Re: [Axiom-developer] LANL NetworkX and algebra dependency graph

[Gabriel Dos Reis]

"Page, Bill" <address@hidden> writes:

| Axiom Developers;
| 
| Over a year ago we were discussing how best to represent Axiom's
| algebra dependency graph. This is a complex problem because Axiom
| has over 1,300 mathematically modules that are tightly integrated
| with each other - even to the point of generating a large number
| cyclical dependency chains in the underlying code. This considerably
| complicates bootstrap process by which Axiom is built and could
| even call into question the fundamental means by which these
| dependency loops are solved in the current build (e.g. It is known
| that just a single iteration of bootstrap -> compile -> bootstrap
| is not sufficient to produce a fixed point solution.)

I believe extending SPAD to understand "extend" (no pun intended)
might help decrease the complexity of the dependency graph to a more
manageable shape.

[...]

| If anyone else is interested in discussing the development and
| use of a graph theory package for Axiom, I would be very glad
| to hear from you.

I'm interested in a graph theory package for Axiom (but I have no
immediate interest in graphics).  I got a copy of "Computational
Discrete Mathematics", by Sriram Pemmaraju and Steven Skiena, to get an
idea of existing work in the area of symbolic computations.  I'm not
done with it yet.

-- Gaby