search

EmuKit / emukit

A Python-based toolbox of various methods in decision making, uncertainty quantification and statistical emulation: multi-fidelity, experimental design, Bayesian optimisation, Bayesian quadrature, etc.
https://emukit.github.io/emukit/
Apache License 2.0
605 stars 128 forks source link

MC estimators for ground truth integrals of models in `quadrature` tests now use same setup as tests #410

Closed mmahsereci closed 2 years ago

mmahsereci commented 2 years ago

Issue #257 , if available:

Description of changes: The 2 outputs of the integrate method of the warped BQ models in the quadrature package are tested by comparing to MC estimators. The MC estimators are pre-computed to speed up tests. So far, the two implementations (tests and MC estimators) were completely separate which could lead to bugs as different models could be used. Now the script computing the MC estimators uses the same setup as the tests, hence when it is re-run it should produce valid estimators and confidence intervals. It's still not super pretty but better than before.

Additionally this PR parameterizes those tests which makes it easier to add new models to the tests. Test themselves are not changed.

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.

mmahsereci commented 2 years ago

I'll merge this again @apaleyes . There are no functional changes in this PR. Just structuring the tests more robustly and more readable.