Fermionic Fock space, Pauli exclusion, and anticommutators

Intuition [Beginner]

A photon is happy to share a state with another photon; a million photons piling into the same laser mode is the physics of coherent light. An electron is not. Two electrons in a helium atom cannot occupy the same single-electron orbital with the same spin. The deepest reason any two atoms in the universe do not collapse into a featureless point is that electrons refuse to share single-particle states. This refusal is the Pauli exclusion principle, proposed by Wolfgang Pauli in 1925 to explain the structure of the periodic table.

The previous unit built the bosonic Fock space — the home of states like photons, where occupation numbers can be any non-negative integer. The fermionic Fock space is its antisymmetric cousin, the home of electrons, protons, neutrons, quarks, and every other half-integer-spin particle. The same skeleton applies: there is a vacuum, there are creation operators that add a particle and annihilation operators that remove one, and there is a one-particle Hilbert space whose multi-particle promotion is the full Fock space.

The single change of rule is dramatic. Swapping the labels on two bosons leaves the state alone; swapping the labels on two fermions multiplies the state by $-1$. Putting two fermions in the same single-particle state then gives a state equal to its own negative, which has to be the zero vector. No two fermions can occupy the same one-particle state. This is exclusion, packaged as a sign.

The algebra of creation and annihilation operators changes correspondingly. For bosons the commutator $a{a}^{\ast }-a^{\ast }a$ was the central identity. For fermions the anticommutator $a{a}^{\ast }+a^{\ast }a$ takes its place. A pair of creation operators for the same one-particle state squares to zero, which is again exclusion expressed in algebra. Three sentences — antisymmetry, anticommutators, and zero squares — generate the entire structure of fermionic many-body quantum mechanics and underwrite the stability of every chunk of matter in the universe.

Visual [Beginner]

Picture the same tower of shelves as the bosonic case, but with a new rule painted on the wall: each shelf has its own occupancy cap, and any single one-particle state can appear on the cap at most once. The vacuum is the empty bottom shelf. The one-particle space is the shelf above it. The two-particle space is the shelf above that, and it is now built from pairs of distinct one-particle states — never two copies of the same state.

Now refine the picture. Each shelf is the antisymmetric $n$-fold tensor power ${\Lambda }^n\mathcal{H}$ of the one-particle Hilbert space $\mathcal{H}$ — the subspace of $n$-fold tensors that change sign under any swap of two slots. The fermionic Fock space is the Hilbert direct sum of all these shelves, with the inner product summing contributions from each level.

The creation operator $a^{\ast }(f)$ wedges the one-particle state $f$ onto the existing antisymmetric product. If $f$ is already on the shelf, the wedge product collapses to zero — exclusion, as a built-in feature of the wedge. The annihilation operator $a(f)$ removes an $f$-occupant and produces a state on the shelf below. Strings of distinct creation operators applied to the vacuum span the entire fermionic Fock space, just as in the bosonic case, except that the order of the operators now matters up to a sign.

Worked example [Beginner]

Consider helium: two electrons orbiting a nucleus of charge $+2e$. The one-particle Hilbert space is the orbital state space, augmented by a two-dimensional spin space — so each one-particle slot has both a spatial label (which orbital) and a spin label ($\uparrow$ or $\downarrow$). For the ground state, both electrons sit in the lowest-energy $1s$ orbital. The Pauli exclusion principle forbids them from also sharing a spin label. They must have opposite spins.

Label the two electrons by their slot, $1$ and $2$. The antisymmetric two-electron ground-state wavefunction is

$${\Psi }_{1s^2}(1,2)=\frac{1}{\sqrt{2}}{\varphi }_{1s}({\mathbf{r}}_1){\varphi }_{1s}({\mathbf{r}}_2)(|\uparrow \downarrow ⟩-|\downarrow \uparrow ⟩),$$

a product of the symmetric two-electron spatial part (both electrons in ${\varphi }_{1s}$) and the antisymmetric spin-singlet $|\uparrow \downarrow ⟩-|\downarrow \uparrow ⟩$. The combination is overall antisymmetric under exchange of the two electrons, as required. Numerically, ${\varphi }_{1s}(\mathbf{r})=(Z^3/\pi a_0^3{)}^{1/2}{e}^{-Zr/a_0}$ with $Z=2$ for helium; the spin singlet is the unique antisymmetric combination of two spin-$1/2$ vectors.

Try instead to put both electrons in $|1s\uparrow ⟩$. The antisymmetric two-particle state would be the slot-swap difference $|1s\uparrow {⟩}_1|1s\uparrow {⟩}_2-|1s\uparrow {⟩}_2|1s\uparrow {⟩}_1$, which is the zero vector. The wavefunction vanishes identically. Two electrons in the same orbital with the same spin is not a possible physical state — exclusion, in numbers. Lithium therefore has one $1s$ electron forced into the next available shell ($2s$) and grows up into the third row of the periodic table; aufbau is the systematic application of this single rule shell by shell.

Check your understanding [Beginner]

Formal definition [Intermediate+]

Let $\mathcal{H}$ be a complex separable Hilbert space — the one-particle Hilbert space. For each integer $n\ge 0$, write ${\mathcal{H}}^{\otimes n}$ for the Hilbert-space $n$-fold tensor power (with ${\mathcal{H}}^{\otimes 0}:=\mathbb{C}$), and let $S_n$ act on ${\mathcal{H}}^{\otimes n}$ by permutation of the tensor factors with unitary $U_{\sigma }$ for $\sigma \in S_n$. The antisymmetrisation projector is

$$P_a^{(n)}:=\frac{1}{n!}\sum _{\sigma \in S_n}\mathrm{sgn}(\sigma )U_{\sigma },$$

where $\mathrm{sgn}(\sigma )\in \{+1,-1\}$ is the sign of the permutation. The antisymmetric $n$-fold tensor power is the closed subspace

$${\Lambda }^n\mathcal{H}:=P_a^{(n)}{\mathcal{H}}^{\otimes n}=\{\xi \in {\mathcal{H}}^{\otimes n}:U_{\sigma }\xi =\mathrm{sgn}(\sigma )\xi \text{ for all }\sigma \in S_n\}.$$

