Re: Automatic Control Package

[Patrick O'Leary]

I'm listening. What did you have in mind?

On Saturday, April 13, 2013 11:09:22 PM UTC-5, jemofthewest wrote:

Hi

My name is Jeremy West.  I'm a Masters student studying Dynamic Systems and Controls but with a strong interest in programming.  I have a Special Topics scheduled for this summer, and I'd like to spend the time coding controls algorithms.  I have contacted Octave's control guy, but honestly Julia seems to be a lot more interesting in many ways.  Is there interest in such a package?  More importantly, is anyone currently working on something similar?  Thanks, and look forward to contributing some way or another.

Jeremy

[Jeremy West]

On 4/14/2013 10:12 AM, Patrick O'Leary wrote:
> I'm listening. What did you have in mind?

Well, I was going to spend some time over the next little bit chatting
with my advisor, picking his brain of what he would think would be top
priorities for general adoption. If anyone on the list has requests...

I know having state-space and transfer function model classes are very
useful. That would naturally lead into model interconnection of some kind.

On an easier note, functions to analyze the controllability and
observability of a model. Controlling and analyzing poles, such as the
lyap or place command. Perhaps estimator or kalman filter design of
some kind. I'm just throwing ideas out there that I know MATLAB/Octave
have, but I don't know what to prioritize first yet.

[Patrick O'Leary]

That all sounds reasonable. I'd say off the top of my head that getting a core set of linear system analysis tools would be a good place to start.

It makes sense to get the classical controls stuff sorted out before tackling Kalman filters or other modern stuff.

The following is just rambling a bit.

Model composition is something to put some thought (and prototypes) into--I'd like to support general, Simulink-style dataflow in the large picture, and support simulating (and analyzing--linearization, Monte Carlo) arbitrarily complex models, so it would be good if the design doesn't close that out.

I haven't put a ton of thought in this direction, and generally leave the practical implementation of control systems up to more capable minds than mine, but I'll try to help where I can.

[Jared Kofron]

I'm also in for helping.  I started doing some work on a DSP package which may have some intellectual overlap - but I haven't touched it in 

a while.  Keep me posted...

[Jameson Nash]

I'm also a masters student in controls. I started the Polynomial package with the intent of using it to write a Controls package (starting with classical control), but didn't get around to it. I'll definitely be following along, and may be able to help some.

-jameson

[Jeremy West]

Thanks for all the feedback! I've made a scaffold repo at
github.com/jemofthewest/Controls.jl. I won't be able to start working
on it too much for the next 2 weeks, finals and all. I'll hit the ground
running soon after that though.

--
Jeremy West
Research Assistant
Intelligent Systems and Controls Laboratory
Michigan Technological University

[Stefan Karpinski]

I'm not sure about the name Controls.jl – sounds like it could be some kind of UI package. As usual, let the naming debates begin.

[Jeremy West]

On Mon, Apr 15, 2013 at 01:30:28PM -0400, Stefan Karpinski wrote:
> I'm not sure about the name Controls.jl – sounds like it could be some kind
> of UI package. As usual, let the naming debates begin.

I'm not particularly attached to the name. If length isn't an issue, it
could be something like AutomaticControls. Or it could be an opaque
acronym, like ACP (automatic controls package). Any suggestions?

[Patrick O'Leary]

Why not just rip off Ogata? "SystemDynamics"

[Stefan Karpinski]

That seems clearer to me.

[Jameson Nash]

Because System Dynamics is the name an unrelated subject in the field of Business (created at MIT apparently http://en.wikipedia.org/wiki/System_dynamics)

Whats wrong with using the standard name for this? I think Controls is correct and will not be confused for a GUI package. Some possible alternatives: AutomaticControl, ControlTheory, and LinearControl

[John Myles White]

I'd say ControlTheory. Googling controls does not return many results for control theory, so it's not specific enough.

 -- John

[Patrick O'Leary]

On Monday, April 15, 2013 4:23:48 PM UTC-5, John Myles White wrote:

I'd say ControlTheory. Googling controls does not return many results for control theory, so it's not specific enough.

Lack of theoreticalness. I have a preference for other alternatives; see below.
 

On Apr 15, 2013, at 5:14 PM, Jameson Nash <vtj...@gmail.com> wrote:

Because System Dynamics is the name an unrelated subject in the field of Business (created at MIT apparently http://en.wikipedia.org/wiki/System_dynamics)

Some of y'all are up there at MIT, can't you fix that?

Whats wrong with using the standard name for this? I think Controls is correct and will not be confused for a GUI package. Some possible alternatives: AutomaticControl, ControlTheory, and LinearControl

I think it's fair to say there isn't a standard name, or we wouldn't be having this discussion right now. The classes I took were titled "System Dynamics and Control", "Linear Control" (but we'll need more packages, which might be okay), and "Robust Control" (which is not currently in scope). I'm leaning towards LinearControl and making more packages later, which at least defers the need to make the decision on the name of an umbrella package.

[Jeremy West]

On 4/15/2013 5:51 PM, Patrick O'Leary wrote:
> I think it's fair to say there isn't a standard name, or we wouldn't
> be having this discussion right now. The classes I took were titled
> "System Dynamics and Control", "Linear Control" (but we'll need more
> packages, which might be okay), and "Robust Control" (which is not
> currently in scope). I'm leaning towards LinearControl and making more
> packages later, which at least defers the need to make the decision on
> the name of an umbrella package.

For the scope of what I'm planning on doing, LinearControl would
actually be quite adequate and the easiest to start with. NonLinear
Controls is a wholly different can of worms.

[Stefan Karpinski]

It seems we have a winner. This round of The Name Game™ is hereby over.