Comparison and integral comparison theorems for linear differential equations (Q1338295)

This paper is concerned with the problem of establishing comparison results for the linear differential equation (1) \(Ly+ q(x) y= 0\), where \(L\) is a linear disconjugate operator of order \(n\), \(n\geq 2\): \(Ly= p_ n(p_{n-1}\cdots (p_ 1(p_ 0 y)')'\cdots)'\). Here \(p_ 1\in C^{n- i}(I)\), \(p_ i(x)> 0\) on \(I= [a_ 0,b_ 0)\) \((a_ 0< b_ 0\leq \infty)\) and \(q\) is a continuous function of one sign. Along with (1) boundary conditions of type (2) \(L_ i y(a)= 0\), \(i\in \{i_ 1,\dots, i_ k\}\), \(L_ j y(b)= 0\), \(j\in \{j_ 1,\dots, j_{n-k}\}\), are considered, where \(1\leq k\leq n-1\), \(0\leq i_ 1<\cdots< i_ k\leq n-1\), \(0\leq j_ 1<\cdots< j_{n-k}\leq n-1\), \(a,b\in I\), \(L_ 0 y= p_ 0 y\), \(L_ i y= p_ i(L_{i-1} y)'\), \(Ly= L_ n y\). For \(a\in I\) we define the \(i\)th extremal point \(\theta_ i(a)\) corresponding to BVP (1), (2) to be the \(i\)th value of \(b\in I\), \(b>a\) for which there exists a nontrivial solution of BVP (1), (2). If \(q(x)\) is one sign, then by \textit{Z. Nehari} [Trans. Am. Math. Soc. 129, 500-516 (1967; Zbl 0183.091)] a necessary condition for the existence of the extremal points is that \(q\) satisfies the parity condition (P) \((-1)^{n-k} q(x)< 0\). Hence through this paper the author assumes that (P) is satisfied. Denote by \(\omega_ i(a)\) the \(i\)th extremal point corresponding to the BVP (3) \(P_ n(P_{n-1}\cdots (P_ 1(P_ 0 u)')'\cdots)'+ Q(x) u=0\), (4) \(M_ i u(a)= 0\), \(i\in \{i_ 1,\dots, i_ k\}\), \(M_ j u(b)= 0\), \(j\in \{j_ 1,\dots, j_{n-k}\}\), (where \(M_ i u\) are defined as in (2) and \(Q\) satisfies the same assumptions as \(q\)). Under conditions of direct type, \(P_ i\leq p_ i\), \(| Q|\geq | q|\), or of integral type, like \[ \int^ x_ a {dt\over P_ i(t)}\geq \int^ x_ a {dt\over p_ i(t)},\quad \int^ x_ a {| Q(t)|\over P_ 0(t) P_ n(t)} dt\geq \int^ x_ a {| q(t)|\over p_ 0(t) p_ n(t)} dt \] for all \(x\in (a,b)\), the author proves some comparison theorems formulated in the extremal points \(\theta_ i(a)\) and \(\omega_ i(a)\) corresponding to the BVPs (1), (2) and (3), (4), respectively. These results generalize and extend those obtained for example by \textit{W. Leighton} and \textit{Z. Nehari} [Trans. Am. Math. Soc. 89, 325-377 (1959; Zbl 0084.081), \textit{G. J. Butler} and \textit{L. H. Erbe} [J. Differ. Equations 47, 214-226 (1983; Zbl 0528.34037)] and \textit{W. J. Kim} [J. Math. Anal. Appl. 105, 187-198 (1985; Zbl 0559.34030)].