Convexifying positive polynomials and sums of squares approximation
We show that if a polynomial f∈R[x1,…,xn] is nonnegative on a closed basic semialgebraic set X={x∈Rn:g1(x)≥0,…,gr(x)≥0}, where g1,…,gr∈R[x1,…,xn], then f can be approximated uniformly on compact sets by polynomials of the form σ0+φ(g1)g1+⋯+φ(gr)gr, where σ0∈R[x1,…,xn] and φ∈R[t] are sums of squares of polynomials. In particular, if X is compact, and h(x):=R2−∣x∣2 is positive on X, then f=σ0+σ1h+φ(g1)g1+⋯+φ(gr)gr for some sums of squares σ0,σ1∈R[x1,…,xn] and φ∈R[t], where ∣x∣2=x12+⋯+xn2. We apply a quantitative version of those results to semidefinite optimization methods. Let X be a convex closed semialgebraic subset of Rn and let f be a polynomial which is positive on X. We give necessary and sufficient conditions for the existence of an exponent N∈N such that (1+∣x∣2)Nf(x) is a convex function on X. We apply this result to searching for lower critical points of polynomials on convex compact semialgebraic sets.