The ABINIT Users Mailing List ( CLOSED )

[Julie Smart <smart.julie@rocketmail.com>]

Thanks for the reply. I do apologize from the list if I didn't think of solving my problem the otherway round. Thanks again.

---

From: matthieu verstraete <matthieu.jean.verstraete@gmail.com>
To: forum@abinit.org
Sent: Wednesday, April 8, 2009 12:03:48 AM
Subject: Re: [abinit-forum] problem when mkmem=0

Hello Julie & co

I have had something similar using mkmem=0 in parallel. I was not able
to debug it properly; I think that the temporary files which are
written to disk with the wavefunctions are somehow mismatched, and end
up with a bad header (containing the wrong number of plane waves) or
the other way around, the file is ok, but the npw in the run is
changed incorrectly and no longer corresponds to that in the file.
This does not happen systematically. Do you have all the temp files
(WF1 WF2...) for all the processors in the same place, or are they on
local disks? You can always try to play with localrdwf, but that
usually doesn't help.

Otherwise, if there is any way you can avoid mkmem=0 it is of course
more efficient if everything is in memory. And this bug is avoided.

Now for the moral of the story (for everyone):

none of us (not even PMA!) are paid to answer the forum, but almost
all of your emails are answered. If you do not receive an answer
relatively quickly, it basically means that there is none, that the
bug is real, and hopefully someday someone will fix it (no one is paid
to do that either). It is useless to send the same message back to the
forum several times, unless you are adding new information for
debugging or diagnostics, so please do not do so. It adds unnecessary
threads to the forum lists, and adds human work for us to check if the
issue has already been raised and/or solved. Check release notes for
new versions of abinit, and if your bug is fixed normally it will be
noted.

The real strategy if you want things fixed is to get a debugger (gdb
dbx idb...), and try to solve some of the problem yourself. For
parallel bugs, see if you can reproduce them on a single 2-core
machine, where you can catch the individual instances of abinip. This
will at least produce valuable information which can be fed back to
the forum (inputs and individual node log files...). And if you are
lucky you'll get your name in Xavier's thank-you list for the next
version of the package! Something to talk about at cocktail parties.

