From: Chris Pine <nemo@...>
Date: 2003-04-19T02:07:00+09:00
Subject: Re: Biased weighted random?

----- Original Message -----
From: "Daniel Carrera" <dcarrera@math.umd.edu>

Now they should never repeat.

Does this look right?
----------------------------

Nope.  :)

See my earlier response.  You met the "does not repeat" requirement, but
your distribution probabilities have been changed.

Mauricio summed it up well:  Given a probability vector (sums to 1), we need
a transition matrix (rows sum to 1) with zeros on the diagonal with the
given vector as an eigenvector of eigenvalue 1.  As he notes, it cannot
always be done.

Problems to work on:

- For which probability vectors does the (or a? (it may not be unique))
specified transition matrix exist?
- What's the matrix?

Working on it,

Chris