Two Characterizations of the Context-Sensitive Languages

An n-dimensional bug-automation is generalization of a finite state acceptor to n-dimensions. With each bug B, we associate the language L(B) which is the set of top rows of the n-dimensional rectangular arrays accepted by B. One-dimensional bugs define trivially the regular sets. Twodimensional bugs define precisely the context-sensitive languages, while bugs of dimension 3 or greater define all the recursively enumerable sets. We consider also finite state acceptors with n two-way non-writing input tapes. For each such machine M, let domain (M) be the set of all strings which are the first component of some n-tuple of tapes accepted by M. For any n ≥ l, the domains of n-tape two-way finite state acceptors are precisely the same as the languages definable by n-dimensional bugs, so as a corollary, the domains of two-tape two-way finite state acceptors are precisely the context-sensitive languages.

[1]  Jeffrey D. Ullman,et al.  Formal languages and their relation to automata , 1969, Addison-Wesley series in computer science and information processing.

[2]  Marvin Minsky,et al.  Computation : finite and infinite machines , 2016 .

[3]  John C. Shepherdson,et al.  The Reduction of Two-Way Automata to One-Way Automata , 1959, IBM J. Res. Dev..

[4]  S.-Y. Kuroda,et al.  Classes of Languages and Linear-Bounded Automata , 1964, Inf. Control..

[5]  M. Blum,et al.  Automata on a 2-Dimensional Tape , 1967, SWAT.