Positive polynomials and numerical approximation
Bruno
Després, Laboratoire J.-L. Lions, Université Paris 6
Abstract:
The design of high order methods with strong control of the
maximum principle appears to be still a core problem for the
numerical approximation of many problems in scientific computing,
including non linear equations. In these directions, a natural
question is having good representations of polynomial with
bounds. I will describe some recent ideas based on the Lukacs
Theorem (and representation as some of squares) in order to design
all such polynomials: it results in a new efficient
Newton-Raphson algorithm which has been adapted for the
numerical solution of the advection equation. The case of polynomial
with two bounds can be treated with an ad-hoc quaternion algebra. A
selection of open problems will be described.