Optimization of neural network simulator via parallelisation

SDDEKit is an algorithmic library written in c backing the neural network simulator of the Virtual Brain's project supporting the calculation of stochastic delay differential equation. Currently, the bottleneck resides on accessing history buffer, which is a look up table that traces the states over the time domain, each look up incurs a cache swap due to unstructured data alignment; and for each single time step, there are thousands of loop up taken place. My proposal is to migrate the whole simulation to a parallel programming platform with reorganised data structure, each thread will handle one input data entry and perform simulation.