Introduction
Constructive quantum field theory was originally shaped by the idea that Euclidean fields should not merely be described in terms of functional integrals, but as probabilistic objects. This has the merit of making it possible to establish criteria for a rigorous definition of Wick rotation, the key tool to reconstruct a corresponding Lorentzian theory. In the Bosonic case this viewpoint is embodied by the Osterwalder–Schrader axioms [13 ] and by Nelson’s construction of quantum fields from Markov fields [11 ] , see also [16 , 8 ] . In this last approach one starts from an Euclidean random field rather than from the canonical commutation relations. This allows to identify a distinguished time-zero algebra by means of conditional expectations recovering in addition the Hamiltonian dynamics from the Euclidean time-translation semigroup.
This programme has acquired renewed momentum in recent years thanks to the interaction between constructive quantum field theory and singular stochastic partial differential equations, which are revolving mainly around regularity structures and paracontrolled calculus [10 , 9 ] . In turn, establishing these new frameworks has allowed to put on firm mathematical grounds the Parisi–Wu idea of stochastic quantization [14 ] , which allows for the construction and analysis of Euclidean models beyond the perturbative level. Yet, most of the advances mentioned above concern Bosonic theories, while Fermionic models are less studied. The main structural difference lies in the anti-commuting nature of the underlying variables. This prevents a direct realization of Fermionic Euclidean fields as random variables on an ordinary commutative probability space, as already stressed in the analysis of Osterwalder and Schrader [12 ] . Consequently, the probabilistic side of the Nelson programme has no immediate Fermionic analogue.
A way around this obstruction has recently emerged in [1 , 5 ] . Herein the idea consists of replacing the commutative probability space by a non-commutative counterpart. This is a unital C ∗ C^* C ∗ -algebra endowed with a state, that is a positive and normalized functional thereon. Grassmann random variables are therefore regarded as algebra homomorphisms from an exterior algebra into this non-commutative probability space. In this setting one can define key probabilistic notions ranging from a Grassmannian white noise to conditional expectations. These allow to devise a natural stochastic language to encode anti-commuting variables, that is a framework in which Nelson’s construction can be revisited for Fermionic fields. On the one hand, it is worth mentioning that this non-commutative setting has already proved useful in the study of weakly coupled subcritical Euclidean Fermionic field theories [6 ] . On the other hand, it suggests that this might be the right framework to analyze interacting two-dimensional models, among which the Thirring model is a natural test case, see [3 ] for a recent analysis in the framework of singular stochastic partial differential equations.
The purpose of the present paper is to carry out part of this programme, focusing on the non-interacting, massive Euclidean Dirac field in two dimensions. More precisely, we construct a Euclidean free Fermi field starting from a Grassmannian white noise on a non-commutative probability space. This reconstruction may be viewed as a non-commutative fermionic analogue of Nelson’s construction from Markov fields [11 ] . Here, the Markov structure comes from an auxiliary stationary Grassmannian Ornstein–Uhlenbeck field φ \varphi φ and its driving white noise. An algebra-valued Euclidean Fermi field ψ \psi ψ , whose mixed two-point functions reproduce the Euclidean Dirac fundamental solution, is derived from φ \varphi φ by applying a differential operator, local in time and space, to its second component. Hence, we do not require ψ \psi ψ itself to be Markovian. Subsequently, we exploit the Markovianity of the auxiliary field and the associated conditional expectation to show that the resulting evolution coincides with the second quantization of the one-particle semigroup. A Markov property has been established also in [7 ] in their gauge-space construction, where the relevant algebra is generated by Euclidean fermionic gauge fields. Their proposal to introduce interactions through a normal central change of state encounters an obstruction. Our noise-based construction offers instead a different possible starting point for interacting models. We restrict the present analysis to the free theory so as to isolate the Markovian construction from renormalization issues. The main results can be summarized as follows:
We construct a Grassmannian white noise and the associated Grassmannian Brownian motion on a non-commutative probability space. The covariance is specified by a bounded operator whose associated two-point function is an antisymmetric bilinear form, as required by the fermionic nature of the variables. The resulting process satisfies the corresponding Gaussian Wick rule and has independent increments.
We define a stationary Grassmannian Ornstein–Uhlenbeck field and, from it, a two-dimensional Euclidean free Fermi field. Its only non-vanishing two-point functions are the mixed ones. These coincide with the integral kernel of the Euclidean Dirac propagator.
We construct zero-time Fermionic fields by conditional expectation onto the algebra generated by the past noise. Their covariance is the reflected Euclidean Dirac kernel and the conditional expectation is compatible with Wick ordering. In other words, Euclidean Wick polynomials are mapped to the corresponding Wick polynomials of the zero-time fields.
We show that Euclidean time translations induce a strongly continuous semigroup on the one-particle zero-time space. Its generator is the positive one-particle Hamiltonian A x = ( − ∂ x 2 + m 2 ) 1 / 2 A_x = (-\partial_x^2 + m^2)^{1/2} A x = ( − ∂ x 2 + m 2 ) 1/2 . On Wick polynomials the induced generator is the corresponding second quantization.
The paper is organized as follows. In the first part of Section 1 we fix the analytic conventions for two-dimensional Euclidean Dirac fields and recall succinctly the Osterwalder–Schrader framework. In Section 2 we introduce the non-commutative probability space, Grassmann random variables, Gaussian covariances and conditional expectations. In Section 3 we discuss the construction of the Grassmannian white noise, the Ornstein–Uhlenbeck field, the Euclidean free Fermi field, the time-zero fields and the Euclidean time-evolution semigroup. The appendix contains the proof of the compatibility between conditional expectation and Wick ordering.
Analytic Structures
In this section we fix the notation and conventions used throughout. We take as background the two-dimensional Euclidean space ( R 2 , δ ) (\mathbb{R}^{2},\delta) ( R 2 , δ ) , with signature ( + , + ) (+,+) ( + , + ) and Cartesian coordinates ( t , x ) (t,x) ( t , x ) . We shall refer to t t t as Euclidean time, in contrast to its Lorentzian counterpart t M t^{M} t M , related to it by the Wick rotation
To simplify the notation we shall denote a generic point of the Euclidean space by x ≡ ( t , x ) x\equiv(t,x) x ≡ ( t , x ) and its Lorentzian counterpart by x M ≡ ( t M , x ) x^{M}\equiv(t^{M},x) x M ≡ ( t M , x ) .
Function spaces. On R 2 \mathbb{R}^{2} R 2 we denote by S ( R 2 ) \mathcal{S}(\mathbb{R}^{2}) S ( R 2 ) the space of rapidly decreasing smooth functions, endowed with its standard locally convex Fréchet topology. Its topological dual, the space of tempered distributions, is denoted by S ′ ( R 2 ) \mathcal{S}'(\mathbb{R}^{2}) S ′ ( R 2 ) . Given τ ∈ R \tau\in\mathbb{R} τ ∈ R , we introduce the subspaces
S τ − ( R 2 ) : = { v ∈ S ( R 2 ) ∣ v ( t , x ) = 0 , for t > τ } , S τ + ( R 2 ) : = { v ∈ S ( R 2 ) ∣ v ( t , x ) = 0 , for t < τ } . (1) \mathcal{S}_{\tau}^{-}(\mathbb{R}^{2}):=\{v\in\mathcal{S}(\mathbb{R}^{2})\mid v(t,x)=0,\text{ for }t>\tau\},\qquad\mathcal{S}_{\tau}^{+}(\mathbb{R}^{2}):=\{v\in\mathcal{S}(\mathbb{R}^{2})\mid v(t,x)=0,\text{ for }t<\tau\}. \tag*{(1)} S τ − ( R 2 ) := { v ∈ S ( R 2 ) ∣ v ( t , x ) = 0 , for t > τ } , S τ + ( R 2 ) := { v ∈ S ( R 2 ) ∣ v ( t , x ) = 0 , for t < τ } . (1)
Equivalently, the elements of S τ + ( R 2 ) \mathcal{S}_{\tau}^{+}(\mathbb{R}^{2}) S τ + ( R 2 ) (resp. S τ − ( R 2 ) \mathcal{S}_{\tau}^{-}(\mathbb{R}^{2}) S τ − ( R 2 ) ) are Schwartz functions supported in the closed half-space { t ≥ τ } \{t\geq\tau\} { t ≥ τ } (resp. { t ≤ τ } \{t\leq\tau\} { t ≤ τ } ). Since the Fourier transform will play an important role, we fix the convention
F : S ( R 2 ) ⟶ S ( R 2 ) , v ( t , x ) ⟼ v ^ ( ω , k ) = 1 2 π ∫ R 2 d t d x v ( t , x ) e − i ( ω t + k x ) . \mathcal{F}:\mathcal{S}(\mathbb{R}^{2})\longrightarrow\mathcal{S}(\mathbb{R}^{2}),\qquad v(t,x)\longmapsto\widehat{v}(\omega,k)=\frac{1}{2\pi}\int_{\mathbb{R}^{2}}\mathrm{d}t\mathrm{d}x\,v(t,x)e^{-i(\omega t+kx)}. F : S ( R 2 ) ⟶ S ( R 2 ) , v ( t , x ) ⟼ v ( ω , k ) = 2 π 1 ∫ R 2 d t d x v ( t , x ) e − i ( ω t + k x ) .
Fourier transforms with respect to only one of the two variables are defined analogously.
Euclidean Dirac Operator – We consider distributional Euclidean spinors and cospinors, namely elements respectively of
S ′ ( R 2 ; C 2 ) , S ′ ( R 2 ; ( C 2 ) ∗ ) . \mathcal{S}'(\mathbb{R}^{2};\mathbb{C}^{2}),\qquad\mathcal{S}'(\mathbb{R}^{2};(\mathbb{C}^{2})^{*}). S ′ ( R 2 ; C 2 ) , S ′ ( R 2 ; ( C 2 ) ∗ ) .
Throughout the paper, using the standard Hermitian pairing, we shall freely identify C 2 \mathbb{C}^{2} C 2 and ( C 2 ) ∗ (\mathbb{C}^{2})^{*} ( C 2 ) ∗ whenever necessary and if no confusion can arise. The Euclidean Dirac operator and its formal adjoint are
D E = γ E μ ∂ μ + m : S ′ ( R 2 ; C 2 ) ⟶ S ′ ( R 2 ; C 2 ) , D E ∗ = − γ E μ ∂ μ + m : S ′ ( R 2 ; C 2 ) ⟶ S ′ ( R 2 ; C 2 ) . (2) D_{E}=\gamma_{E}^{\mu}\partial_{\mu}+m:\mathcal{S}'(\mathbb{R}^{2};\mathbb{C}^{2})\longrightarrow\mathcal{S}'(\mathbb{R}^{2};\mathbb{C}^{2}),\qquad D_{E}^{*}=-\gamma_{E}^{\mu}\partial_{\mu}+m:\mathcal{S}'(\mathbb{R}^{2};\mathbb{C}^{2})\longrightarrow\mathcal{S}'(\mathbb{R}^{2};\mathbb{C}^{2}). \tag*{(2)} D E = γ E μ ∂ μ + m : S ′ ( R 2 ; C 2 ) ⟶ S ′ ( R 2 ; C 2 ) , D E ∗ = − γ E μ ∂ μ + m : S ′ ( R 2 ; C 2 ) ⟶ S ′ ( R 2 ; C 2 ) . (2)
Here m ≥ 0 m\geq0 m ≥ 0 is a mass parameter, while γ E μ ∈ M ( 2 ; C ) \gamma_{E}^{\mu}\in\mathcal{M}(2;\mathbb{C}) γ E μ ∈ M ( 2 ; C ) , μ = 0 , 1 \mu=0,1 μ = 0 , 1 , are Euclidean gamma matrices. They are Hermitian and satisfy the Clifford relations
{ γ E μ , γ E ν } = 2 δ μ ν I 2 , μ , ν = 0 , 1. \{\gamma_{E}^{\mu},\gamma_{E}^{\nu}\}=2\delta^{\mu\nu}\mathbb{I}_{2},\qquad\mu,\nu=0,1. { γ E μ , γ E ν } = 2 δ μν I 2 , μ , ν = 0 , 1.
A convenient choice is given by the Pauli matrices
γ E 0 = σ 1 = ( 0 1 1 0 ) , γ E 1 = σ 2 = ( 0 − i i 0 ) . \gamma_{E}^{0}=\sigma^{1}=\begin{pmatrix}0&1\\1&0\end{pmatrix},\qquad\gamma_{E}^{1}=\sigma^{2}=\begin{pmatrix}0&-i\\i&0\end{pmatrix}. γ E 0 = σ 1 = ( 0 1 1 0 ) , γ E 1 = σ 2 = ( 0 i − i 0 ) .
For later use, we recall the identities
D E D E ∗ = ( − Δ + m 2 ) I 2 , D E ∗ D E = ( − Δ + m 2 ) I 2 . \begin{aligned}
D_{E}D_{E}^{*}&=(-\Delta+m^{2})\mathbb{I}_{2},\\
D_{E}^{*}D_{E}&=(-\Delta+m^{2})\mathbb{I}_{2}.
\end{aligned} D E D E ∗ D E ∗ D E = ( − Δ + m 2 ) I 2 , = ( − Δ + m 2 ) I 2 .
where Δ : = ∂ t 2 + ∂ x 2 \Delta:=\partial_{t}^{2}+\partial_{x}^{2} Δ := ∂ t 2 + ∂ x 2 is the Euclidean Laplacian on R 2 \mathbb{R}^{2} R 2 , while I 2 \mathbb{I}_{2} I 2 denotes the 2 × 2 2\times2 2 × 2 identity matrix. For m > 0 m>0 m > 0 , we denote by G ∈ S ′ ( R 2 ) G\in\mathcal{S}'(\mathbb{R}^{2}) G ∈ S ′ ( R 2 ) the translation-invariant fundamental solution of − Δ + m 2 -\Delta+m^{2} − Δ + m 2 which decays at infinity. Its integral kernel reads
G ( t , x ) = 1 2 π K 0 ( m t 2 + x 2 ) , G(t,x)=\frac{1}{2\pi}K_{0}\left(m\sqrt{t^{2}+x^{2}}\right), G ( t , x ) = 2 π 1 K 0 ( m t 2 + x 2 ) ,
where K 0 K_{0} K 0 is the modified Bessel function of second kind of order 0. Equivalently, in Fourier space, we have
G ^ ( ω , k ) = 1 2 π 1 ω 2 + k 2 + m 2 . \widehat{G}(\omega,k)=\frac{1}{2\pi}\frac{1}{\omega^{2}+k^{2}+m^{2}}. G ( ω , k ) = 2 π 1 ω 2 + k 2 + m 2 1 .
