Status Updates From Nonlinear Time Series Analysis

I while back I arrived at the "advanced topics" and sure enough things got a lot harder. There was one hilarious bit where the author spouts a horrendous mess of incomprehensible point-set topological jargon and then announces that was the version accessible to physicists, avoiding mathematical technicalities...

This is proving to an excellent introduction to a fascinating but notoriously treacherous subject. I wish I had started reading it much earlier.

If I've understood this properly, linear filtering will not remove dynamical noise from a non-linear time series but can remove narrow-spectrum error noise.

Second hand copy: some unmentionable has annotated the text in ink!
Arrrrrrrrrrrghhh!