Closed mmahsereci closed 2 years ago
Merging #423 (3c3f58e) into main (82b58f9) will increase coverage by
0.00%
. The diff coverage is90.32%
.
@@ Coverage Diff @@
## main #423 +/- ##
=======================================
Coverage 89.04% 89.04%
=======================================
Files 136 136
Lines 4746 4703 -43
Branches 697 681 -16
=======================================
- Hits 4226 4188 -38
+ Misses 400 398 -2
+ Partials 120 117 -3
Impacted Files | Coverage Δ | |
---|---|---|
.../point_calculators/quadrature_point_calculators.py | 34.61% <0.00%> (ø) |
|
emukit/model_wrappers/gpy_quadrature_wrappers.py | 82.26% <72.41%> (+1.34%) |
:arrow_up: |
emukit/quadrature/kernels/quadrature_kernels.py | 82.89% <77.27%> (-5.68%) |
:arrow_down: |
emukit/quadrature/measures/integration_measure.py | 72.72% <81.81%> (+2.13%) |
:arrow_up: |
emukit/quadrature/measures/lebesgue_measure.py | 91.30% <91.30%> (ø) |
|
...it/quadrature/acquisitions/uncertainty_sampling.py | 86.20% <100.00%> (-1.68%) |
:arrow_down: |
emukit/quadrature/kernels/__init__.py | 100.00% <100.00%> (ø) |
|
emukit/quadrature/kernels/quadrature_brownian.py | 96.96% <100.00%> (+1.51%) |
:arrow_up: |
emukit/quadrature/kernels/quadrature_matern32.py | 98.00% <100.00%> (+0.04%) |
:arrow_up: |
emukit/quadrature/kernels/quadrature_matern52.py | 98.03% <100.00%> (+0.08%) |
:arrow_up: |
... and 6 more |
Continue to review full report at Codecov.
Legend - Click here to learn more
Δ = absolute <relative> (impact)
,ø = not affected
,? = missing data
Powered by Codecov. Last update 82b58f9...3c3f58e. Read the comment docs.
Issue #, if available:
Description of changes: So far the handling of integration measures was not homogeneous throughout the
quadrature
package. For example the un-normalized Lebesgue measure was represented asmeasure=None
but required integral bounds, while the normalized Lebesgue measure required the instance ofIntegrationMeasure
class and no bounds. While confusing, this also led to many if-else statements in the code.With this PR, integration measures are now handled homogeneously in the Emukit code. Integration bounds are hence only used as convenience interfaces for users where needed.
The nice thing that came for free with his PR is that normalized Lebesgue measures are now available for all kernel embeddings as well (previously only the un-normalized ones were).
By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.