On two ways to use determinantal point processes for Monte Carlo integration

When approximating an integral by a weighted sum of function evaluations, de-terminantal point processes (DPPs) provide a way to enforce repulsion between the evaluation points. This negative dependence is encoded by a kernel.