[Pw_forum] Distribution of wavefunctions

[jqli14 at fudan.edu.cn]

Dear experts,
      When the number of points of  of FFT mesh is divisible by CPU cores , every CPU store the equal number of wavefunctions. If not divisible, there would be many zero values in the variable psic after call invfft ('Wave', psic, dffts) in subroutine sum_band, i.e., wavefunctions in real space. So if I calculate something like <i|j>, I just evaluate it as sum( psi_i*psj ) /(nr1*nr2*nr3) but not  sum( psi_i*psj ) /(nr1x*nr2x*nr3x),  where nr1, nr2, nr3 shown as " Dense  grid:   XXXX G-vectors     FFT dimensions: (nr1, nr2, nr3)" in standard output of QE. Correct me if I am wrong,Thanks!

Best regards!
Jiqiang Li
Fudan University


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20170108/97483449/attachment.html>