Observe that, in the massless case m = 0 m=0 m = 0 , no fundamental solution vanishing at infinity exists in two dimensions and one must instead choose an infrared prescription. We avoid entering into the details leaving this choice understood. Consequently, the Euclidean Dirac fundamental solution is the matrix-valued tempered distribution
S E : = D E ∗ ( G I 2 ) ∈ S ′ ( R 2 ; End ( C 2 ) ) . (3) S_{E}:=D_{E}^{*}(G\mathbb{I}_{2})\in\mathcal{S}'(\mathbb{R}^{2};\operatorname{End}(\mathbb{C}^{2})). \tag*{(3)} S E := D E ∗ ( G I 2 ) ∈ S ′ ( R 2 ; End ( C 2 )) . (3)
whose expression in Fourier space is
S ^ E ( ω , k ) = 1 2 π − i γ E 0 ω − i γ E 1 k + m I 2 ω 2 + k 2 + m 2 . \widehat{S}_{E}(\omega,k)=\frac{1}{2\pi}\frac{-i\gamma_{E}^{0}\omega-i\gamma_{E}^{1}k+m\mathbb{I}_{2}}{\omega^{2}+k^{2}+m^{2}}. S E ( ω , k ) = 2 π 1 ω 2 + k 2 + m 2 − i γ E 0 ω − i γ E 1 k + m I 2 .
Observe that this expression is valid also for m = 0 m=0 m = 0 since, in this case, S E S_{E} S E is locally integrable in a neighbourhood of ( ω , k ) = ( 0 , 0 ) (\omega,k)=(0,0) ( ω , k ) = ( 0 , 0 ) .
Osterwalder–Schrader Distributions and Reconstruction
In this subsection we recall the Osterwalder–Schrader framework for a two-dimensional Bosonic scalar theory. We follow mainly [13 ] and [16 ] . We omit the proofs, in particular of the reconstruction theorem, since they can be found in the mentioned references. Actually our main goal is to fix the functional-analytic setting and to state the class of Euclidean correlation functions to which the theorem applies, as a preparation for the discussion of the Fermionic counterpart in § 3 . We identify the large diagonal
Δ n : = { ( x 1 , … , x n ) ∈ R 2 n ∣ x i = x j for some i ≠ j } . \Delta_n := \{(x_1,\ldots,x_n) \in\mathbb{R}^{2n} \mid x_i = x_j \text{ for some } i \ne j\}. Δ n := {( x 1 , … , x n ) ∈ R 2 n ∣ x i = x j for some i = j } .
At the same time we set
S n : = S ( R 2 n ) and 0 S n : = { u ∈ S n ∣ ∂ α u ∣ Δ n = 0 , for every multiindex α } , \mathcal{S}_n := \mathcal{S}(\mathbb{R}^{2n}) \quad\text{and} \quad{}^{0}\mathcal{S}_n := \{u \in\mathcal{S}_n \mid\partial^\alpha u|_{\Delta_n} = 0,\text{ for every multiindex } \alpha\}, S n := S ( R 2 n ) and 0 S n := { u ∈ S n ∣ ∂ α u ∣ Δ n = 0 , for every multiindex α } ,
where 0 S n {}^{0}\mathcal{S}_n 0 S n is a closed subspace of S n \mathcal{S}_n S n , see [13 ] .
Remark 1 These spaces come accompanied with their topological duals S n ′ ≡ S ′ ( R 2 n ) \mathcal{S}'_n \equiv\mathcal{S}'(\mathbb{R}^{2n}) S n ′ ≡ S ′ ( R 2 n ) and ( 0 S n ) ′ ({}^{0}\mathcal{S}_n)' ( 0 S n ) ′ . Although in [13 ] the latter is considered, in this work we shall assume from the outset that the Schwinger functions are represented by tempered distributions S n ∈ S n ′ S_n \in\mathcal{S}'_n S n ∈ S n ′ satisfying the Osterwalder–Schrader axioms. If one starts instead from an element lying in ( 0 S n ) ′ ({}^{0}\mathcal{S}_n)' ( 0 S n ) ′ , an extension to S n ′ \mathcal{S}'_n S n ′ exists by the Hahn–Banach theorem. Yet, such extension is nonunique and it might not preserve the structural properties encoded by the Osterwalder–Schrader axioms unless these are imposed. Nonetheless, the ambiguity is supported on the large diagonal Δ n \Delta_n Δ n and, once a compatible extension has been fixed, this does not affect the reconstruction.
Two additional ingredients are needed:
the Euclidean time reflection
ϑ : R 2 → R 2 , x = ( t , x ) ↦ ϑ x = ( − t , x ) , (4) \vartheta: \mathbb{R}^2 \to\mathbb{R}^2,\qquad x=(t,x) \mapsto\vartheta x=(-t,x), \tag*{(4)} ϑ : R 2 → R 2 , x = ( t , x ) ↦ ϑ x = ( − t , x ) , (4)
which extends to S n \mathcal{S}_n S n as the map
θ n : S n → S n , u ( x 1 , … , x n ) ↦ ( θ n u ) ( x 1 , … , x n ) = u ( ϑ ( x n ) , … , ϑ ( x 1 ) ) . \theta_n : \mathcal{S}_n \to\mathcal{S}_n,\qquad u(x_1,\ldots,x_n) \mapsto(\theta_n u)(x_1,\ldots,x_n)=u(\vartheta(x_n),\ldots,\vartheta(x_1)). θ n : S n → S n , u ( x 1 , … , x n ) ↦ ( θ n u ) ( x 1 , … , x n ) = u ( ϑ ( x n ) , … , ϑ ( x 1 )) .
the positive-time test functions . Given R + 2 : = [ 0 , ∞ ) × R ∋ ( t , x ) \mathbb{R}^2_+ := [0,\infty) \times\mathbb{R} \ni(t,x) R + 2 := [ 0 , ∞ ) × R ∋ ( t , x ) , we call
S ( R + 2 ) : = S 0 + ( R 2 ) = { f ∈ S ( R 2 ) : supp f ⊂ [ 0 , ∞ ) × R } , (5) \mathcal{S}(\mathbb{R}^2_+) := \mathcal{S}^{+}_0(\mathbb{R}^2) = \{f \in\mathcal{S}(\mathbb{R}^2) : \operatorname{supp} f \subset[0,\infty) \times\mathbb{R}\}, \tag*{(5)} S ( R + 2 ) := S 0 + ( R 2 ) = { f ∈ S ( R 2 ) : supp f ⊂ [ 0 , ∞ ) × R } , (5)
where S 0 + ( R 2 ) \mathcal{S}^{+}_0(\mathbb{R}^2) S 0 + ( R 2 ) is defined as per Equation (1) . For any n ≥ 1 n \ge1 n ≥ 1 , we set
S + , n : = ⨂ j = 1 n S ( R + 2 ) , \mathcal{S}_{+,n} := \bigotimes_{j=1}^{n} \mathcal{S}(\mathbb{R}^2_+), S + , n := j = 1 ⨂ n S ( R + 2 ) ,
together with
S : = C ⊕ ⨁ n ≥ 1 a l g S n , S + : = C ⊕ ⨁ n ≥ 1 a l g S + , n . \mathcal{S} := \mathbb{C} \oplus\bigoplus_{n\ge1}^{\mathrm{alg}} \mathcal{S}_n,\qquad
\mathcal{S}_+ := \mathbb{C} \oplus\bigoplus_{n\ge1}^{\mathrm{alg}} \mathcal{S}_{+,n}. S := C ⊕ n ≥ 1 ⨁ alg S n , S + := C ⊕ n ≥ 1 ⨁ alg S + , n .
The superscript a l g \mathrm{alg} alg denotes the algebraic direct sum, namely only sequences with finitely many nonzero components are admitted.
Definition 1 (Bosonic Osterwalder–Schrader distributions) A collection
( S n ) n ∈ N 0 , S 0 = 1 , S n ∈ S n ′ , (S_n)_{n\in\mathbb{N}_0},\qquad S_0=1,\qquad S_n\in\mathcal{S}'_n, ( S n ) n ∈ N 0 , S 0 = 1 , S n ∈ S n ′ , is called a family of Bosonic Osterwalder–Schrader distributions if
(OS1) Euclidean covariance – For every element of the proper Euclidean isometry group ( a , R ) ∈ E ( 2 ) + = R 2 ⋊ SO ( 2 ) (a,R)\in E(2)^+ = \mathbb{R}^2 \rtimes\operatorname{SO}(2) ( a , R ) ∈ E ( 2 ) + = R 2 ⋊ SO ( 2 ) , it holds
S n ( u ) = S n ( u ( a , R ) ) , u ∈ S n , S_n(u)=S_n(u_{(a,R)}),\qquad u\in\mathcal{S}_n, S n ( u ) = S n ( u ( a , R ) ) , u ∈ S n , where
u ( a , R ) ( x 1 , … , x n ) : = u ( R − 1 ( x 1 − a ) , … , R − 1 ( x n − a ) ) . u_{(a,R)}(x_1,\ldots,x_n):=u(R^{-1}(x_1-a),\ldots,R^{-1}(x_n-a)). u ( a , R ) ( x 1 , … , x n ) := u ( R − 1 ( x 1 − a ) , … , R − 1 ( x n − a )) . (OS2) Reflection positivity – For every finite sequence
( u 0 , u 1 , … , u N ) ∈ S + , (u_0,u_1,\ldots,u_N)\in\mathcal{S}_+, ( u 0 , u 1 , … , u N ) ∈ S + , one has
∑ n , m = 0 N S n + m ( θ n u n ⊗ u m ) ≥ 0. \sum_{n,m=0}^{N} S_{n+m}(\theta_n u_n \otimes u_m)\ge0. n , m = 0 ∑ N S n + m ( θ n u n ⊗ u m ) ≥ 0. The terms with n = 0 n=0 n = 0 or m = 0 m=0 m = 0 are understood in the natural way, employing $S_0=1.
(OS3) Symmetry. Denoting by S n \mathfrak{S}_n S n , n ≥ 1 n \ge1 n ≥ 1 , the permutation group of $n elements, for every π ∈ S n \pi\in\mathfrak{S}_n π ∈ S n ,
S n ( u π ) = S n ( u ) , u ∈ S n , S_n(u^\pi)=S_n(u), \qquad u \in\mathcal{S}_n, S n ( u π ) = S n ( u ) , u ∈ S n ,
where
u π ( x 1 , … , x n ) : = u ( x π ( 1 ) , … , x π ( n ) ) . u^\pi(x_1,\ldots,x_n):=u(x_{\pi(1)},\ldots,x_{\pi(n)}). u π ( x 1 , … , x n ) := u ( x π ( 1 ) , … , x π ( n ) ) .
(OS4) Cluster property – Given a ≡ ( a , 0 ) ∈ E ( 2 ) + a \equiv(a,0) \in E(2)+ a ≡ ( a , 0 ) ∈ E ( 2 ) + a non-vanishing spatial translation, if T λ a T_{\lambda a} T λa denotes the simultaneous translation of all arguments of a test function by λ a \lambda a λa , then
lim λ → + ∞ S n + m ( u ⊗ T λ a u ′ ) = S n ( u ) S m ( u ′ ) , \lim_{\lambda\to+\infty}S_{n+m}(u\otimes T_{\lambda a}u')=S_n(u)S_m(u'), λ → + ∞ lim S n + m ( u ⊗ T λa u ′ ) = S n ( u ) S m ( u ′ ) ,
for all test functions u ∈ S n u \in\mathcal{S}_n u ∈ S n and u ′ ∈ S m u' \in\mathcal{S}_m u ′ ∈ S m .
Remark 2. The relevance of the preceding definition is fully justified by the celebrated Osterwalder-Schrader reconstruction theorem [13 ] . We avoid stating it since it would require a long detour on the formulation of the Wightman axioms for a quantum field theory on Minkowski spacetime [17 ] . We limit ourselves to observing that, given a family ( S n ) n ∈ N 0 (S_n)_{n\in\mathbb{N}_0} ( S n ) n ∈ N 0 of Bosonic Osterwalder–Schrader distributions as per Definition 1, then there exists a unique sequence ( W n ) n ∈ N 0 (W_n)_{n\in\mathbb{N}_0} ( W n ) n ∈ N 0 , W n ∈ S ′ ( R 2 n ) W_n \in\mathcal{S}'(\mathbb{R^{2n}}) W n ∈ S ′ ( R 2 n ) , known as Wightman distributions such that, in addition to a set of structural properties, they are obtained from ( S n ) n ∈ N 0 (S_n)_{n\in\mathbb{N}_0} ( S n ) n ∈ N 0 as boundary values of their analytic continuation in the Euclidean time variables. Hence, up to unitary equivalence, they identify a Hilbert space H \mathcal{H} H , a normalized vector Ω ∈ H \Omega\in\mathcal{H} Ω ∈ H , a unitary representation of the proper orthochronous Poincaré group satisfying the spectrum condition and an operator valued distribution ϕ \phi ϕ such that
W n ( x 1 M , … , x n M ) = ⟨ Ω , ϕ ( x 1 M ) ⋯ ϕ ( x n M ) Ω ⟩ . W_n(x^M_1,\ldots,x^M_n)=\langle\Omega,\phi(x^M_1)\cdots\phi(x^M_n)\Omega\rangle. W n ( x 1 M , … , x n M ) = ⟨ Ω , ϕ ( x 1 M ) ⋯ ϕ ( x n M ) Ω ⟩ .
We conclude this short section remarking that the Osterwalder-Schrader approach is well understood for Bosonic fields also in view of the work by Nelson [11 ] who established a consistent method for defining Bosonic Euclidean fields as random variables on a suitable probability space. However, the scenario for Fermionic fields is more delicate. A notable attempt to address this issue can be ascribed to Osterwalder and Schrader [12 ] , who investigated whether Fermionic correlation functions could be derived from a genuine underlying probability measure. Their analysis highlighted a fundamental obstruction, namely the anti-commutativity of Grassmann variables prevents the definition of positive-definite densities. Consequently, unlike the Bosonic case, Fermionic Euclidean fields lack a realization in terms of Nelson’s stochastic framework and we shall address this issue in the following sections.
Non-commutative probability space and Grassmann random variables
In this section we introduce Grassmann random variables and we start by assigning the underlying building blocks:
a non-commutative probability space as introduced in [1 ] , that is a pair ( A , ω ) (\mathcal{A},\omega) ( A , ω ) , where A \mathcal{A} A is a unital C ∗ C^* C ∗ -algebra while ω : A → C \omega:\mathcal{A}\to\mathbb{C} ω : A → C is a linear, normalized, positive functional, known as a state,
a complex, separable Hilbert space ( V , ( ⋅ , ⋅ ) V ) (V,(\cdot,\cdot)_V) ( V , ( ⋅ , ⋅ ) V ) , which we assume anti-linear in the left variable.
