48
Time to compute reduced model using IRKA
Al40r
B2
Output error (IRKA vs. quasi-active)
120
100
80
co
60
40
20
°0
10 20 30 40 50
к (reduced system size)
-8,
10 20 30 40 50
к (reduced system size)
Figure 2.12: A) Time to compute reduced system, and B) maximum absolute error between
quasi-active and reduced system output using IRKA for cell AR-l-20-04-A (jV = 6726).
in a fraction of the time needed for BT. Notice that although the error in IRKA does
not decrease as consistently as к increases as it did in BT, nor does the error decrease
as rapidly, the accuracy is more than enough for neuroscience applications.
The main difference between IRKA and BT is that IRKA computes a system
of a given size, whereas BT computes full matrices that are then truncated to the
desired size. If a reduced system of a different size is desired, IRKA must recompute
the matrices, whereas BT can just truncate. However, the premium we pay for
immediate BT changes is far too expensive, considering that IRKA’s results are just
as good but are obtained in a fraction of the time. As an example, the reduced
system from IRKA with к = 20 was used to run the same dendritic democratization
parameter sweep as in Figure 2.10. The computed synaptic conductances for the BT
and IRKA systems agree to nearly 5 digits for each compartment, indicating that
IRKA is indeed computing the right reduced system.