Dual quadratic estimates in polynomial and Boolean programming (Q1173717)

The Lagrange function is a source for getting the so-called ``dual bound'' for a wide class of mathematical programming problems. In general, the determination of this bound is a nondifferentiable optimization problem. A dual quadratic bound can be used for polynomial programming problems and for quadratic Boolean problems. In some numerical experiments the relative error was \(3\%\) (the linear relative error was about \(25\%\)).