Title: Ergodic Theory and Control Theory
1Ergodic Theory and Control Theory
Igor Mezic
Department of Mechanical and Environmental
Engineering, University of California, Santa
Barbara
Mohammed Dahleh Symposium, UCSB 2002
Sponsored by AFOSR, NSF, Sloan Foundation
2Introduction
- Control of Hamiltonian systems
Applications -molecular dynamics, vortex
dynamics, satellite
motion control, quantum control, power systems
- Volume-preserving systems on groups.
3Introduction
T
4Ergodicity and Controllability
5Ergodicity and Controllability
Observation 2 For Uw, w irrational, the
system is almost reachable. In
fact, for irrational w the system is
ergodic.
6Ergodicity and Controllability
Theorem Assume A is a compact abelian
topological group, U an input set of
positive Haar measure and g an element
of U such that gn is dense in G. Then
the system TU is controllable.
Remark The element g is an ergodic element. Its
presence in the input set is
necessary Let G-1,1 under multiplication.
Then U1 is an input set of measure ½ but 1 is
not an ergodic element, and the element -1 is NOT
reachable.
7Ergodicity and Controllability
- Stronger results are possible, where only
dim(U)gt0 is necessary.
- Extensions for more general systems e.g. linear
systems - with input constraints continuous-time
Hamiltonian systems - vortex dynamics.
8Hamiltonian systems twist maps
Jim Swifts page http//odin.math.nau.edu/jws/st
d.map/
9Hamiltonian systems twist maps
By Zoran Levnajic (UCSB/Trieste).
10 Hamiltonian systems twist maps
Kolmogorov-Arnold-Moser theorem implies
No controllability!
11Hamiltonian systems twist maps
12Discussion and Conclusions
- Controllability results for Hamiltonian and
- volume-preserving systems extensions for N DOF
- With D. Vainchtein control of (Hamiltonian)
- vortex dynamics using properties of nominal
dynamics. - Control of statistical properties conserved
quantities.
- Linking ergodic theory and control theory
- 1. Control of mixing.
- 2. Nonlinear model validation.
- 3. Ergodicity and controllability.