93
Table 3.8: Accuracy of coarsened full systems (top section) and reduced systems (bottom
section) for the projection interneuron AR-l-20-04-A with the MIG channel model. Simu-
lations were conducted using 2000 random current injections of duration and magnitude as
in §3.4.1, and kv = kf for the reduced systems.
h (μm) |
Speed-up |
% Matched |
% Mismatched |
Γ |
5 |
4.6× |
74.1 |
49.2 |
0.595 |
2 |
2.0× |
77.9 |
42.4 |
0.656 |
ʌ'l) | ||||
60 |
12.3× |
87.8 |
22.3 |
0.821 |
75 |
8.1× |
88.6 |
21.6 |
0.829 |
90 |
5.9× |
82.4 |
12.2 |
0.849 |
105 |
4.3× |
93.9 |
10.2 |
0.917 |
Table 3.9: Specifications and performance of reduced systems (kυ = kf) for the morpholo-
gies shown in Figure 3.10 using the MIG channel model. Cell mp_tb_40984_gcl is marked
with a ‘*’ because the reduced system became unstable before kv was large enough to cap-
ture any significant dynamics, thus this was the only cell for which the reduced system
failed for the MIG channel model.
Cell |
Model |
B |
N |
N∕kv |
Speed-up |
Γ | |
AR-1-20-04-A |
MIG |
35 |
2233 |
120 |
18.6× |
3.3× |
0.933 |
951005a |
MIG |
44 |
1106 |
75 |
14.7× |
4.3× |
0.893 |
12299402 |
MIG |
61 |
5021 |
150 |
33.4× |
4.6× |
0.882 |
100103a |
MIG |
32 |
2707 |
120 |
22.5× |
4.0× |
0.886 |
*mp_tb_40984_gc 1 |
MIG |
54 |
2541 |
60 |
42.4× |
14.7× |
0.471 |
512882 |
MIG |
35 |
4655 |
105 |
44.3× |
9.2× |
0.928 |
P8-DEV66 |
MIG |
47 |
1712 |
75 |
22.8× |
6.7× |
0.856 |
systems are still successful at capturing the spiking dynamics of most cells. However,
larger reduced systems must be used to resolve these dynamics, so smaller speed-ups
are observed. The MIG channel model also shows us a case of utter failure for the first
time: for the retinal ganglion cell in Figure 3.10e, we find that numerical instabilities,
most likely due to poor snapshots, in the reduced system occur for larger dimensions.
More intriguing information
1. Integration, Regional Specialization and Growth Differentials in EU Acceding Countries: Evidence from Hungary2. Testing the Information Matrix Equality with Robust Estimators
3. EXECUTIVE SUMMARIES
4. The name is absent
5. Outsourcing, Complementary Innovations and Growth
6. Environmental Regulation, Market Power and Price Discrimination in the Agricultural Chemical Industry
7. Surveying the welfare state: challenges, policy development and causes of resilience
8. The name is absent
9. The name is absent
10. Magnetic Resonance Imaging in patients with ICDs and Pacemakers