Abstract
We investigate the efficiency of Adaptive Large Neighborhood Search on the Graphics Processing Unit (GPU). We do this by implementing the algorithm for the Distance-constrained Capacitated Vehicle Routing Problem (DCVRP), which we benchmark towards a state of the art CPU implementation. The computational power of the GPU in ordinary computers has increased significantly in recent years. Therefore it is interesting to utilize this computing power. We perform tests on well-known DCVRP instances.
