[Pw_forum] IBM JS20 pw tests problems

[Paolo Giannozzi]

Lazaro Calderin wrote:

> Checking metaGGA...discrepancy in total energy detected
> discrepancy in force detected
> Reference:   0.2336, You got:   0.2338

as I mentioned recently, metaGGA is sometimes nasty from a numerical
point of view. It doesn't look that serious to me.

> Checking noncolin-constrain_atomicnoncolin-constrain_atomic...discrepancy in total energy detected
> Reference:   -55.690313, You got:   -55.687460

this is not so small. Noncolinear calculations with constraints are
also rather nasty. Please have a look at the differences between
noncolin-constrain_atomic.out and noncolin-constrain_atomic.ref
and report suspicious differences

Paolo
-- 
Paolo Giannozzi, Democritos and University of Udine, Italy