Homogeneous threshhold burst - Aggregate Model

from Brown, Cornell, et al. 2006

Author

Haowen He

Published

March 11, 2024

1 Stochastic homogeneous threshold

In this analysis, we build a stochastic aggregate distribution model based on the the single threshold burst function previously reported in \textit{Brown, Cornell et al, 2006}. The dynamics of the planktonic cells vs. aggregates is given by

dM_1 = (N \alpha (N-1) M_{N-1} - \alpha M_1) dt + G_{N-1} dW_{N-1} - G_1 dW_1, dM_n = (\alpha (n-1) M_{N-1} - \alpha n M_n) dt + G_{n-1} dW_{n-1} - G_n dW_n, \text{ for } 1 < n < N,

where N is the maximum aggregate size, \alpha is division rate, i.e., growth rate in this single threshold burst model, dM_1 is the number of new planktonic cells at each time step dt, and dM_n is the number of new aggregates for 1 < n < N. In addition, G_n is the variances of the n-th generation of branching process, and dW_n is Gaussian increment.

1.1 Simulation

Simulation parameters (Initial number of planktonic cells is set to 100)
max aggregate size N step size dt growth rate \alpha
100 t_0 = 1 : 0.01 : 10 = t_f 0 : 0.002 : 0.5

Figure 1 shows the distribution of aggregate size across time.

Code
%}

% max aggregate size
N = 100;      

% simulation time parameters
t0 = 1;
tf = 10;
dt = 0.01;
times = t0:dt:tf;

% division rate
R = 0:0.002:0.5;

% initial conditions
M = zeros(length(times),N-1);
% no aggregates: 100 planktonic cells 
M(1,1) = 100; 
% variance
G = zeros(1,N-1); 

for i = 1:length(R)
    
    % Choose division rate
    a = R(i);

    %% Run simulation
    for t = 1:length(times)-1
        M(M<0) = 0;
        Mt = M(t,:);
        for n = 1:N-2
            G(1,n) = sqrt(a*n*Mt(1,n));
        end
        G(1,N-1) = sqrt(N*a*(N-1)*Mt(1,N-1));
        dW = randn(1,N-1);
        dM1 = (N*a*(N-1)*Mt(1,N-1) - a*Mt(1,1))*dt + G(1,N-1)*dW(1,N-1) - G(1,1)*dW(1,1);
        M(t+1,1) = Mt(1,1) + dM1;
        for n = 2:N-1
            dMn = (a*(n-1)*Mt(1,n-1) - a*n*Mt(1,n))*dt + G(1,n-1)*dW(1,n-1) - G(1,n)*dW(1,n);
            M(t+1,n) = Mt(1,n) + dMn;
        end
    end 

    M(M<0) = 0;
    figure
    hold on
    %bar(M(end,2:end), 'FaceColor', "#74A089")
    %xlabel('Aggregate size')
    %ylabel('Number of aggregates')
    bar(G(end,1:end), 'FaceColor', "#AA9EC9")
    xlabel('Aggregate size')
    ylabel('Variance')
    title(['Division rate = ',num2str(a)])
    saveas(gcf,['Division_rate_',num2str(i), '.png'])
end

Figure 1: Aggregate size vs. Number of aggregates

Figure 2 shows the distribution of variances of the n-th generation of branching process across time.

Code
knitr::include_graphics("ezgif-1-0b5d629daa.gif", error = FALSE)

Figure 2: Aggregate size vs. Variances