Table 2.

This table shows that dynamics of the alanine dipeptide system is not fully reducible to a Markov jump process between six hypersurfaces shown on Fig 5. The probability of jumping to the Milestone i + 1 from the Milestone i depends on the Milestone visited before i. Probabilities (from a long MD trajectory) of jumping from i to i + 1 if the Milestone i − 1 (i + 1) was visited before the hypersurface i are listed in the second (fourth) column. The third and fifth columns list these probabilities as measured by DiM calculation by starting 1000 trajectories from each Milestone. Note that in contrast to DiM, the original Milestoning assumes that P(i → i +1|i − 1 → i) = P(i → i +1|i + 1 → i).

i	P(i → i + 1|i − 1 → i)	NMi−1→iMi→i+1/NMi−1→i	P(i → i + 1|i + 1 → i)	NMi+1→iMi→i+1/NMi+1→i
1	3.9	3.6	8.6	8.3
2	82.4	84.8	89.4	92.0
3	84.9	88.1	91.0	88.0
4	39.0	37.5	49.0	50.0
5	39.2	41.4	60.6	50.5
6	26.3	32.0	35.0	34.1