Halve step in MoreThuente if initial step generates nonfinite values

[anriseth]

Also adds some @inbounds steps to BackTracking

[codecov]

Codecov Report

Merging #73 into master will decrease coverage by 0.22%.
The diff coverage is 80.95%.

@@            Coverage Diff             @@
##           master      #73      +/-   ##
==========================================
- Coverage    62.5%   62.27%   -0.23%     
==========================================
  Files           8        8              
  Lines         600      607       +7     
==========================================
+ Hits          375      378       +3     
- Misses        225      229       +4

Impacted Files	Coverage Δ
src/backtracking.jl	81.81% <100%> (ø)	⬆️
src/morethuente.jl	54.39% <66.66%> (-0.47%)	⬇️

---

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 56155bd...6eb382c. Read the comment docs.

[pkofod]

Is this from from the original ?

[anriseth]

I haven't looked at the original paper or code, however, I hit this problem when running "Exponential". It follows the same approach as hagerzhang and backtracking to ensure the "initial" step length provides finite function values.

…

On Sat, 18 Nov 2017, 18:10 Patrick Kofod Mogensen, ***@***.***> wrote: Is this from from the original ? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#73 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AFV6DW2_3S0Wa4-jA6wkZczQXSmzAKPJks5s3x2KgaJpZM4QiZgv> .