[Wien] :WARN : WARNING: RKmax reduced due to NMATMAX

[Gavin Abo]

Regarding the "foreach" error in WIEN2k 17.1 when starting directly an 
iterative diagonalization, I believe you can safely ignore it if it 
occurs only in cycle 1.  It looks like this is because there is not yet 
a case.vector/up/dn that the vec2old_lapw script can copy to 
case.vector.old/up.old/dn.old.  By ignoring the "foreach" message(s) and 
from what is seen in the terminal, it seems like it is performing full 
diagonalization in cycle 1 and continuing with iterative diagonalization 
[1] after that just fine:

username at computername:~/Desktop/TESTit$ ls
TESTit.struct
username at computername:~/Desktop/TESTit$ init_lapw -b
...
username at computername:~/Desktop/TESTit$ run_lapw -it
hup: Command not found.
  LAPW0 END
foreach: No match.
  LAPW1 END
  LAPW2 END
  CORE  END
  MIXER END
ec cc and fc_conv 0 1 1
in cycle 2    ETEST: 0   CTEST: 0
hup: Command not found.
  LAPW0 END
  LAPW1 END
  LAPW2 END
  CORE  END
  MIXER END
ec cc and fc_conv 0 1 1

A quote from [2], says:

"Often the best performance can be obtained by first running to a crude 
convergence, next doing a ``full diagonalization'' and finally 
continuing iteratively to self consistency. (e. g. use first: run_lapw 
-it -fc 10 and then run_lapw -it -fc 1)."

So, if you happen to start an iterative diagonalization calculation from 
a previous crude calculation, it looks like you will not have the 
"foreach" error:

username at computername:~/Desktop/TESTitcrude$ ls
TESTitcrude.struct
username at computername:~/Desktop/TESTitcrude$ init_lapw -b
...
username at computername:~/Desktop/TESTitcrude$ run_lapw -i 1
hup: Command not found.
  LAPW0 END
  LAPW1 END
  LAPW2 END
  CORE  END
  MIXER END
ec cc and fc_conv 0 1 1

 >   energy in SCF NOT CONVERGED
username at computername:~/Desktop/TESTitcrude$ save_lapw -d crude_calc
...
username at computername:~/Desktop/TESTitcrude$ run_lapw -it
hup: Command not found.
  LAPW0 END
  LAPW1 END
  LAPW2 END
  CORE  END
  MIXER END
ec cc and fc_conv 0 1 1
in cycle 2    ETEST: 0   CTEST: 0
hup: Command not found.
  LAPW0 END
  LAPW1 END
  LAPW2 END
  CORE  END
  MIXER END
ec cc and fc_conv 0 1 1

[1] 
http://zeus.theochem.tuwien.ac.at/pipermail/wien/2007-September/009717.html
[2] 
http://68.98.155.183:81/usersguide/7_SCF_cycle.html#SECTION05420000000000000000

On 11/7/2017 6:43 AM, chin Sabsu wrote:
> Thank you Sir,
>
> runsp_lapw -it gave me an error:  foreach. So I am running with 
> runsp_lapw ....  script.
>
> If I run it with Rkmax=6 then it does not show me any warning. At the 
> FAQ of Wienk page standard RMT for O is 6.5 but it also gave me RKmax 
> reduced warning.
>
> If someone advises me that I can take RKMAX=6 for O2 system then only 
> I can proceed.
>
> I do not have any good facility.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20171110/ebcef966/attachment.html>