The extensive use of weighting patterns in the analysis and synthesis of linear, finite-dimensional systems relies heavily on a realizability result involving a decomposition of the weighting pattern $W( t,\tau )$ into an inner product of two vector terms $\psi ( t )$ and $\beta ( \tau )$. This result, as stated for the continuous-time case, depends on the invertibility of the transition matrix $\Phi ( t,\tau )$ for the general case. Hence, the realizability condition must be modified for the discrete-time case, where the transition matrix is not invertible in general.This paper presents and proves a discrete-time realizability condition which closely parallels the continuous-time results. At the same time, some insight is conveyed regarding the differences between the continuous- and discrete-time cases. Finally, auxiliary results are proved along established lines.
[1]
E. I. Jury,et al.
Theory and application of the z-transform method
,
1965
.
[2]
R. E. Kalman,et al.
Contributions to linear system theory
,
1965
.
[3]
D. Youla.
The Synthesis of Linear Dynamical Systems from Prescribed Weighting Patterns
,
1966
.
[4]
E. K. Blum,et al.
A modification of the Runge-Kutta fourth-order method
,
1962
.
[5]
L. Weiss.
Controllability for various linear and nonlinear system models
,
1970
.
[6]
Michael A. Arbib,et al.
Topics in Mathematical System Theory
,
1969
.