# Re: Distributivity in Tropashko's Lattice Algebra

Date: 15 Aug 2005 12:15:28 -0700

Message-ID: <1124133328.620335.223260_at_f14g2000cwb.googlegroups.com>

vc wrote:

*> Marshall Spight wrote:
*

> > A month or so ago we were discussing relational algebras, and

*> > we were looking at a lattice algebra defined in a paper by
**> > Vadim Tropashko. It had two operators, natural join and inner
**> > union.
**> >
**> > One point that was made at the time was that these two operators
**> > are not distributive. (Although they are commutative, associative,
**> > idempotent, and absorbtive.)
**>
**> If you're taking about http://arxiv.org/ftp/cs/papers/0501/0501053.pdf,
*

Yes.

> then I am not sure whether the new algebra (NA) itself is very

*> interesting as applied to the RM. Clearly, the NA implies some RM
**> extension due to the fact that the NA is defined (sort of) over an
**> infinitely countable set of relations and some relations are infinitely
**> countable themselves.
**>
**> The reason for this kind of infinity is the selection and rename
**> definitions both of which rely on joining with infinite relations, the
**> number of potentially required relations being infinite itself.
*

The reason this doesn't bother me is that I believe the same functionality can be had from functions. Functions can model some infinite relations quite well. For example, the paper mentions the infinite < relation, but the < function would work just as well.

Marshal Received on Mon Aug 15 2005 - 21:15:28 CEST