Skip to content

Bosonic Fock space is the Hilbert space that collects all possible particle-number sectors for identical bosons. If h\mathcal h is the one-particle Hilbert space, the bosonic Fock space is

FB(h)=⨁N=0∞Sym⁡Nh.\mathcal F_B(\mathcal h) = \bigoplus_{N=0}^{\infty} \operatorname{Sym}^N\mathcal h.

Here Sym⁡Nh\operatorname{Sym}^N\mathcal h is the symmetric NN-particle subspace of h⊗N\mathcal h^{\otimes N}. The direct sum lets a state have a definite particle number, or a superposition of different particle-number sectors when the physics allows it.

Helpful background. Occupation-Number Basis supplies mode labels; Direct Sums versus Tensor Products explains the sector construction.

The N=0N=0 sector is the vacuum sector:

Sym⁡0h≅C.\operatorname{Sym}^0\mathcal h \cong \mathbb C.

A normalized basis vector for this one-dimensional sector is written

∣0⟩.\lvert 0\rangle.

The vacuum is not the zero vector. It is a physical no-particle state in Fock space. The zero vector has norm zero and represents no state; the vacuum is normalized and can be acted on by creation operators in later notation.

The vacuum also does not necessarily have zero energy. A Hamiltonian may assign it zero energy by convention, or it may include zero-point energies depending on the model and normal-ordering convention.

The one-particle sector is Sym⁡1h=h\operatorname{Sym}^1\mathcal h=\mathcal h. In an orthonormal two-mode basis a,ba,b, its basis is ∣1,0⟩,∣0,1⟩\lvert1,0\rangle,\lvert0,1\rangle.

The two-particle sector has basis ∣2,0⟩,∣1,1⟩,∣0,2⟩\lvert2,0\rangle,\lvert1,1\rangle,\lvert0,2\rangle. In particle-slot notation, the middle vector is (∣a,b⟩+∣b,a⟩)/2(\lvert a,b\rangle+\lvert b,a\rangle)/\sqrt2, while ∣2,0⟩\lvert2,0\rangle is ∣a,a⟩\lvert a,a\rangle. Repeated occupation is allowed, and no extra particle label is needed.

For general NN, the vectors in Sym⁡Nh\operatorname{Sym}^N\mathcal h satisfy U(π)∣Ψ⟩=∣Ψ⟩U(\pi)\lvert\Psi\rangle=\lvert\Psi\rangle for every permutation. The symmetrizer constructs this subspace from the unrestricted tensor product.

For MM modes, its dimension is (M+N−1N)\binom{M+N-1}{N}. This counts nonnegative solutions of ∑ini=N\sum_i n_i=N: with two modes the dimensions are 1,2,3,4,…1,2,3,4,\ldots as N=0,1,2,3,…N=0,1,2,3,\ldots. Even a single bosonic mode has an infinite-dimensional full Fock space because NN is unbounded.

A vector in bosonic Fock space is a sequence of fixed-number components:

∣Ψ⟩=ψ0⊕ψ1⊕ψ2⊕⋯ ,ψN∈Sym⁡Nh.\lvert\Psi\rangle = \psi_0\oplus\psi_1\oplus\psi_2\oplus\cdots, \qquad \psi_N\in\operatorname{Sym}^N\mathcal h.

The norm is

∥Ψ∥2=∑N=0∞∥ψN∥2.\lVert\Psi\rVert^2 = \sum_{N=0}^{\infty} \lVert\psi_N\rVert^2.

The Hilbert direct sum contains exactly the square-summable sequences, with normalization ∑N∥ψN∥2=1\sum_N\|\psi_N\|^2=1 for a pure state. Finite norm does not guarantee finite mean number: ⟨N⟩=∑NN∥ψN∥2\langle N\rangle=\sum_N N\|\psi_N\|^2 needs a separate convergence condition. A state of definite particle number has only one nonzero component. A superposition of different particle numbers has more than one nonzero component.

