Skip to content
Projects

tree-d-3d

Tree Dynamics in 3D. Barnes-Hut trees and fast multipole methods for charged particles, on NumPy, CuPy or Apple GPUs.

tree-d-3d stands for Tree Dynamics in 3D. For now the particles move in a plane, with the 3D Coulomb field of point charges or the 2D field of line charges. Full 3D, on an octree, is planned.

Flat quadtree

Point charges (3D Coulomb) or line charges (2D Poisson) at any targets, by Barnes-Hut tree walk or direct sum.

Fast multipole method

O(N) fields of line charges with a dual-tree traversal and complex multipole expansions.

Simulation

Second-order time integration with a uniform magnetic field, absorbing walls and objects, and cached fields.

Electrodes

Conductors at prescribed potentials by the panel method, built from any 2D simgeometry shape.

Kernels

Every kernel in Pyccel, NumPy, CUDA and Metal, with the same arguments.