Closed mileslucas closed 4 years ago
Merging #35 into master will increase coverage by
2.74%
. The diff coverage is91.48%
.
@@ Coverage Diff @@
## master #35 +/- ##
==========================================
+ Coverage 83.10% 85.84% +2.74%
==========================================
Files 9 9
Lines 290 318 +28
==========================================
+ Hits 241 273 +32
+ Misses 49 45 -4
Impacted Files | Coverage Δ | |
---|---|---|
src/bounds/ellipsoid.jl | 84.61% <ø> (ø) |
|
src/proposals/Proposals.jl | 86.66% <88.88%> (-1.43%) |
:arrow_down: |
src/interface.jl | 95.29% <89.28%> (-3.20%) |
:arrow_down: |
src/bounds/Bounds.jl | 91.66% <100.00%> (+3.66%) |
:arrow_up: |
src/staticsampler.jl | 80.00% <100.00%> (+46.66%) |
:arrow_up: |
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 22a9768...f809f53. Read the comment docs.
This adds some sampler changes
New Sampler Settings
enlarge
1.2 -> 1.25proposal=:auto
to choose proposal scheme automatically. If ndims <10 choose Uniform, otherwise choose rwalkupdate_interval
automatic defaults based on proposal schememin_ncall
andmin_eff
to allow constraining when the first update occurs. Before this all sampling will be unbounded and uniform, which assists in avoiding over-fitting the bound, especially for MultiEllipsoidThis has already shown beneficial, fixing the testing issues for
RWalk
, and follows a lot of the rules of thumb and defaults from dynesty, which are well-founded.