The Spectral Theorem for Self-Adjoint Operators

The Core Idea: Diagonalization as Decomposition

Imagine you have a complex machine that transforms vectors in a space. If the machine is "self-adjoint," it has a special symmetry: it treats inputs and outputs in a way that preserves the inner product structure. The spectral theorem tells us that such a machine is not arbitrary. It can always be broken down into a simple sequence of steps: first, project the vector onto a specific axis (eigenvector), then scale it by a real number (eigenvalue), and finally sum these scaled projections. In plain language, a self-adjoint operator is just a weighted sum of "directions" in the space. Because these directions are mutually orthogonal, we can analyze the operator by looking at each direction independently. This is why we can always find a basis where the operator looks like a simple diagonal list of numbers, rather than a complicated matrix of mixed terms.

Formal Statement and Consequences

Let \( \hat{A} \) be a self-adjoint operator on a finite-dimensional complex inner-product space \( V \) with \( \dim V = n \). The spectral theorem guarantees three critical properties: the eigenvalues are real, the eigenvectors corresponding to distinct eigenvalues are orthogonal, and there exists an orthonormal basis of eigenvectors. This last point is the most powerful: it means the operator is diagonalizable by a unitary transformation.

Mathematically, if \( \{|e_1\rangle,\dots,|e_n\rangle\} \) is the orthonormal eigenbasis, we can construct a unitary matrix \( \hat{U} \) whose columns are these eigenvectors. The diagonalization is expressed as:

where \( \lambda_k \in \mathbb{R} \). The inverse relation gives the spectral decomposition of the operator itself:

Here, \( |e_k\rangle\langle e_k| \) acts as a projection operator onto the subspace spanned by \( |e_k\rangle \). This form is crucial because it shows that \( \hat{A} \) is a linear combination of orthogonal projectors, which are idempotent and Hermitian.

Handling Degeneracies

In many physical systems, eigenvalues are not unique; multiple linearly independent eigenvectors may share the same eigenvalue \( \lambda \). This is known as degeneracy. The spectral theorem still holds, but the choice of basis within the degenerate subspace is not unique. Any orthonormal basis spanning the eigenspace associated with \( \lambda \) is valid.

Suppose \( \lambda_1 = \lambda_2 = \lambda \). The subspace \( V_\lambda = \mathrm{span}(|e_1\rangle, |e_2\rangle) \) is invariant under \( \hat{A} \). Within this subspace, \( \hat{A} \) acts as \( \lambda \hat{I} \). The projection operator for this degenerate level is \( \hat{P}_\lambda = |e_1\rangle\langle e_1| + |e_2\rangle\langle e_2| \). The spectral sum becomes:

This grouping is physically significant because it allows us to define the probability of measuring the value \( \lambda \) as the norm squared of the projection of the state onto \( V_\lambda \), regardless of the specific basis chosen within that subspace.

Unitary Symmetry and Change of Basis

The unitary matrix \( \hat{U} \) that diagonalizes \( \hat{A} \) represents a change of basis from the standard computational basis to the eigenbasis. Since \( \hat{U} \) is unitary, it preserves inner products and norms, ensuring that the physical content of the state vector is unchanged, only its representation is rotated.

If we have an arbitrary state \( |\psi\rangle \), its components in the eigenbasis are \( c_k = \langle e_k|\psi\rangle \). The expectation value of \( \hat{A} \) in state \( |\psi\rangle \) is:

This equation reveals the statistical nature of quantum measurements: the expectation value is the weighted average of the eigenvalues, where the weights are the probabilities \( |\langle e_k|\psi\rangle|^2 \). The spectral theorem ensures that these probabilities sum to one, as \( \sum |\langle e_k|\psi\rangle|^2 = \langle \psi|\psi\rangle = 1 \).