Estimating Errors in Diffusion Coefficients From Molecular Dynamics Simulations Without Replicates
For each distribution calculated by running independent simulations, the corresponding distribution created using bootstrapping looks very similar. This suggests that, at least for this system, bootstrapping can be a very reliable method of estimating error. However, this simple system is slightly unrealistic, as each particle's trajectory was completely independent of the other particles. Future research will examine how bootstrapping handles a dense, interacting system.