[Wien] [SPAM?] problem in mini for spinpolarized case

Ravindran Ponniah ravindran.ponniah at kjemi.uio.no
Sat Dec 8 18:22:26 CET 2007


Dear Wien users,

As suggested by Laurence, I have now updated SRC_clmaddsub and still the 
mini job crash after 1 structural relaxation step. I will explain the 
problem in detail now.

I have submitted a job as follows

###
# Copy inpufiles to common scratch
cp $WORK/*.in* $RUN
cp $WORK/*.struct $RUN
cp $WORK/*.klist* $RUN
cp $WORK/*.kgen* $RUN
cp $WORK/45k.clmsum $RUN/Mnsp1.clmsum
cp $WORK/45k.clmup $RUN/Mnsp1.clmup
cp $WORK/45k.clmdn $RUN/Mnsp1.clmdn
cp $WORK/.min* $RUN
cp $WORK/*.rsp* $RUN

###
# Run program
# Run program
cp Mnsp1_15.struct Mnsp1.struct
rm *super.clm*
  min -i 20 -s 1 -ex -j 'runsp_lapw -I -i 40 -fc 1.0 -it '
save_lapw Mnsp1_15
mv Mnsp1_15* $WORK
###


The :log file looks

>   (min) options: -i 20 -s 1 -ex -j runsp_lapw -I -i 40 -fc 1.0 -it
>   (min) recover inm-file & call job runsp_lapw -I -i 40 -fc 1.0 -it
>   (runsp_lapw) options: -I -i 40 -fc 1.0 -it
Sat Dec  8 01:56:35 CET 2007> (x) lapw0
Sat Dec  8 01:56:54 CET 2007> (x) lapw1 -c -up
Sat Dec  8 02:05:51 CET 2007> (x) lapw1 -c -dn
Sat Dec  8 02:14:45 CET 2007> (x) lapw2 -c -up
Sat Dec  8 02:16:43 CET 2007> (x) lapw2 -c -dn
Sat Dec  8 02:18:26 CET 2007> (x) lcore -up
Sat Dec  8 02:18:27 CET 2007> (x) lcore -dn
...
...
Sat Dec  8 11:18:03 CET 2007> (x) lcore -up
Sat Dec  8 11:18:04 CET 2007> (x) lcore -dn
Sat Dec  8 11:18:05 CET 2007> (x) mixer
Sat Dec  8 11:18:09 CET 2007> (x) mini Mnsp1 mini
Sat Dec  8 11:18:10 CET 2007> (x) dstart -c
Sat Dec  8 11:18:15 CET 2007> (x) dstart -c -up
Sat Dec  8 11:18:18 CET 2007> (x) dstart -c -dn
>   (min) recover inm-file & call job runsp_lapw -I -i 40 -fc 1.0 -it
>   (runsp_lapw) options: -I -i 40 -fc 1.0 -it
Sat Dec  8 11:18:22 CET 2007> (x) lapw0
Sat Dec  8 11:18:36 CET 2007> (x) lapw1 -c -up
>>> (min) status after runsp_lapw -I -i 40 -fc 1.0 -it \: 9 -> exit
######

The dayfile is

###
Calculating Mnsp1 in /global/work/ravi/92723.snowstorm.uit.no/Mnsp1
on compute-9-7.local with PID 10908

     start       (Sat Dec  8 11:18:22 CET 2007) with lapw0 (40/20 to go)

     cycle 1     (Sat Dec  8 11:18:22 CET 2007)  (40/20 to go)

>   lapw0       (11:18:22) 11.420u 2.875s 0:14.38 99.3% 0+0k 0+0io 0pf+0w
>   lapw1  -c -up       (11:18:36) 0.003u 0.010s 0:00.04 25.0%  0+0k 0+0io 
2pf+0w

>   stop error
###

The uplapw1.error is

###
  'INILPW' - can't open unit: 18
  'INILPW' -        filename: Mnsp1.vspup
  'INILPW' -          status: old          form: formatted
  'LAPW1' - INILPW aborted unsuccessfully.
###

It appears that the case.clmup/dn files are not created from dstart after 
one structural relaxation. ls -lt shows that

