Sebastian Jakobi

FROM EQUIVALENCE TO ALMOST-EQUIVALENCE, AND BEYOND: MINIMIZING AUTOMATA WITH ERRORS

We introduce E-equivalence, which is a straightforward generalization of almost-equivalence. While almost-equivalence asks for ordinary equivalence up to a finite number of exceptions, in E-equival...

THE MAGIC NUMBER PROBLEM FOR SUBREGULAR LANGUAGE FAMILIES

We investigate the magic number problem, that is, the question whether there exists a minimal n-state nondeterministic finite automaton (NFA) whose equivalent minimal deterministic finite automaton (DFA) has α states, for all n and α satisfying n ≤ α ≤ 2n. A number α not satisfying this condition is called a magic number (for n). It was shown that no magic numbers exist for general regular languages, whereas trivial and non-trivial magic numbers for unary regular languages were identified. We obtain similar results for automata accepting subregular languages like, for example, star-free languages, prefix-, suffix-, and infix-closed languages, and prefix-, suffix-, and infix-free languages, showing that there are only trivial magic numbers, when they exist. For finite languages we obtain some partial results showing that certain numbers are non-magic.

Minimal Reversible Deterministic Finite Automata

We study reversible deterministic finite automata (REV-DFAs), that are partial deterministic finite automata whose transition function induces an injective mapping on the state set for every letter of the input alphabet. We give a structural characterization of regular languages that can be accepted by REV-DFAs. This characterization is based on the absence of a forbidden pattern in the (minimal) deterministic state graph. Again with a forbidden pattern approach, we also show that the minimality of REV-DFAs among all equivalent REV-DFAs can be decided. Both forbidden pattern characterizations give rise to NL-complete decision algorithms. In fact, our techniques allow us to construct the minimal REV-DFA for a given minimal DFA. These considerations lead to asymptotic upper and lower bounds on the conversion from DFAs to REV-DFAs. Thus, almost all problems that concern uniqueness and the size of minimal REV-DFAs are solved.

Chop operations and expressions: descriptional complexity considerations

The chop or fusion operation was recently introduced in [S. A. BABU, P. K. PANDYA: Chop Expressions and Discrete Duration Calculus. Modern Applications of Automata Theory, World Scientific, 2010], where a characterization of regular languages in terms of chop expressions was shown. Simply speaking, the chop or fusion of two words is a concatenation were the touching letters are coalesced, if both letters are equal; otherwise the operation is undefined. We investigate the descriptional complexity of the chop operation and its iteration for deterministic and nondeterministic finite automata as well as for regular expressions. In most cases tight bounds are shown. Moreover, we also consider the conversion problem between finite automata, regular expressions, and chop expressions. Again, for most conversions we get tight bounds in order of magnitude. It is worth mentioning that regular expressions can be transformed into equivalent chop expressions of polynomial size, but chop expressions can be exponentially more succinct than regular expressions.

From equivalence to almost-equivalence, and beyond--minimizing automata with errors

We introduce E-equivalence, which is a straightforward generalization of almost-equivalence. While almost-equivalence asks for ordinary equivalence up to a finite number of exceptions, in E-equivalence these exceptions or errors must belong to a (regular) set E. The computational complexity of minimization problems and their variants w.r.t. almost- and E-equivalence are studied. Roughly speaking, whenever nondeterministic finite automata (NFAs) are involved, most minimization problems, and their equivalence problems they are based on, become PSPACE-complete, while for deterministic finite automata (DFAs) the situation is more subtle. For instance, hyper-minimizing DFAs is NL-complete, but E-minimizing DFA s is NP-complete, even for finite E. The obtained results nicely fit to the known ones on ordinary minimization for finite automata. Moreover, since hyper-minimal and E-minimal automata are not necessarily unique (up to isomorphism as for minimal DFAs), we consider the problem of counting the number of these minimal automata. It turns out that counting hyper-minimal DFAs can be done in FP, while counting E-minimal DFA s is #P-hard, and belongs to the counting class #·coNP.

On the complexity of rolling block and alice mazes

We investigate the computational complexity of two maze problems, namely rolling block and Alice mazes. Simply speaking, in the former game one has to roll blocks through a maze, ending in a particular game situation, and in the latter one, one has to move tokens of variable speed through a maze following some prescribed directions. It turns out that when the number of blocks or the number of tokens is not restricted (unbounded), then the problem of solving such a maze becomes PSPACE-complete. Hardness is shown via a reduction from the nondeterministic constraint logic (NCL) of [E. D. Demaine, R. A. Hearn: A uniform framework or modeling computations as games. Proc. CCC, 2008] to the problems in question. By using only blocks of size 2×1×1, and no forbidden squares, we improve a previous result of [K. Buchin, M. Buchin: Rolling block mazes are PSPACE-complete. J. Inform. Proc., 2012] on rolling block mazes to best possible. Moreover, we also consider bounded variants of these maze games, i.e., when the number of blocks or tokens is bounded by a constant, and prove close relations to variants of graph reachability problems.

The Magic Number Problem for Subregular Language Families

We investigate the magic number problem, that is, the question whether there exists a minimal n-state nondeterministic finite automaton (NFA) whose equivalent minimal deterministic finite automaton (DFA) has alpha states, for all n and alpha satisfying n less or equal to alpha less or equal to exp(2,n). A number alpha not satisfying this condition is called a magic number (for n). It was shown in [11] that no magic numbers exist for general regular languages, while in [5] trivial and non-trivial magic numbers for unary regular languages were identified. We obtain similar results for automata accepting subregular languages like, for example, combinational languages, star-free, prefix-, suffix-, and infix-closed languages, and prefix-, suffix-, and infix-free languages, showing that there are only trivial magic numbers, when they exist. For finite languages we obtain some partial results showing that certain numbers are non-magic.

The chop of languages

Abstract We investigate chop operations, which can be seen as generalized concatenation. For several language families of the Chomsky hierarchy we prove (non)closure properties under chop operations and incomparability to the family of languages that are the chop of two regular languages. We also prove non-closure of that language family under Boolean operations and closure under reversal. Further, the representation of a regular language as the chop of two regular expressions can be exponentially more succinct than its regular expression. By considering the chop of two linear context-free languages we already obtain language families that have non-semi-decidable problems such as emptiness or finiteness.

Nondeterministic Biautomata and Their Descriptional Complexity

We investigate the descriptional complexity of nondeterministic biautomata, which are a generalization of biautomata [O. Klima, L. Polak: On biautomata. RAIRO—Theor. Inf. Appl., 46(4), 2012]. Simply speaking, biautomata are finite automata reading the input from both sides; although the head movement is nondeterministic, additional requirements enforce biautomata to work deterministically. First we study the size blow-up when determinizing nondeterministic biautomata. Further, we give tight bounds on the number of states for nondeterministic biautomata accepting regular languages relative to the size of ordinary finite automata, regular expressions, and syntactic monoids. It turns out that as in the case of ordinary finite automata nondeterministic biautomata are superior to biautomata with respect to their relative succinctness in representing regular languages.

Queue Automata of Constant Length

We introduce and study the notion of constant length queue automata, as a formalism for representing regular languages. We show that their descriptional power outperforms that of traditional finite state automata, of constant height pushdown automata, and of straight line programs for regular expressions, by providing optimal exponential and double-exponential size gaps. Moreover, we prove that constant height pushdown automata can be simulated by constant length queue automata paying only by a linear size increase, and that removing nondeterminism in constant length queue automata requires an optimal exponential size blow-up, against the optimal double-exponential cost for determinizing constant height pushdown automata.