Setup milestones & release schedule

[maedoc]

To keep expectations & issues manageable, there should be a clear version schedule, maybe a tick-tock between implementation and application. For instance

• v0.0.x - initial prototyping w/ Jax
• v0.1.x - initial TVB backend + show cases
• v0.2.x - field performance issues
• v0.3.x - stimulation 💉
• v0.4.x - solver performance issues ⏳
• v0.5.x - generalized neural masses 🥅
• v0.6.x - documentation push? 📝
• v0.7.x - multi-connectome cases? ꙮ
• v0.8.x - revisit inversion use cases #5
• ...
• v1.0.0rc - summer break 😮‍💨🍸 first v1 release candidate
• v1.0.0 - API is stable ⛱
• v1.0.x - fix bugs in v1.0.x 👾

@i-Zaak let's identify priorities/dates; it'd be nice to hit v1 by end of September

[maedoc]

Once we have a better sketch, I'll set up the milestones in the tracker here.

[i-Zaak]

Cool! I'd throw in showcase 1 use-case for v0.1. That would be any or all of following:

• paramenter sweep for MPR
• SBI with MPR
• optimization with AdEx

[maedoc]

Feed back following discussion:

• needs bounds for MPR and AdEx
• opt w/ AdEx w/ this model https://github.com/the-virtual-brain/tvb-root/blob/master/tvb_library/tvb/simulator/models/zerlaut.py regional variability w/ some FC like features
  • needs stimulation, w/ explicit time series
• need BOLD (inline?) + FCD for MPR as post processing
  • either Balloon-Windkessel or HRF (how to convert to a/b coefficients for online lfilter)
• BVEP use case incl Proix et al Epileptor
• parameter sweep boilerplate subclassing https://github.com/the-virtual-brain/tvb-inversion/blob/main/tvb_inversion/parameters.py
  • smooth out integration w/ Dask for multimachine use

Next steps:

• update milestone list above
• sideline non-priorities
• start adding milestones + linked issues

[maedoc]

I've mapped out the most relevant bits to milestones + issues.