Orthogonal Bases
What you'll learn: How using bases made of mutually perpendicular vectors simplifies vector space computations dramatically.
The Core Idea
You already know that any basis lets you represent vectors as unique coordinate combinations. But when you choose a basis where all vectors are orthogonal to each other (perpendicular in geometric terms), many computations become remarkably simpler.
An orthogonal basis for a vector space is a basis where every pair of distinct vectors has inner product zero. If those vectors are also unit length (norm = 1), you have an orthonormal basis—the gold standard.
Why Orthogonal Bases Matter
With a general basis, finding coordinates for a vector requires solving a system of equations. With an orthogonal basis, you can compute each coordinate independently using just inner products.
If {v₁, v₂, ..., vₙ} is an orthogonal basis and you want to express vector x in this basis, the coordinate for vᵢ is simply:
coefficient = ⟨x, vᵢ⟩ / ⟨vᵢ, vᵢ⟩
For an orthonormal basis, it's even cleaner (since ⟨vᵢ, vᵢ⟩ = 1):
coefficient = ⟨x, vᵢ⟩
No matrix inversion, no solving systems—just dot products!
Analogy
Think of orthogonal bases like a city grid with streets running perfectly north-south and east-west. To describe any location, you simply measure distance along each perpendicular direction independently. A non-orthogonal basis is like a city with diagonal, intersecting streets—figuring out coordinates requires considering how movements in different directions interfere with each other.
Computational Advantages
- Fast coordinate computation via inner products
- Simplified norm calculations: ||x||² = sum of squared coordinates
- Stability in numerical algorithms
- Direct projections onto subspaces
Key Takeaway: Orthogonal bases transform vector space arithmetic from solving coupled equations into independent inner product calculations, making them computationally superior for nearly every practical application.