How to restart a unfinished GW calculation

[Zafer Kandemir]

Dear Sir,
Thank you for replying. The md4kp42 calculation is pending. I attached a new report file and one log file for md1kp42. I reduced the report file due to the error of too large file. I saw [RD] in the report file. IP*.dat and ndb.* files didn't change. This calculation is running now. Thank you.

Best regards.
Zafer Kandemir

[pyadav]

Dear Daniele,

I followed the following thread for k-points splitting GW calculations into two parts (say I'm in k directory and here itself run two calculations in k1 directory (from 1-5 k-points) and in k2 directory (6-10 k-points). Now if I want to merge these two data then what should I write in the place of "DB1" and "DB2" as mentioned in this quote

Dear Bramhachari,
you need first to generate the ypp input file (ypp -h for a short help) by using the command line:

Code: Select all

ypp -q m 

Next, you need to edit the ypp.in file with the following syntax:

Code: Select all

%Actions_and_names             # [QPDB] Format is "what"|"OP"|"prefactor"|"DB"|. OP can be +/-/x(only for Z)
"E" |"+" |"1" |"DB1" |
"E" |"+" |"1" |"DB2" |
%

Where you substitute the DB1 and DB2 with the name of the QP databases file you want to merge or the path where they are found.
Finally, run ypp to perform the operation, a new database containing the merged data will becreated.
Best,
Daniele

[Daniele Varsano]

Dear Pushpendra,
it should be something like:

Code: Select all

%Actions_and_names             # [QPDB] Format is "what"|"OP"|"prefactor"|"DB"|. OP can be +/-/x(only for Z)
"E" |"+" |"1" |"./k1/ndb.QP" |
"E" |"+" |"1" |"./k2/ndb.QP" |
%

Best,
Daniele

[pyadav]

Thank you, Daniele. It worked.

[pyadav]

Dear Daniele,

With respect to my last concern in this thread

Dear Pushpendra,
it should be something like:

Code: Select all

%Actions_and_names             # [QPDB] Format is "what"|"OP"|"prefactor"|"DB"|. OP can be +/-/x(only for Z)
"E" |"+" |"1" |"./k1/ndb.QP" |
"E" |"+" |"1" |"./k2/ndb.QP" |
%

I got the merged data as "ndb.QP_merged_1_gw_ppa" in the SAVE directory. But when I'm trying to calculate GW bandstructure then in "Fixsymm/SAVE" directory there is no "nbd.QP" or "ndb.QP_merged_1_gw_ppa" file. Also when I'm copying it from the GW SAVE directory then the result of GW band is not matching with a reference. So it is not able to get the GW correction with respect to DFT eigenvalues. Can you please help me with this?

Best,
Pushpendra

[Daniele Varsano]

Dear Pushpendra,
you do not need to remove the TR symmetry anymore.
You can run the ypp interpolation inserting in the ypp input
GfnQPdb= "E < ./$PATH/ndb.QP_merged_1_gw_ppa" where $PATH is SAVE or the directory containing the QP database.
Best,
Daniele

[pyadav]

Thank you, Daniele!!! It seems working now.

Wishes,
Pushpendra

[pyadav]

Dear team Yambo,

I have done a calculation separately for different k (irr. k-points) in YAMBO 5 (Version 5.0.0 Revision 19012 Hash ccf97d835). When I'm trying to merge the data with the following ypp.in input file

QPDBs # [R] Quasiparticle Databases
QPDB_merge # [R] Mergering
%Actions_and_names # [QPDB] Format is "what"|"OP"|"prefactor"|"DB"|. OP can be +/-/x(only for Z)
"E" |"+" |"1" |"k1.out/ndb.QP" |
"E" |"+" |"1" |"k2.out/ndb.QP" |
"E" |"+" |"1" |"k3.out/ndb.QP" |
"E" |"+" |"1" |"k4.out/ndb.QP" |
%

It is showing some error, with first line

Merging...detected 1 family(ies) of QP databases...*** Error in `ypp': free(): invalid next size (fast): 0x000000000632a430 ***

Can someone please help me out with this?

Thank you,
Pushpendra

[claudio]

Dear Pushpendra

it was found a bug in the database merging that has been corrected,
please download the last version of yambo 5.0 from github and try again.

If the error is still there we will check the code again

best
Claudio Attaccalite

[pyadav]

Dear Claudio, thank you for you response!!

-Pushpendra