Measuring a diffusion coefficient by single-particle tracking: statistical analysis of experimental mean squared displacement curves

We provide experimental results on the accuracy of diffusion coefficients obtained by a mean squared displacement (MSD) analysis of single-particle trajectories. We have recorded very long trajectories comprising more than 1.5 × 105 data points and decomposed these long trajectories into shorter segments providing us with ensembles of trajectories of variable lengths. This enabled a statistical analysis of the resulting MSD curves as a function of the lengths of the segments. We find that the relative error of the diffusion coefficient can be minimized by taking an optimum number of points into account for fitting the MSD curves, and that this optimum does not depend on the segment length. Yet, the magnitude of the relative error for the diffusion coefficient does, and achieving an accuracy in the order of 10% requires the recording of trajectories with about 1000 data points. Finally, we compare our results with theoretical predictions and find very good qualitative and quantitative agreement between experiment and theory.