The fermionic Fock space over $\mathcal{H}$ is the Hilbert-space direct sum

$${\mathcal{F}}_a(\mathcal{H}):=\underset{n=0}{\overset{\infty }{⨁}}{\Lambda }^n\mathcal{H},$$

with inner product $⟨({\xi }_n{)}_n,({\eta }_n{)}_n{⟩}_{{\mathcal{F}}_a}={\sum }_{n=0}^{\infty }⟨{\xi }_n,{\eta }_n{⟩}_{{\Lambda }^n\mathcal{H}}$ and norm $\Vert ({\xi }_n{)}_n{\Vert }^2={\sum }_n\Vert {\xi }_n{\Vert }^2<\infty$. The ${\ell }^2$ direct-sum construction makes ${\mathcal{F}}_a(\mathcal{H})$ a complex separable Hilbert space 02.11.08. The summand ${\Lambda }^0\mathcal{H}=\mathbb{C}$ is spanned by the unit vector $\Omega$ called the vacuum.

Antisymmetric inner product

For $f_1,\dots ,f_n,g_1,\dots ,g_n\in \mathcal{H}$, the antisymmetric tensor $f_1\wedge \cdots \wedge f_n:=\sqrt{n!}{P}_a^{(n)}(f_1\otimes \cdots \otimes f_n)$ satisfies

$$⟨f_1\wedge \cdots \wedge f_n,g_1\wedge \cdots \wedge g_n{⟩}_{{\Lambda }^n\mathcal{H}}=\det (⟨f_i,g_j⟩),$$

where $\det$ is the ordinary matrix determinant. The determinant — the antisymmetric analog of the permanent that appeared on the bosonic side — encodes the alternating-sign structure of the wedge product. The normalisation factor $\sqrt{n!}$ in the definition of the wedge product is the convention that gives clean coefficients for the creation operators below.

Slater-determinant basis

Let $\{{\varphi }_i{\}}_{i\ge 1}$ be an orthonormal basis of $\mathcal{H}$. An orthonormal basis of ${\Lambda }^n\mathcal{H}$ is given by the Slater determinants

$$|i_1,\dots ,i_n{⟩}_a:={\varphi }_{i_1}\wedge {\varphi }_{i_2}\wedge \cdots \wedge {\varphi }_{i_n}=\frac{1}{\sqrt{n!}}\det ({\varphi }_{i_k}(x_l){)}_{k,l=1}^n,$$

indexed by strictly increasing tuples $i_1<i_2<\cdots <i_n$. The determinant form on the right is the classical Slater 1929 ^{[Slater 1929]} writing in coordinates: an $n$-particle antisymmetric wavefunction is the determinant of a matrix whose entry $(k,l)$ is the $k$-th orbital evaluated at the $l$-th particle's coordinate. Antisymmetry under particle exchange is then immediate from the determinant's alternation under row swaps.

Creation and annihilation operators

For each $f\in \mathcal{H}$, define the creation operator $a^{\ast }(f):{\Lambda }^n\mathcal{H}\to {\Lambda }^{n+1}\mathcal{H}$ by

$$a^{\ast }(f)(g_1\wedge \cdots \wedge g_n):=f\wedge g_1\wedge \cdots \wedge g_n,$$

with $a^{\ast }(f)\Omega :=f$, extended linearly. The annihilation operator $a(f):{\Lambda }^n\mathcal{H}\to {\Lambda }^{n-1}\mathcal{H}$ is defined as the formal adjoint, acting by

$$a(f)(g_1\wedge \cdots \wedge g_n):=\sum _{k=1}^n(-1{)}^{k-1}⟨f,g_k⟩g_1\wedge \cdots \wedge \widehat{g_k}\wedge \cdots \wedge g_n,$$

(where the caret marks omission), and $a(f)\Omega :=0$. The alternating signs $(-1{)}^{k-1}$ arise from anticommuting the operator past the $k-1$ wedge factors to its left. The map $f\mapsto a^{\ast }(f)$ is complex-linear; the map $f\mapsto a(f)$ is complex-antilinear, exactly as in the bosonic case.

Number operator

The number operator $N$ acts on ${\Lambda }^n\mathcal{H}$ by multiplication by $n$. For an orthonormal basis $\{e_k{\}}_{k\ge 1}$ of $\mathcal{H}$,

$$N=\sum _{k=1}^{\infty }{a}^{\ast }(e_k)a(e_k),$$

where the sum converges on the finite-particle subspace ${\mathcal{F}}_a^{\mathrm{fin}}(\mathcal{H})$ (the algebraic direct sum of the ${\Lambda }^n\mathcal{H}$) and extends to a self-adjoint operator on ${\mathcal{F}}_a(\mathcal{H})$ with spectrum $\sigma (N)={\mathbb{N}}_0$ when $\dim \mathcal{H}=\infty$. The individual mode-occupation operator $n_k:=a^{\ast }(e_k)a(e_k)$ has spectrum $\{0,1\}$ — the Pauli exclusion principle as a spectral statement.

Standard conventions

The Dirac-style field operators for a Dirac field of mass $m$ on Minkowski space are

$$\psi (x)=\sum _s\int \frac{d^3\mathbf{p}}{(2\pi {)}^3\sqrt{2{\omega }_{\mathbf{p}}}}(b_s(\mathbf{p})u_s(\mathbf{p})e^{-ip\cdot x}+d_s^{\ast }(\mathbf{p})v_s(\mathbf{p})e^{+ip\cdot x}),$$

where $b_s,d_s$ are the fermion and antifermion annihilation operators, $u_s,v_s$ are spinor mode functions, $s$ runs over the two spin states, and $p^0={\omega }_{\mathbf{p}}=\sqrt{|\mathbf{p}{|}^2+m^2}$. The CAR for the field reads $\{{\psi }_a(x),{\psi }_b^{†}(y)\}={\delta }_{ab}{\delta }^3(\mathbf{x}-\mathbf{y})$ for equal-time field operators on a Cauchy slice (Tong's QFT notes §5.2 ^[tong]; Peskin-Schroeder Ch. 3 §3.5 ^{[Peskin-Schroeder]}).

