question regarding eptemp ,degaussq and convergene

[eliephys78]

Dear all,

I have a question regarding the two parameters "eptemp" and "dgaussq". What is the criteria for choosing eptemp?

What about degaussq, which is the boradening in phonon calculations? My goal is to calculate the electron phonon coupling. Is this parameter related to convergence? Do i have to vary degaussq and calculate lambda and see if I get the same result? How do I know that my result is a converged one?

Thank you in advance

Regards

[sponce]

Dear eliephys78,

eptemp is the temperature at which you want your results. The temperature eptemp will give the value of the Bose-Einstein n(T) factors.

degaussw is the broadening on the delta of electron properties.
degaussq is the broadening on the delta of phonon properties.

If those parameters are too large, you will wash out important features. If they are too small you will have numerical instabilities.
In theory, those two parameters can be 0 if you have dense enough k/q-points grids.

In my opinion, the correct way to converge them is to do the following:
1) set degaussw and degaussq to relatively large value and then increase the k/q sampling until your results converge
2) decrease degaussw and degaussq and then increase the k/q sampling until your results converge (you will notice that you need larger k/q grid to achieve convergence).
3) repeat until the difference of the converge results (in k/q) converge also in degaussw and degaussq.

Note that if you study for example Eliashberg spectral functions, different value of degaussq will be outputed by default in a2F file.
I usually converge on degaussw only and check that degaussq is relatively stable.

Hope that helps,

Best,

Samuel

[eliephys78]

Thanks very much for your explanation and your time. Will follow your instructions and see what I will get.