Error: Inner matrix dimensions must agree.

Hi, I am using "gamultiobj" to solve my problem. I, however, have no idea why the error message occurred

??? Error using ==> mtimes
Inner matrix dimensions must agree.

Error in ==> col_mop>@(x)x'*(B'*B)*x at 50
fun1 = @(x) x'*(B'*B)*x;

My objective function is coded as follows

fun1 = @(x) x'*(B'*B)*x;
fun2 = @(x) x'*x;
bifun = @(x) [fun1(x) fun2(x)];

The dimension of B'*B is m so x is an m-by-1 vector. The syntax for gamultiobj is

X = gamultiobj(FITNESSFCN,NVARS,A,b,Aeq,beq,LB,UB,options)

I have properly input the number of decision variables (NVARS) as m but the error message shows. I also check the dimensions using the command "whos" which showed what I wanted. Is there anything wrong?

Thanks in advance.

Answer 1

[Paulo Silva]

Error correction...inner matrix dimensions must be same??? how to correct ?

(look at Matt Fig answer on that question)

[Chien-Chia Huang]

Thanks for the information. But I am a bit confused. The command ".*" should be element-wise product, right? For example, [2,3].*[1,4] = [2, 12]. What I am doing is vector-vector (vector-matrix) inner product, for instance, [2,3]'*[1,4] = 2 + 12 = 14. Have I gone wrong?

[Paulo Silva]

I have no clue what you want to do, just pointed you to a similar question, you are correct on those assumptions but you have this x'*(B'*B)*x , some part of those three multiplications doesn't work because of dimensions mismatch, see the size of every vector or array to find where the problem is.

[Matt Fig]

@Chhien-Chia,

Show us the output of the following:

size(x) % I mean the argument to fun1 here...

size(B)

[Chien-Chia Huang]

Thanks a lot, Paulo! I just found out what the problem is. The x in gamultiobj is by default supposed to be a row vector, but I thought intuitively of x as a column vector. The problem was just solved.

[Chien-Chia Huang]

sorry Matt I did not see your reply. The output of size(B) was 5-by-10 and size(x) was 10-by-1. But the requested x in gamultiobj should be 1-by-10. Thanks, anyway. :D