Skip to content

eDMFT parser and revisit DMFT metainfo

I've been working in a new parser for DMFT calculations. This is the first parser without using tight-binding models (unlike w2dynamics and solid_dmft), but instead is based on having a DFT Wien2k calculation, and doing a DMFT on top of it. They can do a Maximum Entropy analytical continuation to compare DFT vs DMFT results (hence the name MaxEnt in the metadata).

The associated new parser (initially reviewed by @ladinesa) can be found in

These are the first set of changes to allocate for the new metainfo.

  • I polished the current run.method.dmft and run.calculation.greens_functions metadata.
  • I also defined maxent_analytical_continuation as a boolean in the results.method.simulation.dmft to indicate if the MaxEnt was performed or not.
  • I defined a new workflow, MaxEnt, which groups DMFT and MaxEnt (which only calculated Σ(ω), hence the name 'MaxEnt Sigma') SinglePoint entries. This workflow also contains the calculation of the G(ω) and the DOS.
  • I extended the metadata schema of , and change the fact that only real or imaginary parts were written down; now the hole complex number is available.
  • I added a new util function, extract_section. I felt traverse_reverse was not the best when resolving multiple calculation sections, as it resolves everything. This function is very useful, I am using widely in the parsers and here can be nice to do it so. But please, let me know if it is good enough for the general utils or not.

There are some issues and todos beyond the scope of this merge request:

  • I have to fix and create some testing.

  • Right now, the plotting of Green's function is broken. In one hand, some entries will have different quantities. For now, I want to keep it simple and show only Σ(iωn) if this quantity is present (i.e. delete regtau plotting).

    On the other hand, I tried defining and in the Js module, but it didn't work; instead is showing the following plots:

Screenshot from 2023-08-10 14-23-51.png

I would like to fix this first, before continuing with more changes into the plotting. @himanel1 can you help me with this simple thing?

  • I have to define the full DFT+DMFT workflow. This can will contain the following tasks: or DFT+(Projection)+DMFT+(MaxEnt). I will do this in a next merge.

  • And after that, I will tackle the full design of the Green's functions plotting section. @wojasadr gave some suggestions and I'd like to try them out; namely, make collapsable plots (at the end, there are going to be so many quantities in the most general workflow entry, so it would be nice if these can be collapse such that the user don't get overwhelm by the amount of plots).

    If it looks nice, this could be then use in other cases.

Closes #1450 (closed)

Edited by Jose Pizarro

Merge request reports