Counterexamples to common slips

• The operators $a(f),a^{\ast }(f)$ are bounded in the fermionic case, with $\Vert a^{\ast }(f)\Vert =\Vert a(f)\Vert =\Vert f\Vert$ (proved in the next section). This is a genuine algebraic contrast with the bosonic side, where the CCR $[a,a^{\ast }]=1$ forces unboundedness. The boundedness lets the fermionic CAR algebra be packaged as a $C^{\ast }$-algebra without domain machinery — Bratteli-Robinson Vol. II §5.2 ^{[Bratteli-Robinson Vol. II]}.

• The antisymmetric tensor power ${\Lambda }^n\mathcal{H}$ vanishes whenever $n>\dim \mathcal{H}$ (with $\dim \mathcal{H}$ counted as a cardinal). For $\mathcal{H}={\mathbb{C}}^d$, the fermionic Fock space is finite-dimensional of dimension ${\sum }_{n=0}^d\left(\genfrac{}{}{0ex}{}{d}{n}\right)=2^d$ — the spin-$1/2$ chain Hilbert space, foreshadowing the Jordan-Wigner transformation.

• The wedge product $f\wedge f=0$ for every $f\in \mathcal{H}$, equivalently $a^{\ast }(f{)}^2=0$. This is the Pauli exclusion principle in algebraic form and is the single most distinctive feature of the fermionic algebra against the bosonic one.

• The CAR algebra is not the same as the CCR algebra with a sign flipped — the bosonic CCR generates a Heisenberg algebra (Lie algebra, with a central extension), while the fermionic CAR generates a Clifford algebra 03.09.02 over ${\mathcal{H}}_{\mathbb{R}}\oplus {\mathcal{H}}_{\mathbb{R}}$ with the inner-product-induced quadratic form. The two algebras live in different universes of representation theory.

Key theorem with proof [Intermediate+]

Theorem (canonical anticommutation relations). For all $f,g\in \mathcal{H}$, the creation and annihilation operators on ${\mathcal{F}}_a(\mathcal{H})$ satisfy

$$\{a(f),a^{\ast }(g)\}=⟨f,g⟩1,\{a(f),a(g)\}=\{a^{\ast }(f),a^{\ast }(g)\}=0,$$

on the finite-particle subspace ${\mathcal{F}}_a^{\mathrm{fin}}(\mathcal{H})$.

Proof. It suffices to compute on an antisymmetric tensor $\xi =g_1\wedge \cdots \wedge g_n\in {\Lambda }^n\mathcal{H}$. Apply $a^{\ast }(g)$ first, then $a(f)$:

$$a^{\ast }(g)\xi =g\wedge g_1\wedge \cdots \wedge g_n\in {\Lambda }^{n+1}\mathcal{H}.$$

Acting by $a(f)$ on this $(n+1)$-fold wedge: the leading slot now carries $g$, with sign $(-1{)}^0=1$, and the remaining slots carry $g_1,\dots ,g_n$ with signs $(-1{)}^k$ for $k=1,\dots ,n$:

$$a(f)a^{\ast }(g)\xi =⟨f,g⟩g_1\wedge \cdots \wedge g_n+\sum _{k=1}^n(-1{)}^k⟨f,g_k⟩g\wedge g_1\wedge \cdots \wedge \widehat{g_k}\wedge \cdots \wedge g_n.$$

Now apply the operators in the reverse order. First $a(f)$:

$$a(f)\xi =\sum _{k=1}^n(-1{)}^{k-1}⟨f,g_k⟩g_1\wedge \cdots \wedge \widehat{g_k}\wedge \cdots \wedge g_n,$$

a vector in ${\Lambda }^{n-1}\mathcal{H}$. Then $a^{\ast }(g)$ wedges $g$ onto the leading slot of each term:

$$a^{\ast }(g)a(f)\xi =\sum _{k=1}^n(-1{)}^{k-1}⟨f,g_k⟩g\wedge g_1\wedge \cdots \wedge \widehat{g_k}\wedge \cdots \wedge g_n.$$

Add the two orderings: the $(-1{)}^k$ and $(-1{)}^{k-1}$ contributions cancel term by term in the sum over $k$, and the surviving term is the leading $⟨f,g⟩g_1\wedge \cdots \wedge g_n$. Therefore

$$a(f)a^{\ast }(g)\xi +a^{\ast }(g)a(f)\xi =⟨f,g⟩\xi ,$$

which is the first CAR identity on the dense finite-particle subspace.

For $\{a^{\ast }(f),a^{\ast }(g)\}=0$: both $a^{\ast }(f)a^{\ast }(g)\xi$ and $a^{\ast }(g)a^{\ast }(f)\xi$ produce the same vector up to a sign, namely $f\wedge g\wedge g_1\wedge \cdots \wedge g_n$ and $g\wedge f\wedge g_1\wedge \cdots \wedge g_n$. The wedge product is antisymmetric, so $f\wedge g=-g\wedge f$, and the two contributions cancel exactly. For $\{a(f),a(g)\}=0$: a parallel double-sum cancellation, where the signs from $(-1{)}^{k-1}$ in $a(f)$ and $a(g)$ combine to give cancelling pairs on the diagonals of the indexing. $\square$

Corollary (Pauli exclusion). For every $f\in \mathcal{H}$, $a^(f)^2 = 0$.*

Proof. Set $g=f$ in the third CAR identity: $\{a^{\ast }(f),a^{\ast }(f)\}=2a^{\ast }(f{)}^2=0$, so $a^{\ast }(f{)}^2=0$. $\square$

Theorem (fermionic bound). For every $f\in \mathcal{H}$, the creation and annihilation operators are bounded with $|a^(f)| = |a(f)| = |f|$.*

