← All problems

The Generalized Lax Conjecture for hyperbolicity cones

14 years old · traced to

A homogeneous polynomial h∈R[x1,…,xn]h\in\mathbb{R}[x_1,\ldots,x_n] is hyperbolic with respect to e∈Rne\in\mathbb{R}^n if h(e)>0h(e)>0 and, for every a∈Rna\in\mathbb{R}^n, the polynomial h(te−a)h(te-a) has only real roots. Its hyperbolicity cone is

C⁡(h,e)={a∈Rn: all zeros of h(te−a) are nonnegative}.\operatorname*{C}(h,e)=\{a\in\mathbb{R}^n:\text{ all zeros of }h(te-a)\text{ are nonnegative}\}.

A cone is spectrahedral if it is defined by a linear matrix inequality, equivalently as the positive-semidefinite solution set of a linear matrix pencil. Generalized Lax Conjecture. Every hyperbolicity cone is spectrahedral. This is a major open problem in the theory of hyperbolic polynomials and convex optimization. The paper proves spectrahedrality for derivative relaxations of the positive semidefinite cone under a representation-theoretic assumption, but the general conjecture remains open.

References

5 papers in this index state this conjecture (2012–2020). The statement above is taken from the most recent of them; the others are arXiv:1611.06104, arXiv:1306.4483, arXiv:1207.3159, arXiv:1204.2997.

Source: Mario Kummer, “Spectral linear matrix inequalities”, arXiv:2008.13452 (2021).

Progress summary

Never refreshed

Nothing recorded yet. Refresh searches the literature and the public web for attempts on this problem, and writes the first summary here.

Solutions 1

Claimed by OpenAI. Claims a degree-twenty homogeneous hyperbolic polynomial in 23 real variables whose cone has no finite real-symmetric linear-matrix-inequality representation in the original coordinates, and a degree-sixteen quotient defining the same cone. This concerns the cone itself, not just determinantal representability of a chosen polynomial.See full solutionHide full solution