Thanks. Witsenhausen counterexample is famous because it showed that problems with certain "information patterns" do not have optimal affine solutions, and that nonlinear controls could perform better. Later research developed proofs about the discrete version of the problem, as well as showing that nonlinear controls can perform infinitely better than linear controls for certain classes of problems.
On Thu, Sep 1, 2011 at 6:32 PM, Max OrHai <[email protected]> wrote: > My "thinking out loud" response would be that classical control theory may > not be very well suited to CS-type problems, which often can't even be > approximated by linear systems. Cybernetic feedback control, a la Weiner, is > IIRC mostly about systems with a few continuous variables, while our > problems more often involve large numbers of discrete variables. But, > there's certainly quite a bit more to control theory than I'm aware of. > > Wolfram MathWorld recommends Zabczyk: > http://www.amazon.com/exec/obidos/ASIN/0817636455 > > There are plenty of other, smaller, less comprehensive intro texts out > there too... sorry I can't recommend one first-hand. > > -- Max > > > > On Thu, Sep 1, 2011 at 1:49 PM, John Zabroski <[email protected]>wrote: > >> Folks, >> >> Lately I've been learning about control theory from research papers. I >> started off with the classical Witsenhausen counterexample paper, and have >> been reading a lot of papers about just that counterexample. I'm really >> interested in control theory problems that overlap with information theory, >> which is just the sort of problem Witsenhausen focused upon. >> >> I'm also wondering if any computer scientists have applied control theory >> to any computational problems. I'm a little stunned that I can't find >> anything relating things like the Actor Model to ideas from control theory. >> >> Just thinking out loud, but also welcoming suggestions! >> >> Cheers, >> Z-Bo >> >> _______________________________________________ >> fonc mailing list >> [email protected] >> http://vpri.org/mailman/listinfo/fonc >> >> > > _______________________________________________ > fonc mailing list > [email protected] > http://vpri.org/mailman/listinfo/fonc > >
_______________________________________________ fonc mailing list [email protected] http://vpri.org/mailman/listinfo/fonc
