Revision 4aef40d869462e0aae0920354df6b08a177a54df authored by st-- on 14 October 2019, 21:36:48 UTC, committed by GitHub on 14 October 2019, 21:36:48 UTC
Fixes a bug in the optimization objective for all `BayesianModel` subclasses when the model has parameters that have a `transform` (e.g. `positive()`, for kernel lengthscales and variances) but no `prior` (i.e. when doing Maximum Likelihood Estimation (MLE) inference). The objective should be independent of the parametrisation of the parameters, i.e. `log_prior()` of the parameters should be zero.
1 parent c5b91dd
File | Mode | Size |
---|---|---|
__init__.py | -rw-r--r-- | 154 bytes |
reference.py | -rw-r--r-- | 2.0 KB |
test_broadcasted_conditionals.py | -rw-r--r-- | 8.5 KB |
test_conditionals.py | -rw-r--r-- | 4.0 KB |
test_config.py | -rw-r--r-- | 2.2 KB |
test_coregion.py | -rw-r--r-- | 7.2 KB |
test_expectations.py | -rw-r--r-- | 10.5 KB |
test_features.py | -rw-r--r-- | 4.1 KB |
test_gplvm.py | -rw-r--r-- | 4.1 KB |
test_kernels.py | -rw-r--r-- | 14.8 KB |
test_kldiv.py | -rw-r--r-- | 5.2 KB |
test_likelihoods.py | -rw-r--r-- | 19.2 KB |
test_logdensities.py | -rw-r--r-- | 4.0 KB |
test_mean_functions.py | -rw-r--r-- | 11.2 KB |
test_method_equivalence.py | -rw-r--r-- | 10.9 KB |
test_methods.py | -rw-r--r-- | 9.4 KB |
test_multioutput.py | -rw-r--r-- | 24.4 KB |
test_multioutput_features.py | -rw-r--r-- | 2.2 KB |
test_notebooks.py | -rw-r--r-- | 3.3 KB |
test_predict.py | -rw-r--r-- | 6.9 KB |
test_printing.py | -rw-r--r-- | 14.1 KB |
test_prior.py | -rw-r--r-- | 2.5 KB |
test_quadrature.py | -rw-r--r-- | 2.9 KB |
test_scaled_euclid_dist.py | -rw-r--r-- | 1.9 KB |
test_training.py | -rw-r--r-- | 2.9 KB |
test_uncertain_conditional.py | -rw-r--r-- | 9.4 KB |
test_variational.py | -rw-r--r-- | 8.1 KB |
Computing file changes ...