These routines were written in March 2003.
They seem to work, but have not been tested extensively.
There has been no serious attempt to optimize them; the emphasis has
been on simplicity and matlab elegance.

There are minor philosophical inconsistencies regarding the state
names:
do the need to be consecutive integers 1,2,... or can they be
arbitrary ints?

The latest approach is to require them to be consec; there remain some
vestiges of the other attempt tho.

the definitions for a DFST are found in
dfsa2dfst_r1(DFSA,c1,c2)

update: 08/04/2003

the union, intersect, regexp2dfsa routines have been used extensively
and seem to work well

update: 02/08/2003
the regexp routines have been generalized and sped up, but NOTE that in
no case is order of operations observed; this should be forced with
parentheses

also, several forays into algebra have been made, cf. dfsa2monoid

05/08/2003
fixed an insidious bug in accept_str
now seems to work fine though still not 100% certain (no time/desire to
rigorously check)