Definition 2 Denoting by Λ V \Lambda V Λ V the exterior algebra associated to V V V , we denote by G ( V ) \mathcal{G}(V) G ( V ) the collection of all Grassmann random variables, namely unital algebra homomorphisms X : Λ V → A X:\Lambda V\to\mathcal{A} X : Λ V → A . Given X ∈ G ( V ) X\in\mathcal{G}(V) X ∈ G ( V ) its law is the family of moments
ω X ( F ) : = ω ( X ( F ) ) , ∀ F ∈ Λ V . \omega^X(F):=\omega(X(F)),\qquad\forall F\in\Lambda V. ω X ( F ) := ω ( X ( F )) , ∀ F ∈ Λ V . Furthermore we say that X ∈ G ( V ) X\in\mathcal{G}(V) X ∈ G ( V ) is continuous if
∥ X ∥ G ( V ) : = sup ∥ v ∥ V = 1 ∥ X ( v ) ∥ A < ∞ . \|X\|_{\mathcal{G}(V)}:=\sup_{\|v\|_V=1}\|X(v)\|_{\mathcal{A}}<\infty. ∥ X ∥ G ( V ) := ∥ v ∥ V = 1 sup ∥ X ( v ) ∥ A < ∞.
The collection of all such random variables is denoted by G − ∞ ( V ) \mathcal{G}^{-\infty}(V) G − ∞ ( V ) .
Observe that one cannot expect G ( V ) \mathcal{G}(V) G ( V ) to be complete, unless V V V is finite dimensional [1 Lemma 7] . The next step consists of establishing a notion of compatibility.
Definition 3 (Compatible Grassmann random variables) Given two complex, separable Hilbert spaces V V V and W W W as well as ( A , ω ) (\mathcal{A},\omega) ( A , ω ) , a non-commutative probability space, we say that two Grassmann random variables X ∈ G ( V ) X\in\mathcal{G}(V) X ∈ G ( V ) and Y ∈ G ( W ) Y\in\mathcal{G}(W) Y ∈ G ( W ) are compatible if the linear map Z : V ⊕ W → A Z:V\oplus W\to\mathcal{A} Z : V ⊕ W → A such that Z ( ( v , 0 ) ) = X ( v ) Z((v,0))=X(v) Z (( v , 0 )) = X ( v ) , if v ∈ V v\in V v ∈ V , while Z ( ( 0 , w ) ) = Y ( w ) Z((0,w))=Y(w) Z (( 0 , w )) = Y ( w ) , if $w\in W, extends to a Grassmann random variable Z : Λ ( V ⊕ W ) → A Z:\Lambda(V\oplus W)\to\mathcal{A} Z : Λ ( V ⊕ W ) → A .
Remark 3. As a consequence of this last definition we can establish a notion of independence. More precisely, two compatible Grassmann random variables X ∈ G ( V , A ) X\in\mathcal{G}(V,\mathcal{A}) X ∈ G ( V , A ) and Y ∈ G ( W , A ) Y\in\mathcal{G}(W,\mathcal{A}) Y ∈ G ( W , A ) as per Definition 3 are said to be independent with respect to the state ω \omega ω if, for all F 1 ∈ Λ ( V ) F_1\in\Lambda(V) F 1 ∈ Λ ( V ) and F 2 ∈ Λ ( W ) F_2\in\Lambda(W) F 2 ∈ Λ ( W )
ω ( X ( F 1 ) Y ( F 2 ) ) = ω ( X ( F 1 ) ) ω ( Y ( F 2 ) ) . (6) \omega(X(F_1)Y(F_2))=\omega(X(F_1))\omega(Y(F_2)). \tag*{(6)} ω ( X ( F 1 ) Y ( F 2 )) = ω ( X ( F 1 )) ω ( Y ( F 2 )) . (6)
Gaussian Grassmann random variables
We enrich the framework considered up to now, assuming that V V V comes endowed also with a real structure χ : V → V \chi: V \to V χ : V → V , compatible with ( ⋅ , ⋅ ) V (\cdot,\cdot)_V ( ⋅ , ⋅ ) V , that is an anti-linear isometry such that
( χ ( v ) , χ ( w ) ) V = ( w , v ) V , χ 2 ( v ) = v , ∀ v , w ∈ V . (\chi(v),\chi(w))_V=(w,v)_V,\qquad\chi^2(v)=v,\quad\forall v,w\in V. ( χ ( v ) , χ ( w ) ) V = ( w , v ) V , χ 2 ( v ) = v , ∀ v , w ∈ V .
On the one hand, this induces a bilinear form
⟨ ⟨ ⋅ , ⋅ ⟩ ⟩ χ : V × V → C , ( v , w ) ↦ ⟨ ⟨ v , w ⟩ ⟩ χ = ( χ ( v ) , w ) V . \langle\langle\cdot,\cdot\rangle\rangle_\chi:V\times V\to\mathbb{C},\quad(v,w)\mapsto\langle\langle v,w\rangle\rangle_\chi=(\chi(v),w)_V. ⟨⟨ ⋅ , ⋅ ⟩ ⟩ χ : V × V → C , ( v , w ) ↦ ⟨⟨ v , w ⟩ ⟩ χ = ( χ ( v ) , w ) V .
On the other hand, given a bounded linear operator C : V → V C:V\to V C : V → V and denoting by C ∗ C^* C ∗ its adjoint with respect to ( ⋅ , ⋅ ) V (\cdot,\cdot)_V ( ⋅ , ⋅ ) V , we can define its χ \chi χ -transpose as
C χ : = χ C ∗ χ , C^\chi:=\chi C^*\chi, C χ := χ C ∗ χ ,
such that
⟨ ⟨ v , C w ⟩ ⟩ χ = ⟨ ⟨ C χ v , w ⟩ ⟩ χ , ∀ v , w ∈ V . \langle\langle v,Cw\rangle\rangle_\chi=\langle\langle C^\chi v,w\rangle\rangle_\chi,\quad\forall v,w\in V. ⟨⟨ v , C w ⟩ ⟩ χ = ⟨⟨ C χ v , w ⟩ ⟩ χ , ∀ v , w ∈ V .
We say that C C C is χ \chi χ -antisymmetric if C χ = − C C^\chi=-C C χ = − C . On top of these data, we consider the antisymmetric Fock space F − ( V ) \mathcal{F}_{-}(V) F − ( V ) , that is the Hilbert space completion of the exterior algebra Λ V \Lambda V Λ V with respect to the scalar product naturally induced from ( ⋅ , ⋅ ) V (\cdot,\cdot)_V ( ⋅ , ⋅ ) V , see [4 ] . Thereon, we define
the creation operator a † : V → B ( F − ( V ) ) a^\dagger:V\to\mathcal{B}(\mathcal{F}_{-}(V)) a † : V → B ( F − ( V )) whose action is completely determined by
a † ( v ) ( u 1 ∧ u 2 ∧ … ∧ u k ) = v ∧ u 1 ∧ u 2 ∧ … ∧ u k , ∀ v , u 1 , … , u k ∈ V , a^\dagger(v)(u_1\wedge u_2\wedge\ldots\wedge u_k)=v\wedge u_1\wedge u_2\wedge\ldots\wedge u_k,\quad\forall v,u_1,\ldots,u_k\in V, a † ( v ) ( u 1 ∧ u 2 ∧ … ∧ u k ) = v ∧ u 1 ∧ u 2 ∧ … ∧ u k , ∀ v , u 1 , … , u k ∈ V ,
the annihilation operator a : V → B ( F − ( V ) ) a:V\to\mathcal{B}(\mathcal{F}_{-}(V)) a : V → B ( F − ( V )) whose action is completely determined by
a ( v ) ( u 1 ∧ u 2 ∧ … ∧ u k ) = ∑ j = 1 k ( − 1 ) j + 1 ( v , u j ) V u 1 ∧ … ∧ u j ^ ∧ … ∧ u k , ∀ v , u 1 , … , u k ∈ V , a(v)(u_1\wedge u_2\wedge\ldots\wedge u_k)=\sum_{j=1}^{k}(-1)^{j+1}(v,u_j)_V u_1\wedge\ldots\wedge\widehat{u_j}\wedge\ldots\wedge u_k,\quad\forall v,u_1,\ldots,u_k\in V, a ( v ) ( u 1 ∧ u 2 ∧ … ∧ u k ) = j = 1 ∑ k ( − 1 ) j + 1 ( v , u j ) V u 1 ∧ … ∧ u j ∧ … ∧ u k , ∀ v , u 1 , … , u k ∈ V ,
where u j ^ \widehat{u_j} u j indicates that the j j j -th element is omitted.
Observe that, being B ( F − ( V ) ) \mathcal{B}(\mathcal{F}_{-}(V)) B ( F − ( V )) a C ∗ C^* C ∗ -algebra, both a a a and a † a^\dagger a † can be extended as algebra homomorphisms so to identify a Grassmann random variable in G ( V , B ( F − ( V ) ) ) \mathcal{G}(V,\mathcal{B}(\mathcal{F}_{-}(V))) G ( V , B ( F − ( V ))) as per Definition 2.
Definition 4 Given a complex, separable Hilbert space ( V , ( ⋅ , ⋅ ) V ) (V,(\cdot,\cdot)_V) ( V , ( ⋅ , ⋅ ) V ) endowed with a real structure χ : V → V \chi:V\to V χ : V → V as well as a bounded, χ \chi χ -antisymmetric linear operator C : V → V C:V\to V C : V → V , we call Gaussian, Grassmann random variable the map X : Λ V → B ( F − ( V ) ) X:\Lambda V\to\mathcal{B}(\mathcal{F}_{-}(V)) X : Λ V → B ( F − ( V )) constructed as the algebra homomorphism induced by
X ( v ) = a † ( C v ) + a ( χ v ) , ∀ v ∈ V . X(v)=a^\dagger(Cv)+a(\chi v),\quad\forall v\in V. X ( v ) = a † ( C v ) + a ( χ v ) , ∀ v ∈ V .
A direct application of the defining properties of the creation and annihilation operators shows that, for any n ∈ N n\in\mathbb{N} n ∈ N and v 1 , … , v 2 n ∈ V v_1,\ldots,v_{2n}\in V v 1 , … , v 2 n ∈ V , the odd moments of X X X vanish, while the even ones are
ω ( X ( v 1 ) X ( v 2 ) … X ( v 2 n ) ) = ∑ σ ∈ P 2 n sgn ( σ ) ∏ i = 1 n ⟨ ⟨ v σ ( 2 i − 1 ) , C v σ ( 2 i ) ⟩ ⟩ χ . (7) \omega(X(v_1)X(v_2)\ldots X(v_{2n}))=\sum_{\sigma\in\mathcal{P}_{2n}}\operatorname{sgn}(\sigma)\prod_{i=1}^{n}\langle\langle v_{\sigma(2i-1)},Cv_{\sigma(2i)}\rangle\rangle_\chi. \tag*{(7)} ω ( X ( v 1 ) X ( v 2 ) … X ( v 2 n )) = σ ∈ P 2 n ∑ sgn ( σ ) i = 1 ∏ n ⟨⟨ v σ ( 2 i − 1 ) , C v σ ( 2 i ) ⟩ ⟩ χ . (7)
where P 2 n \mathcal{P}_{2n} P 2 n denotes the set of ordered pairings of { 1 , … , 2 n } \{1,\ldots,2n\} { 1 , … , 2 n } , namely σ ( 2 i − 1 ) < σ ( 2 i ) \sigma(2i-1)<\sigma(2i) σ ( 2 i − 1 ) < σ ( 2 i ) and σ ( 1 ) < σ ( 3 ) < … < σ ( 2 n − 1 ) \sigma(1)<\sigma(3)<\ldots<\sigma(2n-1) σ ( 1 ) < σ ( 3 ) < … < σ ( 2 n − 1 ) .
More generally, Equation (7) is taken as the defining identity for a Gaussian Grassmann random variable and one can read the construction above as a constructive realization of [1 Lemma 21] . In other words, for every bounded, χ \chi χ -antisymmetric operator C C C , there exists a non-commutative probability space ( A , ω ) (\mathcal{A},\omega) ( A , ω ) and a Gaussian Grassmann random variable whose covariance is uniquely determined by C C C .
Remark 4 The existence of a continuous Gaussian Grassmann variable for any bounded covariance C C C is a distinctive feature of the anti-commutative setting. In the commutative (Bosonic) setting, there is no counterpart to [1 Lemma 21] within the framework of bounded operators. Specifically, for a positive, self-adjoint operator B B B on a Hilbert space V V V , a linear map thereon, satisfying the Gaussian moment condition does not necessarily take values in a C ∗ C^* C ∗ -algebra A \mathcal{A} A .
Grassmannian conditional expectation – To conclude the section we discuss the notion of conditional expectation in the non-commutative setting. Contrary to the commutative framework, this cannot be seen as a byproduct of the Radon-Nikodym theorem [15 Theorem 6.10] , which exists a priori . A case by case analysis is thus necessary. In the following we formalize this statement.
Definition 5 (Conditional expectation) Given a non-commutative probability space ( A , ω ) (\mathcal{A},\omega) ( A , ω ) and a unital C*-subalgebra A ′ ⊆ A \mathcal{A}' \subseteq\mathcal{A} A ′ ⊆ A , a linear map φ : A → A ′ \varphi:\mathcal{A}\to\mathcal{A}' φ : A → A ′ is a conditional expectation of A \mathcal{A} A onto A ′ \mathcal{A}' A ′ with respect to ω \omega ω if it is positive and if, denoting by I A \mathbb{I}_{\mathcal{A}} I A and I A ′ \mathbb{I}_{\mathcal{A}'} I A ′ the unit elements respectively of A \mathcal{A} A and A ′ \mathcal{A}' A ′ ,
φ ( I A ) = I A ′ , φ ( B 1 A B 2 ) = B 1 φ ( A ) B 2 , ω ( φ ( A ) ) = ω ( A ) , ∀ A ∈ A and ∀ B 1 , B 2 ∈ A ′ . \varphi(\mathbb{I}_{\mathcal{A}})=\mathbb{I}_{\mathcal{A}'},\qquad\varphi(B_{1}AB_{2})=B_{1}\varphi(A)B_{2},\qquad\omega(\varphi(A))=\omega(A),\qquad\forall A\in\mathcal{A}\ \text{and}\ \forall B_{1},B_{2}\in\mathcal{A}'. φ ( I A ) = I A ′ , φ ( B 1 A B 2 ) = B 1 φ ( A ) B 2 , ω ( φ ( A )) = ω ( A ) , ∀ A ∈ A and ∀ B 1 , B 2 ∈ A ′ .
As shown in [6 Proposition 2.12] , if such a conditional expectation exists, it is a *-preserving projection of unit norm onto A ′ \mathcal{A}' A ′ . In the following, we are interested in physical systems governed by a suitable dynamics, which is implemented in this framework by a family of conditional expectations. The following definition sets the stage to make this idea precise.
