: Discrepancy in DOS results between ypp and QE

[TusharWaghmre]

Hello everyone,

I am currently encountering an issue where the density of states (DOS) calculated using Yambo's ypp appears significantly smaller compared to the DOS obtained with Quantum ESPRESSO (QE) (by the factor of 100-200). I have come across suggestions in previous discussions that normalizing the data might resolve this discrepancy, but the specifics of how to properly normalize the DOS are not clear to me.

I have attached the gw input file as well as the dos@ypp, and dos@QE output files below for your reference.

Best Regards,
Tushar Waghmare

[Daniele Varsano]

Dear Tushar,

the two plot you are showing have different smearing/broadening and that's why they have different intensity.
In order to compare the two plots, you should use the same broadening and this can be set in the ypp input file, reducing the value of the DOS_broad variable (degauss in dos.x).

Best,

Daniele

[TusharWaghmre]

Dear Daniele,

Thank you for sharing your insights. However, I re-plotted the DOS with reduced broadening, and the improvements were minimal. The calculated DOS is still trailing by a factor of 100. Please see the attached graph with the modifications. How can I resolve this issue?

Additionally, I have a question regarding the broadening factor. Previously, I used a broadening of about 0.2 eV, equivalent to 0.015 Ry, which I applied in SCF calculations in Quantum Espresso. Despite this, the broadening in the ypp plot appears significantly larger than in the QE plot, even though I used similar smearing settings. Why might this be happening?

Best Regards,
Tushar Waghmare

[Daniele Varsano]

Dear Tushar,

the reason is due to different envelope function used in QE and Yambo (Gaussian vs Lorentzian). Actually, you can tune your smearing factor to have equivalent dos profiles. Please note that the absolute value of the DOS is not a physical quantity, what is physical is the area beneath the curve.

Best,

Daniele

[TusharWaghmre]

Dear Daniele,

Thank you for your response and for clarifying the differences in the broadening factors between QE and Yambo. However, I still have a concern regarding the magnitude of the DOS.

The discrepancy becomes significant if we consider the area under the DOS curve to represent the physical quantity of interest. Given that the x-axis (energy range) is the same or nearly the same for both QE and Yambo, the difference in the magnitude of the DOS results in vastly different areas under the curve.

Is there a specific normalization or scaling factor I need to apply to make the results align, or could this indicate an error in my GW calculation? Any guidance you can provide would be greatly appreciated.

Best,
Tushar Waghmare

[Daniele Varsano]

Dear Tushar,

to check if everything is working correctly, you can look at the standard output where the integrated dos value is reported:

Code: Select all

<---> Integrated DOS      

If everything is OK, most probably the discrepancy is due to a renormalization factor. Note that DOS in Yambo should be states/Ha.

Best,

Daniele

[TusharWaghmre]

Dear Daniele,

Thank you for clarifying the DOS calculation units ([states/Ha]) in YAMBO. After converting the data to [states/eV], I noticed a curious discrepancy: when comparing the DOS generated by Quantum ESPRESSO (QE) and the DOS from YPP (without adding any QP corrections), the YPP DOS is consistently half of the QE DOS. This appears to be an exact factor-of-two difference.

Could this be due to spin degeneracy or another detail I might be overlooking? I’ve attached the revised plot for your reference and would appreciate any insights you can offer.

Sincerely,
Tushar Waghmare

[Daniele Varsano]

Dear Tushar,

please have a look to the Integrated dos value reported in the standard output to see if there is some factor missing.
Best,
Daniele