experience with a statistical Expander

[Peter Dobcsanyi]

On Wed, Sep 24, 2003 at 03:45:33PM +0100, Leonard Soicher wrote:
> On Tue, Sep 23, 2003 at 07:22:48PM +0100, Peter Dobcsanyi wrote:
> [...]       
> > #
> > In the Fano plane example, pairwise variances are given by a quasi
> > function:
> > 
> >     <function_on_ksubsets_of_indices domain_base="points"
> >      image_cardinality="1" k="2" n="7" ordered="true"
> >      precision="9">
> >      ...
> > 
> > The "ordered" attribute must be false in this case, since order of quasi
> > functions are not defined.
> > 
> 
> But this is a genuine function (no preimage is "collapsed" to 
> preimage_cardinality or blank) and so according to the ext-rep,
> ordered must be true.

Oops...

I was confusing collapsing into <entire_domain/>, which I also
suggested, with the other kind of collapses.

--             ,
    Peter Dobcsanyi