JeffFessler commented 3 years ago

This PR is intended to address #31 by making the default tolerance depend on the eps() of the data type for the pivot_cache method, instead of being hard-wired to 1e-8. (This refinement is needed in other places too, eventually.) There is also some whitespace cleanup to move closer to https://github.com/invenia/BlueStyle

It reverts to the previous 1e-8 tolerance for non-AbstractFloat types for which eps() is undefined.

codecov[bot] commented 3 years ago

Codecov Report

Merging #32 (867f207) into master (a7ff3bb) will not change coverage. The diff coverage is 100.00%.

@@           Coverage Diff           @@
##           master      #32   +/-   ##
=======================================
  Coverage   89.81%   89.81%           
=======================================
  Files           8        8           
  Lines         481      481           
=======================================
  Hits          432      432           
  Misses         49       49

Impacted Files	Coverage Δ
src/pivot_cache.jl	`93.33% <100.00%> (ø)`

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 a7ff3bb...867f207. Read the comment docs.

JeffFessler commented 3 years ago

I think this is ready to go but I'd like at least one other set of eyes on it before committing...

ahwillia commented 3 years ago

It looks good to me. I only gave a quick skim (am very busy the next few weeks), but looks like @MatFi is on it as well.

JuliaLinearAlgebra / NonNegLeastSquares.jl

tolerance for pivot_cache #32

Codecov Report