Parallel Computation of Vortex Tube Reconnection Using a Graphics Card and the 3D Vortex–in–Cell Method

In this paper we present numerical results for vortex tube reconnection for three different initial configurations: (i) initially straight orthogonally offset tubes, (ii) initially parallel counter-rotating vortices that are sinusoidally perturbed, and (iii) reconnecting vortex rings. Our aim is to detect some universal mechanism for core reconnection. The mixing of the fluid by the reconnection process is demonstrated by tracing passive markers placed initially near the vortex tubes involved in reconnection. For these numerical simulations we use the 3D vortex–in–cell (VIC) method. In order to overcome the problem of the large time required for single- processor calculations, we constructed a numerical code for the multiprocessor environment of a graphics card. We show that the VIC method is very well suited for parallel computation. We carefully tested the method by comparing the numerical results with theoretical results (for motion of a vortex ring and other published results). The computational speed obtained was nearly 50 times greater than for a single processor.