2

I understand that a convex function can not be maximized as there is no such value. However, consider the following function:

$$\begin{array}{ll} \text{maximize} & 3x^2 + 5y^2\\ \text{subject to} & x+y=12\\ & x,y\geq0\end{array}$$

But executing it in CVX and Matlab I get the following error: 

Disciplined convex programming error:   Cannot maximize a(n) convex expression.

But as I have specified the boundaries, should I not get some maximized value in this range?

Rodrigo de Azevedo
  • 23,223
QuestionEverything
  • 582
  • 2
    CVX only solves convex optimization problems, and maximizing a convex function is not a convex optimization problem. In a convex optimization problem you minimize a convex function over a convex set. – littleO Jul 19 '16 at 12:41
  • 1
    http://ask.cvxr.com/t/why-isnt-cvx-accepting-my-model-read-this-first/570 – Michael Grant Jul 19 '16 at 14:44
  • 1
    @Rodrigo: Do we really need a [cvx] tag? Is it really so hard to search for "cvx" in the search line and find more or less everything related to this topic on the site? – Asaf Karagila Jul 19 '16 at 22:04
  • @AsafKaragila Searching for "cvx" returns posts that link to Boyd's book ( https://web.stanford.edu/~boyd/cvxbook/bv_cvxbook.pdf ), too, which is annoying. – Rodrigo de Azevedo Jul 19 '16 at 22:08
  • @Rodrigo: I don't know if that's enough of a reason to add a tag. There are only 55 questions (and 132 posts in general) involving "cvx" at the moment, going through them if one is so inclined shouldn't take long anyway. Do you really expect this to grow so significantly out of control that the fact a link to a book will become an actual obstacle in finding the relevant posts? Not to mention other keywords like "matlab" or "programming" are likely to help in the search regardless. – Asaf Karagila Jul 19 '16 at 22:13
  • @AsafKaragila Decide what you think is best for MSE. I haven't been around long enough to claim to know what benefits the community. – Rodrigo de Azevedo Jul 19 '16 at 22:37
  • 1
    I for one don't think there should be a CVX tag, in part because I don't think this forum should really attempt to accumulate CVX software usage questions. Obviously if a particular mathematical problem someone is working on is readily solved with CVX, it makes sense to say so. But if you look at the CVX forum it's riddled with questions that I don't think would be appropriate on Math.SE, many of which are answered by the FAQ itself (posted above). – Michael Grant Jul 19 '16 at 23:58

2 Answers2

3

CVX can't solve your problem because it's not a convex optimization problem (because you are maximizing, rather than minimizing, a convex function).

littleO
  • 54,048
1

Write $y = 12 - x$. From $x,y \geq 0$, we get $0 \leq x \leq 12$. The objective function becomes

$$3 x^2 + 5 (12 - x)^2 = 720 - 2 x^2$$

Hence, the maximum is $720$, which is attained at $(x,y) = (0,12)$.

Rodrigo de Azevedo
  • 23,223