Duality for infinite Hermite spline interpolation (Q788924): Difference between revisions

This paper considers the problem of interpolating data at a non- decreasing sequence \(x=(x_ i)^\infty_{-\infty}\) by spline functions of degree n with knots at a non-decreasing sequence \(\xi =(\xi_ i)^\infty_{-\infty}\). (As usual, multiplicities in x denote interpolation of consecutive derivatives.) This problem, denoted by (n,x,\(\xi)\), is said to be solvable if for any bounded data there is a unique bounded interpolating spline. The author shows that if the mesh ratios of x and \(\xi\) are finite, then (n,x,\(\xi)\) is solvable if and only if the 'dual' problem \((n,\xi\),x) is solvable. Here the mesh ratio of a non-decreasing sequence \((y_ i)^\infty_{-\infty}\) is defined to be \(\sup \{(y_{i+1}-y_ i)/(y_{j+1}-y_ j):\) integers i,j with \(y_ j<y_{j+1}\}\).