On a method of simplex embeddings in convex programming. (Q1432468)

The authors consider the problem \(\min\{ \varphi(x): x \in R \}\), \(R=\{ x \in E^n: q_j (x)\leq 0, j=1, 2, \dots, m \}\), where \(\varphi(x)\) and \(q_j(x)\), \(j=1, 2, \dots, m\) are quasi--convex functions, and \(R\) satisfies the regularity condition. A well-known approach to approximation of a solution \(x^*\) of the problem consists in constructing a sequence of ellipsoids or polyhedra \(R^k \ni x^*\) such that \(\text{meas}(R^k) \to 0\) as \(k \to \infty\). In this paper the authors suggest a new algorithm of constructing a sequence of approximating polyhedra \(R^k\).