| ... | ... | @@ -56,7 +56,18 @@ program. |
|
|
|
To calculate the self-energy and renormalised spectrum of the system, a fitting procedure of the Green function to Padé polynomials is used, to facilitate the computational effort necessary to complete the calculation. For that, a set of kkrflex_* files similar to step 2 is needed, that contain the information for the host Green function over a large range of energies.
|
|
|
|
- Use the file [emesh.dat](emesh.dat), that describes an energy contour from $-0.3$ to $1.3Ry$ to run one iteration using the host code with the run flag KKRSUSC like in step 2.
|
|
|
|
- Similarly to the susceptibility calculation in step 4, two runs of the code are needed.
|
|
|
|
- Set `lfit=T` to instruct the code to run the fitting procedure and set
|
|
|
|
- Provide an [input.selfe file](kkrsusc/input_selfe) for each applied magnetic field (they should have the same parameters for the computational performance) as well as the {{:kkrsusc:lebedev_ascii.gga.tar.gz|lebedev_ascii.gga file}}{=mediawiki}.
|
|
|
|
- Determine the [exchange and correlation Kernel](exchange_and_correlation_Kernel). To do this run the KKRselfe program for zero applied magnetic field and distribute the [excorr.krnl file](kkrsusc/excorr_krnl) to all other calculations.
|
|
|
|
- For all other folders, run the KKRselfe program to obtain the requested quantities in file, ready to be plotted. |
|
|
|
- Set `lfit=T` to instruct the code to run the fitting procedure and set the numerator polynomial degree `numd` and denominator degree `dend` to proper values, with $dend=numd +1$, to match the $1/\omega$ behaviour of the susceptibility in high frequencies. A typical range for `numd` is between 10-30, with the best fits usually in the region of 17-21.
|
|
|
|
- Set the bias voltage range to match the frequency range defined by `omega min, omegamax` and the corresponding parameters described in [newinpsusc.dat](newinpsusc.dat).
|
|
|
|
- Run the mpi version of the code with `lhdio=T` and `lrestart=F` and `ldos=T` so that the DOS is calculated as well.
|
|
|
|
- Next perform the serial run with `lhdio=F` and `lrestart=T`
|
|
|
|
- After the calculation is done, plot the dos and compare it with the one you get from the impurity code. The goal here is to get a fit as loyal as possible.
|
|
|
|
- Plot the susceptibility in comparison to the one you get from step 4, and once again make sure that there is a good match between them
|
|
|
|
- Finally the files `Sigma_*` contain the information for the real and imaginary parts of the self-energy.
|
|
|
|
| keyword | description | specialties |
|
|
|
|
|---------|-------------|-------------|
|
|
|
|
| `Sigma_imag.dat` | l-decomposed imaginary part of the self-energy | |
|
|
|
|
| `Sigma_real.dat` | l-decomposed real part of the self-energy | |
|
|
|
|
| `Sigma_imag_lw.dat` | l-decomposed imaginary part of the self-energy, in the high frequency regime | |
|
|
|
|
| `Sigma_real_lw.dat` | l-decomposed real part of the self-energy, in the high frequency regime | |
|
|
|
|
- The l-decomposed renormalised spectra are contained in `Renormalized_dos_*`
|
|
|
|
|
