Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Set off-diagonal clover terms to be non-zero in clover invert test to…
… make clover verification non trivial
- Loading branch information
da0baba
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I have code ready for twisted-clover (performing the inversion of the clover term in the GPU and copying back to the CPU, as you suggested), but I think I found a bug in the asymmetric preconditioning, line 336 and 344 of dirac_twisted_clover.cpp, I think it should be kappa2 and not kappa going there.
Until I perform further tests I'm suggesting everybody not to use asymmetric preconditioning. When I find out whether there is a bug or not, I'll upload the code for host twisted-clover and create a hotfix, if necessary.
I'm really annoyed about this...
da0baba
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Glad you found this bug. One of the big motivations for having a proper dslash_test is to prevent these kind of regressions.
da0baba
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.