On the reduction of Holt's problem to a finite interval (Q579865)

Holt's problem: \(x''-p(t)x=0\), \(0<t<\infty\), \(x(0)=b\), \(x(\infty)=0\), \(p(t)=t^ 2(1+p_ 0t^ 2)\), \(p_ 0=2m+1\), m nonnegative integer, is solved using Abramov's variant of the ``chasing'' (or ``factorization'' or ``sweeping'') method, replacing the infinite interval by \(0<t<T\), and the condition \(x(\infty)=0\) by \(x'(t^*)-t^*{\tilde \alpha}(t^*)x(t^*)=0,\) \(0<t^*<T\). This condition appears instead of the usual one: \(x'(T)-T\alpha (T)x(T)=0,\) \({\tilde \alpha}\)(t) being an approximation of \(\alpha\) (t). The method seems advantageous when a small interval \([0,t^*]\) is used. The authors construct an error estimate for both reductions.