“…For example, researchers have shown a toy system where the speedup reaches 1 order of magnitude with a coefficient of variation of 2.9 . It must be noted that the actual speedup depends on the distribution of the unbiased FPT, not only on the coefficient of variation . A method to recover the unbiased MFPT from the restarted one was proposed, but it seems that the theoretical bound for the error remains elusive.…”