[Pw_forum] Problem with tetrahedra method for calculations DOS (fwd)

[Paolo Giannozzi]

I think I figured out the origin of the problem. Array "tetra" passed to
"qexml_read_occ" in line 2010 of routine PW/src/pw_restart.f90 is not
allocated. For obscure reasons, sometimes the code decides that unallocated
== not present (tetra is an optional variable) and continues. Tetrahedra
are allocated and recomputes  later, everything works. Sometimes it tries
to write to an unallocated variable and crashes. Workaround for v.6.1 and
6.2: remove "tetra" from the call.

--- PW/src/pw_restart.f90    2017-11-13 17:54:32.725272296 +0100
+++ PW/src/pw_restart.f90.orig    2017-11-13 17:54:09.630884426 +0100
@@ -2099,7 +2099,7 @@
          f_inp( :, :) = 0.0d0
          !
          CALL qexml_read_occ( LGAUSS=lgauss, NGAUSS=ngauss,
DEGAUSS=degauss, &
-                               LTETRA=ltetra,
NTETRA=ntetra,TFIXED_OCC=tfixed_occ, &
+                               LTETRA=ltetra, NTETRA=ntetra, TETRA=tetra,
TFIXED_OCC=tfixed_occ, &
                                NSTATES_UP=nupdwn(1), NSTATES_DW=nupdwn(2),
INPUT_OCC=f_inp, IERR=ierr )
          !
       ENDIF

Paolo

On Mon, Nov 6, 2017 at 11:22 PM, Andrea Ferretti <andrea.ferretti at unimore.it
> wrote:

>
>
> Quick follow up:
>
> the fact that QE-6.2 does not show the problem is related to the IO
> default, using the new xml schema. When forcing the old QEXML format, the
> problem is reproduced there too.
>
> Andrea
>
>
> ---------- Forwarded message ----------
> Date: Mon, 6 Nov 2017 17:32:22 +0100 (CET)
> From: Andrea Ferretti <andrea.ferretti at unimore.it>
> To: PWSCF Forum <pw_forum at pwscf.org>
> Subject: Re: [Pw_forum] Problem with tetrahedra method for calculations DOS
>
>
>
> I was able to reproduce the problem using QE 6.1 with the attach set of
> input files.
> Bottom line, the variable "tetra" seems not to be properly allocated on
> read when qexml_read_occ is called (by read_occupations).
>
> in the tarball I've included a patched version of PW/src/pw_restart.f90
> where I have attempted a fix.
>
> it seems to work (compare the results obtained with and w/o tetrahedra),
> though I haven't performed extensive testing.
>
> The fix is quite "quick and dirty" style and a better interface with the
> ktetra module would be needed (no advanced tetra method possible, no use of
> tetra_init done). Here I need to refer to someone more expert than me with
> the tetrahedra impementation.
>
> The problem is not there in v6.2 (so I don't know how relevant the fix
> above actually is)
>
> take care
> Andrea
>
>
> I cannot reproduce your problem
>>
>> Paolo
>>
>> On Thu, Nov 2, 2017 at 6:11 PM, Yedu Kondalu <nykondalu at gmail.com> wrote:
>>       Dear Users and Experts,
>>
>>     I am using QE v.6.1.  I ran DOS calculation for Silicon using
>> Occupations='fixed' in 'scf' and 'nscf' calculations
>> and it was finished successfully and dos output file is created. However,
>> when I used occupations='tetrahedra' in nscf
>> (both scf and nscf), the dos.x calculation was  terminated with the
>> following error (I have tried for couple of systems
>> apart from Si). Now, my question is how to use occupations='tetrahedra'
>> in scf and nscf calculations to compute the DOS
>> for any system. Input files are attached.
>>
>> forrtl: severe (174): SIGSEGV, segmentation fault occurred
>> Image              PC                Routine            Line
>> Source
>> dos.x              0000000000687DBD  qexml_module_mp_q        3753
>> qexml.f90
>> dos.x              00000000005270DD  pw_restart_mp_rea        2101
>> pw_restart.f90
>> dos.x              00000000005239EC  pw_restart_mp_pw_        1057
>> pw_restart.f90
>> dos.x              0000000000408A7B  read_xml_file_             98
>> read_file.f90
>> dos.x              000000000040432B  MAIN__                     95
>> dos.f90
>> dos.x              000000000040422C  Unknown               Unknown
>> Unknown
>> libc.so.6          00002AF83A1E9EC5  Unknown               Unknown
>> Unknown
>> dos.x              0000000000404129  Unknown               Unknown
>> Unknown
>>
>> Thank you,
>>
>> Regards
>> N. Yedukondalu,
>> IICT, Hyderabad
>>
>> _______________________________________________
>> Pw_forum mailing list
>> Pw_forum at pwscf.org
>> http://pwscf.org/mailman/listinfo/pw_forum
>>
>>
>>
>>
>> --
>> Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche,
>> Univ. Udine, via delle Scienze 208, 33100 Udine, Italy
>> Phone +39-0432-558216, fax +39-0432-558222
>>
>>
>>
>>
> --
> Andrea Ferretti, PhD
> S3 Center, Istituto Nanoscienze, CNR
> via Campi 213/A, 41125, Modena, Italy
> Tel: +39 059 2055322;  Skype: andrea_ferretti
> URL: http://www.nano.cnr.it
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://pwscf.org/mailman/listinfo/pw_forum
>



-- 
Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche,
Univ. Udine, via delle Scienze 208, 33100 Udine, Italy
Phone +39-0432-558216, fax +39-0432-558222
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20171113/823a29b0/attachment.html>