Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Indistinguishable families of graphs

0 views
Skip to first unread message

Artyom

unread,
Nov 25, 2006, 11:10:14 AM11/25/06
to
Is there some natural not too trivial graph property such that all its
members are look-alikes. More precisely:

Can we find set of graphs S such that for any k there is N satisfying:
any two graphs of order larger than N satisfy the same first-order
formulas with at most k quantifiers, or equivalently are
undistinguashable by a k-round Ehrefeucht-Fraisse game?

Butch Malahide

unread,
Nov 25, 2006, 3:11:17 PM11/25/06
to

I'm not a logician, so I'm just *guessing* that these sets satisfy your
indistinguishability condition. Do they? Are any of them nontrivial?

1. The empty graphs (no edges).

2. The complete graphs K_n.

3. The complete bipartite graphs K_{m,n}.

3. The paths P_n.

4. The cycles C_n.

5. The grid graphs K_m times K_n.

6. The line graphs of complete graphs, L(K_n).

Artyom

unread,
Nov 25, 2006, 3:55:08 PM11/25/06
to
Thaks for reply, Butch!
All your guesses are correct, of course. The main point I want is to
fuse this certain similarity of graphs with some genericity built into
their family. Genericity may be measured like information content of
this set, maybe completeness/hardess for some not too week complexity
class, LOGSPACE at least. Thus listed examples are way too simple. One
example I have in mind are Paley graphs (by the way, do you know what
is the complexity of recognising whether graph is Paley or not?). Or I
can construct such a family straightforward, but more naturality
desired.

On Nov 25, 11:11 pm, "Butch Malahide" <fred.gal...@gmail.com> wrote:
> Artyom wrote:
> > Is there some natural not too trivial graph property such that all its
> > members are look-alikes. More precisely:
>
> > Can we find set of graphs S such that for any k there is N satisfying:
> > any two graphs of order larger than N satisfy the same first-order
> > formulas with at most k quantifiers, or equivalently are

> > undistinguashable by a k-round Ehrefeucht-Fraisse game?I'm not a logician, so I'm just *guessing* that these sets satisfy your

bill

unread,
Nov 25, 2006, 5:15:37 PM11/25/06
to

What is an Ehre(n)feucht-Fraisse game?

regards,

Bill J.

Butch Malahide

unread,
Nov 25, 2006, 5:17:14 PM11/25/06
to
Butch Malahide wrote:

> 5. The grid graphs K_m times K_n.

Oops, those aren't grid graphs, I meant P_m times P_n.

0 new messages