English

Generalized Hex and logical characterizations of polynomial space

Logic 2008-02-03 v1

Abstract

We answer a question posed by Makowsky and Pnueli and show that the logic (±\mboxHEX)[\mboxFOs](\pm\mbox{HEX})^\ast[\mbox{FO}_s], where HEX is the operator (i.e., uniform sequence of Lindstr\"om quantifiers) corresponding to the well-known {\bf PSPACE}-complete decision problem Generalized Hex, collapses to the fragment \mboxHEX1[\mboxFOs]\mbox{HEX}^1[\mbox{FO}_s] and, moreover, that this logic has a particular normal form which results in the problem HEX being complete for {\bf PSPACE} via quantifier-free projections with successor (HEX is the first ``natural'' problem to be shown to have this property). Our proof of this normal form result is remarkably similar to Immerman's original proof that transitive closure logic, (±\mboxTC)[\mboxFOs](\pm\mbox{TC})^\ast[\mbox{FO}_s], has such a normal form; which is surprising given that (±\mboxHEX)[\mboxFOs](\pm\mbox{HEX})^\ast[\mbox{FO}_s] captures {\bf PSPACE} and (±\mboxTC)[\mboxFOs](\pm\mbox{TC})^\ast[\mbox{FO}_s] captures {\bf NL}. We also show that (±\mboxHEX)[\mboxFO](\pm\mbox{HEX})^\ast[\mbox{FO}] does not capture {\bf PSPACE} and that this logic does not have a corresponding normal form.

Cite

@article{arxiv.math/9612228,
  title  = {Generalized Hex and logical characterizations of polynomial space},
  author = {Argimiro A. Arratia-Quesada and Iain A. Stewart},
  journal= {arXiv preprint arXiv:math/9612228},
  year   = {2008}
}
R2 v1 2026-07-22T17:56:34.609Z