I am trying to find a solution (or algorithm) for the following combinatorial problem:
Given an input matrix and a target matrix, find a permutation of the rows and permutation of the columns that minimizes some (predefined) distance between the two matrices.
For example, I have an input matrix \begin{pmatrix} 4 & 2 & 3\\ 4 & 1 & 0\\ 2 & 4 & 1 \end{pmatrix}
and a target matrix
\begin{pmatrix} 5 & 0 & 0\\ 1 & 2 & 1\\ 3 & 2 & 1 \end{pmatrix}
and I want to minimize the element-wise distance between the input matrix I and the target matrix T $\sum_{i}^{n} \sum_{j}^{n} \left | i_{ij} - t_{ij}\right |$
as an example here is a permutation of the rows \begin{pmatrix} 4 & 1 & 0\\ 2 & 4 & 1\\ 4 & 2 & 3 \end{pmatrix}
and here is a permutation of the columns (after the permutation of the rows)
\begin{pmatrix} 4 & 0 & 1\\ 2 & 1 & 4\\ 4 & 3 & 2 \end{pmatrix}
I would be really grateful towards anyone who can point me towards literature for solving this problem.
Here is a simple algorithm that usually reduces, but does not minimize, the distance.
Permute $I$ and $T$ as per the algorithm $f()$ described below:
Calculate the multisets of each row
e.g. \begin{pmatrix} 4 & 2 & 3\\ 4 & 1 & 0\\ 2 & 4 & 1 \end{pmatrix} becomes $(2,3,4),(0,1,4),(1,2,4)$
Sort the rows based on the lexicographical order of the multisets
e.g. $(2,3,4),(0,1,4),(1,2,4)$
becomes
$(0,1,4),(1,2,4),(2,3,4)$
Calculate the multisets of each column
The output of this algorithm is $f(I)∘f`(T)$.