DocumentCode
3015934
Title
Exact solution to Lyapunov´s equation using algebraic methods
Author
Djaferis, T.E. ; Mitter, S.K.
Author_Institution
Massachusetts Institute of Technology, Cambridge, Massachusetts
fYear
1976
fDate
1-3 Dec. 1976
Firstpage
1194
Lastpage
1200
Abstract
Let A´P + PA = -Q be a Lyapunov equation with A being a stability matrix and both A and Q matrices with rational entries. Multiplying A and Q by a suitable positive Integer an equivalent Lyapunov equation A´1P + PA1 = -Q1 is obtained, with A1 and Q1 having integer entries, Let I(x,y) be the ring of polynomials in x and y over the integers I, and E be the set of all square matrices with integer entries. The solution P to this equation is given by: Pu = (emn) = fA 1 (qu(x,y), Q1) P = 1u2 ?? Pu where: qu(x,y) ?? I(x,y) and u ?? I fA 1 : I(x,y) ?? E ?? E defined as fA 1 (h(x,y), M) = ??j,khjk(A´1)j ?? M??A1 k which is a finite sum. The calculation of u and qu(x,y) requires finding the characterstic polynomial of A1, as well as using the Euclidean Algorithm, computations which lead to polynomial coefficient growth. In order to eliminate the space consuming manipulation of large integers in intermediate steps, modular arithimetic is used to obtain the matrix p iPu= (emnmodpi) and p iu=mu modpi with pi a prime, for a sufficient number of primes. The Chinese Remainder Theorem is then applied to obtain the solution P. The algorithm has been programmed on MACSYMA which is a very suitable computer programming system for all the numerical computations involved. Numerical results as well as extensions to solving the Algebraic Riccati Equation are presented.
Keywords
Algebra; Computer science; Digital arithmetic; Equations; Laboratories; Polynomials; Stability; Symmetric matrices;
fLanguage
English
Publisher
ieee
Conference_Titel
Decision and Control including the 15th Symposium on Adaptive Processes, 1976 IEEE Conference on
Conference_Location
Clearwater, FL, USA
Type
conf
DOI
10.1109/CDC.1976.267668
Filename
4045776
Link To Document