Definition 6 (Filtration) Let ( A , ω ) (\mathcal{A},\omega) ( A , ω ) be a non-commutative probability space and let I ⊆ R I\subseteq\mathbb{R} I ⊆ R be an interval. A filtration over I I I is a family ( A t ) t ∈ I (\mathcal{A}_{t})_{t\in I} ( A t ) t ∈ I of C*-subalgebras of A \mathcal{A} A such that A t ⊆ A t ′ ⊆ A \mathcal{A}_{t}\subseteq\mathcal{A}_{t'}\subseteq\mathcal{A} A t ⊆ A t ′ ⊆ A whenever t ≤ t ′ t\leq t' t ≤ t ′ . The triple ( A , ω , ( A t ) t ∈ I ) (\mathcal{A},\omega,(\mathcal{A}_{t})_{t\in I}) ( A , ω , ( A t ) t ∈ I ) is called a filtered non-commutative probability space.
Remark 5 In our analysis we combine all the above structures considering adapted Grassmannian processes ( X t ) t ∈ I (X_{t})_{t\in I} ( X t ) t ∈ I . These are families of Grassmann random variables as per Definition 2 such that the values of X t X_{t} X t lie in the algebra A t \mathcal{A}_{t} A t for every t ∈ I t\in I t ∈ I . We can construct them as follows: starting from a Grassmann random variable X X X , we define ( X t ) t ∈ I (X_{t})_{t\in I} ( X t ) t ∈ I in terms of a family of conditional expectations ( ω t ) t ∈ I (\omega_{t})_{t\in I} ( ω t ) t ∈ I such that, for every s ≤ t s\leq t s ≤ t and for every X ∈ A X\in\mathcal{A} X ∈ A ,
ω s ( ω t ( X ) ) = ω s ( X ) . \omega_{s}(\omega_{t}(X))=\omega_{s}(X). ω s ( ω t ( X )) = ω s ( X ) .
The existence of such a family ( ω t ) t ∈ I (\omega_{t})_{t\in I} ( ω t ) t ∈ I is not guaranteed in general, but it is in the setting of Definition 4, where A ≡ B ( F − ( V ) ) \mathcal{A}\equiv\mathcal{B}(\mathcal{F}_{-}(V)) A ≡ B ( F − ( V )) . This is discussed thoroughly in [6 Appendix A] .
Construction of Fermionic fields
The scope of this section is to present an explicit Fermionic field construction which is analogous to [11 ] , but realized within the setting of Grassmannian probability. While the traditional Osterwalder-Schrader program [12 , 13 ] illustrated the limitations of classical measure theory, these obstacles can be bypassed by replacing the standard Kolmogorov axioms with a non-commutative probabilistic structure. For the sake of clarity, we shall restrict our construction to the two-dimensional case, comprising one spatial and one temporal dimension. While this choice is technically immaterial for the free field, where the Grassmannian Gaussian structure remains formally invariant, it provides a more transparent framework for discussing potential extensions to interacting theories. Here the specificity of the ( 1 + 1 ) (1+1) ( 1 + 1 ) -dimensional spacetime often plays a crucial role in the regularity and renormalizability of the model.
The strategy is the following: Starting from a Grassmannian white noise, we construct the Euclidean and time-zero fields, identifying the physical one-particle and Fock spaces. Using the Markovian structure of this explicit realization, we exploit the conditional expectation ω 0 \omega_{0} ω 0 , which maps translated Euclidean Wick polynomials to time-zero counterparts. We show that the resulting evolution coincides with the second quantization of the one-particle semigroup, thereby identifying the Fock Hamiltonian.
Free fields
We start by fixing the spaces on which the construction is performed. Taking into account the notation and conventions of Section 2, we set
V : = L 2 ( R 2 ; C 4 ) , V:=L^{2}(\mathbb{R}^{2};\mathbb{C}^{4}), V := L 2 ( R 2 ; C 4 ) ,
and, for future reference, we also consider E : = S ( R 2 ; C 4 ) E:=\mathcal{S}(\mathbb{R}^{2};\mathbb{C}^{4}) E := S ( R 2 ; C 4 ) seen as a nuclear Fréchet space, continuously and densely embedded in V V V . Consequently we can choose as underlying non-commutative probability space the pair ( A , ω ) (\mathcal{A},\omega) ( A , ω ) , where A : = B ( F − ( V ) ) \mathcal{A}:=\mathcal{B}(\mathcal{F}_{-}(V)) A := B ( F − ( V )) , while ω \omega ω is the vacuum state thereon. Up to an immaterial choice of basis for C 4 \mathbb{C}^{4} C 4 , we can split V V V as V 1 ⊕ V 2 V_{1}\oplus V_{2} V 1 ⊕ V 2 where
V 1 ≃ V 2 ≃ L 2 ( R 2 ; C 2 ) . V_{1}\simeq V_{2}\simeq L^{2}(\mathbb{R}^{2};\mathbb{C}^{2}). V 1 ≃ V 2 ≃ L 2 ( R 2 ; C 2 ) .
In addition, following Section 2.1, we consider χ \chi χ , the real structure on V V V induced by the standard complex conjugation, as well as the χ \chi χ -antisymmetric bounded linear operator C : V → V C:V\to V C : V → V such that
C : = I L 2 ( R 2 ) ⊗ J where J : = ( 0 I C 2 − I C 2 0 ) . (8) C:=\mathbb{I}_{L^{2}(\mathbb{R}^{2})}\otimes J\quad\text{where}\quad J:=\begin{pmatrix} 0 & \mathbb{I}_{\mathbb{C}^{2}} \\ -\mathbb{I}_{\mathbb{C}^{2}} & 0 \end{pmatrix}. \tag*{(8)} C := I L 2 ( R 2 ) ⊗ J where J := ( 0 − I C 2 I C 2 0 ) . (8)
These data identify a bilinear map
b : V × V → C , ( v , w ) ↦ b ( v , w ) : = ⟨ ⟨ v , C w ⟩ ⟩ χ , (9) b: V \times V \to\mathbb{C},\quad(v,w) \mapsto b(v,w) := \langle\langle v,Cw\rangle\rangle_{\chi}, \tag*{(9)} b : V × V → C , ( v , w ) ↦ b ( v , w ) := ⟨⟨ v , C w ⟩ ⟩ χ , (9)
such that, if v = ( v 1 , v 2 ) ∈ V 1 ⊕ V 2 v=(v^{1},v^{2})\in V_{1}\oplus V_{2} v = ( v 1 , v 2 ) ∈ V 1 ⊕ V 2 and w = ( w 1 , w 2 ) ∈ V 1 ⊕ V 2 w=(w^{1},w^{2})\in V_{1}\oplus V_{2} w = ( w 1 , w 2 ) ∈ V 1 ⊕ V 2 , then
b ( v , w ) = ( χ ( v 1 ) , w 2 ) L 2 ( R 2 ; C 2 ) − ( χ ( v 2 ) , w 1 ) L 2 ( R 2 ; C 2 ) . b(v,w)=(\chi(v^{1}),w^{2})_{L^{2}(\mathbb{R}^{2};\mathbb{C}^{2})}-(\chi(v^{2}),w^{1})_{L^{2}(\mathbb{R}^{2};\mathbb{C}^{2})}. b ( v , w ) = ( χ ( v 1 ) , w 2 ) L 2 ( R 2 ; C 2 ) − ( χ ( v 2 ) , w 1 ) L 2 ( R 2 ; C 2 ) .
Definition 7 (Grassmannian white noise) Given V = L 2 ( R 2 ; C 4 ) V=L^{2}(\mathbb{R}^{2};\mathbb{C}^{4}) V = L 2 ( R 2 ; C 4 ) and A : = B ( F − ( V ) ) \mathcal{A}:=\mathcal{B}(\mathcal{F}_{-}(V)) A := B ( F − ( V )) , a Grassmann random variable ξ ∈ G − ∞ ( V , A ) \xi\in\mathcal{G}^{-\infty}(V,\mathcal{A}) ξ ∈ G − ∞ ( V , A ) as per Definition 2 is called a Grassmannian white noise with covariance C C C as in (8) , if it is Gaussian, as per Definition 4, and if it satisfies
ω ( ξ ( v ) ξ ( w ) ) = b ( v , w ) , v , w ∈ V . \omega(\xi(v)\xi(w))=b(v,w),\qquad v,w\in V. ω ( ξ ( v ) ξ ( w )) = b ( v , w ) , v , w ∈ V .
Remark 6 Exploiting that E ↪ V E\hookrightarrow V E ↪ V , we can also realize a Grassmannian white noise ξ \xi ξ as an A \mathcal{A} A -valued tempered distribution, that is ξ ∈ L ( S ( R 2 ; C 4 ) ; A ) ≡ S ′ ( R 2 ; C 4 ⊗ A ) \xi\in\mathcal{L}(\mathcal{S}(\mathbb{R}^{2};\mathbb{C}^{4});\mathcal{A})\equiv\mathcal{S}'(\mathbb{R}^{2};\mathbb{C}^{4}\otimes\mathcal{A}) ξ ∈ L ( S ( R 2 ; C 4 ) ; A ) ≡ S ′ ( R 2 ; C 4 ⊗ A ) . Furthermore, on account of the decomposition V = V 1 ⊕ V 2 V=V_{1}\oplus V_{2} V = V 1 ⊕ V 2 and given any F ∈ E F\in E F ∈ E , we can write
ξ = ( ξ 1 , ξ 2 ) and F = ( f 1 , f 2 ) , (10) \xi=(\xi^{1},\xi^{2})\quad\text{and}\quad F=(f_{1},f_{2}), \tag*{(10)} ξ = ( ξ 1 , ξ 2 ) and F = ( f 1 , f 2 ) , (10)
where F = ( f 1 , f 2 ) ∈ E F=(f_{1},f_{2})\in E F = ( f 1 , f 2 ) ∈ E . Observe that, henceforth, we shall employ the following notation: Capital letters refer to C 4 \mathbb{C}^{4} C 4 -valued test-functions, while lower-case letters denote C 2 \mathbb{C}^{2} C 2 -valued ones. With the covariance as per (8) , the only non-vanishing two-point correlation functions are the mixed ones, namely for F = ( f 1 , f 2 ) F=(f_{1},f_{2}) F = ( f 1 , f 2 ) , F ′ = ( f 1 ′ , f 2 ′ ) ∈ E F'=(f'_{1},f'_{2})\in E F ′ = ( f 1 ′ , f 2 ′ ) ∈ E ,
ω ( ξ 1 ( f 1 ) ξ 2 ( f 2 ′ ) ) = ( χ f 1 , f 2 ′ ) L 2 ( R 2 ; C 2 ) , ω ( ξ 2 ( f 2 ) ξ 1 ( f 1 ′ ) ) = − ( χ f 2 , f 1 ′ ) L 2 ( R 2 ; C 2 ) , \omega(\xi^{1}(f_{1})\xi^{2}(f'_{2}))=(\chi f_{1},f'_{2})_{L^{2}(\mathbb{R}^{2};\mathbb{C}^{2})},\qquad\omega(\xi^{2}(f_{2})\xi^{1}(f'_{1}))=-(\chi f_{2},f'_{1})_{L^{2}(\mathbb{R}^{2};\mathbb{C}^{2})}, ω ( ξ 1 ( f 1 ) ξ 2 ( f 2 ′ )) = ( χ f 1 , f 2 ′ ) L 2 ( R 2 ; C 2 ) , ω ( ξ 2 ( f 2 ) ξ 1 ( f 1 ′ )) = − ( χ f 2 , f 1 ′ ) L 2 ( R 2 ; C 2 ) ,
whereas
ω ( ξ 1 ( f 1 ) ξ 1 ( f 1 ′ ) ) = 0 , ω ( ξ 2 ( f 2 ) ξ 2 ( f 2 ′ ) ) = 0. \omega(\xi^{1}(f_{1})\xi^{1}(f'_{1}))=0,\qquad\omega(\xi^{2}(f_{2})\xi^{2}(f'_{2}))=0. ω ( ξ 1 ( f 1 ) ξ 1 ( f 1 ′ )) = 0 , ω ( ξ 2 ( f 2 ) ξ 2 ( f 2 ′ )) = 0.
We have all the ingredients to establish an analogue of the Brownian motion associated with the Grassmannian white noise. As a preliminary step, we fix the relevant notation. In the following, since the Euclidean time coordinate plays a distinguished role, we wish to be able to clearly disentangle its role from that of the other coordinate x x x . For this reason we set V x : = L 2 ( R ; C 4 ) V_{x}:=L^{2}(\mathbb{R};\mathbb{C}^{4}) V x := L 2 ( R ; C 4 ) and we define C x : = I L 2 ( R ) ⊗ J C_{x}:=I_{L^{2}(\mathbb{R})}\otimes J C x := I L 2 ( R ) ⊗ J . Starting from (9) , we can induce from b b b a corresponding spatial bilinear map:
b x ( F , F ′ ) : = ( χ F , C x F ′ ) V x , F , F ′ ∈ V x . (11) b_{x}(F,F'):=\left(\chi F,C_{x}F'\right)_{V_{x}},\qquad F,F'\in V_{x}. \tag*{(11)} b x ( F , F ′ ) := ( χ F , C x F ′ ) V x , F , F ′ ∈ V x . (11)
Proposition 1 Given a Grassmannian white noise as per Definition 7 on V ≡ L 2 ( R ; V x ) V\equiv L^{2}(\mathbb{R};V_{x}) V ≡ L 2 ( R ; V x ) with covariance C C C as in (8) , for t ≥ 0 t\geq0 t ≥ 0 and for F ∈ V x F\in V_{x} F ∈ V x , set
B t ( F ) : = ξ ( 1 [ 0 , t ] ⊗ F ) , (12) B_{t}(F):=\xi(\mathbf{1}_{[0,t]}\otimes F), \tag*{(12)} B t ( F ) := ξ ( 1 [ 0 , t ] ⊗ F ) , (12) where 1 [ 0 , t ] \mathbf{1}_{[0,t]} 1 [ 0 , t ] is the characteristic function on the interval [ 0 , t ] [0,t] [ 0 , t ] . Hence ( B t ) t ≥ 0 (B_{t})_{t\geq0} ( B t ) t ≥ 0 is a Brownian motion, namely, for all F , F ′ ∈ V x F,F'\in V_{x} F , F ′ ∈ V x ,
it is a centered Gaussian Grassmannian process with covariance
ω ( B t ( F ) B s ( F ′ ) ) = min { t , s } b x ( F , F ′ ) . \omega(B_{t}(F)B_{s}(F'))=\min\{t,s\}\,b_{x}(F,F'). ω ( B t ( F ) B s ( F ′ )) = min { t , s } b x ( F , F ′ ) . it has independent increments: for $0\leq s\leq t
B t ( F ) − B s ( F ) = ξ ( 1 ( s , t ] ⊗ F ) , B_{t}(F)-B_{s}(F)=\xi(\mathbf{1}_{(s,t]}\otimes F), B t ( F ) − B s ( F ) = ξ ( 1 ( s , t ] ⊗ F ) , and
ω ( ( B t ( F ) − B s ( F ) ) ( B t ( F ′ ) − B s ( F ′ ) ) ) = ( t − s ) b x ( F , F ′ ) . \omega((B_{t}(F)-B_{s}(F))(B_{t}(F')-B_{s}(F')))=(t-s)b_{x}(F,F'). ω (( B t ( F ) − B s ( F )) ( B t ( F ′ ) − B s ( F ′ ))) = ( t − s ) b x ( F , F ′ ) . For any two disjoint bounded intervals I , J ⊂ [ 0 , ∞ ) I,J\subset[0,\infty) I , J ⊂ [ 0 , ∞ )
ω ( ξ ( 1 I ⊗ F ) ξ ( 1 J ⊗ F ′ ) ) = 0. \omega(\xi(\mathbf{1}_{I}\otimes F)\xi(\mathbf{1}_{J}\otimes F'))=0. ω ( ξ ( 1 I ⊗ F ) ξ ( 1 J ⊗ F ′ )) = 0.