Proof. From the CAR with $g=f$, $\{a(f),a^{\ast }(f)\}=a(f)a^{\ast }(f)+a^{\ast }(f)a(f)=\Vert f{\Vert }^{2}1$. Both $a(f)a^{\ast }(f)$ and $a^{\ast }(f)a(f)$ are positive operators (each is $T^{\ast }T$ for $T=a^{\ast }(f)$ or $T=a(f)$ respectively). Therefore each is bounded above by their sum: $0\le a^{\ast }(f)a(f)\le \Vert f{\Vert }^{2}1$. Taking the operator norm,

$$\Vert a(f){\Vert }^2=\Vert a^{\ast }(f)a(f)\Vert \le \Vert f{\Vert }^2,$$

so $\Vert a(f)\Vert \le \Vert f\Vert$. For the matching lower bound, apply $a(f)$ to the one-particle vector $f\in {\Lambda }^1\mathcal{H}$: $a(f)f=⟨f,f⟩\Omega =\Vert f{\Vert }^2\Omega$, giving $\Vert a(f)f\Vert =\Vert f{\Vert }^2\cdot \Vert \Omega \Vert =\Vert f{\Vert }^2$ while $\Vert f{\Vert }_{{\Lambda }^1}=\Vert f{\Vert }_{\mathcal{H}}$. Therefore $\Vert a(f)\Vert \ge \Vert f{\Vert }^2/\Vert f\Vert =\Vert f\Vert$, and combining, $\Vert a(f)\Vert =\Vert f\Vert$. By taking adjoints, $\Vert a^{\ast }(f)\Vert =\Vert a(f)\Vert =\Vert f\Vert$. $\square$

Bridge. The CAR algebra and the fermionic bound together package the entire structure of fermionic many-body quantum mechanics into bounded-operator identities, which lets the CAR algebra be completed in operator norm to a $C^{\ast }$-algebra without the domain machinery the bosonic CCR required. This is exactly the dictionary that lets the antisymmetric tensor algebra of a one-particle Hilbert space generate a many-body theory of indistinguishable fermions; the bridge is then provided by the fermionic second-quantisation functor lifting one-particle dynamics to fermionic Fock-space dynamics, identifying the CAR algebra with the universal irreducible algebra of fermionic creation and annihilation, and feeding into the Dirac-field quantisation programme of the next chapter and the multi-electron atomic-structure programme in section 14.

Exercises [Intermediate+]

A graded set covering CAR computations, the fermionic bound, the Jordan-Wigner transformation, Slater determinants, and the spin-statistics distinction.

Lean formalization [Intermediate+]

lean_status: none — Mathlib provides the antisymmetric tensor power via Mathlib.LinearAlgebra.ExteriorAlgebra.Basic for a vector space over a commutative ring, the Hilbert direct sum via Mathlib.Analysis.InnerProductSpace.l2Space, and inner-product / adjoint infrastructure via Mathlib.Analysis.InnerProductSpace.Adjoint. These suffice to define ${\Lambda }^n\mathcal{H}$ as a closed subspace of the Hilbert tensor power and to assemble ${\mathcal{F}}_a(\mathcal{H})={⨁}_n{\Lambda }^n\mathcal{H}$ as a separable Hilbert space, the antisymmetric inner product as the determinant of the Gram matrix, the Slater-determinant basis of ${\Lambda }^n\mathcal{H}$ indexed by $n$-element subsets of an orthonormal basis, and the finite-dimensional Pauli-matrix representation of the CAR via the Clifford-algebra route. What is not in Mathlib at this date: the creation and annihilation operators $a^{\ast }(f),a(f)$ on the fermionic Fock space as bounded operators with norm equal to $\Vert f\Vert$ satisfying the CAR identities $\{a(f),a^{\ast }(g)\}=⟨f,g⟩$, $\{a(f),a(g)\}=\{a^{\ast }(f),a^{\ast }(g)\}=0$; the CAR $C^{\ast }$-algebra as the unique norm-completed object generated by these relations; the Jordan-Wigner transformation from ${\mathcal{F}}_a({\ell }^2(\mathbb{Z}))$ onto the spin-chain Hilbert space; the fermionic uniqueness theorem (CAR analog of Stone-von Neumann) for finitely many modes; the Dirac field as an operator-valued tempered distribution on the fermionic Fock space over positive-energy spinor solutions of the Dirac equation. This unit is reviewer-attested; see the lean_mathlib_gap block in the frontmatter for the full enumeration.

The CAR algebra as a $C^{\ast }$-algebra [Master]

The fermionic bound $\Vert a^{\ast }(f)\Vert =\Vert f\Vert$ makes the entire fermionic algebraic-Fock-space construction live inside the bounded-operator framework, with no domain machinery needed. Define the CAR algebra ${\mathfrak{A}}_{\mathrm{CAR}}(\mathcal{H})$ as the norm-closure inside $\mathcal{B}({\mathcal{F}}_a(\mathcal{H}))$ of the $\ast$-algebra generated by $\{a(f),a^{\ast }(f):f\in \mathcal{H}\}$. By the fermionic bound and the algebraic CAR identities, ${\mathfrak{A}}_{\mathrm{CAR}}(\mathcal{H})$ is a $C^{\ast }$-algebra — the canonical anticommutation relations $C^{\ast }$-algebra over $\mathcal{H}$, due in this $C^{\ast }$-algebraic form to the Bratteli-Robinson treatment ^{[Bratteli-Robinson Vol. II]}.

Theorem (uniqueness of the CAR algebra). Let $\mathcal{H}$ be a complex separable Hilbert space. The CAR algebra ${\mathfrak{A}}_{\mathrm{CAR}}(\mathcal{H})$ is, up to $$-isomorphism,theunique$C^$-algebrageneratedbyelements${a(f), a^(f)}$($f \in \mathcal{H}$,$a$antilinearin$f$,$a^$linear)satisfyingtheCAR.When$\dim \mathcal{H} = n < \infty$,$\mathfrak{A}{\mathrm{CAR}}(\mathbb{C}^n) \cong M{2^n}(\mathbb{C})$,thefullmatrixalgebraonthe$2^n$-dimensionalfermionicFockspace.When$\dim \mathcal{H} = \infty$,$\mathfrak{A}{\mathrm{CAR}}(\mathcal{H}) \cong \varinjlim_n M{2^n}(\mathbb{C})$,the\ast \ast CAR-AF-algebra\ast \ast (uniformlyhyperfiniteoftype$2^\infty$), and the Fock representation is one of many inequivalent irreducible representations indexed by quasi-equivalence classes of Bogoliubov transformations.

