Improvements on results (1): Phonopy versus Euphonic for phonon processing

[mikibonacci]

in the pro_Euphonic branch, bands and dos are performed using euphonic code. However, we should check if this represents a real advantage or not.

• [x] test performance phonopy vs euphonic, increasing meshes for properties
• [x] define if phonopy should run at the end of a calculation (Harmonic/PhononWorkchain), and for what.
• [x] working result panel within the new logic of exposing outputs
• [x] give to euphonic not the yaml but create the yaml on the fly

Anyway, for the inelastic neutron scattering part, euphonic will be used.

[mikibonacci]

I tested the performances of phonopy and euphonic considering some tasks:

• FC generation (I)
• phonons on mesh, fixed and increasing it (II)
• dos (III)
• phonons dispersion (IV)

I did the tests for Si and SiO2. What I see is that euphonic performs worse in test (II) and moreover phonopy scales better with the size of the system. The DOS is really fast with euphonic, but phonopy is just some 0.01-0.1 seconds for moderate meshes.

So the winner is phonopy.

Results Si:

Results SiO2:

[mikibonacci]

@AndresOrtegaGuerrero, so I would use:

• [x] phonopy for bands, dos, thermo (thermo is ok, I have to implement bands and dos)
• [x] euphonic for INS (in a dedicated branch, but to be adapted)
• [x] aiida-vibroscopy for Raman/IR (already there)