topic MKL does not provide routines in Intel® oneAPI Math Kernel Library

Quadratic programming in intel math kernal lib

jeewantha_d_ — Tue, 04 Apr 2017 15:03:36 GMT

I am a intel fortran commercial user and looking for general Quadratic programming routine, but I was unable to find it in the math library. The existing routines limited to the constrain L1 <= x <<L2. see

https://software.intel.com/en-us/node/471098#7CF8EA20-5C99-4E1D-A8D6-C6225A3F406B

In mathlab, the general solution is

with constrains

Ax<=b Aeq. x =Beq lb<=x<<la

https://www.mathworks.com/help/optim/ug/quadprog.html?requestedDomain=www.mathworks.com

Our constrain requirement is minimization subjected to Ax <= b

Dear customer,

Zhen_Z_Intel — Thu, 06 Apr 2017 08:50:44 GMT

Dear customer,

There's no QP solver in MKL. But what I feel confused is why you are using non-linear least square solver for QP constraints. The hard constraints are linear.

MKL does not provide routines

mecej4 — Thu, 06 Apr 2017 10:58:05 GMT

MKL does not provide routines for general minimization, even for a function of one variable. The TRNLS and TRNLSBC routines can be used for QP with bound constraints, but will not be as efficient as a dedicated QP routine.

See http://plato.asu.edu/sub/pns.html and look for a solver that meets your needs. In particular, consider Gurobi and BPMPD as QP solvers, and Mosek and Knitro as a general optimization package with facilities for solving QP problems.