Proof We observe that (12) is well-posed since 1 [ 0 , t ] ⊗ F ∈ V \mathbf{1}_{[0,t]}\otimes F\in V 1 [ 0 , t ] ⊗ F ∈ V for all finite values of t t t . Furthermore, per construction, B 0 ( F ) = 0 B_{0}(F)=0 B 0 ( F ) = 0 and B t B_{t} B t inherits from ξ \xi ξ the property of being Gaussian and centered. Bearing in mind these observations,
For any s , t ≥ 0 s,t\geq0 s , t ≥ 0 , Definition 7 entails that, for all F , F ′ ∈ V x F,F'\in V_{x} F , F ′ ∈ V x ,
ω ( B t ( F ) B s ( F ′ ) ) = ω ( ξ ( 1 [ 0 , t ] ⊗ F ) ξ ( 1 [ 0 , s ] ⊗ F ′ ) ) = b ( 1 [ 0 , t ] ⊗ F , 1 [ 0 , s ] ⊗ F ′ ) = ∫ R 1 [ 0 , t ] ( λ ) 1 [ 0 , s ] ( λ ) d λ b x ( F , F ′ ) = min { t , s } b x ( F , F ′ ) . \begin{aligned}
\omega(B_{t}(F)B_{s}(F'))&=\omega\left(\xi(\mathbf{1}_{[0,t]}\otimes F)\xi(\mathbf{1}_{[0,s]}\otimes F')\right)=b(\mathbf{1}_{[0,t]}\otimes F,\mathbf{1}_{[0,s]}\otimes F')\\
&=\int_{\mathbb{R}}\mathbf{1}_{[0,t]}(\lambda)\mathbf{1}_{[0,s]}(\lambda)\,d\lambda\; b_{x}(F,F')=\min\{t,s\}\,b_{x}(F,F').
\end{aligned} ω ( B t ( F ) B s ( F ′ )) = ω ( ξ ( 1 [ 0 , t ] ⊗ F ) ξ ( 1 [ 0 , s ] ⊗ F ′ ) ) = b ( 1 [ 0 , t ] ⊗ F , 1 [ 0 , s ] ⊗ F ′ ) = ∫ R 1 [ 0 , t ] ( λ ) 1 [ 0 , s ] ( λ ) d λ b x ( F , F ′ ) = min { t , s } b x ( F , F ′ ) .
If 0 ≤ s ≤ t 0 \le s \le t 0 ≤ s ≤ t , being ξ \xi ξ linear, Equation (12) entails
B t ( F ) − B s ( F ) = ξ ( ( 1 [ 0 , t ] − 1 [ 0 , s ] ) ⊗ F ) = ξ ( 1 ( s , t ] ⊗ F ) . B_t(F)-B_s(F)=\xi\bigl((1_{[0,t]}-1_{[0,s]})\otimes F\bigr)=\xi(1_{(s,t]}\otimes F). B t ( F ) − B s ( F ) = ξ ( ( 1 [ 0 , t ] − 1 [ 0 , s ] ) ⊗ F ) = ξ ( 1 ( s , t ] ⊗ F ) .
Therefore, for all F , F ′ ∈ V x F,F'\in V_x F , F ′ ∈ V x , once more Definition 7 yields
ω ( ( B t ( F ) − B s ( F ) ) ( B t ( F ′ ) − B s ( F ′ ) ) ) = ω ( ξ ( 1 ( s , t ] ⊗ F ) ξ ( 1 ( s , t ] ⊗ F ′ ) ) = b ( 1 ( s , t ] ⊗ F , 1 ( s , t ] ⊗ F ′ ) = ∫ R 1 ( s , t ] ( λ ) d λ b x ( F , F ′ ) = ( t − s ) b x ( F , F ′ ) . \begin{aligned}
\omega\bigl((B_t(F)-B_s(F))(B_t(F')-B_s(F'))\bigr)=\omega\bigl(\xi(1_{(s,t]}\otimes F)\xi(1_{(s,t]}\otimes F')\bigr) \\
&=b(1_{(s,t]}\otimes F,1_{(s,t]}\otimes F')=\int_{\mathbb{R}}1_{(s,t]}(\lambda)\,d\lambda\,b_x(F,F')=(t-s)b_x(F,F').
\end{aligned} ω ( ( B t ( F ) − B s ( F )) ( B t ( F ′ ) − B s ( F ′ )) ) = ω ( ξ ( 1 ( s , t ] ⊗ F ) ξ ( 1 ( s , t ] ⊗ F ′ ) ) = b ( 1 ( s , t ] ⊗ F , 1 ( s , t ] ⊗ F ′ ) = ∫ R 1 ( s , t ] ( λ ) d λ b x ( F , F ′ ) = ( t − s ) b x ( F , F ′ ) .
if I , J ⊂ [ 0 , ∞ ) I,J\subset[0,\infty) I , J ⊂ [ 0 , ∞ ) are two disjoint intervals, then 1 I ( λ ) 1 J ( λ ) = 0 1_I(\lambda)1_J(\lambda)=0 1 I ( λ ) 1 J ( λ ) = 0 . Hence for all F , F ′ ∈ V x F,F'\in V_x F , F ′ ∈ V x , Definition 7 entails
ω ( ξ ( 1 I ⊗ F ) ξ ( 1 J ⊗ F ′ ) ) = b ( 1 I ⊗ F , 1 J ⊗ F ′ ) = ∫ R 1 I ( λ ) 1 J ( λ ) d λ b x ( F , F ′ ) = 0. \omega\bigl(\xi(1_I\otimes F)\xi(1_J\otimes F')\bigr)=b(1_I\otimes F,1_J\otimes F')=\int_{\mathbb{R}}1_I(\lambda)1_J(\lambda)\,d\lambda\,b_x(F,F')=0. ω ( ξ ( 1 I ⊗ F ) ξ ( 1 J ⊗ F ′ ) ) = b ( 1 I ⊗ F , 1 J ⊗ F ′ ) = ∫ R 1 I ( λ ) 1 J ( λ ) d λ b x ( F , F ′ ) = 0.
This entails the sought result. □ \square □
Remark 7 A consequence of point 2 of Proposition 1 is that the Brownian motion B t B_t B t has independent increments in the sense of Remark 3. More precisely, the compatible Grassmann random variables induced by the maps
V x ∋ F ↦ ξ ( 1 I ⊗ F ) , V x ∋ F ↦ ξ ( 1 J ⊗ F ) , V_x\ni F\mapsto\xi(1_I\otimes F),\qquad V_x\ni F\mapsto\xi(1_J\otimes F), V x ∋ F ↦ ξ ( 1 I ⊗ F ) , V x ∋ F ↦ ξ ( 1 J ⊗ F ) ,
satisfy Equation (6) .
By linearity, it is sufficient to prove Equation (6) for monomials. Let n , m ∈ N 0 n,m\in\mathbb{N}_0 n , m ∈ N 0 and let F 1 , … , F n , G 1 , … , G m ∈ V x F_1,\ldots,F_n,G_1,\ldots,G_m\in V_x F 1 , … , F n , G 1 , … , G m ∈ V x . Since ξ \xi ξ is Gaussian, the Wick rule expresses
ω ( ∏ r = 1 n ξ ( 1 I ⊗ F r ) ∏ ℓ = 1 m ξ ( 1 J ⊗ G ℓ ) ) \omega\left(\prod_{r=1}^{n}\xi(1_I\otimes F_r)\prod_{\ell=1}^{m}\xi(1_J\otimes G_\ell)\right) ω ( r = 1 ∏ n ξ ( 1 I ⊗ F r ) ℓ = 1 ∏ m ξ ( 1 J ⊗ G ℓ ) )
as a sum over pairings of all the factors. Every pairing containing a contraction between a factor supported in I I I and one supported in J J J vanishes, since ω ( ξ ( 1 I ⊗ F r ) ξ ( 1 J ⊗ G ℓ ) ) = 0 \omega\bigl(\xi(1_I\otimes F_r)\xi(1_J\otimes G_\ell)\bigr)=0 ω ( ξ ( 1 I ⊗ F r ) ξ ( 1 J ⊗ G ℓ ) ) = 0 . Therefore, only pairings internal to the two families of factors contribute. The Wick sum consequently factorizes, and one obtains
ω ( ∏ r = 1 n ξ ( 1 I ⊗ F r ) ∏ ℓ = 1 m ξ ( 1 J ⊗ G ℓ ) ) = ω ( ∏ r = 1 n ξ ( 1 I ⊗ F r ) ) ω ( ∏ ℓ = 1 m ξ ( 1 J ⊗ G ℓ ) ) . \omega\left(\prod_{r=1}^{n}\xi(1_I\otimes F_r)\prod_{\ell=1}^{m}\xi(1_J\otimes G_\ell)\right)=\omega\left(\prod_{r=1}^{n}\xi(1_I\otimes F_r)\right)\omega\left(\prod_{\ell=1}^{m}\xi(1_J\otimes G_\ell)\right). ω ( r = 1 ∏ n ξ ( 1 I ⊗ F r ) ℓ = 1 ∏ m ξ ( 1 J ⊗ G ℓ ) ) = ω ( r = 1 ∏ n ξ ( 1 I ⊗ F r ) ) ω ( ℓ = 1 ∏ m ξ ( 1 J ⊗ G ℓ ) ) .
By the definition of independence in Equation (6) , the increments of ( B t ) t ≥ 0 (B_t)_{t\ge0} ( B t ) t ≥ 0 over disjoint time intervals are independent.
Remark 8 Observe that, when considering a Grassmannian Brownian motion as per Proposition 1, Equation (10) can be replaced by the decomposition B t = ( B t 1 , B t 2 ) B_t=(B_t^1,B_t^2) B t = ( B t 1 , B t 2 ) . Considering the orthogonal splitting V x = V x 1 ⊕ V x 2 V_x=V_x^1\oplus V_x^2 V x = V x 1 ⊕ V x 2 where V x i ≃ L 2 ( R , C 2 ) V_x^i\simeq L^2(\mathbb{R},\mathbb{C}^2) V x i ≃ L 2 ( R , C 2 ) , i = 1 , 2 i=1,2 i = 1 , 2 , it holds that, for any V x ∋ F = ( f 1 , f 2 ) V_x\ni F=(f_1,f_2) V x ∋ F = ( f 1 , f 2 ) , F ′ = ( f 1 ′ , f 2 ′ ) F'=(f'_1,f'_2) F ′ = ( f 1 ′ , f 2 ′ ) ,
B t ( F ) = B t 1 ( f 1 ) + B t 2 ( f 2 ) , B_t(F)=B_t^1(f_1)+B_t^2(f_2), B t ( F ) = B t 1 ( f 1 ) + B t 2 ( f 2 ) ,
while the two-point correlations are
ω ( B t 1 ( f 1 ) B s 1 ( f 1 ′ ) ) = 0 , ω ( B t 2 ( f 2 ) B s 2 ( f 2 ′ ) ) = 0 , \omega(B_t^1(f_1)B_s^1(f'_1))=0,\qquad\omega(B_t^2(f_2)B_s^2(f'_2))=0, ω ( B t 1 ( f 1 ) B s 1 ( f 1 ′ )) = 0 , ω ( B t 2 ( f 2 ) B s 2 ( f 2 ′ )) = 0 ,
ω ( B t 1 ( f 1 ) B s 2 ( f 2 ′ ) ) = min { t , s } ( χ f 1 , f 2 ′ ) L 2 ( R ; C 2 ) , ω ( B t 2 ( f 2 ) B s 1 ( f 1 ′ ) ) = − min { t , s } ( χ f 2 , f 1 ′ ) L 2 ( R ; C 2 ) . \omega(B_t^1(f_1)B_s^2(f'_2))=\min\{t,s\}(\chi f_1,f'_2)_{L^2(\mathbb{R};\mathbb{C}^2)},\qquad\omega(B_t^2(f_2)B_s^1(f'_1))=-\min\{t,s\}(\chi f_2,f'_1)_{L^2(\mathbb{R};\mathbb{C}^2)}. ω ( B t 1 ( f 1 ) B s 2 ( f 2 ′ )) = min { t , s } ( χ f 1 , f 2 ′ ) L 2 ( R ; C 2 ) , ω ( B t 2 ( f 2 ) B s 1 ( f 1 ′ )) = − min { t , s } ( χ f 2 , f 1 ′ ) L 2 ( R ; C 2 ) .
In the following we introduce one auxiliary structure which is necessary for the construction of a Euclidean free Fermionic field. To this end, we recall that the operator − ∂ x 2 + m 2 -\partial_x^2+m^2 − ∂ x 2 + m 2 with m 2 > 0 m^2>0 m 2 > 0 and domain C 0 ∞ ( R ) C_0^\infty(\mathbb{R}) C 0 ∞ ( R ) admits a unique self-adjoint extension, which has in addition a strictly positive spectrum. Hence, using functional calculus, we can introduce on V x V_x V x the positive, self-adjoint operator A x : = ( − ∂ x 2 + m 2 ) 1 / 2 ⊗ I C 4 A_x:=(-\partial_x^2+m^2)^{1/2}\otimes I_{\mathbb{C}^4} A x := ( − ∂ x 2 + m 2 ) 1/2 ⊗ I C 4 . The associated semigroup e − r A x e^{-rA_x} e − r A x , r ≥ 0 r\ge0 r ≥ 0 is bounded on V x V_x V x and in addition it preserves S ( R ; C 4 ) \mathcal{S}(\mathbb{R};\mathbb{C}^4) S ( R ; C 4 ) . Observe that the construction below uses a white noise constructed out of test functions possibly supported on the full Euclidean time axis, whereas the process B t B_t B t denotes only the increment over [ 0 , t ] [0,t] [ 0 , t ] .
