View Single Post
Old 08-26-2012, 02:21 PM
Andrs Andrs is offline
Join Date: Jul 2012
Posts: 47
Default quadrat prog CVXOPT Python

I have a question about specifying the upper/lower bounds in CVXOPT/Python.
The A and b matrix specify the condition: yTranspose.alpha = 0 and this is clear.
The bounds for "alpha" are defined as: lower bound = 0, upper bound = infinity.
There is a G matrix that defines inequalities (G.alpha<= h).
Could we use the (negative)G matrix defined as an
Identity matrix to define the lower bound. That is, -G<0 (G.alpha<=0)? If G = I than I.alpha = alpha and it would define a bound for alpha.
How about defining the upper bounds as infinite???
Any comments about the use of CVXOPT are really appreciated. How did you reason about specifying the upper/lower bounds....
Reply With Quote