Asymptotic and numerical stability of systems of neutral differential equations with many delays (Q2378240)

The paper is concerned with asymptotic stability of a system of linear neutral differential equations with many delays of the form \[ y'(t)=Ly(t)+\sum_{i=1}^d M_iy(t-\tau_i)+\sum_{i=1}^dN_iy'(t-\tau_i). \] The authors present a new sufficient condition on the matrices \(L, M_i, N_i\) (\(i=1, 2,\dots, d\)) such that the equation is asymptotically stable. The paper begins with a brief summary of existing literature on linear neutral differential equations (NDEs) with a single delay, and indicates progress made with the multiple delay case. Definitions of both delay-dependent and delay-independent asymptotic stability are included in section 2. A lemma, giving a new result concerning the distribution of the zeros of the characteristic function, is presented and proved. Proof by induction is used. A new result, proving delay-dependent asymptotic stability under sufficient conditions on the matrices \(L, M_i, N_i\) is presented and proved in Theorem 2.1. Sufficient conditions for delay-independent asymptotic stability are presented in a corollary. In section 3 the authors include relevant (existing) lemmata and definitions pertaining to numerical stability of NDEs with multiple delays, prior to establishing Theorem 3.1, which gives a full characteristation of the \(NGP_m\)-stability of linear multi-step methods when applied to neutral differential equations with many delays. In section 4 the paper concludes with the results of a numerical experiment, using the two-step backward differentiation formula and the two-step Adams-Bashforth method, to support Theorem 3.1.