Definition 8 We call stationary Grassmannian Ornstein–Uhlenbeck field associated with A x A_x A x the process φ ≡ ( φ t ) t ∈ R \varphi\equiv(\varphi_t)_{t\in\mathbb{R}} φ ≡ ( φ t ) t ∈ R such that, for any t ∈ R t\in\mathbb{R} t ∈ R and for any H ∈ V x H\in V_x H ∈ V x ,
φ t ( H ) : = ξ ( s ↦ 1 ( − ∞ , t ] ( s ) e − ( t − s ) A x H ) . (13) \varphi_t(H):=\xi\bigl(s\mapsto1_{(-\infty,t]}(s)e^{-(t-s)A_x}H\bigr). \tag*{(13)} φ t ( H ) := ξ ( s ↦ 1 ( − ∞ , t ] ( s ) e − ( t − s ) A x H ) . (13)
Remark 9 We highlight that (13) is best understood by means of its formal representation
φ t ( H ) = ∫ − ∞ t ξ s ( e − ( t − s ) A x H ) d s \varphi_t(H)=\int_{-\infty}^{t}\xi_s\left(e^{-(t-s)A_x}H\right)\,\mathrm{d}s φ t ( H ) = ∫ − ∞ t ξ s ( e − ( t − s ) A x H ) d s
The definition is meaningful also because for every t ∈ R t\in\mathbb{R} t ∈ R and for all H ∈ V x H\in V_x H ∈ V x , the map s ↦ 1 ( − ∞ , t ] ( s ) e − ( t − s ) A x H s\mapsto\mathbf{1}_{(-\infty,t]}(s)e^{-(t-s)A_x}H s ↦ 1 ( − ∞ , t ] ( s ) e − ( t − s ) A x H belongs to V ≡ L 2 ( R ; V x ) V\equiv L^2(\mathbb{R};V_x) V ≡ L 2 ( R ; V x ) as one can infer observing that, on ( − ∞ , t ) (-\infty,t) ( − ∞ , t ) the operator e − ( t − s ) A x e^{-(t-s)A_x} e − ( t − s ) A x is bounded with norm controlled by e − m ( t − s ) e^{-m(t-s)} e − m ( t − s ) .
Remark 10 It is worth emphasizing that the process φ \varphi φ can also be viewed as the stationary solution of the Grassmannian Ornstein–Uhlenbeck equation. More precisely, for any H H H lying in the domain of A x A_x A x and for any t ≥ 0 t\geq0 t ≥ 0 , it holds
φ t ( H ) − φ 0 ( H ) = − ∫ 0 t φ s ( A x H ) d s + B t ( H ) , \varphi_t(H)-\varphi_0(H)=-\int_0^t\varphi_s(A_xH)\,\mathrm{d}s+B_t(H), φ t ( H ) − φ 0 ( H ) = − ∫ 0 t φ s ( A x H ) d s + B t ( H ) ,
where
φ 0 ( H ) = ξ ( s ↦ 1 ( − ∞ , 0 ] ( s ) e s A x H ) . \varphi_0(H)=\xi\left(s\mapsto\mathbf{1}_{(-\infty,0]}(s)e^{sA_x}H\right). φ 0 ( H ) = ξ ( s ↦ 1 ( − ∞ , 0 ] ( s ) e s A x H ) .
At a formal notation level, this translates as
d φ t = − A x φ t d t + d B t , φ ∣ t = 0 = φ 0 . \mathrm{d}\varphi_t=-A_x\varphi_t\,\mathrm{d}t+\mathrm{d}B_t,\qquad\left.\varphi\right|_{t=0}=\varphi_0. d φ t = − A x φ t d t + d B t , φ ∣ t = 0 = φ 0 .
In order to make contact with Definition 1, we consider R + 2 : = [ 0 , + ∞ ) × R \mathbb{R}^{2}_{+}:=[0,+\infty)\times\mathbb{R} R + 2 := [ 0 , + ∞ ) × R as well as the test-function space S ( R + 2 ; C 4 ) ≡ S ( R + 2 ) ⊗ C 4 \mathcal{S}(\mathbb{R}^{2}_{+};\mathbb{C}^{4})\equiv\mathcal{S}(\mathbb{R}_{+}^{2})\otimes\mathbb{C}^{4} S ( R + 2 ; C 4 ) ≡ S ( R + 2 ) ⊗ C 4 , see (5) . For F = ( f 1 , f 2 ) ∈ S ( R + 2 ; C 4 ) F=(f_1,f_2)\in\mathcal{S}(\mathbb{R}^{2}_{+};\mathbb{C}^{4}) F = ( f 1 , f 2 ) ∈ S ( R + 2 ; C 4 ) , writing φ = ( φ 1 , φ 2 ) \varphi=(\varphi^1,\varphi^2) φ = ( φ 1 , φ 2 ) , we define the smeared field
φ ( F ) : = ∫ 0 + ∞ φ t ( F ( t , ⋅ ) ) d t , \varphi(F):=\int_0^{+\infty}\varphi_t(F(t,\cdot))\,\mathrm{d}t, φ ( F ) := ∫ 0 + ∞ φ t ( F ( t , ⋅ )) d t ,
or, equivalently,
φ ( F ) = φ 1 ( f 1 ) + φ 2 ( f 2 ) ∈ A , where φ i ( f i ) : = ∫ 0 + ∞ φ t i ( f i ( t , ⋅ ) ) d t , i = 1 , 2. \varphi(F)=\varphi^1(f_1)+\varphi^2(f_2)\in\mathcal{A},\qquad\text{where}\qquad\varphi^i(f_i):=\int_0^{+\infty}\varphi_t^i(f_i(t,\cdot))\,\mathrm{d}t,\quad i=1,2. φ ( F ) = φ 1 ( f 1 ) + φ 2 ( f 2 ) ∈ A , where φ i ( f i ) := ∫ 0 + ∞ φ t i ( f i ( t , ⋅ )) d t , i = 1 , 2.
Definition 9 Given a stationary Grassmannian Ornstein–Uhlenbeck field as per Definition 8, we call Euclidean Fermionic free field the operator-valued tempered distribution ψ = ( ψ 1 , ψ 2 ) ∈ L ( S ( R + 2 ; C 4 ) , A ) \psi=(\psi^1,\psi^2)\in\mathcal{L}(\mathcal{S}(\mathbb{R}^{2}_{+};\mathbb{C}^{4}),\mathcal{A}) ψ = ( ψ 1 , ψ 2 ) ∈ L ( S ( R + 2 ; C 4 ) , A ) such that
ψ 1 : = φ 1 , ψ 2 : = D E ∗ φ 2 , where D E ∗ = − γ E μ ∂ μ + m . \psi^1:=\varphi^1,\qquad\psi^2:=D_E^*\varphi^2,\qquad\text{where}\qquad D_E^*=-\gamma_E^\mu\partial_\mu+m. ψ 1 := φ 1 , ψ 2 := D E ∗ φ 2 , where D E ∗ = − γ E μ ∂ μ + m .
In order to better grasp the meaning of this definition, using the notation established in Section 1.1 and given F = ( f 1 , f 2 ) ∈ S ( R + 2 ; C 4 ) F=(f_1,f_2)\in\mathcal{S}(\mathbb{R}^{2}_{+};\mathbb{C}^{4}) F = ( f 1 , f 2 ) ∈ S ( R + 2 ; C 4 ) , one can write
ψ 1 ( f 1 ) = ∫ R + d k ∫ 0 + ∞ d t ∫ − ∞ t d s ξ ^ i 1 ( k , s ) e − A ( k ) ( t − s ) ( f ^ 1 ) i ( − k , t ) , \psi^1(f_1)=\int_{\mathbb{R}\vphantom{+}}\mathrm{d}k\int_0^{+\infty}\mathrm{d}t\int_{-\infty}^{t}\mathrm{d}s\,\widehat{\xi}^{1}_i(k,s)e^{-A(k)(t-s)}(\widehat{f}_1)_i(-k,t), ψ 1 ( f 1 ) = ∫ R + d k ∫ 0 + ∞ d t ∫ − ∞ t d s ξ i 1 ( k , s ) e − A ( k ) ( t − s ) ( f 1 ) i ( − k , t ) ,
where the i i i -subscript denotes an implicit sum over the two components in C 2 \mathbb{C}^{2} C 2 while A ( k ) = k 2 + m 2 A(k)=\sqrt{k^2+m^2} A ( k ) = k 2 + m 2 . At the same time, setting Γ E ( k ) : = γ E 0 A ( k ) − i γ E 1 k + m I 2 \Gamma_E(k):=\gamma_E^0A(k)-i\gamma_E^1k+mI_2 Γ E ( k ) := γ E 0 A ( k ) − i γ E 1 k + m I 2 , where I 2 I_2 I 2 is the identity matrix acting on C 2 \mathbb{C}^{2} C 2 , it holds
ψ 2 ( f 2 ) = ∫ R d k ∫ 0 + ∞ d t ∫ − ∞ t d s Γ E ( k ) i j ξ ^ i 2 ( k , s ) e − A ( k ) ( t − s ) ( f ^ 2 ) j ( − k , t ) − ∫ R d k ∫ 0 + ∞ d t ( γ E 0 ) i j ξ ^ i 2 ( k , t ) ( f ^ 2 ) j ( − k , t ) , \psi^2(f_2)=\int_{\mathbb{R}}\mathrm{d}k\int_0^{+\infty}\mathrm{d}t\int_{-\infty}^{t}\mathrm{d}s\,\Gamma_E(k)_{ij}\widehat{\xi}^{2}_i(k,s)e^{-A(k)(t-s)}(\widehat{f}_2)_j(-k,t)-\int_{\mathbb{R}}\mathrm{d}k\int_0^{+\infty}\mathrm{d}t\,(\gamma_E^0)_{ij}\widehat{\xi}^{2}_i(k,t)(\widehat{f}_2)_j(-k,t), ψ 2 ( f 2 ) = ∫ R d k ∫ 0 + ∞ d t ∫ − ∞ t d s Γ E ( k ) ij ξ i 2 ( k , s ) e − A ( k ) ( t − s ) ( f 2 ) j ( − k , t ) − ∫ R d k ∫ 0 + ∞ d t ( γ E 0 ) ij ξ i 2 ( k , t ) ( f 2 ) j ( − k , t ) ,
where, also in this case the i , j i,j i , j -subscripts denote an implicit sum over components in C 2 \mathbb{C}^{2} C 2 . Observe that the two-point functions of ψ \psi ψ are determined by the underlying Ornstein–Uhlenbeck covariance. Working at the level of integral kernels,
ω ( ψ i 1 ( x ) ψ j 1 ( y ) ) = 0 , ω ( ψ i 2 ( x ) ψ j 2 ( y ) ) = 0 , \omega(\psi^1_i(x)\psi^1_j(y))=0,\qquad\omega(\psi^2_i(x)\psi^2_j(y))=0, ω ( ψ i 1 ( x ) ψ j 1 ( y )) = 0 , ω ( ψ i 2 ( x ) ψ j 2 ( y )) = 0 ,
ω ( ψ i 1 ( x ) ψ j 2 ( y ) ) = ( S E ) i j ( y − x ) , ω ( ψ i 2 ( x ) ψ j 1 ( y ) ) = − ( S E ) j i ( x − y ) , \omega(\psi^1_i(x)\psi^2_j(y))=(S_E)_{ij}(y-x),\qquad\omega(\psi^2_i(x)\psi^1_j(y))=-(S_E)_{ji}(x-y), ω ( ψ i 1 ( x ) ψ j 2 ( y )) = ( S E ) ij ( y − x ) , ω ( ψ i 2 ( x ) ψ j 1 ( y )) = − ( S E ) j i ( x − y ) ,
where S E S_E S E is as per (3) , and the componentwise convention fixes the two-point kernel as a function of y − x y-x y − x .
Time-zero fields
Following the strategy of Nelson [11 ] , the reconstruction of a Lorentzian theory requires a notion of fields evaluated at Euclidean time zero. In the present setting this can be formulated in terms of conditional expectations associated with the filtration generated by the white noise, hence bypassing the obstruction highlighted in [12 ] . Therefore, let ξ \xi ξ be a Grassmannian white noise as in Definition 7, regarded as an element of L ( S ( R 2 ; C 4 ) ; A ) \mathcal{L}(\mathcal{S}(\mathbb{R}^{2};\mathbb{C}^{4});\mathcal{A}) L ( S ( R 2 ; C 4 ) ; A ) and let B t B_t B t be the associated Grassmannian Brownian motion as in Proposition 1. For any bounded interval I ⊂ R I\subset\mathbb{R} I ⊂ R and for any h ∈ V x h\in V_x h ∈ V x , we set
B I ( h ) : = ξ ( 1 I ⊗ h ) . B_I(h):=\xi(\mathbf{1}_I\otimes h). B I ( h ) := ξ ( 1 I ⊗ h ) .
For t ∈ R t \in\mathbb{R} t ∈ R , set V ≤ t : = L 2 ( ( − ∞ , t ] ; V x ) ⊂ V V_{\leq t}:=L^{2}((-\infty,t];V_{x})\subset V V ≤ t := L 2 (( − ∞ , t ] ; V x ) ⊂ V and define A ≤ t : = { ξ ( F ) , ξ ( F ) ∗ : F ∈ V ≤ t } ′ ′ ⊂ A \mathcal{A}_{\leq t}:=\{\xi(F),\xi(F)^{*}:F\in V_{\leq t}\}^{\prime\prime}\subset\mathcal{A} A ≤ t := { ξ ( F ) , ξ ( F ) ∗ : F ∈ V ≤ t } ′′ ⊂ A , the von Neumann algebra generated by the past fields. Furthermore we assume that there exists a family of conditional expectations, see Definition 5,
ω t : A ⟶ A ≤ t t ∈ R , \omega_{t}:\mathcal{A}\longrightarrow\mathcal{A}_{\leq t}\qquad t\in\mathbb{R}, ω t : A ⟶ A ≤ t t ∈ R ,
such that, for every F ∈ V F\in V F ∈ V
ω t ( ξ ( F ) ) = ξ ( 1 ( − ∞ , t ] F ) . \omega_{t}(\xi(F))=\xi(1_{(-\infty,t]}F). ω t ( ξ ( F )) = ξ ( 1 ( − ∞ , t ] F ) .
Definition 10 (Zero-time fields) We call zero-time Fermionic fields associated to F = ( f 1 , f 2 ) ∈ S ( R + 2 ; C 4 ) F=(f_{1},f_{2})\in\mathcal{S}(\mathbb{R}_{+}^{2};\mathbb{C}^{4}) F = ( f 1 , f 2 ) ∈ S ( R + 2 ; C 4 ) the following elements of A \mathcal{A} A :
ψ 0 1 ( f 1 ) : = ω 0 ( ψ 1 ( f 1 ) ) , ψ 0 2 ( f 2 ) : = ω 0 ( ψ 2 ( f 2 ) ) , ψ 0 ( F ) : = ψ 0 1 ( f 1 ) + ψ 0 2 ( f 2 ) . \psi_{0}^{1}(f_{1}):=\omega_{0}(\psi^{1}(f_{1})),\qquad\psi_{0}^{2}(f_{2}):=\omega_{0}(\psi^{2}(f_{2})),\qquad\psi_{0}(F):=\psi_{0}^{1}(f_{1})+\psi_{0}^{2}(f_{2}). ψ 0 1 ( f 1 ) := ω 0 ( ψ 1 ( f 1 )) , ψ 0 2 ( f 2 ) := ω 0 ( ψ 2 ( f 2 )) , ψ 0 ( F ) := ψ 0 1 ( f 1 ) + ψ 0 2 ( f 2 ) .
