1. 제 목 : A New Active Set Algorithm for Large-Scale Box Constrained Optimization
2. 발표자 : Hongchao Zhang (University of Minnesota, USA)
3. 일 시 : 2006년 7월 6일(목) 10:00 ~ 12:00
4. 장 소 : 경북대 자연대 3호관 291호
5. 초청교수 : 한 동 석 교수
6. 강사약력
2006-present Institute for Mathematics and its Applications (IMA),
University of Minnesota, USA
2006 Ph.D. University of Florida, Mathematics, USA
2001 M.S. Chinese Academy of Science, Mathematics
1998 B.S. Shandong University—China
7. 내용요약 :
A new active set algorithm (ASA) is developed for box constrained optimization which consists of a nonmonotone gradient projection step, an unconstrained optimization step and a set of rules for branching between these two steps. Global convergence to a stationary point is established. This algorithm eventually reduces box constrained optimization to unconstrained optimization even without assuming the strict complementarity condition. A specific implementation of ASA is given which exploits the recently developed cyclic Barzilai-Borwein algorithm for the gradient projection step and the recently developed conjugate gradient algorithm CG_DESCENT for unconstrained optimization. Extensive numerical results are provided.
※ 주최 : 전자전기컴퓨터학부, 전자기술연구소,
경북대학교 BK21(정보기술연구인력양성사업단),
경북대학교 기초과학연구소, 경북대학교 BK21(수리계산)
|
