Mon Apr 1 23:53:39 EDT 2013

Root locus

Just a random thought: computing root locus paths directly is
difficult since it requires the factorization of a polynomial at each
point.  However, when a single root is known, working the other way
around might be a lot more trivial: compute how coefficients move when
we nudge the poles, then integrate this according to the desired
direction of the coefficients.

This could probably be used in some kind of parameter optimization
procedure!  Compute function (functional) of pole path, then solve or

[1] http://en.wikipedia.org/wiki/Inverse_functions_and_differentiation