The Identity Matrix

The square matrix with ones on the diagonal and zeros elsewhere acts as the multiplicative identity for matrices.

The neutral element

The identity matrix I_n is the n-by-n matrix with 1 on every diagonal entry and 0 everywhere else. It plays the role that the number 1 plays for scalars: for any conformable matrix A, both A I = A and I A = A. Multiplying a vector by I leaves it unchanged, so I represents the transformation that does nothing.

Why it matters

The identity is the reference point for the inverse: a matrix A is invertible when there exists A^{-1} with A A^{-1} = A^{-1} A = I. It also appears in eigenvalue analysis through the expression A - lambda I, whose determinant gives the characteristic polynomial, and in regularization, where adding a small multiple of I to a matrix improves its conditioning.

Kronecker delta

The entries of the identity are the Kronecker delta: delta_ij equals 1 when i = j and 0 otherwise. This compact symbol is standard in physics and tensor notation, where it selects diagonal terms and contracts indices.

Scaling the identity

A scalar multiple cI is a scalar matrix; it scales every vector uniformly by c. Scalar matrices are the only matrices that commute with every other matrix of the same size, a fact tied to Schur's lemma in representation theory.

import numpy as np
I = np.eye(4)
A = np.random.rand(4, 4)
print(np.allclose(A @ I, A))   # True
print(np.allclose(I @ A, A))   # True

In iterative solvers used for large physics simulations, the identity anchors preconditioners: the closer a preconditioned operator is to I, the faster the iteration converges.