Proof sketch. The finite-mode case is direct: the CAR with $n$ generators $c_k,c_k^{\ast }$ produces the algebra $M_{2^n}(\mathbb{C})$ via the Jordan-Wigner identification with the spin-chain Hilbert space ${⨂}_k{\mathbb{C}}^2$ of dimension $2^n$. Uniqueness for finite $n$ is the finite-dimensional Stone-von Neumann analog. For infinite $n$, the algebra is the inductive limit (uniformly hyperfinite of type $2^{\infty }$, classified by Glimm 1960), and the Fock representation is the cyclic-and-separating representation generated by the vacuum-state on the quasi-free state $\omega (a^{\ast }(f)a(g))=0$, $\omega (a(f)a^{\ast }(g))=⟨f,g⟩$ — the Fock state in the $C^{\ast }$-algebraic sense. The GNS construction over the Fock state recovers the Fock representation. Other quasi-free states (Bogoliubov-rotated, KMS at finite temperature) give inequivalent representations, classified by the Shale-Stinespring criterion in fermionic form: a Bogoliubov transformation $(a(f),a^{\ast }(f))\mapsto (Ua(f)+V{a}^{\ast }(\overline{f}),U^{\ast }{a}^{\ast }(f)+V^{\ast }a(\overline{f}))$ with $U^{\ast }U+V^T\overline{V}=1$ is implementable by a unitary on the original Fock space iff $V$ is Hilbert-Schmidt. $\square$

The $C^{\ast }$-algebraic packaging makes the fermionic case structurally cleaner than the bosonic case (where the unbounded operators forced the Weyl exponentiated form for the $C^{\ast }$-algebra), and is the workhorse of fermionic many-body statistical mechanics: Bratteli-Robinson §5.2 ^{[Bratteli-Robinson Vol. II]} develops the entire equilibrium-state theory (KMS states at inverse temperature $\beta$ and chemical potential $\mu$, Fermi-Dirac distribution recovery, BCS pairing) inside this framework.

Connection to QFT and the Dirac field [Master]

In quantum field theory, the one-particle Hilbert space ${\mathcal{H}}_1$ for a spin-$1/2$ fermion of mass $m$ is taken to be the positive-energy spinor space — the space of $L^2$-sections of the spinor bundle over Minkowski space supported on positive-energy solutions of the Dirac equation $(i{\gamma }^{\mu }{\partial }_{\mu }-m)\psi =0$, where ${\gamma }^{\mu }$ are the Dirac gamma matrices generating a Clifford algebra 03.09.02 over the Minkowski metric. The fermionic Fock space ${\mathcal{F}}_a({\mathcal{H}}_1)$ is the free fermion Hilbert space, and the free Dirac field is the operator-valued tempered distribution

$$\psi (x)=\sum _{s=1}^2\int \frac{d^3\mathbf{p}}{(2\pi {)}^3\sqrt{2{\omega }_{\mathbf{p}}}}(b_s(\mathbf{p})u_s(\mathbf{p})e^{-ip\cdot x}+d_s^{\ast }(\mathbf{p})v_s(\mathbf{p})e^{+ip\cdot x})$$

acting on ${\mathcal{F}}_a({\mathcal{H}}_1)$, with $u_s,v_s$ the four-component positive- and negative-energy spinor mode functions and $b_s,d_s$ the particle and antiparticle annihilation operators. The CAR for the field reads

$$\{{\psi }_a(x),{\psi }_b^{†}(y)\}={\delta }_{ab}{\delta }^3(\mathbf{x}-\mathbf{y})$$

for equal-time field operators on a Cauchy slice (Glimm-Jaffe §6.3 ^{[Glimm-Jaffe 1987]}; Tong's QFT §5.2 ^[tong]). Like the bosonic Klein-Gordon field, $\psi (x)$ is well-defined only as a distribution: smeared against a Schwartz spinor-valued test function $f\in \mathcal{S}({\mathbb{R}}^4,{\mathbb{C}}^4)$, $\psi (f)=\int \psi (x)f(x)d^{4}x$ is a densely-defined bounded operator on ${\mathcal{F}}_a({\mathcal{H}}_1)$.

The Dirac field on the fermionic Fock space satisfies a fermionic-statistics version of the Wightman axioms — strictly, the Wightman framework as packaged in Streater-Wightman 1964 ^{[Streater-Wightman 1964]} includes the spin-statistics theorem as an output: the requirement of local commutativity must be relaxed to local anti-commutativity for half-integer-spin fields, and the proof in Ch. 4 derives this from the joint requirements of Lorentz invariance, positive-energy spectrum, and the existence of a unique Lorentz-invariant vacuum. Bosonic Wightman axioms with half-integer-spin field give a contradiction (the would-be Wightman two-point function fails positivity); fermionic Wightman axioms with integer-spin give a contradiction (the field anticommutator at spacelike separation does not vanish). The fermionic Fock space is therefore the unique kinematic substrate of any rigorous quantisation of half-integer-spin fields.

The CPT theorem in Streater-Wightman Ch. 5 takes the spin-statistics result as an input and proves that any Lorentz-invariant local quantum field theory satisfying the Wightman axioms (with the appropriate bosonic / fermionic statistics) is invariant under the combined action of charge conjugation, parity, and time reversal — one of the deepest theorems in mathematical physics, and the rigorous output of the framework introduced in this unit's bosonic sibling and extended here.

