[EM] Taylor or McLaurin polynomial for the complicated functions would reduce the numerical work.

Michael Ossipoff email9648742 at gmail.com
Tue Jul 24 09:15:48 PDT 2012


I probably won't do test-simulations of Weighted-Ballot-Free, because,
with such a method, unbias ( in the form of equal expected s/q, in
every interval between two successive integers in q's range) is a
certainty, assuming that the probability-density function, F(q), being
used is accurate.

The real numerical work would be in finding the rounding-point, R, for
WBF, in each interval.

But, for the reason given in the first paragraph, there's no need to
do that calculation unless WFB is actually in use.

Well, ok, if WBF is being proposed and considered, then it might be of
interest to tell people what a WFB allocation would look like, in
comparison to those of BF, Webster, Hill, and the fixed-rounding-point
methods based on average s/q in each interval, and also those based on
expected s/q with some nonuniform probability-density function.

If there's interest, and if someone is willing to deal with the
probably-more-realistic log-normal distribution for the
probability-density (or the more complicated one that Kristofer
suggested), and program it, to do WFB apportionment for the 2010
census, then I'd participate in the discussion.

Mike Ossipoff.

More information about the Election-Methods mailing list