Re: qp(): how do I get a simple inequality constraint?

[Leo Butler]

forkandwait <address@hidden> writes:

> Leo Butler <l.butler <at> cmich.edu> writes:
>
>> 
>> forkandwait <forkandwait <at> gmail.com> writes:
>> >
>> > min (p'*G*p)/2 + f'*p
>> >
>> > subject to C*p <= b
>> 
>> In my first post, I should have said to use A_lb, A_in and A_ub to
>> enforce the inequallity on your slack variable, as you divined.
>>
>
> Like this? --
>
> phat = qp(x0, G, f, [], [], [], [], [], C, b) 

Let me explain with an example:

Suppose you want to minimize

x^2 + y^2 + x subject to

x <= 0

We can do the following:

H = 2*eye(2);
q = [1;0];
A = [];
b = [];
lb = [];
ub = [];
A_lb = [];
A_ub = [0];
A_in = [1,0];

[x, obj, info, lambda] = qp ([0;0], H, q, A, b, lb, ub, A_lb, A_in,
A_ub)

You can verify that the objective function is
(x+1/2)^2 + y^2 - 1/4
which has a global minimum at x=-1/2, and that qp finds that.

--------------------

[You can probably ignore the following comment]
Alternatively, *if we could only put lower bounds on variables*, we
could add a slack variable, s, and re-write the program as

x^2+y^2+x s.t. x+s=0, s>=0.

Here:

H = 2*[1,0,0;0,1,0;0,0,0];
q = [1;0;0];
A = [1,0,1];
b = [0];
lb = [];
ub = [];
A_lb = [0];
A_ub = [];
A_in = [0,0,1];

[x, obj, info, lambda] = qp ([0;0;0], H, q, A, b, lb, ub, A_lb, A_in,
A_ub)

Again, qp finds the global minimum.





>
> This is what I have, and I get a very different answer than the book.  See 
> below
> for my inputs (phat is the result, book_phat is what the book has, etc).
>
> I will continue hacking and reading (can you recommend a book?).
>
> (The difficulty in posing the question is that the algorithm munges a bunch of
> stuff before calling qp, so determining whether the problem is in the set up 
> or
> my call to qp() is not so trivial, at least for an optimization ignoramus like
> myself.)
>
> The full qp reference is this:  qp(x0, H, q, A, b,lb, ub, A_lb, A_in, A_ub)
>
> My inputs and outputs:
>
> phat =
>    0.00000
>    1.00000
>    0.00000
>    1.00000
>    0.01673
>    1.00000
>
> book_phat = 
>    0.00000
>    1.00000
>    0.45000
>    0.54000
>    0.01000
>    0.97000
>
> G =
>   1.4151e+002  0.0000e+000  0.0000e+000  0.0000e+000  3.0723e+003  0.0000e+000
>   0.0000e+000  1.4151e+002  6.6074e+002  0.0000e+000  0.0000e+000  0.0000e+000
>   0.0000e+000  6.6074e+002  3.0851e+003  0.0000e+000  0.0000e+000  0.0000e+000
>   0.0000e+000  0.0000e+000  0.0000e+000  3.0851e+003  0.0000e+000  1.4345e+004
>   3.0723e+003  0.0000e+000  0.0000e+000  0.0000e+000  6.6701e+004  0.0000e+000
>   0.0000e+000  0.0000e+000  0.0000e+000  1.4345e+004  0.0000e+000  6.6701e+004
>
> f =
>   -5.1391e+001
>   -2.7590e+002
>   -1.2882e+003
>   -5.1457e+004
>   -1.1157e+003
>   -2.3926e+005
>
> C =
>   -1   0   0   0   0   0
>    0  -1   0   0   0   0
>    0   0  -1   0   0   0
>    0   0   0  -1   0   0
>    0   0   0   0  -1   0
>    0   0   0   0   0  -1
>    1   1   0   0   0   0
>    0   0   1   1   0   0
>    0   0   0   0   0   1
>
> b =
>    0
>    0
>    0
>    0
>    0
>    0
>    1
>    1