OffDiagonalElement

OffDiagonalElement is the term used in matrix algebra to denote any entry of a square matrix whose row index does not equal its column index. For a matrix A = [a_{ij}] of size n×n, elements with i ≠ j are considered off-diagonal. The diagonal elements are a_{ii} (i = 1, ..., n).

Examples illustrate the concept clearly: in a 3×3 matrix, a12, a13, a21, a23, a31, a32 are off-diagonal, while the diagonal elements are a11, a22, and a33.

Properties and structure: In a symmetric matrix, off-diagonal elements satisfy a_{ij} = a_{ji}, forming symmetric pairs. In a skew-symmetric matrix, a_{ij} = -a_{ji}, and diagonal elements are zero. Off-diagonal entries often carry the information about interactions or couplings between different coordinates or variables represented by the matrix.

Applications: In graph theory, the adjacency matrix of a simple graph uses off-diagonal entries to indicate the presence of edges between distinct vertices, with diagonals typically zero. In statistics, the off-diagonal entries of a correlation or covariance matrix represent the pairwise relationships between different variables. In optimization, physics, and numerical analysis, off-diagonal terms reflect coupling between components, affecting eigenvalues, stability, and dynamics.

Computation and notation: The off-diagonal part of A can be written as A_off = A − diag(a_{11}, ..., a_{nn}). The set of off-diagonal elements is {a_{ij} | i ≠ j}, and the sum of all off-diagonal elements equals the sum of all entries minus the trace ∑_i a_{ii}.