Course contentsShow
← Mathematics
Lesson 2223 of 3,12146. Linear Algebra — Inner Products and OrthogonalityFree lesson

Normalizing to Obtain an Orthonormal Basis

Convert an orthogonal basis to an orthonormal basis by normalizing each vector to unit length.

Normalizing to Obtain an Orthonormal Basis

You'll learn how to convert an orthogonal basis into an orthonormal basis by scaling each vector to unit length.

The Core Idea

After applying the Gram-Schmidt algorithm, you have a set of orthogonal vectors—meaning they're all perpendicular to each other. That's great! But for many applications, we want something even nicer: an orthonormal basis, where each vector not only points in a perpendicular direction but also has exactly length 1.

The good news? This final step is simple: just divide each orthogonal vector by its own length (its norm).

The Normalization Process

Given an orthogonal basis {v₁, v₂, ..., vₖ}, you create an orthonormal basis {u₁, u₂, ..., uₖ} by normalizing each vector:

uᵢ = vᵢ / ||vᵢ||

In other words, for each vector vᵢ, compute its norm ||vᵢ|| = √⟨vᵢ, vᵢ⟩, then divide the vector by that scalar.

Why This Works

Think of each orthogonal vector as an arrow pointing in the "right direction" but possibly the "wrong length." Dividing by the norm is like scaling the arrow down (or up) until it has exactly length 1, without changing which direction it points.

Example in ℝ²

Suppose Gram-Schmidt gave you orthogonal vectors v₁ = (3, 0) and v₂ = (0, 4).

  • ||v₁|| = 3, so u₁ = (3, 0)/3 = (1, 0)
  • ||v₂|| = 4, so u₂ = (0, 4)/4 = (0, 1)

Now {u₁, u₂} is orthonormal: perpendicular and unit length.

Key Takeaway: To convert an orthogonal basis to an orthonormal basis, divide each vector by its norm—this preserves orthogonality while standardizing all lengths to 1.