The parallel simulation of concentrating of weakly magnetic nanoparticles by high gradient magnetic field using GPUs

The concentrating of weakly magnetic nanoparticles dispersed in a fluid by using high gradient magnetic field is simulated based on a parallel algorithm running on GPUs. The algorithm can support varying thread size per block. In our model, the high gradient magnetic field is produced by placing cylindrical wires traversely to a background uniform magnetic field. The problem of particle concentrating is considered in two dimension according to the feature of high gradient magnetic field around each wire. So, the particles transport is simulated on a two dimensional domain enclosing a representative wire. The transport equation of weakly magnetic nanoparticle of known magnetic property is solved numerically by assigning initial and boundary values. The simulation result shows the feasibility of concentrating weakly magnetic nanoparticles by using high gradient magnetic field. We adapt the work in [1] for this GPU architecture. The shared memory is used for the computation. The performance results show significant speedup over the traditional sequential version and openMP version.