How can I use the iteration to store variables? It gives error Subscript indices must either be real positive integers or logicals.

I wrote the code for 2 agents, but facing problem in storing variables in iteration. As the error shows Subscript indices must either be real positive integers or logicals. I want to rum the loop for 50 iterations.

t0=0;
tf=20;
tspan=[t0:1:tf];

x0=[3;0;0];

u1(1)=3; u2(1)=-3; 
E1(1)=2; E2(1)=2;
itt=50;N=4;
y1=zeros
y1(1)=3; y2(1)=-1;
a=[0 -1;-1 0];

% For 1 agent
for i=1:itt
    
    options=odeset('RelTol',1e-3,'AbsTol',1e-6);
    [t,x]=ode23(@new,tspan,x0,options)
    
    y1_est(i)=y1(i-1)+E1(i)*(du1(i));
    
    err1(i)=y1(i)-y1_est(i);

    
    if (sign(E1(i))<0)
        E1(i)=E1(1);
    else
        E1(i)=E1(i-1)+(eta*du1(i-1)*err1(i-1))/(mu+du1(i-1)^2)
    end
       
 


end

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
function xdot = new(t,x)
x1=x(1);
x2=x(2);
x3=x(3);


X1=[x1;x2;x3];



A=[0.5 0.035 0.025;0.0255 0.6 -0.99;0.75 0.03 0.025];
B=[0.2;0.2;0];
C=[1 0 1];
y1=C*X1;

xdot1 =A*X1+B*u1;


xdot =[xdot1];
end

Answer 1

[Geoff Hayes]

Shivanshu - presumably the error is with

for i=1:itt
    
    options=odeset('RelTol',1e-3,'AbsTol',1e-6);
    [t,x]=ode23(@new,tspan,x0,options)
    
    y1_est(i)=y1(i-1)+E1(i)*(du1(i));

because on the first iteration of the list, i is 1 yet the code tries to access y1(i-1)...hence the error. I think that you need to reconsider your how you are accessing data from these arrays. I was going to suggest starting from 2 (i.e. for i=2:itt) but I suspect that will lead to other issues.

[Yokuna]

Yes, I corrected it. Now, it gives onother issue like

"Index exceeds matrix dimensions"

[Geoff Hayes]

How did you correct it? If you changed the starting index to a 2, then this will cause problems elsewhere in the code. I think you need to reconsider how your arrays are initialized and how you want to use them. For example,

for i=1:itt-1
    
    options=odeset('RelTol',1e-3,'AbsTol',1e-6);
    [t,x]=ode23(@new,tspan,x0,options)
    
    y1_est(i)=y1(i-1)+E1(i)*(du1(i));
    y2_est(i)=y2(i-1)+E2(i)*(du2(i));
    
    err1(i)=y1(i)-y1_est(i);
    err2(i)=y2(i)-y2_est(i);

Do you expect both y1(i-1) and y1(i) to be populated already? If so, why?

[Yokuna]

I corrected the code and got a diiferent error at the last step. err1(i)=y1(i)-y1_est(i); NEW1 must return a column vector.

New code:

for i=2:itt-1
    
    options=odeset('RelTol',1e-3,'AbsTol',1e-6);
    [t,x]=ode23(@new1,tspan,x0,options)

    y1(i)=y1;
    y2(i)=y2;

    E1(i)=E1(i-1)+(eta*du1(i-1)*err1(i-1))/(mu+du1(i-1)^2)
    E2(i)=E2(i-1)+(eta*du2(i-1)*err2(i-1))/(mu+du2(i-1)^2)


    J1(i)=a(1,2)*(y2(i)-y1(i));
    
    J2(i)=a(2,1)*(y1(i)-y2(i));
    
    u1(i)=u1(i-1)+row*(E1(i)/(lamda+E1(i)^2))*J1(i);
    u2(i)=u2(i-1)+row*(E2(i)/(lamda+E2(i)^2))*J2(i);
    
    du1(i+1)=u1(i)-u1(i-1);
    du2(i+1)=u2(i)-u2(i-1);

    y1_est(i)=y1(i-1)+E1(i)*(du1(i));
    y2_est(i)=y2(i-1)+E2(i)*(du2(i));
    
    err1(i)=y1(i)-y1_est(i);
    err2(i)=y2(i)-y2_est(i);
    
    end

[Geoff Hayes]

I think that you will need to repost or attach all of your code. What is new1 as that seems to be different than the new function you had before.

