View source for Lls2qp

[[Category:Utility Functions]]
==Purpose==

Converts an lls problem to a new problem based on the formula below.  Only the objective function is affected. The 

problem can be of any type with an LLS objective.


<math>
\begin{array}{lc}
\min\limits_{x} & f(x) = \frac{1}{2} || C x - d || = \\& 0.5 * (y - Cx)'*(y - Cx) = 0.5 * (y'y - 2y'Cx + x'C'Cx) 

=\\& 0.5 * y'y + 0.5 * x'Fx + c'x\\\\& F = C'C\\& c' = -y'C\\& const = \frac{1}{2} y'y\\
\end{array}
</math>

==Calling  Syntax==

<source lang="matlab">
qpProb = lls2qp(Prob, IntVars)
</source>

==Inputs==

{| class="wikitable"
!Input||Description
|-
|''Prob.LS.C''||The linear matrix in 0''.''5 ''* <nowiki>||y - Cx||</nowiki>''.
|-
|''Prob.LS.y''||The constant vector in 0''.''5 ''* <nowiki>||y - Cx||</nowiki>''.
|}

==Outputs==

{| class="wikitable"
!Output||Description
|-
|''qpProb''||The converted problem.
|}

==Description==

If the problem is a linear least squares problem a qp problem is created. The new problem may have integer variables. Create the problem with llsAssign then use this routine.

If the problem has nonlinear constraints an nlp is created. The new problem may have integer variables. Create the problem with conAssign or minlpAssign, the set the fields Prob.LS.C and Prob.LS.y