1
$\begingroup$

Suppose we want to solve the following optimization problem:

\begin{equation*} \begin{aligned} & \underset{x,y,z}{\text{minimize}} && x(a-y) \\ & \text{subject to} && (1-x)f(z)+xf(y)=b \end{aligned} \end{equation*}

where $0 \leq x \leq 1$, $0 \leq z \leq a \leq y \leq 1$, and $a$ and $b$ are some constants in $[0,1]$. $x$, $y$, and $z$ are variables to be optimized. Moreover, $f(\cdot)$ is defined on $[0,1]$, and is convex-$\cap$ and decreasing on $[0,1]$.

But I am not sure if this problem is solvable by some standard convex optimization problem. Anyone has some idea? Thanks!

$\endgroup$
3
$\begingroup$

Convex optimization problems require affine equality constraints, co convexity of $f$ doesn't really help you here directly. Moreover, you do not have a convex objective function.

For more on relaxing non-affine equality constraints, see Boyd & Vandenberghe Exercise 4.6 (the book is availiable as a free pdf on Stephen Boyd's site).

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.