Working at the level of integral kernels and using the notation introduced in the previous subsection, one obtains that, for F = ( f 1 , f 2 ) ∈ S ( R + 2 ; C 4 ) F=(f_{1},f_{2})\in\mathcal{S}(\mathbb{R}_{+}^{2};\mathbb{C}^{4}) F = ( f 1 , f 2 ) ∈ S ( R + 2 ; C 4 )
ψ 0 1 ( f 1 ) = ∫ R d k ∫ 0 + ∞ d t ∫ − ∞ 0 d s ξ ^ i 1 ( k , s ) e − A ( k ) ( t − s ) ( f ^ 1 ) i ( − k , t ) , \psi_{0}^{1}(f_{1})=\int_{\mathbb{R}}dk\int_{0}^{+\infty}dt\int_{-\infty}^{0}ds\,\widehat{\xi}_{i}^{1}(k,s)e^{-A(k)(t-s)}(\widehat{f}_{1})_{i}(-k,t), ψ 0 1 ( f 1 ) = ∫ R d k ∫ 0 + ∞ d t ∫ − ∞ 0 d s ξ i 1 ( k , s ) e − A ( k ) ( t − s ) ( f 1 ) i ( − k , t ) ,
ψ 0 2 ( f 2 ) = ∫ R d k ∫ 0 + ∞ d t ∫ − ∞ 0 d s Γ E ( k ) i j ξ ^ i 2 ( k , s ) e − A ( k ) ( t − s ) ( f ^ 2 ) j ( − k , t ) , (14) \psi_{0}^{2}(f_{2})=\int_{\mathbb{R}}dk\int_{0}^{+\infty}dt\int_{-\infty}^{0}ds\,\Gamma_{E}(k)_{ij}\widehat{\xi}_{i}^{2}(k,s)e^{-A(k)(t-s)}(\widehat{f}_{2})_{j}(-k,t), \tag*{(14)} ψ 0 2 ( f 2 ) = ∫ R d k ∫ 0 + ∞ d t ∫ − ∞ 0 d s Γ E ( k ) ij ξ i 2 ( k , s ) e − A ( k ) ( t − s ) ( f 2 ) j ( − k , t ) , (14)
where, once more A ( k ) = ( k 2 + m 2 ) 1 / 2 A(k)=(k^{2}+m^{2})^{1/2} A ( k ) = ( k 2 + m 2 ) 1/2 , while Γ E ( k ) = γ E 0 A ( k ) − i γ E 1 k + m I 2 \Gamma_{E}(k)=\gamma_{E}^{0}A(k)-i\gamma_{E}^{1}k+mI_{2} Γ E ( k ) = γ E 0 A ( k ) − i γ E 1 k + m I 2 . We can also identify the covariance of the zero-time fields, but, to this end, we need some preliminary notation. Setting x , y ∈ R + 2 x,y\in\mathbb{R}_{+}^{2} x , y ∈ R + 2 with x = ( t , x ) x=(t,x) x = ( t , x ) and y = ( t ′ , y ) y=(t',y) y = ( t ′ , y ) , we define the reflected Euclidean Dirac kernel by
S E ϑ ( x , y ) : = S E ( x − ϑ ( y ) ) , S_{E}^{\vartheta}(x,y):=S_{E}(x-\vartheta(y)), S E ϑ ( x , y ) := S E ( x − ϑ ( y )) ,
where ϑ \vartheta ϑ is the Euclidean time-reflection as in (4) .
Proposition 2 For any F = ( f 1 , f 2 ) F=(f_{1},f_{2}) F = ( f 1 , f 2 ) and H = ( h 1 , h 2 ) ∈ S ( R + 2 ; C 4 ) H=(h_{1},h_{2})\in\mathcal{S}(\mathbb{R}_{+}^{2};\mathbb{C}^{4}) H = ( h 1 , h 2 ) ∈ S ( R + 2 ; C 4 ) and given the zero-time Fermionic fields as per Definition 10, it holds
ω ( ψ 0 1 ( f 1 ) ψ 0 2 ( h 2 ) ) = ∑ i , j = 1 2 ⟨ ( S E ϑ ) i j , ( f 1 ) i ⊗ ( h 2 ) j ⟩ , (15) \omega(\psi_{0}^{1}(f_{1})\psi_{0}^{2}(h_{2}))=\sum_{i,j=1}^{2}\left\langle(S_{E}^{\vartheta})_{ij},(f_{1})_{i}\otimes(h_{2})_{j}\right\rangle, \tag*{(15)} ω ( ψ 0 1 ( f 1 ) ψ 0 2 ( h 2 )) = i , j = 1 ∑ 2 ⟨ ( S E ϑ ) ij , ( f 1 ) i ⊗ ( h 2 ) j ⟩ , (15) where
( S E ϑ ) i j ∈ S ′ ( R + 2 × R + 2 ) (S_{E}^{\vartheta})_{ij}\in\mathcal{S}'(\mathbb{R}_{+}^{2}\times\mathbb{R}_{+}^{2}) ( S E ϑ ) ij ∈ S ′ ( R + 2 × R + 2 ) are the components of the reflected Euclidean Dirac kernel. At the same time,
ω ( ψ 0 1 ( f 1 ) ψ 0 1 ( h 1 ) ) = 0 , ω ( ψ 0 2 ( f 2 ) ψ 0 2 ( h 2 ) ) = 0. \omega(\psi_{0}^{1}(f_{1})\psi_{0}^{1}(h_{1}))=0,\qquad\omega(\psi_{0}^{2}(f_{2})\psi_{0}^{2}(h_{2}))=0. ω ( ψ 0 1 ( f 1 ) ψ 0 1 ( h 1 )) = 0 , ω ( ψ 0 2 ( f 2 ) ψ 0 2 ( h 2 )) = 0.
Proof. Combining (14) and the two-point correlation functions in Remark 6, one obtains
ω ( ψ 0 1 ( f 1 ) ψ 0 2 ( h 2 ) ) = ∫ R d k ∫ 0 + ∞ d t ∫ 0 + ∞ d t ′ ∫ − ∞ 0 d s e − A ( k ) ( t − s ) e − A ( k ) ( t ′ − s ) Γ E ( − k ) i j ( f ^ 1 ) i ( − k , t ) ( h ^ 2 ) j ( k , t ′ ) . \omega\left(\psi_0^1(f_1)\psi_0^2(h_2)\right)=\int_{\mathbb{R}}dk\int_0^{+\infty}dt\int_0^{+\infty}dt'\int_{-\infty}^0ds\,e^{-A(k)(t-s)}e^{-A(k)(t'-s)}\Gamma_E(-k)_{ij}(\widehat{f}_1)_i(-k,t)(\widehat{h}_2)_j(k,t'). ω ( ψ 0 1 ( f 1 ) ψ 0 2 ( h 2 ) ) = ∫ R d k ∫ 0 + ∞ d t ∫ 0 + ∞ d t ′ ∫ − ∞ 0 d s e − A ( k ) ( t − s ) e − A ( k ) ( t ′ − s ) Γ E ( − k ) ij ( f 1 ) i ( − k , t ) ( h 2 ) j ( k , t ′ ) .
Taking the integral over s s s yields
ω ( ψ 0 1 ( f 1 ) ψ 0 2 ( h 2 ) ) = ∫ R d k ∫ 0 + ∞ d t ∫ 0 + ∞ d t ′ Γ E ( − k ) i j 2 A ( k ) e − A ( k ) ( t + t ′ ) ( f ^ 1 ) i ( − k , t ) ( h ^ 2 ) j ( k , t ′ ) . \omega\left(\psi_0^1(f_1)\psi_0^2(h_2)\right)=\int_{\mathbb{R}}dk\int_0^{+\infty}dt\int_0^{+\infty}dt'\,\frac{\Gamma_E(-k)_{ij}}{2A(k)}e^{-A(k)(t+t')}(\widehat{f}_1)_i(-k,t)(\widehat{h}_2)_j(k,t'). ω ( ψ 0 1 ( f 1 ) ψ 0 2 ( h 2 ) ) = ∫ R d k ∫ 0 + ∞ d t ∫ 0 + ∞ d t ′ 2 A ( k ) Γ E ( − k ) ij e − A ( k ) ( t + t ′ ) ( f 1 ) i ( − k , t ) ( h 2 ) j ( k , t ′ ) .
This is the spatial Fourier representation of the reflected kernel S E 0 ( x , y ) = S E ( y − ϑ ( x ) ) S_E^0(\mathrm{x},\mathrm{y})=S_E(\mathrm{y}-\vartheta(\mathrm{x})) S E 0 ( x , y ) = S E ( y − ϑ ( x )) appearing on the right-hand side of Equation (15) . The remaining identity is instead a by-product of Remark 6. □
To conclude this subsection we establish an interplay between time-zero fields and Wick polynomials. We recall that the latter are maps W n ψ ∈ L ( S ( R + 2 ; C 4 ) ⊗ b n ; A ) W_n^\psi\in L(S(\mathbb{R}_+^2;\mathbb{C}^4)^{\otimes_b n};\mathcal{A}) W n ψ ∈ L ( S ( R + 2 ; C 4 ) ⊗ b n ; A ) , n ≥ 1 n\geq1 n ≥ 1 , recursively defined as follows: Given F 1 , … , F n ∈ S ( R + 2 ; C 4 ) F_1,\ldots,F_n\in S(\mathbb{R}_+^2;\mathbb{C}^4) F 1 , … , F n ∈ S ( R + 2 ; C 4 ) ,
W 1 ψ ( F 1 ) : = ψ ( F 1 ) , W 2 ψ ( F 1 ⊗ F 2 ) : = ψ ( F 1 ) ψ ( F 2 ) − ω ( ψ ( F 1 ) ψ ( F 2 ) ) I A , (16) W_1^\psi(F_1):=\psi(F_1),\qquad W_2^\psi(F_1\otimes F_2):=\psi(F_1)\psi(F_2)-\omega(\psi(F_1)\psi(F_2))\mathbb{I}_{\mathcal{A}}, \tag*{(16)} W 1 ψ ( F 1 ) := ψ ( F 1 ) , W 2 ψ ( F 1 ⊗ F 2 ) := ψ ( F 1 ) ψ ( F 2 ) − ω ( ψ ( F 1 ) ψ ( F 2 )) I A , (16)
where ω \omega ω is the state of the underlying non-commutative probability space, while I A \mathbb{I}_{\mathcal{A}} I A is the unit element of the underlying algebra. For ease of notation, we shall omit henceforth to write explicitly the identity element leaving it implicit. For n ≥ 3 n\geq3 n ≥ 3 , the Wick rule yields
W n ψ ( F 1 ⊗ ⋯ ⊗ F n ) : = ψ ( F 1 ) W n − 1 ψ ( F 2 ⊗ ⋯ ⊗ F n ) − ∑ j = 2 n ( − 1 ) j ω ( ψ ( F 1 ) ψ ( F j ) ) W n − 2 ψ ( F 2 ⊗ ⋯ F ^ j ⋯ ⊗ F n ) , (17) W_n^\psi(F_1\otimes\cdots\otimes F_n):=\psi(F_1)W_{n-1}^\psi(F_2\otimes\cdots\otimes F_n)-\sum_{j=2}^{n}(-1)^j\omega(\psi(F_1)\psi(F_j))W_{n-2}^\psi(F_2\otimes\cdots\widehat{F}_j\cdots\otimes F_n), \tag*{(17)} W n ψ ( F 1 ⊗ ⋯ ⊗ F n ) := ψ ( F 1 ) W n − 1 ψ ( F 2 ⊗ ⋯ ⊗ F n ) − j = 2 ∑ n ( − 1 ) j ω ( ψ ( F 1 ) ψ ( F j )) W n − 2 ψ ( F 2 ⊗ ⋯ F j ⋯ ⊗ F n ) , (17)
where the tilde denotes an omitted factor. For simplicity and to make contact with the notation often used in the theoretical physics literature, we shall also write
: ψ ( F 1 ) ⋯ ψ ( F n ) : = W n ψ ( F 1 ⊗ ⋯ ⊗ F n ) . (18) :\psi(F_1)\cdots\psi(F_n):=W_n^\psi(F_1\otimes\cdots\otimes F_n). \tag*{(18)} : ψ ( F 1 ) ⋯ ψ ( F n ) := W n ψ ( F 1 ⊗ ⋯ ⊗ F n ) . (18)
Remark 12 In the following we will also consider Wick polynomials constructed out of zero-time Fermionic fields. Similarly, they can be defined recursively setting
W 1 ψ 0 ( F 1 ) = : ψ 0 ( F 1 ) : = ψ 0 ( F 1 ) , W 2 ψ 0 ( F 1 ⊗ F 2 ) = : ψ 0 ( F 1 ) ψ 0 ( F 2 ) : = ψ 0 ( F 1 ) ψ 0 ( F 2 ) − ω ( ψ 0 ( F 1 ) ψ 0 ( F 2 ) ) , (19) W_1^{\psi_0}(F_1)=:\psi_0(F_1):=\psi_0(F_1),\qquad W_2^{\psi_0}(F_1\otimes F_2)=:\psi_0(F_1)\psi_0(F_2):=\psi_0(F_1)\psi_0(F_2)-\omega(\psi_0(F_1)\psi_0(F_2)), \tag*{(19)} W 1 ψ 0 ( F 1 ) =: ψ 0 ( F 1 ) := ψ 0 ( F 1 ) , W 2 ψ 0 ( F 1 ⊗ F 2 ) =: ψ 0 ( F 1 ) ψ 0 ( F 2 ) := ψ 0 ( F 1 ) ψ 0 ( F 2 ) − ω ( ψ 0 ( F 1 ) ψ 0 ( F 2 )) , (19)
while, for n ≥ 3 n\geq3 n ≥ 3 , the Wick rule once more yields
W n ψ 0 ( F 1 ⊗ ⋯ ⊗ F n ) = : ψ 0 ( F 1 ) ⋯ ψ 0 ( F n ) : W_n^{\psi_0}(F_1\otimes\cdots\otimes F_n)=:\psi_0(F_1)\cdots\psi_0(F_n): W n ψ 0 ( F 1 ⊗ ⋯ ⊗ F n ) =: ψ 0 ( F 1 ) ⋯ ψ 0 ( F n ) :
: = ψ 0 ( F 1 ) : ψ 0 ( F 2 ) ⋯ ψ 0 ( F n ) : − ∑ j = 2 n ( − 1 ) j ω ( ψ 0 ( F 1 ) ψ 0 ( F j ) ) : ψ 0 ( F 2 ) ⋯ ψ 0 ( F j ) ^ ⋯ ψ 0 ( F n ) : . (20) :=\psi_0(F_1):\psi_0(F_2)\cdots\psi_0(F_n):-\sum_{j=2}^{n}(-1)^j\omega(\psi_0(F_1)\psi_0(F_j)):\psi_0(F_2)\cdots\widehat{\psi_0(F_j)}\cdots\psi_0(F_n):. \tag*{(20)} := ψ 0 ( F 1 ) : ψ 0 ( F 2 ) ⋯ ψ 0 ( F n ) : − j = 2 ∑ n ( − 1 ) j ω ( ψ 0 ( F 1 ) ψ 0 ( F j )) : ψ 0 ( F 2 ) ⋯ ψ 0 ( F j ) ⋯ ψ 0 ( F n ) : . (20)
We can now establish the sought interplay between Wick ordering and time-zero fields. To favor a smooth reading of the main body of this manuscript, we postpone the proof of the following proposition to Appendix A. Observe that ω 0 \omega_0 ω 0 is obtained from the conditional expectations ω t \omega_t ω t setting t = 0 t=0 t = 0 .
