Suggestion: tell us what need to implement and how to implement correcly in a TODO list

[wizardforcel]

According to some students in engineering, there are still big gaps between simupy and simulink.

The simulation lib is a huge work so we need a leading and overall planning.

So I think we need:

(1) Compare docs of simupy and simulink to figure out what need to implement.

(2) For every missed component (classes, methods, ...), write a spec (roughly what to produce in which conds) to tell us how to implement them correctly.

The spec can partially ensure its quality without industry applications. It can also help someone else who want to write simulation projections.

Paper links in simulink docs are handy and worth refering to (and c code generated by simulink, I think).

The above two can be summarized as a TODO list to help developement.

[astrojuanlu]

simupy is a very appealing project (probably the only one of its kind written in pure Python) but activity has been a bit sparse. @sixpearls do you have any sort of roadmap in mind, so potential contributors can prioritize pull requests?

[sixpearls]

Hi all, thank you for expressing interest in this project! Active development has been slow for a while as I finished up my doctorate and figured out my employment situation. I do have a roadmap in mind. I plan to work through some of the last API design details, start working on implementing them, and document the plan so potential contributors can make useful contributions.

[samiit]

I have been looking for a graphical model building workspace, if that is available in Python. Does simupy have this implemented somewhere?

[jjuch]

Hello everyone,

I'm currently working on a more block-like implementation for nonlinear control loops. The toolbox is developed based on the simupy library, but has a strong focus on control engineering applications. The Github repo can be found here. The basics are there, but it still needs a lot of development. A next step is to implement more advanced closed-loop configurations in combination with more visuals. A lot of people are interested in the click and drag features of simulink. I don't want to go that far, but a visualised closed-loop overview scheme with some information on hover, would be nice.

Sorry for hijacking this discussion, but it seemed relevant.

Kind regards,
Jasper

[sixpearls]

@samiit A graphical interface is not something I will be implementing, but would gladly support other developers in implementing as this is a commonly sought feature.

[jjuch]

@samiit A graphical interface is not something I will be implementing, but would gladly support other developers in implementing as this is a commonly sought feature.

Always looking for help! ;-) If anyone's interested, please contact me.

[sixpearls]

I created a roadmap for version 2.0 wiki page. I'll leave this issue open for high level discussion/planning. Feel free to open new issues to discuss something specific.

[isudos]

Hi, I've just tested a bit and found out couple of issues that can be useful to mention.
The first one is inability to use any other out of box scipy ODE integrators like 'vode' or 'lsoda'. The problem is that these integrators don't provide solout callback passing. These integrators can be somehow useful for stiff problems that occur quite often in practice. Actually this issue is main stopper for me now.
Another problem happens in Ipython and is quite tricky though reproducible in different python enviroments. symbolic.DynamicSystem refuses to be initialized unless you declare it's dynamic symbols right in the previous separate cell.