sci.op-research

On Jan 30, 10:09 pm, nav...@gmail.com wrote:
> Hi

Asumming you mean

min || A x - b ||_2
st. C x <= d


Then that can easily be formulated as conic quadratic optimization
problems
(aka second cone problem). Such problems can be solved by MOSEK,
see

http://www.mosek.com/

Using conic quadratic optimization is way more flexible in the type of
constraints
that you can hand than SVD. See section 7.4 at

http://www.mosek.com/fileadmin/products/5_0/tools/doc/html/capi/index.html


> I need to find x that will minimize Ax-b=0, under the inequality
> constraints Cx> Actually the constraints in my problem are only upper and lower bounds
> to x values.
> x is 4x1 vector, A is about 100x4 (and b is of course 100x1(.
> What is the appropriate algorithm?
> Is there any C / C++ code available?
>
> I succeeded solving the non-constrained problem with SVD, but some
> times it give non-legal solution.
>
> Thanks a lot in advance
> Ariel