Closed st-- closed 2 years ago
Merging #447 (0902e84) into master (2162032) will increase coverage by
22.21%
. The diff coverage isn/a
.
@@ Coverage Diff @@
## master #447 +/- ##
===========================================
+ Coverage 70.94% 93.16% +22.21%
===========================================
Files 52 52
Lines 1253 1259 +6
===========================================
+ Hits 889 1173 +284
+ Misses 364 86 -278
Impacted Files | Coverage Δ | |
---|---|---|
src/test_utils.jl | 94.11% <ø> (ø) |
|
src/utils.jl | 91.35% <0.00%> (+2.46%) |
:arrow_up: |
src/kernels/transformedkernel.jl | 72.72% <0.00%> (+6.06%) |
:arrow_up: |
src/chainrules.jl | 83.95% <0.00%> (+14.81%) |
:arrow_up: |
src/basekernels/periodic.jl | 100.00% <0.00%> (+44.44%) |
:arrow_up: |
src/basekernels/exponential.jl | 100.00% <0.00%> (+50.00%) |
:arrow_up: |
src/basekernels/polynomial.jl | 100.00% <0.00%> (+65.00%) |
:arrow_up: |
src/basekernels/fbm.jl | 100.00% <0.00%> (+71.42%) |
:arrow_up: |
src/basekernels/matern.jl | 100.00% <0.00%> (+75.00%) |
:arrow_up: |
src/basekernels/rational.jl | 100.00% <0.00%> (+80.00%) |
:arrow_up: |
... and 8 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 2162032...0902e84. Read the comment docs.
The
basekernels
tests are currently by far the slowest, taking almost twice the time of the next-slowest test. A quick profiling suggests that the large kernelmatrix computations (by default on the order of 1000x1000) are at least somewhat to blame. The validity of the tests should not depend on how large they are, so this PR significantly reduces that.