ndb.elph_gkkp database error

[martinspenke]

Dear all,
when trying to produce the ndb.elph_gkkp database with ypp_ph in yambo_4.1.1, i end up with the following error :

Code: Select all

 <01s> [05.03] RL shells
 <01s> [05.04] K-grid lattice
 <01s> [05.05] Energies [ev] & Occupations
 <01s> [06] == Electron-Phonon Interface: PW->Yambo Databases ==
 <01s> PW(ELPH) databases ...[PHONON] ...found 8 NOT Q-grid compatible
 <02s> ELPH databases (WRITE) |                                        | [000%] --(E) --(X)forrtl: severe (24): end-of-file during read, unit 41, file /tmp1/spenke/pho/ab_08/elph_dir/././s.dbph_000002
Image              PC                Routine            Line        Source             
ypp_ph             00000000007271F9  Unknown               Unknown  Unknown
ypp_ph             000000000074BDF4  Unknown               Unknown  Unknown
ypp_ph             000000000040F789  Unknown               Unknown  Unknown
ypp_ph             000000000040CD91  Unknown               Unknown  Unknown
ypp_ph             000000000040B3AA  Unknown               Unknown  Unknown
ypp_ph             0000000000406058  Unknown               Unknown  Unknown
libc.so.6          00007F269D969A15  Unknown               Unknown  Unknown
ypp_ph             00000000004052A9  Unknown               Unknown  Unknown

For small systems such as silicon, every thing is fine, but as soon as i turn to a bit larger system, i get the above error.

Bests
Martin

[amolina]

Dear Martin,
In principle yambo_ph should work fine for large systems. It seems the problem is in the reading of the s.dbph #2 What is the size of this file?
Have you checked that the QE run has finished well?
Alejandro.

[martinspenke]

Dear Alejandro,

I attach my input files, maybe you find something wrong or you may want to run it yourself.
It is not that much time consuming.

My problem is that simply the data conversion from QE to YAMBO fails.

The size of the file is just 15 mb for each k point.
Yes, QE run finished correctly.

I obtain the same error also on other systems.
The problem is ypp_ph and not yambo_ph.

Further as i checked on silicon, the error regarding secant solver is still available.

Best wishes
Martin

[amolina]

Dear Martin,

sorry for the delay. I have some problems to generate the s.dbph files but it is another issue.

I have generated just the first two and I have generated the gkkp netcdf files with yambo (Version 3.4.2 Revision 4706) without any problem.

Maybe something went wrong during the calculation of the s.dbph files, even they were printed.

I suggest you to generate them in QE following the tutorial we did in yambopy:
http://yambopy.readthedocs.io/en/latest ... raction-si

Pay attention when creating the dyn files in QE. It seems that if you use ibrav=0 the dyn files are not compatible with the elphon.f90 subroutine.

Please, keep me inform in case it doesn't work.

Alejandro.

[martinspenke]

Dear Alejandro,

Thanks for your reply; however i still get the same error with yambo_4.1.1 rev 112 and yambo_3.4.2 , even if i use ibrav= 1, for instance.

Code: Select all

 <01s> [05.05] Energies [ev] & Occupations
 <01s> [06] == Electron-Phonon Interface: PW->Yambo Databases ==
 <01s> PW(ELPH) databases ...[PHONON] ...found 2 NOT Q-grid compatible
 <02s> ELPH databases (WRITE) |                                        | [000%] --(E) --(X)forrtl: severe (24): end-of-file during read, unit 41, file /tmp1/martin/pho/ab_08/elph_dir/././s.dbph_000002
Image              PC                Routine            Line        Source             
ypp_ph             00000000007271F9  Unknown               Unknown  Unknown
ypp_ph             000000000074BDF4  Unknown              Unknown  Unknown
ypp_ph             000000000040F789  Unknown               Unknown  Unknown

Any way this is how it is till i find a solution.