###
-rw-rw-r--  1 ravi ravi       0 Dec  8 11:18 lapw0.error
-rw-rw-r--  1 ravi ravi   11990 Dec  8 11:18 :log
-rw-rw-r--  1 ravi ravi     398 Dec  8 11:18 Mnsp1.dayfile
-rw-rw-r--  1 ravi ravi       0 Dec  8 11:18 Mnsp1.dmatdn_old
-rw-rw-r--  1 ravi ravi       0 Dec  8 11:18 Mnsp1.dmatud_old
-rw-rw-r--  1 ravi ravi       0 Dec  8 11:18 Mnsp1.dmatup_old
-rw-rw-r--  1 ravi ravi   61880 Dec  8 11:18 Mnsp1.output0
-rw-rw-r--  1 ravi ravi  764260 Dec  8 11:18 Mnsp1.vns
-rw-rw-r--  1 ravi ravi   31473 Dec  8 11:18 Mnsp1.vsp
-rw-rw-r--  1 ravi ravi     563 Dec  8 11:18 uplapw1.def
-rw-rw-r--  1 ravi ravi     282 Dec  8 11:18 uplapw1.error
-rw-rw-r--  1 ravi ravi    3535 Dec  8 11:18 Mnsp1.scf0
-rw-rw-r--  1 ravi ravi       0 Dec  8 11:18 dndstart.error
-rw-rw-r--  1 ravi ravi     965 Dec  8 11:18 lapw0.def
-rw-rw-r--  1 ravi ravi       0 Dec  8 11:18 Mnsp1.clmdn
-rw-rw-r--  1 ravi ravi  848898 Dec  8 11:18 Mnsp1.clmsum
-rw-rw-r--  1 ravi ravi       0 Dec  8 11:18 Mnsp1.clmup
-rw-rw-r--  1 ravi ravi     453 Dec  8 11:18 Mnsp1.in2c
-rw-rw-r--  1 ravi ravi     177 Dec  8 11:18 Mnsp1.in2_ls
-rw-rw-r--  1 ravi ravi     453 Dec  8 11:18 Mnsp1.in2_st
-rw-rw-r--  1 ravi ravi     276 Dec  8 11:18 Mnsp1.in2_sy
-rw-rw-r--  1 ravi ravi  145524 Dec  8 11:18 Mnsp1.outputddn
-rw-rw-r--  1 ravi ravi       0 Dec  8 11:18 Mnsp1.vnsdn
-rw-rw-r--  1 ravi ravi       0 Dec  8 11:18 Mnsp1.vspdn
-rw-rw-r--  1 ravi ravi  795665 Dec  8 11:18 new_super.clmdn
-rw-rw-r--  1 ravi ravi     403 Dec  8 11:18 dndstart.def
-rw-rw-r--  1 ravi ravi     142 Dec  8 11:18 Mnsp1.in0_std
-rw-rw-r--  1 ravi ravi  145524 Dec  8 11:18 Mnsp1.outputdup
-rw-rw-r--  1 ravi ravi  848898 Dec  8 11:18 old.clmdn
-rw-rw-r--  1 ravi ravi       0 Dec  8 11:18 updstart.error
-rw-rw-r--  1 ravi ravi  795665 Dec  8 11:18 new_super.clmup
-rw-rw-r--  1 ravi ravi       0 Dec  8 11:18 dstart.error
-rw-rw-r--  1 ravi ravi  145524 Dec  8 11:18 Mnsp1.outputd
-rw-rw-r--  1 ravi ravi       0 Dec  8 11:18 Mnsp1.test
-rw-rw-r--  1 ravi ravi  848898 Dec  8 11:18 old.clmup
-rw-rw-r--  1 ravi ravi     403 Dec  8 11:18 updstart.def
-rw-rw-r--  1 ravi ravi  795665 Dec  8 11:18 new_super.clmsum
-rw-rw-r--  1 ravi ravi     346 Dec  8 11:18 dstart.def
-rw-rw-r--  1 ravi ravi       0 Dec  8 11:18 mini.error
-rw-rw-r--  1 ravi ravi  848898 Dec  8 11:18 Mnsp1_1.clmdn
-rw-rw-r--  1 ravi ravi  848898 Dec  8 11:18 Mnsp1_1.clmsum
-rw-rw-r--  1 ravi ravi  848898 Dec  8 11:18 Mnsp1_1.clmup
-rw-rw-r--  1 ravi ravi    3159 Dec  8 11:18 Mnsp1_1.struct
###

