NLSolvers.jl
NLSolvers.jl copied to clipboard
do not allocate in `x_converged`
it changes y = x .- z
for an iterator. ideally, a custom struct could be provided for this, i imagine some problems could arise if using CUArrays?
I suppose preallocating y is an alternative ?
it is an alternative. Ideally, there should be a way to calculate norm(dx)
without allocating somewhere, but the last time i asked, that was the agreed-upon solution
tests failed it seems
ah that was the anderson test
Codecov Report
:exclamation: No coverage uploaded for pull request base (
master@8d4792b
). Click here to learn what that means. The diff coverage isn/a
.
@@ Coverage Diff @@
## master #54 +/- ##
=========================================
Coverage ? 76.47%
=========================================
Files ? 54
Lines ? 2776
Branches ? 0
=========================================
Hits ? 2123
Misses ? 653
Partials ? 0
Looks like CI passed