Re: warning: sqp: QP subproblem failed to converge in 200 iterations

[Urs Hackstein]

Just for the record: I tried to change all entries "200" to "500" in qp.m (https://bitbucket.org/carandraug/octave/src/13d1e9bfa362/scripts/optimization/qp.m?at=default), but it raises a new error
error: __qp__: operator *: nonconformant arguments (op1 is 6x6, op2 is 7x1)
error: called from:
error:   /home/hackstein/.../qp.m at line 393, column 26

which I don't understand. 7x1 seems correct as we have seven variables and starting point (2,4,6,1,2,6,50), but 6x6? Any idea?

Or if I want to follow Louis' hint: Where have I to write optimset(MaxIter,500) so that it sets the options for Maxiter in qp.m higher?

2013/4/23 Urs Hackstein <address@hidden>

sqp(x0, phi, g, h, lb, ub, 500) sets the options for Maxiter in sqp higher, but it is not sqp causing the problem, it is the qp-subroutine called by sqp. Or am I missing something?

2013/4/23 louis scott <address@hidden>

see http://www.gnu.org/software/octave/doc/interpreter/Quadratic-Programming.html

and http://www.gnu.org/software/octave/doc/interpreter/Linear-Least-Squares.html#doc%2doptimset

you want to set the options for MaxIter higher.

Something like optimset(MaxIter,201)

On Tue, Apr 23, 2013 at 10:38 AM, Urs Hackstein <address@hidden> wrote:

Hello,

I am dealing with an iteration where in each step a minimization problem

min_{a_i<=x_i<=b_i})(real(b)-real(f(x_1,x_2,x_3,x_4,x_5,x_6,x_7)).^2+(imag(b)-imag(f(x_1,x_2,x_3,x_4,x_5,x_6,x_7)).^2) is solved using sqp.

However, I receive every time the following error:

warning: sqp: QP subproblem failed to converge in 200 iterations

How can I adjust the number of iterations of qp?

(It should converge as it does if I compute f in a different way, but this way is slightly faster.)

Best regards,

Urs Hackstein

_______________________________________________
Help-octave mailing list
address@hidden
https://mailman.cae.wisc.edu/listinfo/help-octave