SOS-convexity
A multivariate polynomial is SOS-convex if its Hessian matrix H can be factored as H = STS where S is a matrix which entries are polynomials in x. In other words, the Hessian matrix is a SOS matrix polynomial.
An equivalent definition is that the form defined as g = yTHy is a sum of squares of forms.
Connection with convexity
If a polynomial is SOS-convex, then it is also convex. Since establishing whether a polynomial is SOS-convex amounts to solving a semidefinite programming problem, SOS-convexity can be used as a proxy to establishing if a polynomial is convex. In contrast, deciding if a generic quartic polynomial of degree four is convex is a NP-hard problem.The first counterexample of a polynomial which is convex but not SOS-convex was constructed by Amir Ali Ahmadi and Pablo Parrilo in 2009. The polynomial is a homogeneous polynomial that is sum-of-squares and given by:
In the same year, Grigoriy Blekherman proved in a non-constructive manner that there exist convex forms that is not representable as sum of squares. An explicit example of a convex form that is not a sum of squares was claimed by James Saunderson in 2021.