Cheers,

Matthieu


>
> Dear all,
>
> Please find a message on behalf of Julie who is experiencing difficulties.
> Anyone with a similar problem, workaround or a suggestion? Julie, what
> version of abinit are you using - did you try the latest 5.7.3?
>
> Kind regards,
>
> Alain
>
> ...
> I have put the below question in the forum some months ago, but I have
> received no solution for the "BUG" I have received putting mkmem=0 in the
> calculation of the imaginary part of dielectric function.
>
> My main job is to work on optical dilectric properties, now I am haulted. I
> really need you to kindly help me overcomethis problem.
>
> Regards,
> Julie
>
> The reply of my email from Mr. PMA finding the BUG: (My email and the out
> put comes the following)
>
> Hi,
>
> Unfortunately I can not explain what is going on with your case ...
> However, the warnings you get seem not related to the error encountered by
> abinip with your case. When Abinit is run in parallel, the error messages
> and the "real" logs are written to log files, one per process. If I am
> correct (since i have modified a little bit you input, not run the same
> version of abinit, unlikely used the same number of processors), in you case
> the real error message is of the following kind :"""
>  rwwf : BUG -
>   Reading option of rwwf. One should have npw=np
>
> w1
>   However, npw= 6868, and npw1= 6820.
>   Action : contact ABINIT group.
>
>  rwwf : BUG -
>   Reading option of rwwf. One should have npw=npw1
>   However, npw= 6851, and npw1= 6820.
>   Action : contact ABINIT group.
>
>  rwwf : BUG -
>   Reading option of rwwf. One should have npw=npw1
>   However, npw= 6863, and npw1= 6820.
>   Action : contact ABINIT group.
>
> ...
> """
> Each process compute its own number of plane wave and all of them are
> differents. Does anyone knows  this bug already ?
>
> By the way, do you know you could shorten quite a bit the SCF run by using
> iprcel1 48, iscf1 7, diecut1 1 and nline1 9 ?
>
> regards
>
> PMA
>
> My email:
>
>> Hello,
>>
>>
>> I have a problem calculating the imaginary part of dielectric function. I
>> receive a warning in the log that stops me in the second NSC loop, please
>> find it in the enclosed.
>>
>> I have increased the nband by adding the frequency/energy I needed, but I
>> do not know why I am receiving such a stopping warning:
>>
>> hdr_check: WARNING -
>> -P-0000   k pt num    1 input nband=      44 not equal disk file
>> nband=      22
>> -P-0000
>> -P-0000  hdr_check: WARNING -
>> -P-0000   k pt num    2 input nband=      44 not equal disk file
>> nband=      22
>> -P-0000
>> -P-0000  hdr_check: WARNING -
>> -P-0000   k pt num    3 input nband=      44 not equal disk file
>> nband=      22
>> -P-0000
>> -P-0000  hdr_check: WARNING -
>> -P-0000   k pt num    4 input nband=      44 not equal disk file
>> nband=      22
>> -P-0000
>> -P-0000  hdr_check: WARNING -
>> -P-0000   k pt num    5 input nband=      44 not equal disk file
>> nband=      22
>> -P-0000
>> -P-0000  hdr_check: WARNING -
>> -P-0000   k pt num    6 input nband=      44 not equal disk file
>> nband=      22
>> -P-0000
>> -P-0000  hdr_check: WARNING -
>> -P-0000   k pt num    7 input nband=      44 not equal disk file
>> nband=      22
>> -P-0000
>> -P-0000  hdr_check: WARNING -
>> -P-0000   k pt num    8 input nband=      44 not equal disk file
>> nband=      22
>> -P-0000
>> -P-0000  hdr_check: WARNING -
>> -P-0000   k pt num    9 input nband=      44 not equal disk file
>> nband=      22
>> -P-0000
>> -P-0000  hdr_check: WARNING -
>> -P-0000   k pt num   10 input nband=      44 not equal disk file
>> nband=      22
>> -P-0000
>> -P-0000  hdr_check: WARNING -
>> -P-0000   k pt num   11 input nband=      44 not equal disk file
>> nband=      22
>>
>> -P-0000  inwffil about to open unit  11 for file=t1x_WF1_P-0000
>> -P-0000  wfconv:    22 bands initialized randomly with npw=  6820, for
>> ikpt=     1
>> -P-0000  wfconv:    22 bands initialized randomly with npw=  6854, for
>> ikpt=     2
>> -P-0000  wfconv:    22 bands initialized randomly with npw=  6868, for
>> ikpt=     3
>> -P-0000  wfconv:    22 bands initialized randomly with npw=  6873, for
>> ikpt=     4
>> -P-0000  wfconv:    22 bands initialized randomly with npw=  6872, for
