99
Chapter 4
The Reduced Strong-Weak Model
In the previous two chapters I introduced two different techniques to reduce the
dimension of neuronal models, and in this chapter I will fuse them together in a
model that uses both techniques where they are most advantageous. If the ion chan-
nel models were purely linear (that is, if the dendrites were purely passive), then the
linear reduction techniques would be able to reproduce to multi-digit accuracy the
somatic voltage. As explained in §3.4.3, many real cells have weakly excitable distal
dendrites, and these branches behave like passive cables, whereas the proximal den-
drites are active cables. It makes sense, therefore, to consider a model in which the
weakly excitable dendrites are modeled as a linear system, while the strongly excitable
dendrites are modeled as a nonlinear system, and then to apply the corresponding
reduction techniques to each part.
This reduced strong-weak model, while not a cure-all for neuronal simulations,
opens doors to accurate reduced simulation of certain cells that otherwise would
be out of reach. Since the linear reduction techniques work when there are only a
few observables, we do not aim to accurately reduce cells with non-tufted or weblike