First order conditions for semidefinite representations of convex sets defined by rational or singular polynomials
First order conditions for semidefinite representations of convex sets defined by rational or singular polynomials
复制标题
DOI:
10.1007/s10107-009-0339-9
复制
发表时间:
2008-06
影响因子:
2.7
通讯作者:
Jiawang Nie
中科院分区:
文献类型:
--
作者:
Jiawang Nie
A set is calledsemidefinite representableorsemidefinite programming(SDP)representableif it equals the projection of a higher dimensional set which is defined by someLinear Matrix Inequality(LMI). This paper discusses the semidefinite representability conditions for convex sets of the form. Here,is a convex domain defined by some “nice” concave polynomialsgi(x) (they satisfy certain concavity certificates), andf(x) is a polynomial or rational function. Whenf(x) is concave over, we prove thathas some explicit semidefinite representations under certain conditions calledpreordering concavityorq-module concavity, which are based on the Positivstellensatz certificates for the first order concavity criteria:$$f(u) + \nabla f(u)^T(x-u) -f(x) \geq 0, \quad \forall \, x, u \in \mathcal {D}.$$Whenf(x) is a polynomial or rational function having singularities on the boundary of, aperspective transformationis introduced to find some explicit semidefinite representations forunder certain conditions. In the special casen= 2, if the Laurent expansion off(x) around one singular point has only two consecutive homogeneous parts, we show thatalways admits an explicitly constructible semidefinite representation.