Is there a way to break down or approximate an $m \times n$ matrix into a product of $m \times k$, $k \times k$ and $k \times n$ matrices, with a certain amount of error? I am not looking for SVD because here, unlike in SVD, $k> m,n$. Also, m,k,n $\neq 1$. For example, say a $4 \times 3$ matrix into $4\times 8$, $8 \times 8$ and a $8 \times 3$ matrix. I have been scouring the internet to see if there exists a matrix decomposition of sorts to do this. I would like to know if there is any such in existence.
I am basically trying to solve for X in matrix equation of the format : $AXB=C$ where X is a diagonal matrix. Matrices $A,B$ and $C$ are known. They can be square or rectangular, complex valued matrices.
Thanks.