563). |
| 18 | +* Turn off `fast_computations` setting in linear_operator by default (#1547). |
| 19 | + |
| 20 | +#### Compatibility |
| 21 | +* Require linear_operator == 0.3.0 (#1538). |
| 22 | +* Require pyro-ppl >= 1.8.4 (#1606). |
| 23 | +* Require gpytorch == 1.9.1 (#1612). |
| 24 | + |
| 25 | +#### New Features |
| 26 | +* Add `eta` to `get_acquisition_function` (#1541). |
| 27 | +* Support 0d-features in `FixedFeatureAcquisitionFunction` (#1546). |
| 28 | +* Add timeout ability to optimization functions (#1562, #1598). |
| 29 | +* Add `MultiModelAcquisitionFunction`, an abstract base class for acquisition functions that require multiple types of models (#1584). |
| 30 | +* Add `cache_root` option for qNEI in `get_acquisition_function` (#1608). |
| 31 | + |
| 32 | +#### Other changes |
| 33 | +* Docstring corrections (#1551, #1557, #1573). |
| 34 | +* Removal of `_fit_multioutput_independent` and `allclose_mll` (#1570). |
| 35 | +* Better numerical behavior for fully Bayesian models (#1576). |
| 36 | +* More verbose Scipy `minimize` failure messages (#1579). |
| 37 | +* Lower-bound noise in`SaasPyroModel` to avoid Cholesky errors (#1586). |
| 38 | + |
| 39 | +#### Bug fixes |
| 40 | +* Error rather than failing silently for NaN values in box decomposition (#1554). |
| 41 | +* Make `get_bounds_as_ndarray` device-safe (#1567). |
| 42 | + |
| 43 | + |
5 | 44 | ## [0.8.0] - Dec 6, 2022 |
6 | 45 |
|
7 | 46 | ### Highlights |
|
0 commit comments