If you take into acocunt the assumption made by Sebastian 1.b which says

for the purpose of this exercise consider that a block diagonal matrix must contain at least 2x2 size blocks in the diagonal

then diagonal matrix it is not block-diagonal matrix because it doesn't have any 2x2 sub-matrices.

Thank, you're right

On the other hand, What do you think about the necessity of having a square matrix in order for it to be simetric?
I remember this from the school,
but I don´t know if Sebastian considered that B.trasponse() = B is equivalent to symetric.

Thank in avance

Pretty sure your definition of symmetric is correct and the right one to use.

@amer I think you are right, it should be a square. When Sebastian talks about symmetrical matrix he refers to Omega and not to B matrix. If you compute the trasponse of any non squared matrix you get

matrix_non_squared.transpose() != matrix_non_squared

Actually I feel the opposite. I am more confident on A and C. About B, the robot can see max 3 landmarks but it can also sees just 1. Think about for each position X0 the robot sees just L0, in X1 it sees just L1, in the X2 it sees just L2. In addition, let's suppose we have the same number of position and landmarks (so N == M). How will be the matrix B in this case ?

I think that Sebastian means as block diagonal also the k-diagonal matrices, a generalization of let's say tri-diagonal. So you can have off-diagonal members and a matrix with 2x2 blocks that overlap eachother on the diagonal.