To modernize the package the building tool should be changed to flit as explained in the aiida-core issue this allows to put all the information of the project in the pyproject.toml

With the release of aiida-2.x several changes, specially pertaining the repository, have been introduced.

To ensure that the plugin is compatible one needs to take care of issues with the opening of files from the repository, entrypoints, etc as indicated in the aida-core wiki

Many of the workflows right one are very application specific, e.g. trying to make the phase diagram of a skyrmion specifically instead of just a general phase diagram.

Several workflows have their visualization tools inside the workflows. They should be moved to different modules unless it is strictly necessary to have them for them to be part of the workflow.
The visualization tools then should be adapted so that a node can be passed as input and that the appropriate plots are then produced from them.

Plotting and visualization is a post processing operation not suitable for aiida unless you are storing the plots in the database.

This would imply that one also needs to add a extra installation option where the dependencies are mentioned there.

What do you think @MXJK851 , @anbergman ?

The inputs sometimes have nested orm types. This is unnecessary. One should use flat orm types.

To minimize the problems resulting from the development one needs to add a test suite to the package.

The tests should cover the following areas:

1. Parsing of the output files
2. Generation of the input files for inpsd.dat
3. Base calculation
   • MC
   • SD
4. Workchains

The CI/CD should also be used for the generation of the documentation in readthedocs and pre-commit checking.

The restart workflow can be used to introduce convergence tests to check that the calculation has reached a ground state.

This could be added for the MC method, where one could check that the average magnetization, total energy, so on are checked so that one can see if the calculation has converged. If it has not one could then restart the calculation from the last iteration and it could be submitted again but eliminating any thermalization steps.

What do you think @MXJK851 , @anbergman ?