Re: formal (theoretical) treatment of database indices

Personally, I do not know any of such models exist. I think the problem here is what to consider to be a data model.

For conceptual level we have a Codd's definition where a data model should have at least three parts to define model objects, integrity rule, data operations. In addition to that Codd added (in RM/2) security and some other things (can't recall right now).

For internal level (storage)... It makes sense to define objects, data operations. C. Date defines internal level as unbounded linear space, a set of mappings to physical storage, indexes, a method to specify order of records, etc.

"falcon" <shahbazc_at_gmail.com> wrote in message news:1129342077.604687.6740_at_o13g2000cwo.googlegroups.com...
> Exactly! That is what I was trying to say. Now is this mathematical
> abstraction the same as the mathematics of B-tree optimization or
> bitmap compresson (etc.) or is there more general and fundamental work?
>
Received on Thu Oct 20 2005 - 22:58:43 CEST