xxii
List of Figures
6.31 Score is payoff returned playing IPD game. . . . . . . . . . . 172
7.1
Industrial applications of FLAME. . . . . . . . . . . . . . . 184
7.2
Comparing real and simulated data of wound healing in 2D.
Adapted from [192]. . . . . . . . . . . . . . . . . . . . . . . . 185
7.3
3D model snapshots of wound healing at different time steps
of the simulation. . . . . . . . . . . . . . . . . . . . . . . . . 186
7.4
Calling Copasi from FLAME C code. . . . . . . . . . . . . . 187
7.5
Movement of proteins within a Drosophila embryo. A structured view. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
7.6
Agent activities during one iteration. . . . . . . . . . . . . . 190
7.7
Bicoid concentration profiles jointly in A-P axis and developmental time, shows a deterministic model output as an average
value of stochastic model (A), to one stochastic simulation (B)
and the results of one agent-based model run (C). . . . . . . 198
7.8
One realization of stochastic simulation using Gillespie Algorithm at different time points: 60 (A), 100 (B), 144 (C)
and 180 (D) min. Blue histograms show number of Bicoid
molecules along anterior and posterior axis in embryo. Red
lines show average amount of molecules from deterministic
reaction diffusion model. Bicoid intensity at 144 min (C) is
the peak stage and will degrade after mRNA regulation. Red
histograms show number of Bicoid molecules along anterior
and posterior axis in embryo resulting from average 20 runs
of the agent-based model simulation. . . . . . . . . . . . . . 199
7.9
The agent-based modeling simulation result with stochastic
model. The circle shows missing data points in agent-based
results using same initial settings in both models. . . . . . . 200
7.10 Zoom in to find shortest possible error between simulated results in agent-based, stochastic and original datasets. . . . . 200
7.11 Using shortest possible error between the simulated results.
201
7.12 Ant simulation. . . . . . . . . . . . . . . . . . . . . . . . . . 204
7.13 Sequential trails for drug therapy. . . . . . . . . . . . . . . . 225
7.14 Testing drug effect on cancer cells. . . . . . . . . . . . . . . 235
7.15 Multiple views of the same cancer model. . . . . . . . . . . . 236
8.1
Testing low and high level functions. . . . . . . . . . . . . . 238
8.2
Screenshot of Weka analyzing cancer output. . . . . . . . . . 239
8.3
Plots of no-hope cells with correct and incorrect codes. . . . 243
8.4
Simulation times of models. . . . . . . . . . . . . . . . . . . 245
9.1
Diagram of ABM-based decision support system. . . . . . . 277
9.2
Patient flow in green zone versus resource usage. . . . . . . 278
9.3
Part of a Concoursia simulation of a London main station. . 280
List of Figures
6.31 Score is payoff returned playing IPD game. . . . . . . . . . . 172
7.1
Industrial applications of FLAME. . . . . . . . . . . . . . . 184
7.2
Comparing real and simulated data of wound healing in 2D.
Adapted from [192]. . . . . . . . . . . . . . . . . . . . . . . . 185
7.3
3D model snapshots of wound healing at different time steps
of the simulation. . . . . . . . . . . . . . . . . . . . . . . . . 186
7.4
Calling Copasi from FLAME C code. . . . . . . . . . . . . . 187
7.5
Movement of proteins within a Drosophila embryo. A structured view. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
7.6
Agent activities during one iteration. . . . . . . . . . . . . . 190
7.7
Bicoid concentration profiles jointly in A-P axis and developmental time, shows a deterministic model output as an average
value of stochastic model (A), to one stochastic simulation (B)
and the results of one agent-based model run (C). . . . . . . 198
7.8
One realization of stochastic simulation using Gillespie Algorithm at different time points: 60 (A), 100 (B), 144 (C)
and 180 (D) min. Blue histograms show number of Bicoid
molecules along anterior and posterior axis in embryo. Red
lines show average amount of molecules from deterministic
reaction diffusion model. Bicoid intensity at 144 min (C) is
the peak stage and will degrade after mRNA regulation. Red
histograms show number of Bicoid molecules along anterior
and posterior axis in embryo resulting from average 20 runs
of the agent-based model simulation. . . . . . . . . . . . . . 199
7.9
The agent-based modeling simulation result with stochastic
model. The circle shows missing data points in agent-based
results using same initial settings in both models. . . . . . . 200
7.10 Zoom in to find shortest possible error between simulated results in agent-based, stochastic and original datasets. . . . . 200
7.11 Using shortest possible error between the simulated results.
201
7.12 Ant simulation. . . . . . . . . . . . . . . . . . . . . . . . . . 204
7.13 Sequential trails for drug therapy. . . . . . . . . . . . . . . . 225
7.14 Testing drug effect on cancer cells. . . . . . . . . . . . . . . 235
7.15 Multiple views of the same cancer model. . . . . . . . . . . . 236
8.1
Testing low and high level functions. . . . . . . . . . . . . . 238
8.2
Screenshot of Weka analyzing cancer output. . . . . . . . . . 239
8.3
Plots of no-hope cells with correct and incorrect codes. . . . 243
8.4
Simulation times of models. . . . . . . . . . . . . . . . . . . 245
9.1
Diagram of ABM-based decision support system. . . . . . . 277
9.2
Patient flow in green zone versus resource usage. . . . . . . 278
9.3
Part of a Concoursia simulation of a London main station. . 280