Synthesis. The fermionic Fock space identifies the many-body quantum mechanics of identical fermions with a single, fixed, second-quantised representation of the canonical anticommutation relations on a separable Hilbert space — and the CAR algebra in its $C^{\ast }$-algebraic form pins down the Fock representation as the unique cyclic representation generated by the vacuum-quasi-free state. The foundational reason the construction supports both atomic-structure physics and relativistic QFT is that the antisymmetric tensor algebra of a one-particle Hilbert space generates a Pauli-exclusion-respecting many-body theory automatically, with the fermionic bound $\Vert a^{\ast }(f)\Vert =\Vert f\Vert$ giving bounded operators throughout and the Jordan-Wigner transformation identifying the CAR algebra with the spin-chain Hilbert space in the 1D lattice case. This is exactly the central insight Pauli 1925 ^{[Pauli 1925]}, Dirac 1926 ^{[Dirac 1926]}, Fermi 1926 ^{[Fermi 1926]}, and Jordan-Wigner 1928 ^{[Jordan-Wigner 1928]} introduced — putting these together with the spin-statistics theorem of Pauli 1940 ^{[Pauli 1940]} and Streater-Wightman 1964 ^{[Streater-Wightman 1964]}, the bridge is the operator-valued tempered distribution $\psi (x)$ that satisfies the fermionic Wightman axioms and identifies the free Dirac field with the second-quantised Dirac equation; this pattern recurs in every textbook free-fermion-field construction, builds toward the Standard Model where every matter field is half-integer-spin and quantised this way, and appears again in the Lieb-Thirring stability of matter, the BCS theory of superconductivity, the quantum Hall effect, and the entire $C^{\ast }$-algebraic formulation of equilibrium quantum statistical mechanics for fermionic systems.

Connections [Master]

• Bosonic Fock space and second quantisation 12.13.01. Direct sibling. The fermionic construction differs from the bosonic only in the replacement of the symmetric projector $P_s^{(n)}$ by the antisymmetric projector $P_a^{(n)}$ and of the symmetric tensor power $S^n\mathcal{H}$ by the antisymmetric ${\Lambda }^n\mathcal{H}$. The commutation relation $[a,a^{\ast }]=1$ becomes the anticommutation relation $\{a,a^{\ast }\}=1$; the unbounded creation and annihilation operators of the bosonic side become bounded fermionic ones with $\Vert a^{\ast }(f)\Vert =\Vert f\Vert$; the symmetric inner product as a permanent becomes the antisymmetric one as a determinant; the bosonic Stone-von Neumann uniqueness becomes the fermionic CAR-algebra uniqueness; the Klein-Gordon free field becomes the free Dirac field. Read this unit in parallel with its sibling.

• Hilbert-space formalism of quantum mechanics 12.02.01 pending. Direct prerequisite. The fermionic Fock space is a Hilbert space and the creation / annihilation operators are bounded self-adjoint and anti-self-adjoint combinations on it; the Dirac-von Neumann postulates apply verbatim, with the additional structure of the antisymmetrisation projector and the CAR algebra. The exclusion principle is expressed inside the Dirac-von Neumann framework as the spectral statement $\sigma (N_k)=\{0,1\}$ for each mode-occupation observable $N_k=a^{\ast }(e_k)a(e_k)$.

• Stern-Gerlach and spin-1/2 12.01.02 pending. Chapter seed. The Pauli matrices ${\sigma }_x,{\sigma }_y,{\sigma }_z$ on ${\mathbb{C}}^2$ developed there are the smallest-possible finite-dimensional CAR representation, realised on ${\mathcal{F}}_a(\mathbb{C})=\mathbb{C}\oplus \mathbb{C}={\mathbb{C}}^2$: ${\sigma }_{+}=\left(\begin{array}{cc}0& 1\\ 0& 0\end{array}\right)$ acts as the single creation operator $a^{\ast }$, ${\sigma }_{-}={\sigma }_{+}^{†}$ acts as $a$, $\{{\sigma }_{+},{\sigma }_{-}\}=1$, and ${\sigma }_{+}^2={\sigma }_{-}^2=0$. The Stern-Gerlach experiment was historically the first probe of half-integer angular momentum and, via spin-statistics, the first probe of fermionic statistics; the Jordan-Wigner transformation then promotes this single-site picture to a 1D fermionic lattice via the spin-chain Hilbert space ${⨂}_n{\mathbb{C}}^2$.

• Clifford algebra 03.09.02. The CAR algebra is a Clifford algebra. Concretely, for a real Hilbert space ${\mathcal{H}}_{\mathbb{R}}$ underlying $\mathcal{H}$ with the real-bilinear form $B(f,g)=\mathrm{Re}⟨f,g⟩$, the operators $\gamma (f):=a^{\ast }(f)+a(f)$ for $f\in {\mathcal{H}}_{\mathbb{R}}$ satisfy $\gamma (f)\gamma (g)+\gamma (g)\gamma (f)=2B(f,g)1$, the defining identity of the Clifford algebra $\mathrm{Cl}({\mathcal{H}}_{\mathbb{R}},B)$. The fermionic Fock space is the spin representation of this Clifford algebra (Lawson-Michelsohn Ch. I ^{[Lawson-Michelsohn]}); the Pauli matrices on ${\mathbb{C}}^2$ generate the Clifford algebra ${\mathrm{Cl}}_2\cong M_2(\mathbb{C})$ corresponding to the spin-$1/2$ one-particle space.

• Hilbert space 02.11.08, inner product space 02.11.07. Substrate. The antisymmetric tensor power ${\Lambda }^n\mathcal{H}$ uses the determinant-of-Gram-matrix inner product, an inner-product-space construction; the fermionic Fock space is the ${\ell }^2$ direct sum of these, a separable Hilbert space; the fermionic bound and the CAR identities are bounded-operator statements that rely on Hilbert-space completeness throughout.

• Bounded linear operators 02.11.01. The fermionic creation and annihilation operators are bounded with $\Vert a^{\ast }(f)\Vert =\Vert a(f)\Vert =\Vert f\Vert$ — a definitive structural contrast with the bosonic case, where $a^{\ast }(f)$ is unbounded. The CAR algebra is therefore a $C^{\ast }$-algebra in its raw algebraic form, without exponentiation; the Weyl-form work-around of the bosonic case is unnecessary on the fermionic side. The norm equality $\Vert a^{\ast }(f)\Vert =\Vert f\Vert$ is the bound underlying every fermionic-many-body estimate.

