Fix klu_refactor issue

[geofjamg]

Signed-off-by: Geoffroy Jamgotchian [email protected]

Please check if the PR fulfills these requirements (please use '[x]' to check the checkboxes, or submit the PR and then click the checkboxes)

• The commit message follows our guidelines
• Tests for the changes have been added (for bug fixes / features)
• Docs have been added / updated (for bug fixes / features)

Does this PR already have an issue describing the problem ? If so, link to this issue using '#XXX' and skip the rest
No

What kind of change does this PR introduce? (Bug fix, feature, docs update, ...)
Bug fix

What is the current behavior? (You can also link to an open issue here)
In some cases, klu_refactor may lead to very inaccurate factorization and serious numerical issues.

What is the new behavior (if this is a feature change)?
A klu_rgrowth function is used to determine if a whole factorization has to be done again (klu_factor) or if klu_refactor result is fine.

Does this PR introduce a breaking change or deprecate an API? If yes, check the following:

• The Breaking Change or Deprecated label has been added
• The migration guide has been updated in the github wiki (What changes might users need to make in their application due to this PR?)

Other information:

(if any of the questions/checkboxes don't apply, please delete them entirely)

[mathbagu]

The documentation says:

Computes the reciprocal pivot growth, rgrowth = min j ((max i |c ij |)/(max i |u ij |)), where c ij is a
scaled entry in a diagonal block of the block triangular form. In MATLAB notation:
rgrowth = min (max (abs (R\A(p,q) - F)) ./ max (abs (U)))
where the factorization is L*U + F = R \ A(p,q).

[mathbagu]

@geofjamg I won't be able to release this new version, as we cannot generate the Mac version. Maybe we can try to a Github actions script that:

• build on the three OS (Linux, Mac and Windows)
• upload everything as package
• create the jar with the three versions
  and upload the jar manually on maven central?

[zamarrenolm]

@geofjamg I won't be able to release this new version, as we cannot generate the Mac version. Maybe we can try to a Github actions script that:

• build on the three OS (Linux, Mac and Windows)
• upload everything as package
• create the jar with the three versions
  and upload the jar manually on maven central?

I have been able to build the updated math library on Mac OS anc check it against the small-grid-hvdc test case using the OpenLoadFlow implementation.

With the updated math version I obtained the same values with dense and sparse matrix implementations (diffs in resulting voltages < 6e-9). With the previous version the differences were greater (up to 10 kV).