ypp, Bands interpolation problem

[Daniele Varsano]

Dear Fabio,
my impression is that the interpolator get confused when applying qp corrections as your system became a metal:

Code: Select all

 <01s> [WARNING]Fit order on E [c] reduced to 0
 <01s> [WARNING]Global and response function energy levels have different metallic character

This can be seen inspecting the report file: K point n.19:

QP [eV] @ K [19] (iku):-.368E-16 0.577 0.00
B=17 Eo= -0.92 E= 3.69 E-Eo= 4.61 Re(Z)=0.86 Im(Z)=-.1215E-2 nlXC=-7.113 lXC=-12.18 So=0.2640
B=18 Eo= -0.76 E= 3.84 E-Eo= 4.60 Re(Z)=0.87 Im(Z)=-.1206E-2 nlXC=-7.017 lXC=-12.14 So=0.1952
B=19 Eo= 0.76 E= 5.42 E-Eo= 4.66 Re(Z)=0.84 Im(Z)=-.1426E-2 nlXC=-6.458 lXC=-12.74 So=-.7376
B=20 Eo= 0.77 E= 5.42 E-Eo= 4.65 Re(Z)=0.84 Im(Z)=-.1428E-2 nlXC=-6.426 lXC=-12.74 So=-.7729
B=21 Eo= 2.13 E= 6.34 E-Eo= 4.20 Re(Z)=0.86 Im(Z)=-.1104E-2 nlXC=-5.351 lXC=-11.64 So=-1.418

Is this behavior expected for this system?
Best,

Daniele

[Fabiof]

Dear Daniele,

so you are saying that with the qp corrections the MoS2 became a metal. That is not supposed to happen.

In the report file K19, how can you see the metal behavior? I see that Eo and E is increasing with the number of band.

Thanks

Fabio,

[Daniele Varsano]

Dear Fabio,
Sorry you are right, I was reading badly the output file.
Anyway, the output you post does not seem to correspond to the ypp.in file you uploaded:
I can see in the post:

Code: Select all

 <01s> [QP@G] E<../PPA100-5/ndb.QP[ PPA XG:285 Xb:1  100 Scb:1  100]

while in the ypp.in the qp correction is commented, and anyway pointing to another file.
In the report you attached you calculated qp energies for bands 17-21, while in the input
you are asking for bands: 1-25.
What happen if you reduce the interpolated bands to the ones you calculated?

Most importantly I can see that you are using a 3.4.1 input file running a 4.0.2 release, some variables are changed,
for instance the:

Code: Select all

% BANDS_range

I think it changed the name in:

Code: Select all

%QPkrange 

as in the yambo.in qp calculation.
I suggest you to generate an ypp input file from the last release, and use the ypp of the last release.



Best,
Daniele

[Fabiof]

Dear Daniele,

sorry i attached the wrong ypp file.

Below is the ypp file that i am using. I reduced the band range to 17-21, and i still get the same problem.

Code: Select all

electrons                    # [R] Electrons (and holes)
bnds                         # [R] Bands
cooIn= "rlu"                 # Points coordinates (in) cc/rlu/iku/alat
BANDS_steps= 20              # Number of divisions
% INTERPGrid
-1 |-1 |-1 |                             # Interpolation BZ Grid
%
ShellFac= 20.00000           # The bigger it is a higher number of shells is used
GfnQPdb= "E < ../PPA100-5/ndb.QP" # [EXTQP G] Database
%QPkrange                    #  generalized Kpoint/Band indices
  1| 31| 17| 21|
%
%BKpts                       # K points of the bands circuit
0.00000 |0.00000 |0.00000 |
0.33333 |0.66667 |0.00000 |
0.00000 |0.50000 |0.00000 |
0.00000 |0.00000 |0.00000 |
%

Thanks,

Fabio

[Daniele Varsano]

Dear Fabio,
we will need to reproduce the error, as also the occupied bands (for instance at gamma) looks me they are wrong.
In the input file you have 31 K points, because this is the number resulting after having removed the TR?

Best,
Daniele

[Fabiof]

Dear Daniele,

yes it is the number after removing TR.

Thanks,

Fabio

[Davide Sangalli]

Dear Fabio,
sorry for going back to the same previous issue, but just to be sure this is a different problem.
viewtopic.php?f=9&t=1051&start=10#p5033

We have released a new version of yambo, but the bug-fix to ypp was done too close to the release.
Thus it did not actually enter it (it will from the next one).
So can you confirm me that you are anyway using a version with the fix.

I.e. with the line

Code: Select all

if(abs(int_sop(1,1,1)+1._SP)<zero) int_sop=-int_sop

in src/interpolate/bz_interp_setup.F

Otherwise can you check if the problem persist also fixing the src/interpolate/bz_interp_setup.F subroutine ?

Best,
D.

[Fabiof]

Dear Davide

The version 4.0.2 that i'm using has the bug fixed. Line 74

Code: Select all

if(abs(int_sop(1,1,1)+1._SP)<zero) int_sop=-int_sop

When i said that there is no problem with COHSEX, i was wrong. It is getting the error also.

Im testing the non relativistic system to see if the problem persists.

Thanks,

Fabio

[Fabiof]

Dear all,

I just would like to report that when testing the MoS2 non relativistic, i could reproduce the previously error with a PPA calculation when using the the following parallel scheme

Code: Select all

SE_CPU= "2 2 6"                # [PARALLEL] CPUs for each role
SE_ROLEs= "b.qp.b"             # [PARALLEL] CPUs roles (q,qp,b)

But if i change it to

Code: Select all

SE_CPU= "4 6"                # [PARALLEL] CPUs for each role
SE_ROLEs= "b.qp"             # [PARALLEL] CPUs roles (q,qp,b)

then i don't have the error. It seems that i get different HF energy values with these two parallel schemes. This is not supposed to happen right?


Thanks,

Fabio

[Daniele Varsano]

No,
it should not happen, but this:

SE_CPU= "2 2 6" # [PARALLEL] CPUs for each role
SE_ROLEs= "b.qp.b" # [PARALLEL] CPUs roles (q,qp,b)

it is not correct. You repeat b twice.
Best,

Daniele