AndiK hat geschrieben:Should I analyze the STR file and create an algorithm that guesses from speed and acceleration which way the train took?
A Zusi "Strecke" is, mathematically, a directed graph. After parsing the STR file, it should be possible to determine ALL possible paths between the starting-point and the end-point. Of course, these points have to be known!
In most cases, all paths should differ only marginally in length and altitude profile. After all, different platforms of the same station usually lie on the same altitude level.
Should this not be the case for some special track configuration, you could determine the path the train took by computing the length of the path and comparing it to the driven distance. If there are multiple paths of matching length, then you could use the altitude of either of them, or their average. If, for example, a crossover of an S-Bahn-track over another is missing in the resulting track profile -- that's not a major problem.
This sounds complicated, but the traversal algorithm should be relatively compact, below 100 lines of codes in C++, perhaps.
However, there is no mathematical proof of correctness for that algorithm. There might be pathological STR files that result in totally incorrect values as they misguide the algorithm. Also, finding all paths in a directed graph is computationally intensive (as I mentioned, it might take a few seconds).
I agree that this is still a lot more than is required to output only the data collected during the simulation -- I assume that (using the right library
) this is just one function call (or at most a handful of calls). So your statement about that "big bunch of work" is probably right -- but in my opinion such an algorithm would be
feasible. I haven't tried it yet, though.
- Christopher
.
!!!