The Kähler Structure of the Total Least Squares Problem, Brockett’s Steepest Descent Equations, and Constrained Flows
In this paper we show how the Total Least Squares identification problem may be viewed as a steepest descent problem on a Kähler manifold. The Kähler structure gives a method of explicitly deriving the steepest descent equations from a corresponding Hamiltonian flow associated with the problem. In the line-fitting case the steepest descent flow itself is shown also to be equivalent to the flow of a constrained Hamiltonian system — the Toda system.
KeywordsSteep Descent Symplectic Form Gradient Flow Toda Lattice Grassmann Manifold
Unable to display preview. Download preview PDF.
- 1.R. A. Abraham and J. E. Marsden, “Foundation of Mechanics,” Benjamin Cummings, 1978.Google Scholar
- 4.A. Bloch, Steepest Descent, Linear Programming and Hamiltonian Flows, to appear in Cont Math. A.M.S.Google Scholar
- 5.A. Bloch, R. W. Brockett and T. Ratiu, A New Formulation of the Generalized Toda Lattice Equations and their Fixed Point Analysis via the Moment Map, to appear.Google Scholar
- 6.R. W. Brockett, A Geometrical Matching Problem, to appear in J. of Linear Algebra and its Applications.Google Scholar
- 8.C. I. Byrnes and J. C. Willems, Least Squares Estimation, Linear Programming and Momentum, preprint.Google Scholar
- 12.J. Moser, Finitely Many Mass Points on the Line under the Influence of an Exponential Potential, Battelles Recontres, Springer Notes in Physics (1976), 417–497.Google Scholar
- 13.A. Pressley and G. Segal, Loop Groups, Oxford University Press, 1986.Google Scholar