• Lie algebra representation 07.06.01. Although the CAR algebra is a Clifford algebra (not a Lie algebra), the bilinear combinations $a^{\ast }(f)a(g)$ (one creation, one annihilation, equal in number) close under the commutator, giving a Lie algebra isomorphic to $\mathfrak{gl}(\mathcal{H})$ (or, for self-adjoint combinations, $\mathfrak{u}(\mathcal{H})$). The number operator $N$, the total angular momentum, and every conserved quantum-number operator are bilinear in $a^{\ast },a$ and live inside this Lie subalgebra — this is the framework for second-quantised one-particle dynamics lifting to fermionic Fock space.

• Stone-von Neumann theorem and the Weyl algebra (proposed successor 12.03.03). The fermionic-side uniqueness theorem (CAR analog of Stone-von Neumann) is sketched in Exercise 6 and developed in full in the dedicated successor unit, where the CCR and CAR uniqueness results are treated in parallel.

• Wightman axioms and the free scalar field (proposed successor 12.04.01). The fermionic version of the Wightman axioms requires the local commutator to be replaced by the local anticommutator for half-integer-spin fields, and the free Dirac field on the fermionic Fock space is the canonical example verifying the axioms — the input to the spin-statistics theorem and to the CPT theorem in the rigorous QFT framework.

• Dirac field (proposed successor 12.04.03). The free Dirac field $\psi (x),{\psi }^{†}(x)$ as an operator-valued tempered distribution on the fermionic Fock space over positive-energy spinor solutions of the Dirac equation is the half-integer-spin counterpart to the Klein-Gordon free field on the bosonic side; the dedicated unit develops the construction and the verification of the fermionic Wightman axioms.

• Multi-electron atoms (proposed cross-section 14.04.01). Atomic-structure physics — helium, lithium, the periodic table — is built directly on Slater determinants, the orthonormal basis of ${\Lambda }^n\mathcal{H}$ introduced here. The aufbau principle is the systematic application of the Pauli exclusion principle proved here: each one-particle orbital can be doubly occupied (with opposite spins), once spin is included as part of the one-particle label.

Historical & philosophical context [Master]

The Pauli exclusion principle was introduced by Wolfgang Pauli in 1925, in the paper Über den Zusammenhang des Abschlusses der Elektronengruppen im Atom mit der Komplexstruktur der Spektren ^{[Pauli 1925]}, appearing in Zeitschrift für Physik 31, 765. The paper proposed the Ausschließungsprinzip — exclusion principle — as a phenomenological rule to explain the regularities of atomic spectra and the closing of electron shells in the periodic table, with each one-particle quantum-number tuple admitting at most one electron. Pauli had no microscopic justification at the time; the rule was an empirical scaffolding to organise spectroscopic data and predict the noble-gas configurations.

The quantum-mechanical content emerged over the next three years. Dirac 1926 ^{[Dirac 1926]} introduced Fermi-Dirac statistics in Proc. Roy. Soc. A 112, 661, proving that the antisymmetry of the many-particle wavefunction was the formal incarnation of exclusion. Fermi 1926 ^{[Fermi 1926]} independently developed the same statistics in the ideal-gas context. Slater 1929 ^{[Slater 1929]} introduced the Slater determinant construction in Phys. Rev. 34, 1293, packaging the antisymmetric many-electron wavefunction as a determinant of single-particle orbitals — the calculational backbone of atomic and molecular physics ever since. Jordan-Wigner 1928 ^{[Jordan-Wigner 1928]} in Z. Phys. 47, 631 introduced the canonical anticommutation relations as the operator-algebra incarnation of fermionic statistics, and the Jordan-Wigner transformation linking the CAR algebra on a 1D lattice to spin-Pauli operators on a 1D chain — a transformation that remains central in modern condensed-matter physics and quantum-computing simulation of fermionic systems.

The connection between spin and statistics was originally a phenomenological observation: half-integer-spin particles (electrons, protons, neutrons) obey Fermi-Dirac statistics, integer-spin particles (photons, helium-4) obey Bose-Einstein statistics. The mathematical underpinning came from Fierz 1939 ^{[Fierz 1939]} and Pauli 1940 ^{[Pauli 1940]} in Phys. Rev. 58, 716, where Pauli proved the spin-statistics theorem for free fields: in any Lorentz-invariant local quantum field theory, integer-spin fields must use canonical commutation relations and half-integer-spin fields must use canonical anticommutation relations, with the "wrong" choice giving a non-positive-definite Hilbert space inner product. The rigorous extension to all Wightman-axiomatic quantum field theories was given by Burgoyne 1958 and Streater-Wightman 1964 ^{[Streater-Wightman 1964]}, establishing spin-statistics as one of the load-bearing structural theorems of relativistic QFT.

The post-war development of the fermionic Fock space tracked the bosonic case: Cook 1953 supplied the bosonic rigorous framework, and the parallel fermionic framework was developed by Segal, Berezin, and others through the 1960s. Berezin's Method of Second Quantization ^{[Berezin 1966]} introduced the Grassmann-variable / Berezin-integral formalism for fermionic path integrals, the calculational tool that has dominated fermionic QFT ever since. Bratteli-Robinson Vol. II ^{[Bratteli-Robinson Vol. II]} in the 1980s packaged the CAR algebra as the workhorse $C^{\ast }$-algebra of fermionic quantum statistical mechanics, developing KMS states, quasi-free states, Bogoliubov-Shale-Stinespring criteria, and the BCS pairing mechanism inside this framework.

