I am now struggling with the following problem:
\begin{alignat}{3} \max_{\mathbf x} & \quad& \mathbf c^T \mathbf x && \\ \text{subject to} &\quad& ||\mathbf x||_2 &=& 1 \\ &\quad& \mathbf{Ax} &\leq& \mathbf b \end{alignat}
If I replace the norm constraint by $||\mathbf x||_2 \leq 1$, then everything is easy as I only need to maximise a linear function subject to convex constraints. Many algorithms could be used to solve it. Yet, things become highly nonconvex after the inequality is replaced by equality.
I was wondering if there are any tools to solve it. Any direct calculations or iterative algorithms are welcome.
Thanks.