[EM] Question regarding Copeland variant

[Warren Smith]

Hello Kristofer,

your "iterated Copeland"
method in the limit of a large number N of iterations,
is going to end up caring about the eigenvector of greatest eigenvalue of
the matrix M where
M[ab]=2 if a beats b
M[ab]=1 if a and b tied
M[ab]=0 if b beats a
Let this "frobenius eigenvector" be x.

It will elect as winner, the guy g such that x[g] is maximized.

In a paper of mine I discussed Keener's eigenvector-based voting system
and a related (which seemed a little better) idea based on "sinkhorning."
See paper #88 here
http://www.math.temple.edu/~wds/homepage/works.html

I had not viewed it in terms of this kind of "iteration" (which consists of
just matrix multiplication by M); but it can be viewed that way, which
is another interesting way to look at it.

-- 
Warren D. Smith
http://RangeVoting.org  <-- add your endorsement (by clicking
"endorse" as 1st step)
and
math.temple.edu/~wds/homepage/works.html