coulamb cutoff , chi2 value

[sohan]

Dear sir,
In order to calculate GW-BSE SHG of a material, should I do GW corrections after breaking symmetry (or before?), and use the db for further SHG calculation?

[claudio]

Dear Sohan

you can do the GW calculation before breaking the symmetries, Yambo automatically will remove
symmetries also in the Quasi-Particle database

best
Claudio

[sohan]

Dear Sir,
During collisions and tdBSE calculation it takes a whole of time to complete, I wonder how will I able to converge the parameters efficiently with calculation taking this much time. i haven't used coulomb cutoff? do adding this will speed up calculation, also I'm using a 2d material which have vacuum of 20A in z so adding using 'slab z' is efficient?
I am using 16x16x1 with 60 bands. Also do I need to converge collisions and SHG together? Also, can converged k grid and NLbands form IPA can be used tdBSE SHG?
Eagerly waiting for response.

[Daniele Varsano]

Dear Sohan,
the coulomb cutoff will not speed up the calculation itself. Anyway, when using it you can reduce the amount of vacuum.
You are using a large k point grid, I suggest you to reduce it and check if such a large grid is really needed.
I cannot tell you more as you did not post any report/log file, so I do not know how many CPUs you are using, which parallelization strategy, and I do not how much "much time" is.

Best,

Daniele

[sohan]

Dear sir,
thank you for swift reply. Sir previously the collisions calculations were getting stuck at 62% when using mpirun, but on single core with yambo_rt it does run well give results.
I'll share the report as shg calculation get over.
In the meantime, sir I want to know the essential flags that should be there in QE scf and nscf input for 2d system before going for GW and SHG with YAMBO. I'm using SG15 PPs.

&control
calculation = 'scf',
verbosity = 'high',
outdir = './tmp/'
prefix = 'AB',
pseudo_dir = './',
tprnfor = .true.,
tstress = .true.,
etot_conv_thr = 1.0D-6,
forc_conv_thr = 1.0D-5,
/
&system
ibrav = 0,
nat = 4,
ntyp = 2,
ecutwfc = 90,
assume_isolated = '2D'
/
&electrons
mixing_beta = 0.7,
conv_thr = 1.D-7,
/
&ions
/
&cell
cell_dofree= '2Dxy'
/
ATOMIC_SPECIES

CELL_PARAMETERS (angstrom)

ATOMIC_POSITIONS (crystal)

K_POINTS (automatic)
8 8 1 0 0 0

Sir, when using bigger cutoff ie. 90 Ry it give G=G1-G2 error during yambo setup after fixing sym , while with smaller cutoff ie. 70Ry runs well and give o/p?

thanking you,
sohan

[Daniele Varsano]

Dear Sohan,

the flag assume_isolated = '2D' is correct.
About G=G1-G2 error what kind of error do you receive?

Best,
Daniele

[sohan]

Dear sir,
the G error is been fixed in 5.3.0. Sir i am attaching a image telling a procedure of g0w0 BSe SHG, please tell me if this is the way fwd to the same?

[sohan]

Dear Daniele sir,
Is this the right way to G0W0 BSE SHG (referring to attachment in previous response)? ,
Also, values of collbands , gwbands(qp_bands) and NL bands should be same? For instance if from ipa I got converged band range as 6 to 24 then , should I put the same values for gw bands , collbands and NLbands in g0w0 bse shg?

[claudio]

Dear Sohan

yes it is the right path to do it,
the only minor correction, I advice you to use ypp_nl to remove symmetries instead of ypp
because it is in double precision.

best regards
Claudio Attaccalite

PS if you do not mind, I will put you scheme in the wiki webpage

[sohan]

Dear Sir,
Yes, you can put on the wiki page. Sir, if I want to calculate G0W0 BSe SHG for band range of 6-24 (is converged for IPA), do I need to calculate QP corrections and collisions for same range or I don't require to take this large range for them?
thanking you,
sohan