A funnction that is a linear combination of Convex functions where the weights can be negative. Equivalently, it is a linear combination of convex and concave functions.

If it is a function to be optimized, there is an efficient method, where we approximate the concave functions in the linear combination by a hyperplane (linear approx, using Taylor expansion). We then have a convex function overall, which we can optimize using Convex optimization methods. We then re-approximate the concave functions linearly in the newly found point, and repeat!