Convergence and Bandstructure with GW correction

[Daniele Varsano]

Dear Rupesh,
ypp has the same logic of yambo:

Code: Select all

ypp -s b

will create the input file (ypp.in) for a band structure interpolation. Edit the input file and then run:

Code: Select all

ypp

Use the -F option if you input file is different from ypp.in (default).
Best,
Daniele

[rupesh]

Hi Dr.Daniele ,

Yes sir, I command on terminal ypp -s b but no file was created . I just got on terminal

Code: Select all

<---->[01] Game Over.

I request you if you have sample (default) ypp input files please post. Because in documentation I didnt find the init file for ypp. I think problem is that I am not able to create and initialize ypp.
Thank you !!

Regards,
Rupesh

[Daniele Varsano]

An input file is reported in the documentation:
http://www.yambo-code.org/input_file/ypp/ypp_bands.php

Anyway if any file is produced, there is something wrong as it means ypp is not able to read the databases.
Did you compile ypp with the same option (configure) you used for yambo?
Are you running it in the directory *containing* the SAVE directory?

Best,
Daniele

[rupesh]

Dear Daniele,
Yes, Iam running it in same directory containing SAVE directory. It was compiled automaticaly when I was compiling and make YAMBO. But I other packages like p2y had compiled manually by me . But I check its makefile and tried to make it (ypp), In terminal I got these error

Code: Select all

ELPH_gkkp_expand.o: In function `elph_gkkp_expand_':
ELPH_gkkp_expand.f90:(.text+0x24): undefined reference to `__elph_MOD_ph_modes'
ELPH_gkkp_expand.f90:(.text+0x33): undefined reference to `__elph_MOD_elph_nb'
ELPH_gkkp_expand.f90:(.text+0xbf): undefined reference to `__elph_MOD_elph_gkkp'
ELPH_gkkp_expand.f90:(.text+0xc9): undefined reference to `__elph_MOD_elph_gkkp'
ELPH_gkkp_expand.f90:(.text+0xf0): undefined reference to `__elph_MOD_elph_gkkp'
ELPH_gkkp_expand.f90:(.text+0x11b): undefined reference to `__elph_MOD_elph_nb'
ELPH_gkkp_expand.f90:(.text+0x226): undefined reference to `__elph_MOD_elph_gkkp'
ELPH_gkkp_expand.f90:(.text+0x232): undefined reference to `__elph_MOD_elph_gkkp'
ELPH_gkkp_expand.f90:(.text+0x24b): undefined reference to `__elph_MOD_elph_gkkp'
ELPH_gkkp_expand.f90:(.text+0x708): undefined reference to `__elph_MOD_elph_gkkp'
ELPH_gkkp_expand.f90:(.text+0x70f): undefined reference to `__elph_MOD_elph_gkkp'
ELPH_gkkp_expand.o:ELPH_gkkp_expand.f90:(.text+0x716): more undefined references to `__elph_MOD_elph_gkkp' follow
collect2: ld returned 1 exit status
make: *** [ypp_ph] Error 1

I am not able to figure it now. But I checked that there is netcdf library path included in Makefile of ypp. Please suggest solution . Is there any option to compile ypp separately (without disturbing installed YAMBO ) like p2y with separate configure file?
Thank you !

Regards,
Rupesh

[Daniele Varsano]

Dear Rupesh,
you need to configure the code once for all.
Next you can compile the different packages:

Code: Select all

make yambo
make interfaces
make ypp

If you compile yambo using netcdf and ypp without them (and viceversa), it will not work producing the Game Over you experienced.
My suggestion is to do everything by scratch, a make clean_all, relaunch the configure script and the make commands.

Best,
Daniele

[rupesh]

Dear Dr. Daniele,


I have recompile and reconfigure the ypp ,now its working fine, Very Very Thank you for the guidance ! I have certain doubts please help me in each

• 1) I need to plot the GW corrected bandstructure of material Silicon, Black Phosphorus and Phosphorene , I have done GW corrections and now need to plot so using ypp I am doing it. I have run ypp it created one file o.bands-sp1 ,in it there is energy levels according to direction provided in ypp.in, now afterward do I have to plot it by gnuplot or any other software or Is there any ploting tool available which directly plot it for me giving o.bands-sp1 as input file ?
  2) I have attached my ypp.in and o.bands-sp1 file , why in the output file at the end portion of it is showing ony one band circuit point , though I had given all 5 points in ypp.in file?
  3) For Black Phosphorus being base-centered orthorhombic structure many reference paper have taken 9X6X8 k-points Grid for scf and non-scf calculations , but as for Yambo symmetric k-points required how should I come over it (i.e if I choose 8X8X8 symetric k-grid , Will there be significant change in band structure?) ?
  4) Same for Phosphorene being 2-D pluckered layer I was adviced to take (10X8X1) k-points grid , How should I solve it with Yambo ?

Please help!!

Regards,
Rupesh.

[Daniele Varsano]

Dear Rupesh,
-1. Gnuplot is a very ewell documented software. You can plot the band structure by writing a very simple script to do the job e.g. writing a file
named script.gnu:

Code: Select all

p "o-.bands-sp1" u 0:4 w l t "band 1",\\
rep "o.bands-sp1" u 0:5 w l t "band 2"
etc. etc.

you can run it entering in gnuplot by >load "script.gnu"

-2. OK, this is just wrong mirroring of the input file, does not affect anything, we will correct it, thanks for noticing.
-3/4. What does it means symmetric k-grid? Where did you read yambo needs nxnxn mohnkhrost-pack grid? All the grid you mentioned (9x6x8; 10x8x1) will work in yambo. In the previous post I was mentioning uniform q-grids which is a totally different stuff. In previous calculations you did not have uniform gris as in your input there were k-points you input by hand and the resulting q-grid was not uniform. Monkhorst Pack grids (ixjxk) in {automatic} mode should give uniform q-point grids.
Anyway about point 3, convergence with respect k-points and convergences in general, it is something you need to check carefully.
Best,

Daniele

[rupesh]

Dear Dr.Daniele,

Thanks a lot !!

Regards,
Rupesh

[rupesh]

Hi Dr. Daniele,

I tried to plot the output I attached in last mail by gnuplot with commands

Code: Select all

 plot  "o.bands-sp1_01" u 0:4 w lines , "o.bands-sp1_01" u 0:5 w lines ,"o.bands-sp1_01" u 0:6 w lines , "o.bands-sp1_01" u 0:7 w lines ,"o.bands-sp1_01" u 0:8 w lines, "o.bands-sp1_01" u 0:9 w lines

but it is not correct, its just replecating the same pattern again and again. I have attached plot and o.bands-sp1_01 file here. ypp.in is same as before.
Please suggest where I am going wrong ?
Thank you !

Regards,
Rupesh

[Daniele Varsano]

Dear Rupesh,
please first note that you are plotting KS band structure and not GW band structure. As documented here you need to add this input variable if you want the GW corerction:

Code: Select all

GfnQPdb= "E < SAVE/ndb.QP"

Please read carefully the documentation!!!
Next I do not think that gnuplot is replicating the pattern (see last part of the bands structure). Gnuplot data are nice as you have there all the informations:

Code: Select all

kx, ky, kz, band1, band2 .....

so you have there all the information to understand what is going on. Most probably you are considering paths in direction where the BS is equivalent.Check your paths and which are your high symmetry points (here I think that (0,0,0) and (0,0,1) are the same point and so on....)
Best,
Daniele