Philosophically, the bosonic / fermionic distinction is one of the few absolute dichotomies in nature. There is no continuous interpolation between Bose-Einstein and Fermi-Dirac statistics in three spatial dimensions; particles are either bosons or fermions, and the spin-statistics theorem ties the distinction to the relativistic kinematics of half-integer vs integer spin. In two spatial dimensions the rigid dichotomy gives way to a continuum of anyonic statistics (Wilczek 1982), with arbitrary phases $e^{i\theta }$ under exchange — the foundation of the fractional quantum Hall effect and topological quantum computing. The fermionic Fock space and the Pauli exclusion principle are also the foundational input for the stability of matter programme (Dyson-Lenard 1967; Lieb-Thirring 1975 ^{[Lieb-Thirring 1975]}; Lieb-Seiringer monograph ^{[Lieb-Seiringer 2010]}), which proves that the total ground-state energy of a system of $N$ electrons and nuclei is bounded below by $-cN$ for a universal constant $c$ — without fermionic antisymmetry the bound fails and ordinary matter is unstable. Exclusion is the structural reason matter does not collapse.

Bibliography [Master]

Primary literature (cite when used; not all currently in reference/):

• Pauli, W., Z. Phys. 31, 765 (1925). Über den Zusammenhang des Abschlusses der Elektronengruppen im Atom mit der Komplexstruktur der Spektren — originator paper for the Pauli exclusion principle. [Need to source.]
• Dirac, P. A. M., Proc. Roy. Soc. A 112, 661 (1926). On the theory of quantum mechanics — Fermi-Dirac statistics. [Need to source.]
• Fermi, E., Rend. Lincei 3, 145 (1926); Z. Phys. 36, 902 (1926). Fermi statistics for the ideal gas. [Need to source.]
• Jordan, P. & Wigner, E., Z. Phys. 47, 631 (1928). Canonical anticommutation relations and the Jordan-Wigner transformation. [Need to source.]
• Slater, J. C., Phys. Rev. 34, 1293 (1929). The Slater determinant. [Need to source.]
• Fierz, M., Helv. Phys. Acta 12, 3 (1939). Precursor to the spin-statistics theorem. [Need to source.]
• Pauli, W., Phys. Rev. 58, 716 (1940). The spin-statistics theorem (first published form). [Need to source.]
• Lieb, E. H. & Thirring, W., Phys. Rev. Lett. 35, 687 (1975); erratum 35, 1116. The Lieb-Thirring inequality for fermionic kinetic energy. [Need to source.]

Canonical monographs (math-physics lineage):

• Streater, R. F. & Wightman, A. S., PCT, Spin and Statistics, and All That (Benjamin, 1964; Princeton Landmarks reprint, 2000). Ch. 4 — spin-statistics theorem. [Need to source — shared with 12.03.01.]
• Reed, M. & Simon, B., Methods of Modern Mathematical Physics, Vol. II: Fourier Analysis, Self-Adjointness (Academic Press, 1975). §X.7 — fermionic Fock space. [Need to source — shared with 12.03.01.]
• Bratteli, O. & Robinson, D. W., Operator Algebras and Quantum Statistical Mechanics, Vol. II, 2nd ed. (Springer, 1997). §5.2 — CAR-algebra. [Need to source — shared with 12.03.01.]
• Glimm, J. & Jaffe, A., Quantum Physics: A Functional Integral Point of View, 2nd ed. (Springer, 1987). Ch. 6 — Dirac field on fermionic Fock space. [Need to source — shared with 12.03.01.]
• Lawson, H. B. & Michelsohn, M.-L., Spin Geometry (Princeton, 1989). Ch. I — Clifford algebras and the CAR. [Need to source.]
• Berezin, F. A., The Method of Second Quantization (Academic Press, 1966). Part II — fermionic case; Grassmann variables. [Need to source.]

Modern textbooks (mathematics-side and physics-side):

• Folland, G. B., Quantum Field Theory: A Tourist Guide for Mathematicians (AMS Math. Surveys 149, 2008). Ch. 4 — antisymmetric Fock space. [Need to source — shared with 12.03.01.]
• Hall, B. C., Quantum Theory for Mathematicians (Springer GTM 267, 2013). Ch. 19 — identical particles, Slater determinants. [Need to source — shared with 12.03.01.]
• Chatterjee, S., Introduction to Quantum Field Theory for Mathematicians (Stanford lecture notes, 2022). Lecture 2 (fermionic Fock space); Lecture 5 (Dirac field). [Need to source — shared with 12.03.01.]
• Sakurai, J. J. & Napolitano, J., Modern Quantum Mechanics, 2nd ed. (Pearson/Cambridge, 2011). Ch. 7 — identical particles, second quantisation. [Need to source — shared with 12.03.01.]
• Griffiths, D. J., Introduction to Quantum Mechanics, 2nd ed. (Pearson, 2005). Ch. 5 §5.2 — identical particles, helium Slater determinant. [Need to source — shared with 12.03.01.]
• Peskin, M. E. & Schroeder, D. V., An Introduction to Quantum Field Theory (Westview, 1995). Ch. 3 §3.5 — Dirac-field quantisation. [Need to source.]
• Tong, D., Quantum Field Theory, DAMTP Cambridge lecture notes (raw/pdfs/qft/qft.pdf). §4-§5 — Dirac equation and Dirac-field quantisation. [Have.]
• Lieb, E. H. & Seiringer, R., The Stability of Matter in Quantum Mechanics (Cambridge, 2010). Chs. 3, 7 — Pauli exclusion and stability of matter. [Need to source.]

Reference / extended:

• Susskind, L. & Friedman, A., Quantum Mechanics: The Theoretical Minimum (Basic Books, 2014). Lecture 7 (entanglement) and Lecture 10. [Need to source — shared with 12.03.01.]

---

Section 2 third production unit, sibling to 12.03.01 (bosonic Fock space, Cycle 10). Builds on 12.02.01 (Hilbert-space formalism), 12.01.02 (Stern-Gerlach, chapter seed), 03.09.02 (Clifford algebra), and the direct bosonic sibling. Sets up successor units on the CAR / CCR uniqueness comparison (12.03.03), the fermionic Wightman axioms (12.04.01), the free Dirac field (12.04.03), and the cross-section atomic-structure handoff (14.04.01). Citations marked TODO_REF indicate sources expected in reference/ but not yet acquired; replace with [ref: ...] when sources arrive.