Finding Convenient Bases
You'll learn how to strategically choose bases that make matrix representations of linear transformations much simpler to work with.
The Core Idea
After mastering change of basis and similarity transformations, you now face a practical question: which basis should you pick? Not all bases are created equal—some make computations dramatically easier.
Think of it like choosing the right coordinate system for a physics problem. You could describe planetary motion using Cartesian coordinates anchored to an arbitrary point, but using polar coordinates centered on the sun makes everything cleaner. Same physics, much simpler math.
What Makes a Basis "Convenient"?
A convenient basis typically produces a matrix representation that is:
- Diagonal (best case—eigenvalue basis when the operator is diagonalizable)
- Triangular (upper or lower, when diagonalization isn't possible)
- Block diagonal (when the space decomposes into invariant subspaces)
- Sparse (many zero entries)
Strategy
-
Look for eigenvectors first: If your transformation has enough linearly independent eigenvectors, use them as your basis. The matrix becomes diagonal.
-
Exploit orthogonality: If working in an inner product space, an orthonormal basis often simplifies computations (projection formulas become trivial, norms are preserved).
-
Use invariant subspaces: If certain subspaces map to themselves, choose basis vectors adapted to that structure.
-
Consider the problem context: For rotations, use a basis aligned with the rotation axis. For projections, use a basis that includes the projection direction.
Key Takeaway: The "right" basis transforms complex matrix operations into simple ones—choosing bases aligned with the transformation's natural geometry (eigenvectors, invariant subspaces, symmetries) minimizes computational complexity and reveals structural insights.