Closed tyronerees closed 5 years ago
Merging #34 into master will increase coverage by
0.01%
. The diff coverage is100%
.
@@ Coverage Diff @@
## master #34 +/- ##
==========================================
+ Coverage 84.39% 84.41% +0.01%
==========================================
Files 4 4
Lines 2564 2566 +2
==========================================
+ Hits 2164 2166 +2
Misses 400 400
Impacted Files | Coverage Δ | |
---|---|---|
libRALFit/src/ral_nlls_workspaces.f90 | 79.02% <ø> (ø) |
:arrow_up: |
libRALFit/src/ral_nlls_internal.f90 | 85.1% <100%> (+0.01%) |
: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 a8d5b7c...4cb57c1. Read the comment docs.
Fixes #24
When the subproblem is being scaled, we are currently solving it in the S-norm, where S =
diag(sqrt(scal))
.There is a need for both ||d||_2 (which was needed for the convergence test), and ||d||_S, which is needed when we increase the TR radius.
I've created two variables in the workspace, so that each can be used when appropriate. This makes ||d||_2 available without any degradation in the performance solver.