[Pw_forum] question on output for nscf calculation and high verbosity

Giovanni Cantele giovanni.cantele at spin.cnr.it
Thu Jul 7 13:34:23 CEST 2016


Dear all,

I’ve run an nscf calculation (calculation = ’nscf’) with pw.x, 5.4.0 version. The setting verbosity    = 'high’ has been included in the input.

The run corresponds to the eigenvalues calculation on a 128x128x1 k-point grid.

At the beginning of the output I find
     number of k points=  4161  Marzari-Vanderbilt smearing, width (Ry)=  0.0150

Indeed, just after this line 4161 k-points are printed both in cartesian and crystal coordinates.

The strange thing is that ate the end of the band structure calculation I find:
     Computing kpt #:   417
     total cpu time spent up to now is    25778.5 secs

     ethr =  8.33E-10,  avg # of iterations = 25.6

     total cpu time spent up to now is    25778.9 secs

     End of band structure calculation

So only the calculation on 417 out of the 4161 k-points declared at the beginning is printed in output.
Even more oddly, after "End of band structure calculation”, the eigenvalues are printed.

The command 
grep ". k = “ MYOUTPUT | wc
returns 1431 lines so I get eigenvalues on 1431 k-points.

The “417” could be understood (I tried to browse the source code but I cannot find the proof of what I’m saying!!!)
saying that because I’m using 10 pools and 417 ~ 4161 / 10, the section "Computing kpt #:” is printed only
by the first pool, either because there is a bug or because I’m wrongly using pools.

The 1431 is instead more difficult to understand.

Just to complete the information, the calculation is obviously done an ALL 4161 k-points, as testified by the presence
of the K04161 directory within prefix.save.

Any hint is appreciated! How to obtain the print of all eigenvalues in output?

Last question: If in this case I would like not to repeat the calculation, apart from writing a script to read all the K*/eigenval.xml,
is there any other way to get all the already computed eigenvalues to compute the DOS?

Thanks,

    Giovanni

-- 

Giovanni Cantele, PhD
CNR-SPIN
c/o Dipartimento di Fisica
Universita' di Napoli "Federico II"
Complesso Universitario M. S. Angelo - Ed. 6
Via Cintia, I-80126, Napoli, Italy
e-mail: giovanni.cantele at spin.cnr.it
Phone: +39 081 676910
Skype contact: giocan74

ResearcherID: http://www.researcherid.com/rid/A-1951-2009
Web page: http://people.na.infn.it/~cantele






More information about the users mailing list