Bests
Martin

[amolina]

Dear Martin,

I attach the files I have used to generate the gkkp database. Can you make a try? Be sure your dyn files looks like that:
Dynamical matrix file

2 24 0 11.7335741 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000
1 'O ' 14582.561007571081
2 'H ' 918.68110941480279

and NOT like that:
Dynamical matrix file

2 24 0 11.7335741 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000
Basis vectors
1.000000000 0.000000000 0.000000000
0.000000000 1.000000000 0.000000000
0.000000000 0.000000000 1.000000000
1 'O ' 14582.561007571081
2 'H ' 918.68110941480279

This addition of the Basis vectors gave me a problem and maybe generate empty gkkp files. This seems a problem of compatibility inside QE to be solved.

Please, tell me if it works.

Alejandro.

PS: To make the ph.in run I use the TEST.save from scf folder. To make the elph.in run I use the TEST.save from the nscf folder.

[martinspenke]

Dear Alejandro,

I did exactly what you told me; however i still get the same error.

Interestingly, if you look at my elph database, from whatever reason, the size of my generated s.dbph_000002 is smaller than s.dbph_000001 !

Code: Select all

 29M -rw-r--r-- 1  martin 29M Dez 18 15:04 s.dbph_000001
 15M -rw-r--r-- 1 martin 15M Dez 18 15:05 s.dbph_000002

Further QE_6.0.0 and QE_5.3.0 both automatically generate (i can not control this):
Dynamical matrix file

2 24 0 11.7335741 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000
Basis vectors
1.000000000 0.000000000 0.000000000
0.000000000 1.000000000 0.000000000
0.000000000 0.000000000 1.000000000
1 'O ' 14582.561007571081
2 'H ' 918.68110941480279

I then deleted these lines by hand :

Basis vectors
1.000000000 0.000000000 0.000000000
0.000000000 1.000000000 0.000000000
0.000000000 0.000000000 1.000000000

However, it did not help me.
I still end up with the same error.

Bests
Martin

[amolina]

Hello,
I also deleted these lines by hand and the size of my files is the same...

Code: Select all

-rw-r--r-- 1 amolinasanchez clusterusers  29M Dec 18 14:45 s.dbph_000001
-rw-r--r-- 1 amolinasanchez clusterusers  15M Dec 18 14:45 s.dbph_000002

My gkkp files have the same size:

Code: Select all

-rw-r--r-- 1 amolinasanchez clusterusers  304 Dec 18 14:45 db.elph_gkkp
-rw-r--r-- 1 amolinasanchez clusterusers  11M Dec 18 14:45 db.elph_gkkp_fragment_1
-rw-r--r-- 1 amolinasanchez clusterusers  11M Dec 18 14:45 db.elph_gkkp_fragment_2

I proceed in the following way.

Code: Select all

p2y -O test
cp s.dbph* test/
yambo
ypp_ph -F ypp.in

ypp.in is this

Code: Select all

gkkp                         # [R] Electron-Phonon databases
DBsPATH= "."                 # Path to the PW el-ph databases
PHfreqF= "none"              # PWscf format file containing the phonon frequencies
PHmodeF= "none"              # PWscf format file containing the phonon modes
#GkkpExpand                  # Expand the gkkp in the whole BZ

I have tried the stable version and the developer version. I really don't know what can be. Maybe the yambo installation? which compiler are you using?

Cheers,

Alejandro

[martinspenke]

Dear Alejandro,

I already did it in the same way as you did;
However i noticed now that when running yambo , i get a gamma point only message:

Code: Select all

<02s> [03] Transferred momenta grid
<02s> [RL indx] Q-grid:Electron-Phonon databases
<02s> [RL indx] Q BZ pts are /= from X grid BS pts. Gamma point only.
<02s> [04] External corrections

I am using a uniform grid, maybe QE shifts internally the Q-grid ? Because in the ph.in file we have this "1" numbers:

Code: Select all

               trans = .true.
     electron_phonon = 'dvscf'
 /
 2
         0.0000000   0.0000000   0.0000000  1
         0.0000000   0.0000000  -0.5000000  1

The code was compiled using mpiifort compiler.
The compilation of the code should be fine, since i could generate the databases for Silicon as in the tutorial.

Bests
Martin

[amolina]

Dear Martin,
can you upload the r_setup file? I want to check if it is the same that I have.

Regarding the gamma point message, I think you got (me too) due to put the s.dbph files in the same folder than SAVE when doing the setup.
Alejandro.