How to iterate my algorithm?

2 次查看（过去 30 天）

Thomas Veith 2019-6-6

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/465876-how-to-iterate-my-algorithm

评论： Thomas Veith 2019-6-6

So I have the code below, and I think I need to put another for-loop around it, but I'm not exactly sure how to get where I want to go.

In the example below, I solve my ODE for ten 'individuals', sort by MSE, delete the bottom half of results (IE, the ones with greatest MSE), genereate parameters for five 'new individuals', then solve again and sort my results by MSE. I'd like to do this a total of ten times before I finally end with a 10x6 numeric array sorted by MSE.

Any advice?

Thanks in advance.

load('patientdata1.mat')
%%assign number of individuals and setup vector for parameters
n = 10 ; 
parameters = zeros(n,6); 
%%initialize parameters
for k=1:n ; 
    alpha0=1;psa0=29;gamma=rand;psi=rand;beta=rand;mse=0; 
    parameters(k,:) = [alpha0,psa0,gamma,psi,beta,mse]; 
end;
%%solve ODE
for k = 1:size(parameters,1) 
    tspan=linspace(0,7,50);
    ODE = @(t,x,gamma,psi,beta) [-gamma*x(1) + psi*x(1); beta*x(2) - x(1)*x(2)]; 
    [t,x(:,:,k)] = ode45(@(t,x) ODE(t,x,parameters(k,3),parameters(k,4),parameters(k,5)), tspan, [parameters(k,1) parameters(k,2)]); 
end
%%access all x2 columns from x
x2 = x(:,[2],:);
%%grab last entry from each individual in x2
lp=x2(end,1,:);
lastposition=lp(:);
%%calulate mse
msematrix=[lastposition-patientdata1(9,2)].^2;
%%update parameters matrix with mse values and sort by mse
parameters(:,6)=msematrix(:);
parameters=sortrows(parameters,6);
%%delete bottom 5 rows
parameters([6:10],:)=[];
%%initialize 5 new parameter sets
>> parameters([6:10],:)=[alpha0,psa0,rand,rand,rand,mse;alpha0,psa0,rand,rand,rand,mse;alpha0,psa0,rand,rand,rand,mse;alpha0,psa0,rand,rand,rand,mse;alpha0,psa0,rand,rand,rand,mse];
%%solve ODE again with new paremeters matrix
for k = 1:size(parameters,1) 
    tspan=linspace(0,7,50);
    ODE = @(t,x,gamma,psi,beta) [-gamma*x(1) + psi*x(1); beta*x(2) - x(1)*x(2)]; 
    [t,x(:,:,k)] = ode45(@(t,x) ODE(t,x,parameters(k,3),parameters(k,4),parameters(k,5)), tspan, [parameters(k,1) parameters(k,2)]); 
end
%%access all x2 columns from x
x2 = x(:,[2],:);
%%grab last entry from each individual in x2
lp=x2(end,1,:);
lastposition=lp(:);
%%calulate mse
msematrix=[lastposition-patientdata1(9,2)].^2;
%%update parameters matrix with mse values and sort by mse
parameters(:,6)=msematrix(:);
parameters=sortrows(parameters,6);

2 个评论
显示无隐藏无

Geoff Hayes 2019-6-6

在 MATLAB Online 中打开

Thomas - try putting the loop after where you populate/initialize the parameters

load('patientdata1.mat')
%%assign number of individuals and setup vector for parameters
n = 10 ; 
parameters = zeros(n,6); 
%%initialize parameters
for k=1:n ; 
    alpha0=1;psa0=29;gamma=rand;psi=rand;beta=rand;mse=0; 
    parameters(k,:) = [alpha0,psa0,gamma,psi,beta,mse]; 
end;
% iterate n times
for v = 1:n
    %%solve ODE
    for k = 1:size(parameters,1) 
        tspan=linspace(0,7,50);
        ODE = @(t,x,gamma,psi,beta) [-gamma*x(1) + psi*x(1); beta*x(2) - x(1)*x(2)]; 
        [t,x(:,:,k)] = ode45(@(t,x) ODE(t,x,parameters(k,3),parameters(k,4),parameters(k,5)), tspan, [parameters(k,1) parameters(k,2)]); 
    end
    % etc.
end