dimension error with * operator

This line of code: asynch = C1(:)*N(:)*C(:)'/(delta-2) throws an error: Error using * Inner matrix dimensions must agree. That would seem to be right C1 is a 1X100 double and N is a 100x100 double. So I changed the code to T = transpose(C1); asynch = C1(:)*N(:)*T(:)'/(delta-2). This makes T a 100X1 double and the code should run but I get the same error. What's wrong. Thanks

[OCDER]

synch = C1(:)*C1(:)'/(delta-2)
                   ^
This works because of the transpose operator.
C1(:)  = Nx1 vector
C1(:)' = 1xN vector
so C1(:)*C1(:)' works as a (Mx1) * (1xM) = MxM vector (inner dimension agrees)

In your case, you have a matrix N that is 100x100.
N = 100x100 matrix
N(:) = 10000x1 vector

C(:)*N(:) will NOT WORK because it's a (Mx1) * (10000x1) = ????

Answer 1

[OCDER]

Although N is a 100x100 matrix, when you do N(:), you make it into a 10000x1 vector. Get rid of the "(:)" when you want to proper matrix math.

Answer 2

[Pratik Bajaria]

Hello,

Try .* operator. It is the normal matrix multiplication, the one you are trying is doing a Hadamard multiplication.

Regards, Pratik

Answer 3

[Douglas Brenner]

The paper from which I got the code uses * and that works in this line: synch = C1(:)*C1(:)'/(delta-2). Not only do I not get an error but I get what I think is the right answer. For asynch = C1(:)*N(:)*C(:)'/(delta-2), I think the problem is that N is 2D and C1 is one D but I got that code from the paper too.