V3 Matrix from Scattering Rates

[ejmeitz]

I was wondering if there is a way to get the V3 matrix from the linewidth calculation (here)? or at least a list of the unique elements that ALAMODE uses internally.

Thanks!

[ttadano]

Yes, it is possible to compute the V3 matrix elements using anphon, as described below. However, this option is not officially supported and may change without prior notice, so please be careful.

In the input file, please add the following options:

&general
 PREFIX = test
 MODE = RTA
 FCSXML = file_with_IFC3.xml
/

&cell
... (skipped)
/

&kpoint
2 # uniform mesh mode
4 4 4 # make it denser to get more points
/

&analysis
 PRINTV3 = 1 # Flag to compute |V3(-q, q', q'')|^2
 KS_INPUT = mode.in
/

For the KS_INPUT tag, please provide a file name containing information about the target phonon modes for which you want to compute |V3|.

In the file mode.in, please write the number of the target phonon modes and their momentum and branch index (q1, q2, q3, j) as follows:

3
0 0 0 4
0 0 0 5
0 0 0 6

In this case, the |V3(-q, q', q'')|^2 elements are computed for the 4-6th branches at the gamma point.

The index j starts from 1 and ends at 3*N (N is the number of atoms in the primitive cell).

If you want to compute the |V3(-q, q', q'')|^2 elements at other momentum points, say q=(0.5, 0, 0), please change the first three columns accordingly.

Note: The q-points in the KS_INPUT file must be a member of the uniform q-points constructed from the &kpoint-field information.

[ejmeitz]

Thanks, is there no way to get just the raw value and not the magnitude? If I need to I'm willing to write my own code that manually intercepts those values from ALAMODE.

Like I see this function in the source but I don't think the AnharmonicCore class is exposed anywhere I can construct.

[ttadano]

If you are willing to change the code yourself, please modify here.