Oscillation and nonoscillation criteria for nth-order linear differential equations (Q1084248)

According to a well-known result of \textit{E. Hille} [Trans. Am. Math. Soc. 64, 234-252 (1948; Zbl 0031.35402)], the second order differential equation \(y''+py=0\), where p is positive and continuous on [0,\(\infty)\), is oscillatory if \(\omega_*=\liminf_{x\to \infty}x^ 2p(x)>1/4\) and nonoscillatory if \(\omega^*=\limsup_{x\to \infty}x^ 2p(x)<1/4.\) \textit{M. Hanan} [Pac. J. Math. 11, 919-944 (1961; Zbl 0104.309)], \textit{H. Howard} [Trans. Am. Math. Soc. 96, 296-311 (1960; Zbl 0094.285)] and \textit{Z. A. Chanturiya} [On the oscillation of solutions of higher order linear differential equations, Report 16, Seminar of I. Vekua Institute of Applied Mathematics, Tbilisi, U.S.S.R. (1982)] obtained similar results for the third-, fourth- and nth-order equations, respectively. In all of these results, the constant \(M_ n=\max_{0\leq x\leq n-1}\prod^{n- 1}_{j=0}| x-j|\) plays an important role. We not only extend the above results to the more general class of differential equations (E) \(L_ n(r_ n,...,r_ 0)y+py=0,\) where the differential operator \(L_ n\) is defined successively by \[ L_ 0(r_ 0)y=r_ 0(x)y,\quad L_ k(r_ k,...,r_ 0)y=r_ k(x)\frac{d}{dx}L_{k-1}(r_{k-1},...,r_ 0)y, \] k\(=1,2,...,n\), but establish improved criteria of the solution space using the values of the local maxima and minima of the polynomial function \(P_ n(x)=x(x-1)...(x-n+1).\) The functions \(r_ 0,...,r_ n\) in (E) are assumed to be continuous and positive on an interval [a,\(\infty)\) and p is continuous and has constant sign on [a,\(\infty)\). Equation (E) is said to be k-(n-k) disfocal on an interval I if, for every pair of points b,c\(\in I\), \(b<c\), the only solution satisfying the condition \(L_ 0y(b)=...=L_{k-1}y(b)=0=L_ ky(c)=...=L_{n-1}y(c)\) is the trivial solution; otherwise, it is said to be k-(n-k) focal on I. When n is even and \(p>0\), a main theorem states that (E) is \((2j+1)-(n- 2j-1)\) disfocal, \(j=0,1,...,m-1\), (n-2m)/2,...,(n-2)/2, on [a,\(\infty)\) if \(r_ i\geq K_ i\) for some positive constant \(K_ i\), \(i=1,2,...,n- 1\), and \((r_ nr_ 0K_ 1...K_{n-1})^{-1}(x-a)^ np\leq - M_{n,2m-2},\) for some m, \(1\leq m\leq (\sigma +2)/2\), on [a,\(\infty)\), where \(\sigma =[n/2]-1-((-1)^{[n/2]}+(-1)^ n)/2\) and \(M_{n,i}=\max_{i\leq x\leq i+1}| P_ n(x)|.\) On the other hand, (E) is \((2j+1)-(n-2j-1)\) focal, \(j=m,...,(n-2m-2)/2\), on \([a_ 1,\infty)\) for any \(a_ 1\geq a\) if \(r_ i\leq K_ i'\) for some positive constant \(K_ i'\), \(i=1,2,...,n-1\), and if \(-M_{n,2m}+\epsilon \leq (r_ nr_ 0K_ 1'...K'_{n-1})^{-1}(x-a)^ np\) for some \(\epsilon >0\) and some m, \(0\leq m\leq \sigma /2\), on [a,\(\infty)\). Also discussed are the other cases of (E) and the special case \(r_ i=1\), \(i=0,1,...,n\).