greg@endor.harvard.edu (Greg) (10/19/86)
Does anyone have references, knowledge, or ideas about the following class of problems: Let k be a non-negative integer. Suppose you have n points in the plane {(x_i,y_i)} such that if i>j, x_i>x_j. There is any easy theorem that states that there is a function f such that: 1) The k'th derivative of f is defined and continuous. 2) For all i f(x) restricted to [x_i,x_i+1] is a polynomial of degree k+1. 3) f(x_i) = y_i for all i. My question is about possible generalizations (they would come in handy for some computer graphics stuff that I'm doing). Let T be a triangulation of the interior of a polyhedron P in R^n. Let g be a function from the vertices of T to the reals, and let k be a non-negative integer. Is there always a function f such that: 1) The k'th derivative of f is defined and continuous. 2) For any n-dimensional simplex S of T, f restricted to S is a polynomial of degree k+1. 3) For any vertex v of T, f(v) = y(v)? For an even broader generalization, we may allow the faces of the simplices of T to be algebraic surfaces of some degree (of course, we would need to allow f to have degree >k+1 as well) rather than flat, and we could allow T to be embedded in an algebraic surface of some degree rather than R^n. ---- gregregreg
greg@endor.harvard.edu (Greg) (10/21/86)
In article <471@husc6.HARVARD.EDU> greg@endor.UUCP (Greg) writes: >Let T be a triangulation of the >interior of a polyhedron P in R^n. Let g be a function from the vertices of T >to the reals, and let k be a non-negative integer. Is there always a function >f from P to R such that: > >1) The k'th derivative of f is defined and continuous. >2) For any n-dimensional simplex S of T, f restricted to S is a polynomial of >degree k+1. >3) For any vertex v of T, f(v) = y(v)? Here are the essentials of an article that I stupidly posted to net.math rather than sci.math. I also wish to make a minor correction: I'm convinced that if f has degree k+1, f usually doesn't have enough degrees of freedom to satisfy all of the above conditions. So I'll change the question some: For what integer d (which may depend on k, n, and T) can we always find an f with the above properties such that f restricted to S is a polynomial of degree d? ---- Greg