QP

From OpenOpt

Jump to: navigation, search
Quadratic Problems (QP)
 \mathbf{\frac{1}{2} x^T Hx + f^T x \rightarrow min}
subjected to
\mathbf{lb \le x \le ub}
\mathbf{A x \le b}
\mathbf{A_{eq} x = b_{eq}}


Attention all these solvers listed below can handle only convex QPs, for non-convex local optimum try using converter to NLP, if global optimum is required - try using a GLP solver on the obtained NLP

QP solvers connected to OpenOpt:

Solver License Made by Info
cvxopt_qp GPL3 Lieven Vandenberghe, Joachim Dahl requires CVXOPT installed
qlcp MIT Enzo Michelangeli and IT Vision Ltd qlcp code is included into OO (requires ver >= 0.32). Currently it handles dense probs only and involves inversion of H matrix (no walk-around is implemented yet).
(since v. 0.33)
cplex
  • commercial
  • full version free for education
  • free 90-days trial with limitations up to 500 vars/cons
IBM (after ILOG acquisition) requires cplex and its Python API installed
converter to nlp Dmitrey Example: r = p.solve('nlp:ipopt', plot=1). Can handle x0. Recommended solvers (mb require installation, see NLP doc page): ipopt, algencan. For ralg reducing p.ftol and p.xtol sometimes is helpful


See also: QCQP, MIQP, MIQCQP, SOCP, SDP, NLP

Retrieved from "http://openopt.org/QP"
Personal tools
    Latest OOSuite 0.37
    Next release:

    2012-03-15


    Development