Orthonormalization

Gram-Schmidt orthonormalization is a process for making vectors into orthonormal vectors.

Vectors

Two vectors a and b can be orthonormalized into A and B.

Orthogonality is a property of two vectors, not one. Therefore a needs no transformation and becomes A.

The process of transforming b into B is simply the subtraction of all components of a from b. This is a linear combination and does not change the column space of a system that includes both a and b. Projections are a complimentary idea; p is the component of a that estimates b. The process of orthonormalization is the same as computing projections but the error term e is the desired result. Recall that e = b - ax̂ and x̂ = (A^Tb)/(A^TA). Therefore, B = b - A (A^Tb)/(A^TA).

To transform another vector c into being orthogonal to both A and B, apply the same process for each component: C = c - A (A^Tc)/(A^TA) - B (B^Tc)/(B^TB).

The orthogonal vectors are then normalized by scaling to their Euclidean distances, as A/||A|| and B/||B||.

Matrices

The process applied to vectors is also applicable to the columns in a matrix. Instead of vectors a and b, use v₁ and v₂ in V. The process yields u₁ and u₂ in U. Then the columns are normalized into Q like q₁ = u₁/||u₁||.

To re-emphasize, this is a linear combination generalized as A = QR, and does not change the column space of A.

Note that Q is a matrix with orthonormal columns; it must also be square to be called an orthogonal matrix.

CategoryRicottone