MATLAB: “Subscripted assignment dimension mismatch” when using a for loop

arrayforloopMATLABmatrixr2012a

I'm having a problem with the following iteration

for i=L:N
    e(it,1) = x(i) - flipud(y(i-L+1:i,1))' * g(:,it); %Here is the error
    g(:,it+1) = g(:,it) + mu * flipud(y(i-L+1:i,1)) * e(it,1);    
    it = it + 1;
end

When I execute the whole code I get "Subscripted assignment dimension mismatch." error in the signaled line. I have reviewed some other posts about this error, but I don't understand how can I solve it in my code. The variable 'e' is expected to be a matrix which saves a line in every iteration.

Best Answer

There is a mismatch here:

flipud(y(i-L+1:i,1))' * g(:,it);

You have a L*1 matrix (flipud(y(i-L+1:i,1))) and you are transposing that so it would be 1*L. So, g(:,it) must be a L*1 matrix, otherwise you will get this error.

Related Solutions

MATLAB: In an assignment A(I) = B, the number of elements in B and I must be the same.

I cannot follow what you are doing, and I cannot run your code.

However, the most likely problem is that you need to vectorize your code, since otherwise MATLAB will interpret multiply, divide, and exponentiation operations as matrix rather than element-wise array operations. That means using the ‘dot operator’, substituting ‘.*’ for ‘*’, ‘./’ for ‘/’, and ‘./’ for ‘^’.

Use the vectorize (link) fiunction, and see the documentation on Array vs. Matrix Operations (link) for details.

MATLAB: Coupled ODE error- too many inputs

Try it with an explicit instead of an anonymous function:

function dy = fcn(t, y)
g=9.8; l1=1; l2=1; m1=1; m2=1;
dy = [(l2*y(3)-l1*y(4)*cos(y(1)-y(2)))/((l1^2)*l2*(m1+m2*(sin(y(1)-y(2))^2))); ...
  (l1*(m1+m2)*y(4)-l2*m2*y(3)*cos(y(1)-y(2)))/(l1*(l2^2)*m2*(m1+m2*(sin(y(1)-y(2))^2))); ...
  -(m1+m2)*g*l1*sin(y(1))-((y(3)*y(4)*sin(y(1)-y(2)))/(l1*l2(m1+m2*(sin(y(1)-y(2))^2))))+ ...
  (((l2^2)*m2*(y(3)^2)+(l1^2)*(m1+m2)*(y(4)^2)-l1*l2*m2*y(3)*y(4)*cos(y(1)-y(2)))* ...
  sin(2*(y(1)-y(2))))/(2*(l1^2)*(l2^2)*(m1+m2*(sin(y(1)-y(2))^2))^2); ...
  -m2*g*l2*sin(y(2))+((y(3)*y(4)*sin(y(1)-y(2)))/(l1*l2(m1+m2*(sin(y(1)-y(2))^2)))) - ...
  (((l2^2)*m2*(y(3)^2)+(l1^2)*(m1+m2)*(y(4)^2)-l1*l2*m2*y(3)*y(4)*cos(y(1)-y(2)))* ...
  sin(2*(y(1)-y(2))))/(2*(l1^2)*(l2^2)*(m1+m2*(sin(y(1)-y(2))^2))^2)];
end

And in the main code:

[t,yi] = ode45(@fcn,[0,1],[45,0,0,0]);

If this works, --- wow, no, it does not work. As soon as your write this as a function, you might see this part (twice!):

... l2(m1+m2*(sin(y(1)-y(2))^2) ...

This should be most likely:

l2 * (m1+m2*(sin(y(1)-y(2))^2)
   ^

With your code "l2" is treated as a function and this might cause the "Too many input arguments" error. Do you have a "l2.m" file in your path? If not, I'd expect a message message like "no such function found for input argument double" or "indices exceed array dimensions". So maybe there is another error in addition.

Such typos are very hard to identify an anonymous functions.

A simplified version of the function:

function dy = fcn(t, y)
g = 9.8; l1 = 1; l2 = 1; m1 = 1; m2 = 1;
c = cos(y(1) - y(2));
s = sin(y(1) - y(2));
d = m1 + m2 * s ^ 2;
f = y(3) * y(4) * s / (l1 * l2 * d);
h = (l2^2 * m2      * y(3)^2 + ...
     l1^2 * (m1+m2) * y(4)^2 - ...
     l1 * l2 * m2   * y(3) * y(4) * c) * ...
    sin(2 * (y(1) - y(2))) / (2 * l1^2 * l2^2 * d^2);
dy = [(y(3) - l1/l2 * y(4) * c) / (l1^2 * d); ...
  ((m1+m2) * y(4) - l2/l1 * m2 * y(3) * c) / (l2^2 * m2 * d); ...
  -(m1+m2) * g * l1 * sin(y(1)) - f + h; ...
  -m2      * g * l2 * sin(y(2)) + f - h];
end

Looks less cruel in my eyes. Do you see the magic power of spaces? Now a typo like "l2()" would be easier to find and if unnecessary parentheses are removed, the meaning of the required ones gets more obvious.

Note: I did not run it to compare the results, so maybe I made a typo during inserting temporary variables. Check this by your own.

Afterwards it might be more useful to provide the parameters (but only the parameters, not the equation to integrate) by an anonymous function:

g=9.8; l1=1; l2=1; m1=1; m2=1;
diffeq = @(t,y) fcn(t, y, g, m1, m2, l1, l2);
...
function dy = fcn(t, y, g, m1, m2, l1, l2)
dy = ...

Best Answer

Related Solutions

MATLAB: In an assignment A(I) = B, the number of elements in B and I must be the same.

MATLAB: Coupled ODE error- too many inputs

Related Question