sum b(i,j,k)=1
sum b(i,j,k)=0
These lines don't do what you think they do. They are equivalent to:
sum('b(i,j,k)=1')
sum('b(i,j,k)=0')
which means they're adding up the ASCII values of the characters in those two char vectors.
There's an easier way to create your desired array. Let's start off with a 10-by-10-by-10 array filled with 0's just as you did.
A = zeros(10, 10, 10);
Now, for all the columns and pages of A, we want to fill in rows 1 and 10 with the value 1. This uses array indexing with element positions. When you read this, read : as "all".
A([1 10], :, :) = 1; % Elements in rows 1 and 10, all columns, all pages of A
We do the same for all rows and pages of A, filling in columns 1 and 10.
A(:, [1 10], :) = 1; % Elements in all rows, columns 1 and 10, all pages of A
And the same for all rows and columns, filling in pages 1 and 10.
A(:, :, [1 10]) = 1; % Elements in all rows, all columns, pages 1 and 10 of A
Now if we've done it right, A should have an 8-by-8-by-8 array of 0's still in the middle. Let's make sure it has the right number of 0's.
nnz(A == 0) % 512 == 8*8*8
The only other values in A are an outer "shell" of 1's.
nnz(A == 1) % 1000 - 512 = 488
