Problems with quadratic objective function should be solved with CPXqpopt

[ghost]

According to the CPLEX manual, QPs with quadratic objective (and only linear constraints) should also be solved by CPXqpopt. However, the current implementation only calls CPXqpopt if the problem has a quadratic constraint (i.e. if has_qc equals true which is only set in add_qconstr!.

[joehuchette]

Were you noticing any issues with this? We test this behavior and it seems to work as expected. This is certainly something change, though; thanks for the report!

[ghost]

Actually yes. When solving some QP cplex always used the simplex method
and not the barrier, although I set
CPX_PARAM_QPMETHOD=CPLEX.CPX_ALG_BARRIER. This led to some numerical
problems in my case. But I'm not sure if the computed solution was
correct when there were no numerical problems.

Am 26.09.2014 um 17:44 schrieb Joey Huchette:

Were you noticing any issues with this? We test this behavior and it seems to work as expected. This is certainly something change, though; thanks for the report!

---

Reply to this email directly or view it on GitHub:
#28 (comment)