I am able to create case.clmup/dn files from dstart interactively after 
the job crash. I am unable to understand why case.clmup/dn files were not 
created after one structural relaxation (through mini) eventhough dstart 
-c -up and dstart -c -dn ran successfully (see dstart.error is empty). But, if I run 
dstart interactively after the job crash I am able to create case.clmup/dn 
files?!

Could you kindly explain to me why mini crash after 1 structural 
relaxation step in spin-polarized cases?

Best regards
Ravi


On Thu, 6 Dec 2007, Laurence Marks wrote:

> The lapw0 program looks for non-zero case.clmup and case.clmdn
> (MnCosp.clmup & dn) in order to know if it should run spin-polarized
> or not. If spin-polarized then it creates the files case.vspup/dn for
> the spherical parts of the potential, and case.vnsup/dn for the
> non-spherical. If not spin polarized it will only create case.vsp and
> case.vns. From the looks of it lapw0 ran non-polarized. Please check
> if you have non-zero length case.clmup & clmdn. If they are zero
> (which I suspect) have a look at the error files for dstart since this
> probably crashed. You can then try running them by hand (x dstart -c ;
> x dstart -c -up ; x dstart -c -dn) to see if this works or crashes for
> some reason. If it crashes (which I suspect it will) you will have to
> find out why -- perhaps rerun lstart (x lstart) again.
>
> If this works, copy them to new_super.clmsum, new_super.clmup,
> new_super.clmdn and then copy your previous density files (probably
> MnCosp_1.clmsum etc) to MnCosp.clmsum & clmup/clmdn. Then restart the
> minimization but add "-NI" so it will keep going from the previous
> cycle.
>
> This will put you back with appropriate files, although it might not
> cure all the problems.
>
> N.B., as I may have mentioned earlier, check the directory
> SRC_clmaddsub since the earliest version had a bug for spin-polarized
> cases.
>
> On Dec 6, 2007 8:29 AM, Ravindran Ponniah
> <ravindran.ponniah at kjemi.uio.no> wrote:
>>
>> Dear Wien users,
>>
>> As suggested by Peter Blaha I have now done structural relaxation by rm
>> *super.clmsum/up/dn before the mini. i.e.
>>
>> ##
>> rm *super.clm*
>> cp MnCosp_15.struct MnCosp.struct
>>   min -i 20 -s 1 -ex -j 'runsp_lapw -I -i 40 -fc 1.0 -it '
>> save_lapw MnCosp15
>> mv MnCosp15* $WORK
>> ###
>>
>> This time the mini compleated the one step in structural relaxation
>>
>>
>> ##
>>>   (min) options: -i 20 -s 1 -ex -j runsp_lapw -I -i 40 -fc 1.0 -it
>>>   (min) recover inm-file & call job runsp_lapw -I -i 40 -fc 1.0 -it
>>>   (runsp_lapw) options: -I -i 40 -fc 1.0 -it
>> Wed Dec  5 20:43:04 CET 2007> (x) lapw0
>> Wed Dec  5 20:43:21 CET 2007> (x) lapw1 -c -up
>> Wed Dec  5 20:51:03 CET 2007> (x) lapw1 -c -dn
>> Wed Dec  5 20:58:41 CET 2007> (x) lapw2 -c -up
>> Wed Dec  5 21:00:50 CET 2007> (x) lapw2 -c -dn
>> Wed Dec  5 21:02:26 CET 2007> (x) lcore -up
>> Wed Dec  5 21:02:27 CET 2007> (x) lcore -dn
>> Wed Dec  5 21:02:28 CET 2007> (x) mixer
>> ...
>> ...
>> Thu Dec  6 01:45:30 CET 2007> (x) lapw0
