Adaptive fast multipole methods on the GPU

Re-assign community