The direct sum is different from a tensor product over NN. It is not a space with every particle-number sector simultaneously occupied as separate subsystems. It is a Hilbert space whose alternatives are different total particle numbers.

Choose an orthonormal mode basis {∣φi⟩}\{\lvert\varphi_i\rangle\} for h\mathcal h. A bosonic occupation-number basis vector is

∣n1,n2,n3,…⟩B,\lvert n_1,n_2,n_3,\ldots\rangle_B,

with

ni=0,1,2,…,N=∑ini.n_i=0,1,2,\ldots, \qquad N=\sum_i n_i.

For fixed NN, these vectors form an orthonormal basis of Sym⁡Nh\operatorname{Sym}^N\mathcal h; their closed linear span is the whole sector. Allowing all finite NN gives an orthonormal basis of the full Fock space. Even when there are only finitely many modes, a general Fock vector may be an infinite square-summable superposition of these basis vectors. No basis vector with an actual infinite occupation is added by this completion.

Once creation operators are introduced, the normalized occupation vector is written

∣n1,n2,…⟩B=∏i(ai†)nini!∣0⟩,\lvert n_1,n_2,\ldots\rangle_B = \prod_i \frac{(a_i^\dagger)^{n_i}}{\sqrt{n_i!}} \lvert0\rangle,

where only finitely many nin_i are nonzero in the finite-particle sector. The factorials are the normalization factors associated with repeated bosonic occupation of the same mode.

The definite-occupation basis vectors themselves are treated in Number States.

  1. For two bosonic modes aa and bb, list the occupation basis states in the N=3N=3 sector.
Solution

The nonnegative occupations must satisfy na+nb=3n_a+n_b=3. The basis states are

∣3a,0b⟩B,∣2a,1b⟩B,∣1a,2b⟩B,∣0a,3b⟩B.\lvert3_a,0_b\rangle_B, \qquad \lvert2_a,1_b\rangle_B, \qquad \lvert1_a,2_b\rangle_B, \qquad \lvert0_a,3_b\rangle_B.
  1. If there are MM bosonic modes and total particle number N=2N=2, how many occupation basis states are there?
Solution

One can put both bosons in the same mode, giving MM states, or put them in two distinct modes, giving M(M−1)/2M(M-1)/2 states. The total is

M+M(M−1)2=M(M+1)2.M+\frac{M(M-1)}{2} = \frac{M(M+1)}{2}.

This is also the dimension of Sym⁡2CM\operatorname{Sym}^2\mathbb C^M.

  1. Finite norm, divergent mean. In a single mode let ∣Ψ⟩=C∑n≥0(n+1)−1∣n⟩\lvert\Psi\rangle=C\sum_{n\ge0}(n+1)^{-1}\lvert n\rangle. Does it belong to Fock space? Does it have finite mean number?
Solution

Yes: ∑n≥0(n+1)−2=π2/6\sum_{n\ge0}(n+1)^{-2}=\pi^2/6, so C=6/πC=\sqrt6/\pi normalizes it. But ⟨N⟩=C2∑n≥0n/(n+1)2\langle N\rangle=C^2\sum_{n\ge0}n/(n+1)^2 diverges as a harmonic series. A normalized state need not lie in the domain needed for every unbounded observable.

  • A. Altland and B. Simons, Condensed Matter Field Theory, 2nd ed., Cambridge University Press, 2010.
  • L. E. Ballentine, Quantum Mechanics: A Modern Development, 2nd ed., World Scientific, 2014.
  • P. A. M. Dirac, The Principles of Quantum Mechanics, 4th ed., Oxford University Press, 1958.
  • A. L. Fetter and J. D. Walecka, Quantum Theory of Many-Particle Systems, McGraw-Hill, 1971.
  • R. Shankar, Principles of Quantum Mechanics, 2nd ed., Springer, 1994.