md_summary prints null instead of default values.

[alinelena]

  timestep: 1.0
  traj_append: false
  traj_every: 100
  traj_file: null
  restart_stem: null
  restarts_to_keep: 4
  rotate_restart: true
  seed: null
  ensemble: npt
  equil_steps: 0
  file_prefix: null

[ElliottKasoar]

My initial thought was that things like file_prefix are not readily accessible, as it's not even a simple default value, but actually we can access all of the attributes of the ensemble when we create the summary, just before we call .run, so this shouldn't be too bad to fix.

Only issue is that it depends if we consider the summary to be from the CLI perspective (it doesn't set these) or the Python perspective.

It's not possible at the moment, but if, for example, we wanted to use the summary as a new config for a different structure (so overwrite the structure only), we may want to keep file_prefix as None, rather than using the file_prefix that's been created for the first structure.

[ElliottKasoar]

Maybe what we should have is

• inputs

  • All CLI options that you could take and use directly as a --config
  • I still think this should probably be the CLI input/default, rather than anything that is inferred in the Python
  • We could even allow it to be read directly from summaries

• info

  • Extra details e.g. formula, n_atoms, start/end time etc.)

• Maybe also outputs?

  • Files created
  • Slightly trickier in the case of optional minimization etc.

The first two would basically be rearranging what we already have, so maybe worth doing on their own, then outputs might be nice as an alternative to outputting the inferred file_prefix.