>> ikpt=     5
>> -P-0000  wfconv:    22 bands initialized randomly with npw=  6850, for
>> ikpt=     6
>> -P-0000  wfconv:    22 bands initialized randomly with npw=  6844, for
>> ikpt=     7
>> -P-0000  wfconv:    22 bands initialized randomly with npw=  6856, for
>> ikpt=     8
>> -P-0000  wfconv:    22 bands initialized randomly with npw=  6869, for
>> ikpt=     9
>> -P-0000  wfconv:    22 bands initialized randomly with npw=  6854, for
>> ikpt=    10
>> -P-0000  wfconv:    22 bands initialized randomly with npw=  6845, for
>> ikpt=    11
>> -P-0000  wfconv:    22 bands initialized randomly with npw=  6841, for
>> ikpt=    12
>> -P-0000  wfconv:    22 bands initialized randomly with npw=  6863, for
>> ikpt=    13
>> -P-0000  wfconv:    22 bands initialized randomly with npw=  6866, for
>> ikpt=    14
>> -P-0000  wfconv:    22 bands initialized randomly with npw=  6850, for
>> ikpt=    15
>> -P-0000  wfconv:    22 bands initialized randomly with npw=  6848, for
>> ikpt=    16
>> -P-0000  wfconv:    22 bands initialized randomly with npw=  6842, for
>> ikpt=    17
>> -P-0000  wfconv:    22 bands initialized randomly with npw=  6852, for
>> ikpt=    18
>> -P-0000  wfconv:    22 bands initialized randomly with npw=  6837, for
>> ikpt=    19
>> -P-0000  leave_test : synchronization done...
>> -P-0000  leave_test : exiting...
>> STOP 1
>> STOP 1
>> STOP 1
>> STOP 1
>>
>> the input is:
>>
>>  ndtset 6
>>
>> #First dataset : SC run with kpoints in the IBZ
>>    iscf1  3
>>   nband1  22
>>   nstep1 500
>>  kptopt1 1
>>  nbdbuf1 0
>>  prtden1 1   getden1 0   getwfk1 0
>>
>> #Second dataset : NSC run with large number of bands, and points in the
>> IBZ
>>     iscf2 -2
>>    nband2  44
>>    nstep2 500
>>   kptopt2  1
>>   getwfk2  1   getden2 1
>>
>> #Third dataset : NSC run with large number of bands, and points in the the
>> full BZ
>>     iscf3 -2
>>    nband3  44
>>    nstep3  500
>>   kptopt3  2
>>   getwfk3  2   getden3 1
>>
>>
>> #Fourth dataset : ddk response function along axis 1
>>    iscf4 -3
>>   nband4  44
>>   nstep4  1  nline4  0
>>  kptopt4  2
>>
>>    nqpt4  1  qpt4  0.0d0 0.0d0 0.0d0
>>   rfdir4  1 0 0
>>  rfelfd4  2
>>  getwfk4  3
>>
>> #Fifth dataset : ddk response function along axis 2
>>    iscf5 -3
>>   nband5  44
>>   nstep5  1  nline5  0
>>  kptopt5  2
>>
>>    nqpt5  1  qpt5  0.0d0 0.0d0 0.0d0
>>   rfdir5  0 1 0
>>  rfelfd5  2
>>  getwfk5  3
>>
>> #Sixth dataset : ddk response function along axis 3
>>    iscf6 -3
>>   nband6  44
>>   nstep6  1  nline6  0
>>  kptopt6  2
>>
>>    nqpt6  1  qpt6  0.0d0 0.0d0 0.0d0
>>   rfdir6  0 0 1
>>  rfelfd6  2
>>  getwfk6  3
>>
>>
>> #Data common to all datasets
>>  nshiftk 1
>>  shiftk  0.0 0.0 0.5
>>  ngkpt  12 12 8
>>
>>  acell   6.0445081230E+00  6.0445081230E+00  9.7638554549E+00 Bohr
>>  amu 65.39  15.9994
>>  iprcel 45
>>  ecut 60
>>  iscf 3
>>  ixc 3
>>  natom  4  nbdbuf 2
>>  ntypat  2
>>  angdeg 90 90 120
>>  xred         3.3333333333E-01  6.6666666667E-01 -3.3300232113E-16
>>               6.6666666667E-01  3.3333333333E-01  5.0000000000E-01
>>               3.3333333333E-01  6.6666666667E-01  3.7863682474E-01
>>               6.6666666667E-01  3.3333333333E-01  8.7863682474E-01
>>  tnons        0.0000000  0.0000000  0.0000000     0.0000000  0.0000000
>> 0.5000000
>>               0.0000000  0.0000000  0.5000000     0.0000000  0.0000000
>> 0.0000000
>>               0.0000000  0.0000000  0.0000000     0.0000000  0.0000000
>> 0.5000000
>>               0.0000000  0.0000000  0.5000000     0.0000000  0.0000000
>> 0.0000000
>>               0.0000000  0.0000000  0.0000000     0.0000000  0.0000000
>> 0.5000000
>>               0.0000000  0.0000000  0.5000000     0.0000000  0.0000000
>> 0.0000000
>>  typat  1 1 2 2  tolwfr  1.e-18
>>  znucl  30 8
>> mkmem 0
>>
>> I'd be grateful to have your kind reply soon.
>>
>> Regards,
>> Julie
>>
>>
>
>
>
>
>



--
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Dr. Matthieu Verstraete

European Theoretical Spectroscopy Facility (ETSF)
Dpto. Fisica de Materiales,
U. del Pais Vasco,
Centro Joxe Mari Korta, Av. de Tolosa, 72,  Phone: +34-943018393
E-20018 Donostia-San Sebastian, Spain        Fax  : +34-943018390

Mail : matthieu.jean.verstraete@gmail.com
http://www-users.york.ac.uk/~mjv500