Scenery reconstruction for random walk on random scenery systems (Q2142707): Difference between revisions

\textit{S. A. Kalikow}'s famous paper in ergodic theory [Ann. Math. (2) 115, 393--409 (1982; Zbl 0523.28018)] has a huge influence on this work on scenery reconstruction. The scenery reconstruction problem is all about what can be said if one has partial information. Unlike in other papers, the author here deals with the task of handling the reconstruction problem where the record does not omit the walk channel. That is, errors are allowed to creep into the record. So when the size of an alphabet grows to infinity, the additional information that one gets on the scenery from the steps becomes less important. The main theorem in this paper is about this fact and the author remarks that an extension of this particular result may be an eye-opener to stimulate readers to look for supporting results from ergodic theory. A real novelty of this paper lies in the fact that it deals with the harder problem of allowing an adversary to cunningly introduce errors in the worst possible way, only restricting the number of errors as a proportion of the record. In that case, how to reconstruct some scenery, even if the errors are placed in the most irksome and confusing places? This type of attempt was not considered before explicitly. Overall, the results obtained are highly complex and require an advanced level of comprehension of the subject in order to appreciate the work.