[Pw_forum] pwscf 3.* good for production?

[Paolo Giannozzi]

On Oct 2, 2006, at 21:01 , Antonio Suriano wrote:

> i am moving from pwsc 2.* to 3.* to take advantage of parallel
> diagonalization and other stuff.

for parallel (subspace) diagonalization I would suggest to use the
CVS version

> Is it numerically stable?

only on Monday and Wednesday. Avoid calculations on Tuesday
and Friday. On Thursday, it depends on the phase of the moon.

> Results are consistent to previous version?

no, they are completely different, so that we can re-write old papers
with new results. The ground state structure of Silicon, for istance, is
a fullerene in v.3.0 and a nanotube in v.3.1. It is so much better than
the diamond structure, sooo boring.

Paolo

Disclaimer: only the first sentence is serious