Ph · Physics

Why Do the Projectors of an Orthonormal Basis Sum to Identity?

Published By the 9to5Science editorial team

At a glance

  1. 1. Define each projector

    For a normalized basis vector, the rank-one projector is |e_n><e_n|.

  2. 2. Apply the sum

    Acting on any vector, each projector extracts its coefficient along one basis direction.

  3. 3. Use completeness

    A complete orthonormal basis reconstructs every vector from those components.

  4. 4. Check convergence

    In an infinite-dimensional space, the partial sums converge strongly to identity.

Advertisement

For an orthonormal basis, each rank-one projection keeps only the component of a vector along one basis direction. Adding those projections restores all the components. That is why the sum acts exactly like the identity operator.

The important condition is that the basis be complete: its vectors must span the whole space, or be dense in it for an infinite-dimensional Hilbert space. An incomplete orthonormal set sums to a projection onto its closed linear span.

The rank-one projector

Let {∣en⟩}\{|e_n\rangle\} be an orthonormal basis. Orthonormality means

⟨em∣en⟩=δmn,\langle e_m|e_n\rangle=\delta_{mn},

where δmn\delta_{mn} is 1 when m=nm=n and 0 otherwise. The projection onto the direction ∣en⟩|e_n\rangle is

Pn=∣en⟩⟨en∣.P_n=|e_n\rangle\langle e_n|.

To see what this operator does, apply it to an arbitrary vector ∣ψ⟩|\psi\rangle:

Pn∣ψ⟩=∣en⟩⟨en∣ψ⟩.P_n|\psi\rangle=|e_n\rangle\langle e_n|\psi\rangle.

The scalar ⟨en∣ψ⟩\langle e_n|\psi\rangle is the coefficient of ∣ψ⟩|\psi\rangle along ∣en⟩|e_n\rangle. So PnP_n extracts that component and returns it as a vector in the direction ∣en⟩|e_n\rangle.

Proof using the basis expansion

Because the basis is complete, every vector can be expanded in it:

∣ψ⟩=∑n∣en⟩⟨en∣ψ⟩.|\psi\rangle=\sum_n |e_n\rangle\langle e_n|\psi\rangle.

The sum of the projectors therefore acts on an arbitrary vector as

(∑n∣en⟩⟨en∣)∣ψ⟩=∑n∣en⟩⟨en∣ψ⟩=∣ψ⟩.\left(\sum_n |e_n\rangle\langle e_n|\right)|\psi\rangle =\sum_n |e_n\rangle\langle e_n|\psi\rangle =|\psi\rangle.

Since the operator leaves every vector unchanged, it is the identity operator:

∑n∣en⟩⟨en∣=I.\boxed{\sum_n |e_n\rangle\langle e_n|=I.}

This is called the completeness relation or resolution of the identity. The proof is not specific to quantum mechanics; it is the operator form of reconstructing a vector from its coordinates in an orthonormal basis.

A concrete two-dimensional example

Take the standard basis of C2\mathbb{C}^2:

∣e1⟩=(10),∣e2⟩=(01).|e_1\rangle=\begin{pmatrix}1\\0\end{pmatrix}, \qquad |e_2\rangle=\begin{pmatrix}0\\1\end{pmatrix}.

Their projectors are

P1=∣e1⟩⟨e1∣=(1000),P2=∣e2⟩⟨e2∣=(0001).P_1=|e_1\rangle\langle e_1| =\begin{pmatrix}1&0\\0&0\end{pmatrix}, \qquad P_2=|e_2\rangle\langle e_2| =\begin{pmatrix}0&0\\0&1\end{pmatrix}.

Adding them gives

P1+P2=(1001)=I.P_1+P_2=\begin{pmatrix}1&0\\0&1\end{pmatrix}=I.

For example, on ∣ψ⟩=(3,−2)T|\psi\rangle=(3,-2)^T, the first projector gives (3,0)T(3,0)^T and the second gives (0,−2)T(0,-2)^T. Their sum is the original vector (3,−2)T(3,-2)^T.

What changes in infinite dimensions?

For a finite basis, the operator sum is an ordinary finite sum. In an infinite-dimensional Hilbert space, the expression is interpreted through its action on vectors. If the basis is countable, define the partial sums

SN=∑n=1N∣en⟩⟨en∣.S_N=\sum_{n=1}^{N}|e_n\rangle\langle e_n|.

For every vector ∣ψ⟩|\psi\rangle, completeness gives

lim⁡N→∞SN∣ψ⟩=∣ψ⟩\lim_{N\to\infty} S_N|\psi\rangle=|\psi\rangle

in the Hilbert-space norm. This is strong operator convergence. It does not generally mean that SNS_N converges to II in operator norm: in an infinite-dimensional space, each finite-rank SNS_N leaves some directions out, and ∥I−SN∥=1\|I-S_N\|=1.

In a nonseparable Hilbert space, a basis may be uncountable. The same statement is understood as a limit over finite subsets of basis vectors; for any particular vector, at most countably many of its basis coefficients are nonzero.

Common mistakes

  • Using an incomplete orthonormal set: If the vectors have closed linear span VV, then ∑n∣en⟩⟨en∣\sum_n |e_n\rangle\langle e_n| is the projector onto VV, not the identity on the whole space.
  • Forgetting normalization: The formula ∣en⟩⟨en∣|e_n\rangle\langle e_n| is the projector onto the basis direction when ⟨en∣en⟩=1\langle e_n|e_n\rangle=1. For a nonzero, unnormalized vector ∣v⟩|v\rangle, the projector is ∣v⟩⟨v∣/⟨v∣v⟩|v\rangle\langle v|/\langle v|v\rangle.
  • Assuming any basis works in the same form: The simple sum uses an orthonormal basis. For a nonorthogonal basis, the vectors are not mutually orthogonal, so these rank-one operators do not isolate the coordinates in this way; a dual basis or a different completeness formula is needed.
  • Treating an infinite sum as an ordinary matrix sum: In infinite dimensions, specify the sense of convergence. The usual completeness statement guarantees convergence when the partial sums act on each vector.

In quantum mechanics, the relation is useful whenever states or operators are expanded in a complete set of states. Inserting I=∑n∣en⟩⟨en∣I=\sum_n |e_n\rangle\langle e_n| into an expression lets you sum over the possible basis components without changing the expression’s value.

References

  1. Introduction to Quantum Mechanics, David J. Griffiths and Darrell F. Schroeter
  2. Linear Algebra Done Right, Sheldon Axler

Spotted an error? Send a correction. Read how we write and check articles in our editorial policy.