complexity of regular expressions.

["Shai Rubin" <shai@xxxxxxxxxxx>]

I wonder if someone has good references for one (or more) of the following
problems.

1. Measuring the "size" of regular languages: by size I mean the number of
words in the language.
For example, take the two languages L1=(a)+  and L2=(aa)+. L2 is a subset of
L1. However,
although both have infinite number of words, L2 has "half"  the words than
L1 has.
I'm looking for a work that measures this kind of complexity.

2. Sampling words from a regular language.  I'm not even sure if I can talk
about "uniform" sampling, but any reference that address such a question
will be helpful.

3. Splitting a regular language into equivalence classes. For example, the
language (a)+ can be split into two classes: (aa)+ and a(aa)*


Thanks,

Shai