The name is absent

91

Table 3.7: Specifications and performance of reduced systems (fc_υ = feʃ) for the morpholo-
gies shown in Figure 3.8 using the HH or HHA channel models. Coincidence factors shown
correspond to using Algorithm 3. Here B = number of branches in each cell, and for each
cell we set h = 1 μm to be the desired compartment size.

3.4.3 Cells With Weakly Excitable Dendrites

Up to this point we have considered ion channel distributions (HH and HHA)
that lead to strongly excitable dendrites. However, real neurons often have weakly
excitable dendrites due in large part to the increase of the density of K⁺ channels
with distance from the soma. In order to assess how well the POD and DEIM capture
the spiking dynamics of these weakly excitable cells, we use the MIG channel model
(see Table B.3) with the previously considered morphologies.

In addition to considering the full model versus the reduced model, we also offer
a comparison of the accuracy of coarsened models. That is, we use larger values of
h for the full system and compare the resulting spike trains to those computed by
the full system using the fine reference value of h = 1 μm. As shown in Table 3.8,
coarsening of cell AR-1-20-04-A yields less accurate results than the reduced systems,
and the reduced systems are also much faster. This presents a strong argument in

<<   <   98   99   100   101   102   103   104   >   >>