"An n! lower bound on formula size"

We introduce a new Ehrenfeucht-Fraisse game for proving lower bounds on the size of first-order formulas. Up until now such games have only been used to prove bounds on the operator depth of formulas, not their size. We use this game to prove that the CTL/sup +/ formula Occur/sub n//spl equiv/E[Fp/s...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Adler, M., Immerman, N.
Format: Tagungsbericht
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:We introduce a new Ehrenfeucht-Fraisse game for proving lower bounds on the size of first-order formulas. Up until now such games have only been used to prove bounds on the operator depth of formulas, not their size. We use this game to prove that the CTL/sup +/ formula Occur/sub n//spl equiv/E[Fp/sub 1//spl and/Fp/sub 2//spl and//spl middot//spl middot//spl middot//spl and/F/sub n/] which says that there is a path along which the predicates p/sub 1/ through p/sub n/ occur in some order; requires size n! to express in CTL. Our lower bound is optimal. It follows that the succinctness of CTL+ with respect to CTL is exactly /spl Theta/(n). Wilke (1999) had shown that the succinctness was at least exponential. We also use our games to prove all optimal /spl Theta/(n) lower bound on the number of boolean variables needed for a weak reachability logic (/spl Rscr//spl Lscr//sup w/) to polynomially embed the language LTL. The number of booleans needed for full reachability logic RC and the transitive closure logic FO/sup 2/(TC) remain open (Immerman and Vardi, 1997; Alechina and Immerman, 2000).
ISSN:1043-6871
2575-5528
DOI:10.1109/LICS.2001.932497