Agents in Biology
197
TABLE 7.1: Comparing building simulations in MATLAB and FLAME.
Objective
Stochastic
Agents
Total simulation time
200 min for one realization. Total time step is
around 3 x 106 (stochastically).
12000 time steps.
Actual simulation time
CPU time: 795.02 seconds.
5 hours.
Memory usage Approx 420 MB .
Approx 30 GB.
Results format
produced
3.1 × 106 by 100 matrix in
MATLAB.
120,000 xml files which
are later parsed to produce excel sheets to plot
graphs.
Model writing
time
1 week. Understanding
Gillespie Algorithm and
implementation in MATLAB.
1 month, involves understanding the model description and converting
to what happens in one iteration.
Global values
which can easily be changed
All decay, production and
diffusion rates highlighted
in starting conditions.
All decay, production and
diffusion rates highlighted
in starting conditions.
Simulation
tool used
MATLAB.
FLAME serial version run
on a MAC laptop.
Results measured
The results have measured
every minute according to
all the compartments as
shown in Figure 7.7. In
Figure 7.8, the protein distributed at t = 60 minutes (9.3 × 10
5 iteration),
t = 100 min (1.6 ×
10
6 iteration), t = 144
min (2.2 × 10
6 iteration),
t = 180 min (2.8 ×
10
6 iteration), t = 200
min (3.1 × 10
6 iteration).
As number of proteins per
time step across compartments, and protein distributions at times 60 min
(3600 iteration step (it)),
100 min (6000 it), 144 min
(8640 it), 180 min (10800
it), 200 min (12000 it).
Average over
runs
One realization was taken.
The averaged stochastic
model is shown by PDE in
[120].
Model was run 20 times
and the average was
taken.
Due to the large number of cases, a minimum square distance was used
to calculate the error rate between the results of each of the cases with the
stochastic results, shown in Figure 7.10. The best case which was able to
197
TABLE 7.1: Comparing building simulations in MATLAB and FLAME.
Objective
Stochastic
Agents
Total simulation time
200 min for one realization. Total time step is
around 3 x 106 (stochastically).
12000 time steps.
Actual simulation time
CPU time: 795.02 seconds.
5 hours.
Memory usage Approx 420 MB .
Approx 30 GB.
Results format
produced
3.1 × 106 by 100 matrix in
MATLAB.
120,000 xml files which
are later parsed to produce excel sheets to plot
graphs.
Model writing
time
1 week. Understanding
Gillespie Algorithm and
implementation in MATLAB.
1 month, involves understanding the model description and converting
to what happens in one iteration.
Global values
which can easily be changed
All decay, production and
diffusion rates highlighted
in starting conditions.
All decay, production and
diffusion rates highlighted
in starting conditions.
Simulation
tool used
MATLAB.
FLAME serial version run
on a MAC laptop.
Results measured
The results have measured
every minute according to
all the compartments as
shown in Figure 7.7. In
Figure 7.8, the protein distributed at t = 60 minutes (9.3 × 10
5 iteration),
t = 100 min (1.6 ×
10
6 iteration), t = 144
min (2.2 × 10
6 iteration),
t = 180 min (2.8 ×
10
6 iteration), t = 200
min (3.1 × 10
6 iteration).
As number of proteins per
time step across compartments, and protein distributions at times 60 min
(3600 iteration step (it)),
100 min (6000 it), 144 min
(8640 it), 180 min (10800
it), 200 min (12000 it).
Average over
runs
One realization was taken.
The averaged stochastic
model is shown by PDE in
[120].
Model was run 20 times
and the average was
taken.
Due to the large number of cases, a minimum square distance was used
to calculate the error rate between the results of each of the cases with the
stochastic results, shown in Figure 7.10. The best case which was able to