These fine trains compute the constant e by long
division, using the Taylor expansion factored as (...(((1 / n + 1)
/ (n - 1) + 1) / (n - 2) + 1)...)/2 + 2.
This simulation demonstrates that the algorithm is indeed
parallelizable using a locking protocol which keeps the
trains from crashing. This has been used to
compute e to ten million digits in one-sixth the time
of the single-threaded version, using spinlocks since the
threads naturally keep some distance.
* Note that log10(N!) must be greater than
the number of segments times the number of stations for the
computation to converge.