home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!mcsun!uknet!stl!sbil!wet!spitws75!tdk
- From: tdk@spitws75.sbil.co.uk (Tim Kempster)
- Newsgroups: comp.theory
- Subject: Constructive proof of directed graph unconectivity
- Keywords: CON NLOG NLOG-Complete
- Message-ID: <1992Sep3.161109.22586@sbil.co.uk>
- Date: 3 Sep 92 16:11:09 GMT
- Sender: news@sbil.co.uk
- Reply-To: tdk@spitws75.sbil.co.uk
- Organization: Salomon Brothers International Limited
- Lines: 28
-
- I am trying to find an alogrithm with NLOG complexity for the
- problem defined below.
-
- UNCON
- INSTANCE: A directed graph G=(V,A), two vertices u,v \in V
- QUESTION: Is there no path from u to v in G ?
-
-
- This is the complement of the problem CON which has the same instance
- and asks, "Is there a path u to v ? ". This problem has an NLOG
- solution and is infact NLOG complete w.r.t LOG space reduction.
-
- Immerman and Szelepcsenyi have shown NLOG = CO-NLOG so combining these
- results we see UNCON is indeed in NLOG.
-
- Is there a direct constructive proof of this fact ?
-
- That is is there a Turing Machine algorithm (non-deterministic
- log space) which can resolve this question. I was thinking along the lines
- of transforming an instance of UNCON into CON by taking the complement
- of G or the complement of the transitive closure ? A bit vaugue I know.
-
- The proof that CON is in NLOG gives the algorithm for a TM explicitly.
-
-
- Can anyone help ?
-
- Tim Kempster
-