How to restrict tensorprod (with contraction) to only give certain results?

I have vector of matrices A and a vector of vectors M. I want to do matrix multiplication of matrix in A and corresponding vector in M where the result in in R:

A=((1 2 M=((1 R=((5

3 4) 2) 11)

(5 6 (3 (39

7 8)) 4)) 53))

I have three possible solutions:

A=cat(3,[1 2;3 4],[5 6;7 8]) % vector of matrices

M=cat(3,[1;2],[3;4]) % vector of vectors

% 1

N=permute(M,[2 1 3]) % make rows of the vectors

C=A.*N

R1=sum(C,2)

% 2

R2=pagemtimes(A,M)

% 3

R3=tensorprod(A,M,2,1) % contraction acts as matrix multiplication

This gives:

R1(:,:,1) =

5

11

R1(:,:,2) =

39

53

R2(:,:,1) =

5

11

R2(:,:,2) =

39

53

R3(:,:,1,1) =

5 17

11 23

R3(:,:,1,2) =

11 39

25 53

So R1 and R2 I must reshape, but the solution is there. But I have a question about the tensorprod. Because it calculates four vectors. So can I do something with tensorprod so that it gives the upper left and lower right vectors?

Answer 1

[Matt J]

No, you cannot. pagemtimes is the appropriate solution.