>> Thu Dec  6 01:45:45 CET 2007> (x) lapw1 -it -c -up
>> Thu Dec  6 01:52:03 CET 2007> (x) lapw1 -it -c -dn
>> Thu Dec  6 01:58:17 CET 2007> (x) lapw2 -c -up
>> Thu Dec  6 02:04:08 CET 2007> (x) lapw2 -c -dn
>> Thu Dec  6 02:08:14 CET 2007> (x) lcore -up
>> Thu Dec  6 02:08:15 CET 2007> (x) lcore -dn
>> Thu Dec  6 02:08:17 CET 2007> (x) mixer
>> Thu Dec  6 02:08:21 CET 2007> (x) mini MnCosp mini
>> Thu Dec  6 02:08:24 CET 2007> (x) dstart -c
>> Thu Dec  6 02:08:29 CET 2007> (x) dstart -c -up
>> Thu Dec  6 02:08:32 CET 2007> (x) dstart -c -dn
>>>   (min) recover inm-file & call job runsp_lapw -I -i 40 -fc 1.0 -it
>>>   (runsp_lapw) options: -I -i 40 -fc 1.0 -it
>> Thu Dec  6 02:08:36 CET 2007> (x) lapw0
>> Thu Dec  6 02:08:51 CET 2007> (x) lapw1 -c -up
>>>>> (min) status after runsp_lapw -I -i 40 -fc 1.0 -it \: 9 -> exit
>> ####
>>
>> and then after lapw0 it crashed with the error message
>>
>> ##
>>   MIXER END
>>>>  (mini) arrived at end -> exit
>> DSTART ENDS
>> DSTART ENDS
>> DSTART ENDS
>>   LAPW0 END
>> LAPW1 - Error
>> ...
>> ##
>>
>> The uplapw1.error file show that
>>
>> ##
>>   'INILPW' - can't open unit: 18
>>   'INILPW' -        filename: MnCosp.vspup
>>   'INILPW' -          status: old          form: formatted
>>   'LAPW1' - INILPW aborted unsuccessfully.
>> ##
>>
>> I have no case.vaspup file in the working directory. see
>>
>> -rw-rw-r--  1 ravi ravi 31473 MnCosp.vsp
>> -rw-rw-r--  1 ravi ravi     0 MnCosp.vspdn
>> -rw-rw-r--  1 ravi ravi 31473 MnCosp.vspdn_old
>> -rw-rw-r--  1 ravi ravi 31473 MnCosp.vspup_old
>> -rw-rw-r--  1 ravi ravi 31533 MnCosp.vspdn_st
>> -rw-rw-r--  1 ravi ravi 31533 MnCosp.vsp_st
>>
>> Do you know why lapw1 -up crashed in the second structural relaxation
>> step? (I have no problem in structural relaxation for nonspinpolarized
>> cases).
>>
>> Best regards
>> Ravi
>>
>> On Tue, 4 Dec 2007, Peter Blaha wrote:
>>
>>> In the next version these problems will be fixed. (min_lapw will always
>>> create with "x dstart -super" the necessary files automatically). I hope
>>> to release this soon.
>>>
>>> In the present version, the required densities of superposed atoms is
>>> done only at the end of init_lapw (and in clmextrapol for the new struct
>>> file).
>>> In your case, the new volume does not have a proper superposed density
>>> and thus leads to the crash.
>>>
>>> At the moment I'd simply remove *super.clmsum/up/dn before the next
>>> volume starts.
>>>
>> _______________________________________________
>> Wien mailing list
>> Wien at zeus.theochem.tuwien.ac.at
>> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>>
>
>
>
> -- 
> Laurence Marks
> Department of Materials Science and Engineering
> MSE Rm 2036 Cook Hall
> 2220 N Campus Drive
> Northwestern University
> Evanston, IL 60208, USA
> Tel: (847) 491-3996 Fax: (847) 491-7820
> email: L-marks at northwestern dot edu
> Web: www.numis.northwestern.edu
> Commission on Electron Diffraction of IUCR
> www.numis.northwestern.edu/IUCR_CED
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
>


More information about the Wien mailing list