search

RobotLocomotion / drake

Model-based design and verification for robotics.
https://drake.mit.edu
Other
3.36k stars 1.27k forks source link

Relocate //systems/analysis:lyapunov? #10761

Closed jwnimmer-tri closed 5 years ago

jwnimmer-tri commented 5 years ago

To avoid unwanted accidental dependencies like #10705, I wonder if //systems/analysis:lyapunov should be relocated? Would //systems/optimization:lyapunov be a more meaningful home? As it stands now, it's the only thing in analysis that brings in any solvers, and the solvers package is a major subsystem, where we should be careful about spuriously adding into build targets.

WDYT @RussTedrake @avalenzu? I think the main question is whether the new home would be more sensible or not from a discoverability / expectations perspective. If it's best placed where it currently lives, then I can look at the dependency hazard mitigations separately, but if moving it makes sense that's the easy answer.

avalenzu commented 5 years ago

Hmm. I'm not sure I agree with that move. Yes, SampleBasedLyapunovAnalysis() solves an optimization problem, but it is a way of analyzing a system. I don't think that we should let the simulator push all usage of solvers out of analysis.

RussTedrake commented 5 years ago

I agree with andres. I’d be more inclined to move simulator out of analysis, since it feels more out of place than the optimization approaches (for which there will be many more).

jwnimmer-tri commented 5 years ago

OK, I'll definitely keep :lyapunov where it is. Next time I come back to this question, I'll try to figure out the most sensible way to help control / reduce dependency waste.