There is an inequality
$$\sum^{K}_{j=1} \mathbf h_k^H \mathbf F_j \mathbf h_k+\sigma^2_{a_{k}} \ge \frac{P}{1-\rho_k}.$$
The variable is $\mathbf F_j$, a $4 \times 4$ matrix, and $\rho_k$ is a value which is larger equal than $0$ but less equal than $1$ , that is, $0 \le \rho_k \le 1$.
If $P $ is a positive value,then equations on both sides of inequality is convex,but if $P$ is negative,the right hand side,$\frac{P}{1-\rho_k}$, will become concave.
How do I rewrite the formula to let left side is concave and the right side is convex,that is,
concave $\ge$ convex , when the $P$ is negative ?