qexml error when porting from PW

jmullen · Post by **jmullen** » Tue Oct 06, 2009 7:42 pm

Hello All,

I am running silicon as a test system for yambo/espresso-4.1. I get the following error when executing p2y

<---> RL vectors...
[ERROR] STOP signal received while in :[01] P(W) 2 Y(ambo) Ver. 4.0
[ERROR]Error in qexml_read_planewaves IOTK error ierr: 1

I am running the sample system in the espresso-4.1 example01 for silicon and have added
'force_symmorphic=.true.'
nosym=.true.

Any advice?

Thanks

Post by **andrea marini** » Tue Oct 06, 2009 8:08 pm

jmullen wrote: I am running silicon as a test system for yambo/espresso-4.1. I get the following error when executing p2y

<---> RL vectors...
[ERROR] STOP signal received while in :[01] P(W) 2 Y(ambo) Ver. 4.0
[ERROR]Error in qexml_read_planewaves IOTK error ierr: 1

I am running the sample system in the espresso-4.1 example01 for silicon and have added
'force_symmorphic=.true.'
nosym=.true.

Jeff, please post the whole input file. We need first to reproduce the error in order to fix it. It could be everything.

Thanks

Post by **Conor Hogan** » Tue Oct 06, 2009 8:40 pm

Hi Jeff,
It's also quite possible that 4.1 doesn't work with p2y for version 4.0 (or more explicitly, with the qexml_4.0.F library), I'll have a look at it over the next days, it's been on my list of things to do for a while.

I assume you link with the iotk from 4.1, yes? See also the other thread in this forum on PW 4.1 - using the iotk from 4.0.5 worked.

Conor

jmullen · Post by **jmullen** » Wed Oct 07, 2009 3:16 am

I have done a few tests with the following combinations

1) espresso v 4.0.5 with the current version of yambo compiled against the 4.0.5 iotk
2) espresso v 4.1 with the current version of yambo compiled against the 4.0.5 iotk

First tests of Si with option 1 first informs me to use 'nosym=.true', which I expected. After adding the flag and rerunning, the error I get is the one from my first message about qexml. After this, I move to option 2 first with nosym=.true. then adding the 'force_symmorphic=.true.' and get the same error. The odd thing about this is I have been running a few other configurations with option 1 and am not getting this message. The other configurations (graphene and some others) complete with no issue.

The Si input file is below, with the nosym and force_symmorphic flags included. As I said before, it is just the example from the espresso 4.0.5 package.

Thanks again for the help

&control
calculation = 'scf'
restart_mode='from_scratch',
prefix='silicon',
pseudo_dir = '.',
/
&system
ibrav= 2, celldm(1) =10.20, nat= 2, ntyp= 1,
ecutwfc =18.0,nosym=.true., force_symmorphic=.true.
/
&electrons
mixing_mode = 'plain',
mixing_beta = 0.7,
conv_thr = 1.0d-8
/
ATOMIC_SPECIES
Si 28.086 Si.vbc.UPF
ATOMIC_POSITIONS
Si 0.00 0.00 0.00
Si 0.25 0.25 0.25
K_POINTS {automatic}
4 4 4 0 0 0

Post by **Conor Hogan** » Wed Oct 07, 2009 10:33 am

Hi Jeff,
I compiled QE 4.1 and yambo (linked to iotk in 4.1) on an IBM SP6, and ran a bulk silicon test fine, somewhat surprisingly.
It's not completely clear to me what procedure you have followed. Normally one would run a 'scf' calculation to make the charge, and then a 'bands' calculation to generate the bandstructure to be exported to yambo. In the 'bands' calculation you have to put 'wf_collect = .true.' to make the 'silicon.save' directory - its not clear to me if you have done this.

In the scf run, you can use the symmorphic symmetries.
In the nscf/bands, you should remove them with force_symmorphic = .true.
This will give you an output from PW with 24 symmetries, which is what we want (check the number in the data-file.xml file).
Hence there is no need for nosym=.true. (unless you are forcing a list of points by hand, I would not use nosym = .true. - its behaviour has changed a lot in recent versions of PW)

So, with difference from your input, I tried:
nosym = .false.
force_symmorphic = .true.
wf_collect = .true.
Kpoints = 444 111

Please try this two step approach (see the other notes in the documentation about other important flags for the band structure part, precision of diagonalization etc) and let me know if it works.
If not, post the exact scf and bands files, and specify the compiler/architecture you are using - linking IOTK is a little sensitive to the compiler sometimes.

All the best
Conor

jmullen · Post by **jmullen** » Wed Oct 07, 2009 1:36 pm

Well, turns out I forgot the wf_collect in this file and now p2y works fine whether I use QE 4.0.4, 4.0.5, or 4.1. Of course, now I have core dumps occurring before completion of the GW calculation, but I will attempt to resolve this before posting in another thread.

Sorry to bother for such a simple problem

Post by **Conor Hogan** » Wed Oct 07, 2009 1:46 pm

Ha, no worries Jeff - your timing was good since I just got access to an SP6 with 4.1 preinstalled on it....so no time wasted. Anyway the posts always help other people with similar problems.
All the best,
Conor

Fariba Nazari · Post by **Fariba Nazari** » Tue Nov 03, 2009 12:09 pm

Dear All,

I have compiled yambo which is linked to pwscf 4.0.5. every thing goes well. I have susue 11. on the computer. The lib is
intel/mkl/emt64. The compilers are ifort and icc. There is no mpi. I have tried to run one of the pwscf examples but running p2y stops with

the following error.

Error in qexml_read_ions IOTK error ierr: 1

I have tried some suggestion form forum but the problem remains. Would you please give me suggestion.

Regards
Fariba Nazari
IASBS

Post by **Conor Hogan** » Tue Nov 03, 2009 1:22 pm

Dear Fariba,
Can you attach all log and output files from the run, the config.log file from the compilation, and also attach the data-file.xml file in the *.save directory? (gzipped please)
Which version of ifort is it?

I'm afraid the linking with iotk can be quite tricky, and not easy to debug.

Fariba Nazari · Post by **Fariba Nazari** » Wed Nov 04, 2009 12:48 pm

Which version of ifort is it? 11.

Yambo Community Forum