Produce matrices through for loop

2 次查看（过去 30 天）

Ali Esmaeilpour 2019-5-27

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/464248-produce-matrices-through-for-loop

评论： Ali Esmaeilpour 2019-5-31

Hello people! I want to produce N number of matrices while k=1:N and I want the result in a way like h(k). I mean h(1),h(2) etc which are matrices that depend on k. I mean creating them through a for loop. I tried the following code but it failed to produce N matrices and it just gave one matrix:

clc;
clear;
close all;
hx = [-1/sqrt(5);-2/sqrt(5)];
hu = [0;0];
N = 25;
Ek1 = zeros(N+1,1);
Ek2 = ones(N+1,1);
Ek3 = ones(N,1);
Ek4 = zeros(N,1);
h = zeros(102,1);
for k=1:N
    if k==1
        h(:,:) = [kron(Ek2,hx);kron(Ek3,hu)];
    else
        h(:,:) = [kron(Ek1,hx);kron(Ek4,hu)];
    end
end

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Stephen23 2019-5-28

编辑：Stephen23 2019-5-28

" I want to produce N number of matrices..."

Then the best** solution is to use indexing, exactly as your question already shows

** best in the sense simplest, neatest, easiest to write, easiest to debug, and most efficient.

Note that dynamically access variable names is one way that beginners force themselves into writing slow, complex, buggy code that is hard to debug. Read this to know why:

https://www.mathworks.com/matlabcentral/answers/304528-tutorial-why-variables-should-not-be-named-dynamically-eval

请先登录，再进行评论。

请先登录，再回答此问题。

采纳的回答

Geoff Hayes 2019-5-27

1
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/464248-produce-matrices-through-for-loop#answer_376805

在 MATLAB Online 中打开

Ali - if the output matrix of each iteration is of dimension 102x1, then you could store each output as a column in a 102xN matrix. For example,

    h = zeros(102,N);
    for k=1:N
        if k==1
            h(:,k) = [kron(Ek2,hx);kron(Ek3,hu)];
        else
            h(:,k) = [kron(Ek1,hx);kron(Ek4,hu)];
        end
    end

21 个评论
显示 19更早的评论隐藏 19更早的评论

Ali Esmaeilpour 2019-5-28

编辑：Geoff Hayes 2019-5-29

在 MATLAB Online 中打开

when I use your h(:,k) code separatly the result is h 22*5 and when I copy it to my main program and run h is 22*2 I dont know why and I want it 22*1 I mean for k==1 I need 22*1 matrix I wanna be sure that it replaces a 22*1 matrix when k==1 itself but I think it doesn't do that. I give you my full optimization program maybe you can get better picture of my problem. the following program has dimension problem because of h, but I just wanna show you h 22*2 in my main code and the usage of my question in my main code:

clc;
clear;
close all;
%% Time structure
dt=0.01;          %sampling time
%% System structure
A = [1.02  -0.1
     0.1   0.98];
 
B = [0.5 0
    0.05 0.5];
G = [0.3 0
     0 0.3];
 
Qf = [50 0
      0 50];
[A, B] = c2d(A,B,dt);
[nx,nu] = size(B);
%% MPC parameters
Q  = eye(2);
R  = eye(2)*50;
N = 5;
%% Building block matrices
I = eye(2,2);
q = 2;
m = 2;
Qbar = blkdiag(Q,Q,Q,Q,Qf,R,R,R,R);
Fq = blkdiag(sqrt(Q),sqrt(Q),sqrt(Q),sqrt(Q),sqrt(Q),sqrt(Qf),sqrt(R),sqrt(R),sqrt(R),sqrt(R),sqrt(R));
Gxx = zeros(2*N , size(A,1));
for i = 1:N
    Gxx(q*i-q+1:q*i , :) = A^i;
end
Gxu = zeros(2*N , 2*N);
for i = 1:N
   for j = 1:i
      Gxu(q*i-q+1:q*i , m*j-m+1:m*j) = A^(i-j)*(B); 
   end    
end
Gxw = zeros(2*N , 2*N);
for i = 1:N
    for j = 1:i
    Gxw(q*i-q+1:q*i , m*j-m+1:m*j) = A^(i-j)*(I);    
    end
end
%% Solve using YALMIP
K = sdpvar(repmat(10,1,N),repmat(22,1,N));
alpha = 0.975;
r = 1.96;
Sigmaw = eye(10,10);
Sigma0 = eye(2,2);
xbar0 = ones(2,20);
Cs = [Gxx*xbar0;zeros(12,20)];
hx = [-1/sqrt(5);-2/sqrt(5)];
hu = [0;0];
Ahat = [Gxx*xbar0 eye(10,2);zeros(12,20) zeros(12,2)];
Bhat = [Gxu;eye(12,10)];
Abar = [eye(11,10);zeros(11,10)];
M = Gxx*Sigma0*Gxx' + Gxw*Sigmaw*Gxw';
Mhat = [M zeros(10,12);zeros(12,10) ones(12,12)];
Euhat = ones(22,20);
Ek = [zeros(12,10);eye(10,10)];
g = 3;
Ek1 = zeros(N+1,1);
Ek2 = ones(N+1,1);
Ek3 = ones(N,1);
Ek4 = zeros(N,1);
constraint=[];
 
objective=0;
for k = 1:N   
    if k==1
        h(:,k) = [kron(Ek2,hx);kron(Ek3,hu)];    
    else
        h(:,k) = [kron(Ek1,hx);kron(Ek4,hu)];    
    end
    
    b = sqrt(h'*((Abar+Bhat*K{k}*Ek)*M*(Abar+Bhat*K{k}*Ek)')*h);
    
    objective = objective + 0.5*trace(Fq'*(Ahat+Bhat*K{k})*Mhat*(Ahat+Bhat*K{k})'*Fq);
    
    constraint = [constraint, h'*(Cs+Bhat*K{k}*Euhat)+r*(sqrt(h'*(Abar+Bhat*K{k}*Ek)*M*(Abar+Bhat*K{k}*Ek)'*h))<=g, [(h'*(Abar+Bhat*K{k}*Ek)*M*(Abar+Bhat*K{k}*Ek)'*h) (h'*(Abar+Bhat*K{k}*Ek));((Abar+Bhat*K{k}*Ek)'*h) (Gxx*Sigma0*Gxx' + Gxw*Sigmaw*Gxw')]>=0];
end
options = sdpsettings('solver','sedumi');
optimize (constraint,objective,options);
K = value(K);