Algebraic-exponential data recovery from moments (Q908214)

Given a bounded open subset \(G\) of \(\mathbb{R}^{n}\) with algebraic boundary \(\Gamma =\partial G\) of degree \(d,\) the authors describe an algorithmic procedure for obtaining a real polynomial \(p\in I(\Gamma )\) (i.e. vanishing on \(\Gamma \)) by means of finitely many moments of the Lebesgue measure on \(G\) (up to the order \(3d\) for the general case, and \(2d\) for the convex case). The main result is Theorem 2.2, which uses a Whitney's generalization of Stokes's theorem applied over a semi-algebraic triangulation of \(\Gamma\). The authors prove this result under the technical assumption that \(x=0\) does not lie on the Zariski closure of \(\Gamma \) (i.e. the polynomials of the ideal \(I(\Gamma )\) do not vanish at \(0\)), and treat the general case by a change of variable (Section~2.3). In addition to the Lebesgue measure, the authors' technique is easily adapted to include measures of the form \(d\mu :=\exp (p(x))dx\) where \(p\in \mathbb{R}[x]\) (in this case the required number of moments takes into account the degree of \(p\)). The authors also deal with general (non-algebraic) boundaries, using approximations and illustrate their results with several examples.