32.9. Summary of Key Concepts

Eigenspaces are null spaces of (A−λI)

An eigenspace is defined as the set of all vectors that, when multiplied by a matrix A minus a scalar λ times the identity matrix I, result in the zero vector. This means that the eigenspace represents all the vectors that are 'stretched' or 'squashed' by the matrix A by the factor of eigenvalue λ, without changing the direction of these vectors. So, if you take the matrix A and subtract λI, and look at the vectors that give you zero when multiplied by that matrix, those vectors belong to the eigenspace associated with the eigenvalue λ.

Think of an eigenspace like the direction of a boat when it encounters a strong current. The current (the matrix) pushes the boat an exact amount (the eigenvalue), but if the boat is positioned just right, it continues to float straight (eigenspace) without sinking (resulting in zero).

Basis of eigenvectors is a linearly independent set spanning the eigenspace

The basis of eigenvectors refers to a set of eigenvectors that are linearly independent, meaning no eigenvector in the set can be written as a combination of the others. This basis spans the entire eigenspace for the specific eigenvalue λ, which means any vector in that eigenspace can be expressed as a combination (linear combination) of these basis vectors. This concept is crucial as it allows us to simplify complex systems into manageable components.

Imagine a musical band where each instrument plays a distinct part but together creates a harmonious song. Each instrument represents an independent eigenvector, and when combined, they create the full musical piece which represents the eigenspace. Just as every song requires its unique set of instruments to be created, each eigenspace requires its basis vectors.

If eigenvectors form a basis for Rn, the matrix is diagonalizable

A matrix is said to be diagonalizable if it has enough eigenvectors to form a complete basis for Rn, where n is the number of dimensions of the space. This means that if you can find n linearly independent eigenvectors corresponding to n eigenvalues, you can express the matrix in a diagonal form (D) through a similarity transformation (A = PDP−1). This simplification makes computations involving the matrix much easier, especially when dealing with powers of the matrix or exponentials.

Consider sorting a deck of playing cards. If you can sort them by suit and rank (like finding a basis), you can easily access any specific card you need (diagonalizing the problem). Without this organization, finding a specific card would take a lot longer—that's why diagonalization simplifies many operations involving matrices.

For symmetric matrices, eigenvectors can form an orthonormal basis

In the case of symmetric matrices, a special property comes into play: the eigenvectors corresponding to distinct eigenvalues are not only linearly independent but are also orthogonal to each other. Moreover, they can be normalized to have unit length, forming an orthonormal basis. An orthonormal basis has the advantage of simplifying various operations, especially in terms of calculating projections and transformations in multi-dimensional spaces.

Think of orthonormal vectors as the axes on a graph (like X, Y, and Z axes in 3D space). Each axis is not only independent but also at right angles to each other, giving you a straightforward way to describe the position of a point in that space. Just like with orthonormal axes, you can easily manipulate data points represented in this basis.

--