As a method option to FindMinimum "QuadraticProgramming" is still a valid option. It has the advantage of being significantly faster than the default for such problems ("InteriorPoint") in my experience. It has the disadvantage that it appears to crash the kernel (V10.1 Win64) on some problems and its output is a little erratic (does not appear to be optimal in some cases in comparison to interior point solutions).

However my question is, is this still a supported method, as it appears to have been removed from the documentation (except for one mention in one of the FindMaximum warning messages)?

  • 1
    $\begingroup$ For completeness, can you give some examples where the method crashes or does not give a true optimum? $\endgroup$ – J. M. is in limbo May 19 '15 at 15:58

Since the QuadraticProgramming method is not documented, it is also not supported (however, bug reports on it are welcome!)

It has been implemented some years ago using a rather dated by now version of the COIN-CLP library, which has a number of known problems. Future versions will hopefully bring optimization improvements and use more modern libraries.

  • $\begingroup$ I'll attempt to add some examples when I have time. $\endgroup$ – Ymareth May 19 '15 at 16:36

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.