We discovered a programming error and an error in some of the input files, which led us to recommend a reweighting factor that gives a positive and far from optimal time-step error. The reweighting factor in Eq. (9) of the original paper1 should be replaced by Δw=eTeff(S(R))+S(R))/2 With the corrected program and this choice of reweighting factor, we find time-step errors (shown in Figs. 1.4, the updated arXiv version of the paper,1 and the supplementary material), which are similar to the corresponding figures in the paper. For small time steps, they are nearly identical. The conclusions of the paper are unchanged. We thank Claudia Filippi for the derivations in Eqs. (23).(30). (Figure Presented). © 2021 American Institute of Physics Inc.. All rights reserved.