[Geoff Hayes]

Shivanshu's answer moved here

It runs for 2 iterations but then it says "NEW1 must return a column vector". Not getting the reason for such an error.

t0=0;
tf=20;
tspan=[t0:1:tf];

x0=[3;0;0;0;0;1];

u1(1)=3; u2(1)=-3; 
E1(1)=2; E2(1)=2;
J1(1)=0; J2(1)=0;
du1(1)=3; du2(1)=3; 
du1(2)=3; du2(2)=3; 
row=0.45; lamda=1; mu=1; eta=1; err=10^-5;
itt=50;N=2;
y1=zeros
y1(1)=3; y2(1)=-1;
a=[0 -1;-1 0];
% y1(2)=6; y2(2)=3;
err1(1)=0;err2(1)=0;
y1_est(1)=0; y1_est(1)=0;

for i=2:1:itt-1
    
    options=odeset('RelTol',1e-3,'AbsTol',1e-6);
    [t,x]=ode23(@new1,tspan,x0,options)
    y1(i)=Y1
    y2(i)=Y2

    E1(i)=E1(i-1)+(eta*du1(i-1)*err1(i-1))/(mu+du1(i-1)^2)
    E2(i)=E2(i-1)+(eta*du2(i-1)*err2(i-1))/(mu+du2(i-1)^2)


    J1(i)=a(1,2)*(y2(i)-y1(i));
    
    J2(i)=a(2,1)*(y1(i)-y2(i));
    
    u1(i)=u1(i-1)+row*(E1(i)/(lamda+E1(i)^2))*J1(i);
    u2(i)=u2(i-1)+row*(E2(i)/(lamda+E2(i)^2))*J2(i);

    if i>3
        du1(i)=u1(i-1)-u1(i-2);
        du2(i)=u2(i-1)-u2(i-2);
    end

    y1_est(i)=y1(i-1)+E1(i)*(du1(i));
    y2_est(i)=y2(i-1)+E2(i)*(du2(i));
    
    err1(i)=y1(i)-y1_est(i)
    err2(i)=y2(i)-y2_est(i)

end

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

function xdot = new1(t,x)
    N=4;
    
    global u1
    global u2
    
    global Y1
    global Y2
    
    x1=x(1);
    x2=x(2);
    x3=x(3);
    x4=x(4);
    x5=x(5);
    x6=x(6);
    
    
    X1=[x1;x2;x3];
    X2=[x4;x5;x6];
    
    
    A=[0.5 0.035 0.025;0.0255 0.6 -0.99;0.75 0.03 0.025];
    B=[0.2;0.2;0];
    C=[1 0 -1];
    Y1=C*X1;
    Y2=C*X2;
    
    
    xdot1 =A*X1+B*u1;
    xdot2 =A*X2+B*u2;
    
    xdot =[xdot1;xdot2];
end

[Geoff Hayes]

I can't run the above code since I don't know what Y1 and Y2 should be. Consider nesting your function new1 into a main function so that you don't require global variables.

[Geoff Hayes]

If I just use dummy values for Y1 and Y2, then I see the same error as you. It appears that the output from new1 is no longer a column vector/array but a matrix

xdot =

    2.1000    2.2440
    0.6765    0.8205
    2.2500    2.2500
   -0.5750   -0.7190
   -1.5900   -1.7340
    0.0250    0.0250

and this may be related to how your u1 and u2 are computed. I recommend using the MATLAB debugger to step through the code to determine why this happens.

[Yokuna]

Got this. Just putting u(i-1) instead of u in new1 file gives the result.