Tue Oct 21 10:46:34 EDT 2014

Circuits: is connection essential?

Trying to get a good intuition about the bridge between circuits
(impedance) and feed-forward difference equations.

Suppose the space->time approximation to linearize the connection
matrix is appropriate, is the "network quality" essential?

Basically, at each time step there is the solution of a system of
linear equations.  Is the "global nature" of this equation essential
to the behavior?

In other words: can such a system be factored to localize (pipeline!)

The idea of "pushing things into the past" is quite central when
thinking about distributed computation.

In practice: the connection is a matrix where coefficients might
depend on past behaviour (making them essentially constants).
What about factoring this matrix?

I.e. for linear systems it's possible to decouple.
For nonlinear ones there might be an essential "global effect".

Nonlinearity globalizes.