P-complete

Greedy Alphabet Reducing (GAR)

Category: Miscellaneous

Input

A finite alphabet $\Sigma = \{a_1, \ldots, a_n\}$ with $n \geq 2$, two sets $P, N \subseteq \Sigma^\ast$ with $P \cap N = \varnothing$, and a positive integer $k$ less than or equal to $n$.

Problem

Is the size of the alphabet $\Gamma$ obtained by the greedy alphabet reducing algorithm less than $k$?

Definitions

The greedy alphabet reducing algorithm is given below:

\[ \begin{aligned} &\text{begin} \\ &\quad \Gamma \leftarrow \Sigma \\ &\quad \psi(a) \leftarrow a \text{ for each } a \in \Sigma \\ &\quad \text{repeat} \\ &\qquad \text{for each } (a, b) \in \Gamma \times \Gamma \text{ with } a \neq b \\ &\qquad\qquad \text{let } \phi_{a\to b} \text{ be a mapping obtained by letting } \psi(\sigma) = b, \\ &\qquad\qquad\qquad \text{for all } \sigma \in \Sigma \text{ with } \psi(\sigma) = a \\ &\qquad\qquad \text{choose } \phi_{a\to b} \text{ satisfying } \tilde{\phi}_{a\to b}(P) \cap \tilde{\phi}_{a\to b}(N) = \varnothing \\ &\qquad\qquad\qquad \text{and minimizing } \left|\tilde{\phi}_{a\to b}(P) \cup \tilde{\phi}_{a\to b}(N)\right| \\ &\qquad\qquad \text{(Note: } \tilde{\phi}_{a\to b} \text{ denotes the natural homomorphism associated with } \phi_{a\to b}\text{.)} \\ &\qquad\qquad \text{remove } a \text{ from } \Gamma \\ &\quad \text{until no more replacements are found} \\ &\text{end.} \end{aligned} \]

Status

$\P$-complete (Shimozono and Miyano [1]).

Remarks

The reduction to show hardness is from NANDCVP. Shimozono and Miyano also show that the general Alphabet Indexing Problem is $\NP$-complete, and the Alphabet Indexing Local Search Problem in the weighted case is $\mathsf{PLS}$-complete and in the unweighted case is $\P$-complete [1].

References

  1. S. Shimozono and S. Miyano (1994). Finding alphabet indexing for decision trees over regular patterns: an approach to bioinformatical knowledge acquisition.
  2. Greenlaw, R., Hoover, H. J., and Ruzzo, W. L. (1995). Limits to Parallel Computation: P-Completeness Theory. Oxford University Press. Problem A.12.7.