Stability and comparison of systems in partially ordered space
Alexey G.
Mazko
Various
natural systems are positive and monotone. Positivity of a dynamic system
is equivalent to positivity of some operator describing its motion with respect
to a cone of phase space. Similarly, monotonicity of the system means that the
motion operator is monotone. These properties of systems should be taken into
consideration and be used in analysis and synthesis problems especially in
stability and spectral characteristics investigation, in numerical procedures
of construction of the solutions and appropriate controls etc. Stability
investigation of linear autonomous positive systems is reduced to solving
algebraic equations defined by operator coefficients of the systems. Well-known
Lyapunov and Riccati differential equations are positive systems concerning the
cone of symmetric positive semi-definite matrices.
In present paper, we study
positivity, monotonicity and stability
conditions for the classes of differential systems
,
,
where
is a linear bounded
operator in Banach space partially ordered by normal reproducing cone and
is a nonlinear
operator function. In robust stability problem, we assume that
,
.
Positivity of the system can be utilized for
estimation of its solutions. In the case of monotone evolutional operator, we
have two-sided estimations for the solution
in terms of solutions
of auxiliary linear systems.
The main research results are the
positivity conditions and algebraic criteria for asymptotic stability of linear
systems formulated in terms of the monotone and monotonically invertible
operators. Stability investigation of the linear positively reducible systems
and time-varying systems described by functionally commutative operators is
reduced to solving algebraic equations with monotonically invertible operators.
The methods for robust stability analysis and analogs of comparison systems in
partially ordered space are developed.
We have some examples of positive and monotone systems with respect to the cones of nonnegative vectors and positive semi-definite matrices. The matrix differential equation
,
where
and
, is positive with respect to the cone of positive
semi-definite symmetric matrices. It can be used as a comparison system in
stability problem for more complicated dynamic systems.
Well-known
criteria for a mean quadratic asymptotic stability of the Ito stochastic
systems are corollaries of obtained results.