A penalty approach for solving nonsmooth and nonconvex MINLP problems

This paper presents a penalty approach for globally solving nonsmooth and nonconvex mixed-integer nonlinear programming (MINLP) problems. Both integrality constraints and general nonlinear constraints are handled separately by hyperbolic tangent penalty functions. Proximity from an iterate to a feas...

ver descrição completa

Detalhes bibliográficos
Autor principal: Costa, M. Fernanda P. (author)
Outros Autores: Rocha, Ana Maria A. C. (author), Fernandes, Edite Manuela da G. P. (author)
Formato: conferencePaper
Idioma:eng
Publicado em: 2018
Assuntos:
DIRECT
MINLP
Oracle
Penalty function
Science & Technology
Texto completo:http://hdl.handle.net/1822/55224
País:Portugal
Oai:oai:repositorium.sdum.uminho.pt:1822/55224
Descrição
Resumo:This paper presents a penalty approach for globally solving nonsmooth and nonconvex mixed-integer nonlinear programming (MINLP) problems. Both integrality constraints and general nonlinear constraints are handled separately by hyperbolic tangent penalty functions. Proximity from an iterate to a feasible promising solution is enforced by an oracle penalty term. The numerical experiments show that the proposed oracle-based penalty approach is effective in reaching the solutions of the MINLP problems and is competitive when compared with other strategies.

Registros relacionados