Re: [eigen] Eigen MINRES

[ Thread Index | Date Index | More Archives ]


I just add a quick look at the code/doc, and indeed it seems pretty much OK to be considered for 3.5. I spotted a few issues though:

- complex support: we should either fix support for complex or make sure that we get a proper static assertion for them + clear warning in the doc
- extend unit test for indefinite cases.


On Fri, Jul 20, 2018 at 7:03 AM David Hyde <dabh@xxxxxxxxxxxx> wrote:

Hi there,

Thanks for all the great work on Eigen, and for getting the recent MINRES fix in there.

Just wondering, what would it take for the MINRES solver to be moved from unsupported into mainline Eigen?  Seems like the solver has been in unsupported for several years with almost no issues reported, so maybe it is stable enough to be considered supported?  Are there any specific steps/TODOs that need to be addressed to get it supported?  If it's a question of needing a maintainer who's willing to answer questions about / fix bugs with the solver, I'd be willing to volunteer.  I use this solver a lot in my work, so it would be great to get it to be a first-class citizen in Eigen.

Thanks again!


Mail converted by MHonArc 2.6.19+