Changed the control notebook to use Linearizer - Githubissues

pydy / pydy-tutorial-human-standing

PyDy tutorial materials for MASB 2014, PYCON 2014, and SciPy 2014/2015.

http://pydy.org

160 stars 71 forks source link

Changed the control notebook to use Linearizer #81

Closed jcrist closed 9 years ago

jcrist commented 9 years ago

Notebook 9 (control) now uses the Linearizer class to perform the linearization.

TODO:

[ ] More exercises to replace the ones I removed?
[ ] odeint currently times-out after the constants are changed, due to the controller not properly stabilizing the system (I think). Not sure what's up, the matrices generated by linearizing this way are equivalent to before.

Closes #77.

moorepants commented 9 years ago

@jcrist I moved the milestone for finishing this repo to june 18th (thursday). We've got a bit more to do to release PyDy 0.3.0, one more major issue is all that is left. So I think that will be finished up in a day. I've got lots of neglected work piling up so I need to attend to some of those.

I'll look at this when I get a chance, but feel free to make some executive decisions here. I also can use more help. SciPy asked that we have everything done and shared with them by the 19th.

moorepants commented 9 years ago

This looks fine. Are there no exercises for this notebook now? We should have something so that they can play with the controller weights at least. I don't think changing the model constants is that useful.

jcrist commented 9 years ago

There are exercises. View it here. The exploratory exercise is the only one that doesn't work with the default weights, although that could be changed. Sorry I haven't been working on this lately.

moorepants commented 9 years ago

Hmm, the default weights should guarantee that the solution converges to zero but it may take a while if the weights are bad. I'll play with it. Let's just merge this and open a new issue.

moorepants commented 9 years ago

Or do you want to fix it in this PR?

jcrist commented 9 years ago

Well, too late now:) But no worries, I can fix it in a separate PR. I think it's a numerics issue - the graph gets kind of wonky.

moorepants commented 9 years ago

Sorry, trigger happy :) Just open a new PR.