Linearized oscillations for difference equations (Q1199187)

The authors consider the nonlinear difference equation \[ X_{n+1}=X_ n+f(X_{n-k_ 1},X_{n-k_ 2},\dots,X_{n-k_ m}), \] where \(k_ 1,k_ 2,\dots,k_ m\) are positive integers, \(n=0,1,2,\dots\), and \(f\) is a real-valued function of \(m\) variables, \(f(u_ 1,u_ 2,\dots,u_ m)\geq 0\) or \(\leq 0\) according as \(u_ 1,u_ 2,\dots,u_ m\geq 0\) or \(\leq 0\), respectively. It is further assumed that \(f\) has continuous first partial derivatives \(D_ if\), and there exists \(\delta>0\), such that for all \(u_ 1,u_ 2,\dots,u_ m\in[-\delta,\delta]\), \(D_ i f(0,0,\dots,0)=p_ i>0\), \(i=1,2,\dots,m\) and \(\sum_{i=1}^ m (p_ i+k_ i)\neq 1\). Further \(f(u_ 1,u_ 2,\dots,u_ m)\leq\sum_{i=1}^ m p_ i u_ i\), \(u_ i\in[0,\delta]\) or \(\geq \sum_{i=1}^ m p_ i u_ i\), \(u_ i\in[-\delta,0]\), \(i=1,2,\dots,m\). Theorem: Under the above conditions, everey solution of the difference equation oscillates if and only if every solution of the associated linear equation \(y_{n+1}-y_ n+\sum_{i=1}^ m p_ i y_{n-k_ i}=0\), \(n=0,1,2,\dots\), oscillates.It is significant that, for oscillation, the characteristic equation \(\lambda-1+\sum_{i=1}^ m p_ i \lambda^{-k_ i}=0\) of the associated linear equation has no positive roots. The authors also present an application to equations in neural networks. Amongst the research papers cited, the authors refer to \textit{I. Györi} and the second author, Differ. Integral Equ. 2, No. 2, 123-131 (1989; Zbl 0723.34058).