IP absorption

[Daniele Varsano]

Dear Fadil,

it is not easy to help you spot the reason for the discrepancy.
in IP the dielectric function reduces to:

eps_M=1-vX^0

you can check that you build the response function with the same time ordering of Yambo (T-ordered in your input) and same damping (0.050000 eV).
Next you can try to follow step by step the yambo building, which is mainly in:
src/pol_function/X_irredux.F

Best,
Daniele

[Fadil]

Dear Daniele,

Thank you for your response.
I am extracting the dipole matrix elements for all the bands and k-points and used the formula what you wrote.
I took q=10^-5 and the same damping.
Can you please explain me shortly, what should I do to obtain the response function with the same time ordering of Yambo?
Shall I change the formula or should I change the 'DipApproach' to built the response function with the same time ordering with yambo?
I checked 'src/pol_function/X_irredux.F', but it is very hard to follow.

Best

[Daniele Varsano]

Dear Fadil,

Can you please explain me shortly, what should I do to obtain the response function with the same time ordering of Yambo?

time-ordered and retarded response functions differ by the sign of i0+ in eq. 8 of the paper you mentioned.
You can also consider the resonant response function considering only the first term if the Equation.

You can select the ordering in the input file with the variable: GrFnTpXd, in your input is set to Time-ordered "T".
The damping in the image you posted does not seem to be the same in the two cases.

Best,
Daniele

[Fadil]

Dear Daniele,

The damping in the image you posted does not seem to be the same in the two cases.

Thank you for noticing, I am aware of it, it was for 100 meV damping. I checked with 50 meV, still they are not matching.
But I noticed that transitions with similar energy are grouped as given in Appendix A.1 in the paper.
Could this be the reason for the inconsistency in both results?
Is it possible to extract grouped oscillator strengths and their energies (as we obtain in the bse calculations)?

Best

[Daniele Varsano]

Dear Fadil,

indeed it could be the case.
What you can do is to change the way the poles are grouped by changing the CGrdSp variable.
You can activate this variable in input by adding -V resp in the command line to generate the input file.

Best,
Daniele

[Fadil]

Dear Daniele,

Changing GrFnTpXd did not make noticeable effect in the IP spectrum.
I tried also to cancel grouping by CGrdSpXd=0.000. But difference in the obtained spectra are insignificant.
Still I am getting different results with yambo calculated dielectric function when I calculate the spectrum with dipole matrix elements.

Just to confirm; in the IP (eps_M=1-vX^0)
X^0 is exactly the Eq. 8 of 'yambo: An ab initio tool for excited state calculations', right?
Can you please confirm it?

Best

[Daniele Varsano]

Dear Fadil,

yes, you can find its definition in src/pol_function
OPTICS_driver.F

Code: Select all

Epsilon_ii(:,1:2)=1._SP+Alpha(:,1:2)

where:

Code: Select all

Alpha(:,1)=-X_par(1)%blc(i_G_shift,i_G_shift,:)*4.0_SP*pi/Q_plus_G_sq_modulus

The X_par is calculated in X_irredux.F

Best,

Daniele

[Fadil]

Dear Daniele,

Thank you for your quick reply. If I am not wrong the noninteracting response function, \Chi_{GG'}^ {0}(q,\omega), is X_par in X_irredux.F
If I want to print out it I need to extract;
X_par%blc(ig_row,ig_col,iw)
from X_irredux.F right?

Best

[sabrine]

Dear yambo developers and users,

I want to calculate the optical dipole matrix element (ODME) between the lowest conduction band (c1) and the highest conduction band (cm) <k, cm|vec(p)| k,c1>.

The values I find for this optical dipole matrix element are zero, which makes sense because it is a transition between empty states!

So, I'm wondering if there is a way in Yambo to populate the c1 band in order to look for these transitions?

I created the scf and nscf with q.e 7.0, then used Yambo 5.1 to perform a post-processing calculation with Independent particle approximation and yambopy to extract the ODME values.

Best regards.

[Daniele Varsano]

Dear Dr.Sabrine Ayari,

the fact they are empty states does not mean that the dipole element between these states is zero.
In an IP calculation these elements are not calculated as they do not contribute to the absorption.
You can evaluate them by setting in input the keyword "DipBandsALL".
In this way, Yambo calculates all the elements, including cc and vv states. They are stored in the dipole database and you can read them by using yambopy (see e.g. viewtopic.php?t=1889).

Best,
Daniele