Proposition 3 For any n ∈ N n\in\mathbb{N} n ∈ N and for any F 1 , … , F n ∈ S ( R + 2 ; C 4 ) F_1,\ldots,F_n\in S(\mathbb{R}_+^2;\mathbb{C}^4) F 1 , … , F n ∈ S ( R + 2 ; C 4 ) ,
ω 0 ( W n ψ ( F 1 ⊗ ⋯ ⊗ F n ) ) = W n ψ 0 ( F 1 ⊗ ⋯ ⊗ F n ) . \omega_0\left(W_n^\psi(F_1\otimes\cdots\otimes F_n)\right)=W_n^{\psi_0}(F_1\otimes\cdots\otimes F_n). ω 0 ( W n ψ ( F 1 ⊗ ⋯ ⊗ F n ) ) = W n ψ 0 ( F 1 ⊗ ⋯ ⊗ F n ) .
Remark 13 In Nelson’s Markov-field approach to Bosonic quantization [11 ] , usually one considers the conditional expectation onto the time-zero algebra rather than onto the larger algebra generated by the entire past. Indeed, for observables supported at nonnegative times, the Markov property implies that conditioning on the entire past is equivalent to conditioning on the field at time zero. An analogous statement holds in the Fermionic framework presented here. Let
R 0 H ( s ) : = 1 ( − ∞ , 0 ] ( s ) e s A k H , K 0 : = R 0 ( V x ) ⊥ ≤ 0 , R_0H(s):=\mathbf{1}_{(-\infty,0]}(s)e^{sA_k}H,\qquad\mathcal{K}_0:=R_0(\mathcal{V}_x)^{\perp_{\leq0}}, R 0 H ( s ) := 1 ( − ∞ , 0 ] ( s ) e s A k H , K 0 := R 0 ( V x ) ⊥ ≤ 0 ,
and set A 0 : = { ξ ( v ) , ξ ( v ) ∗ : v ∈ K 0 } ′ ′ ⊂ A ≤ 0 \mathcal{A}_0:=\{\xi(v),\xi(v)^*:v\in\mathcal{K}_0\}''\subset\mathcal{A}_{\leq0} A 0 := { ξ ( v ) , ξ ( v ) ∗ : v ∈ K 0 } ′′ ⊂ A ≤ 0 . The orthogonal decomposition
V ≤ 0 = K 0 ⊕ K 0 ⊥ \mathcal{V}_{\leq0}=\mathcal{K}_0\oplus\mathcal{K}_0^\perp V ≤ 0 = K 0 ⊕ K 0 ⊥
defines a vacuum conditional expectation
E 0 : A ≤ 0 ⟶ A 0 . E_0:\mathcal{A}_{\leq0}\longrightarrow\mathcal{A}_0. E 0 : A ≤ 0 ⟶ A 0 .
Consequently,
ω ~ 0 : = E 0 ∘ ω 0 \widetilde{\omega}_0:=E_0\circ\omega_0 ω 0 := E 0 ∘ ω 0
is a conditional expectation onto the time-zero algebra. Since
W n ψ 0 ( F 1 ⊗ ⋯ ⊗ F n ) ∈ A 0 , W_n^{\psi_0}(F_1\otimes\cdots\otimes F_n)\in\mathcal{A}_0, W n ψ 0 ( F 1 ⊗ ⋯ ⊗ F n ) ∈ A 0 ,
Proposition 3 gives
ω ~ 0 ( W n ψ ( F 1 ⊗ ⋯ ⊗ F n ) ) = ω 0 ( W n ψ ( F 1 ⊗ ⋯ ⊗ F n ) ) = W n ψ 0 ( F 1 ⊗ ⋯ ⊗ F n ) . \widetilde{\omega}_0\left(W_n^\psi(F_1\otimes\cdots\otimes F_n)\right)=\omega_0\left(W_n^\psi(F_1\otimes\cdots\otimes F_n)\right)=W_n^{\psi_0}(F_1\otimes\cdots\otimes F_n). ω 0 ( W n ψ ( F 1 ⊗ ⋯ ⊗ F n ) ) = ω 0 ( W n ψ ( F 1 ⊗ ⋯ ⊗ F n ) ) = W n ψ 0 ( F 1 ⊗ ⋯ ⊗ F n ) .
Thus, on the Wick polynomials considered here, conditioning on the entire past is equivalent to conditioning on the time-zero algebra. The two conditional expectations do not, however, coincide on arbitrary elements of A ≤ 0 \mathcal{A}_{\leq0} A ≤ 0 .
The physical one-particle space at time zero
In this subsection we show how the field ψ 0 = ( ψ 0 1 , ψ 0 2 ) \psi_0 = (\psi_0^1,\psi_0^2) ψ 0 = ( ψ 0 1 , ψ 0 2 ) , that is the zero-time Fermionic field in Definition 10, can be used to build a version of the time zero physical space of Fermionic fields. In comparison with [11 ] , as a last step in our analysis of Dirac fields, we need to establish an Euclidean time evolution of the zero-time fields. In the following we shall consider all Wick polynomials as per Remark 12 and we shall use all notations and conventions established in Section 3.2. Given F = ( f 1 , f 2 ) ∈ S ( R + 2 ; C 4 ) F=(f_1,f_2)\in\mathcal{S}(\mathbb{R}_+^2;\mathbb{C}^4) F = ( f 1 , f 2 ) ∈ S ( R + 2 ; C 4 ) we define
Q ~ 1 ( f ~ 1 ) ( k ) : = ∫ 0 + ∞ e − A ( k ) t f ~ 1 ( k , t ) d t and Q ~ 2 ( f ~ 2 ) ( k ) : = ∫ 0 + ∞ e − A ( k ) t Γ E ( − k ) f ~ 2 ( k , t ) d t . (21) \widetilde{Q}_1(\widetilde{f}_1)(k):=\int_0^{+\infty}e^{-A(k)t}\widetilde{f}_1(k,t)\,\mathrm{d}t
\quad\text{and}\quad
\widetilde{Q}_2(\widetilde{f}_2)(k):=\int_0^{+\infty}e^{-A(k)t}\Gamma_{\mathrm{E}}(-k)\widetilde{f}_2(k,t)\,\mathrm{d}t. \tag*{(21)} Q 1 ( f 1 ) ( k ) := ∫ 0 + ∞ e − A ( k ) t f 1 ( k , t ) d t and Q 2 ( f 2 ) ( k ) := ∫ 0 + ∞ e − A ( k ) t Γ E ( − k ) f 2 ( k , t ) d t . (21)
Observe that Q 1 ( f 1 ) , Q 2 ( f 2 ) ∈ S ( R ; C 2 ) Q_1(f_1),Q_2(f_2)\in\mathcal{S}(\mathbb{R};\mathbb{C}^2) Q 1 ( f 1 ) , Q 2 ( f 2 ) ∈ S ( R ; C 2 ) and in the last equation, we use Γ E ( − k ) \Gamma_{\mathrm{E}}(-k) Γ E ( − k ) for consistency with respect to (14) where the test function appears as f ~ 2 ( − k , t ) \widetilde{f}_2(-k,t) f 2 ( − k , t ) . We set
S ( R + 2 ; C 4 ) ∋ F ↦ Q ( F ) : = ( Q 1 ( f 1 ) , Q 2 ( f 2 ) ) E 0 : = Q ( S ( R + 2 ; C 4 ) ) ⊂ S ( R ; C 4 ) . (22) \mathcal{S}(\mathbb{R}_+^2;\mathbb{C}^4)\ni F\mapsto Q(F):=(Q_1(f_1),Q_2(f_2))
\qquad
\mathcal{E}_0:=Q\bigl(\mathcal{S}(\mathbb{R}_+^2;\mathbb{C}^4)\bigr)\subset\mathcal{S}(\mathbb{R};\mathbb{C}^4). \tag*{(22)} S ( R + 2 ; C 4 ) ∋ F ↦ Q ( F ) := ( Q 1 ( f 1 ) , Q 2 ( f 2 )) E 0 := Q ( S ( R + 2 ; C 4 ) ) ⊂ S ( R ; C 4 ) . (22)
It is important to note that the inclusion defining E 0 \mathcal{E}_0 E 0 is strict as one can infer from (21) since the matrix Γ E ( − k ) \Gamma_{\mathrm{E}}(-k) Γ E ( − k ) has vanishing determinant for every k ∈ R k\in\mathbb{R} k ∈ R . Hence Q 2 Q_2 Q 2 is not surjective on the whole S ( R ; C 2 ) \mathcal{S}(\mathbb{R};\mathbb{C}^2) S ( R ; C 2 ) and, therefore, one is forced to work with Ran ( Q ) \operatorname{Ran}(Q) Ran ( Q ) . In order to better understand the construction we define the map
Ψ 0 : E 0 → A , Q ( F ) ↦ Ψ 0 ( Q ( F ) ) : = ψ 0 ( F ) . \Psi_0:\mathcal{E}_0\to\mathcal{A},\qquad Q(F)\mapsto\Psi_0(Q(F)):=\psi_0(F). Ψ 0 : E 0 → A , Q ( F ) ↦ Ψ 0 ( Q ( F )) := ψ 0 ( F ) .
Proposition 4 For every F , G ∈ S ( R + 2 ; C 4 ) F,G\in\mathcal{S}(\mathbb{R}_+^2;\mathbb{C}^4) F , G ∈ S ( R + 2 ; C 4 ) ,
⟨ ψ 0 ( F ) Ω , ψ 0 ( G ) Ω ⟩ F − ( V ) = ( Q ( F ) , 1 2 A x Q ( G ) ) V x . \langle\psi_0(F)\Omega,\psi_0(G)\Omega\rangle_{\mathcal{F}_-(V)}
=
\left(Q(F),\frac{1}{2A_x}Q(G)\right)_{V_x}. ⟨ ψ 0 ( F ) Ω , ψ 0 ( G ) Ω ⟩ F − ( V ) = ( Q ( F ) , 2 A x 1 Q ( G ) ) V x .
Consequently ker ψ 0 = ker Q \ker\psi_0=\ker Q ker ψ 0 = ker Q , and the vacuum-vector map q ↦ Ψ 0 ( q ) Ω q\mapsto\Psi_0(q)\Omega q ↦ Ψ 0 ( q ) Ω is an isometry when E 0 \mathcal{E}_0 E 0 is endowed with the inner product
⟨ q , r ⟩ 0 : = ( q , 1 2 A x r ) V x , q , r ∈ E 0 . (23) \langle q,r\rangle_0:=\left(q,\frac{1}{2A_x}r\right)_{V_x},
\qquad q,r\in\mathcal{E}_0. \tag*{(23)} ⟨ q , r ⟩ 0 := ( q , 2 A x 1 r ) V x , q , r ∈ E 0 . (23)
Proof For q ∈ E 0 q\in\mathcal{E}_0 q ∈ E 0 , set
( R q ) ( s ) : = 1 ( − ∞ , 0 ] ( s ) e s A x q . (24) (Rq)(s):=\mathbf{1}_{(-\infty,0]}(s)e^{sA_x}q. \tag*{(24)} ( R q ) ( s ) := 1 ( − ∞ , 0 ] ( s ) e s A x q . (24)
(14) gives ψ 0 ( F ) = ξ ( R Q ( F ) ) \psi_0(F)=\xi(RQ(F)) ψ 0 ( F ) = ξ ( R Q ( F )) . Hence, using ξ ( v ) Ω = a ∗ ( C v ) Ω \xi(v)\Omega=a^*(Cv)\Omega ξ ( v ) Ω = a ∗ ( C v ) Ω and C ∗ C = I C^*C=\mathbb{I} C ∗ C = I ,
⟨ ψ 0 ( F ) Ω , ψ 0 ( G ) Ω ⟩ = ( R Q ( F ) , R Q ( G ) ) V = ∫ − ∞ 0 ( e s A x Q ( F ) , e s A x Q ( G ) ) V x d s = ( Q ( F ) , 1 2 A x Q ( G ) ) V x . \begin{aligned}
\langle\psi_0(F)\Omega,\psi_0(G)\Omega\rangle
&=(RQ(F),RQ(G))_V\\
&=\int_{-\infty}^{0}\left(e^{sA_x}Q(F),e^{sA_x}Q(G)\right)_{V_x}\,\mathrm{d}s\\
&=\left(Q(F),\frac{1}{2A_x}Q(G)\right)_{V_x}.
\end{aligned} ⟨ ψ 0 ( F ) Ω , ψ 0 ( G ) Ω ⟩ = ( R Q ( F ) , R Q ( G ) ) V = ∫ − ∞ 0 ( e s A x Q ( F ) , e s A x Q ( G ) ) V x d s = ( Q ( F ) , 2 A x 1 Q ( G ) ) V x .
Since A x ≥ m > 0 A_x\geq m>0 A x ≥ m > 0 , the last quadratic form vanishes precisely when Q ( F ) = 0 Q(F)=0 Q ( F ) = 0 , proving ker ψ 0 = ker Q \ker\psi_0=\ker Q ker ψ 0 = ker Q . □ \square □
Thus since ker ( Q ) = ker ψ 0 \ker(Q)=\ker\psi_0 ker ( Q ) = ker ψ 0 and Q Q Q is continuous, we endow E 0 \mathcal{E}_0 E 0 with the standard quotient topology, upon the identification E 0 ≃ S ( R + 2 ; C 4 ) / ker ψ 0 \mathcal{E}_0\simeq\mathcal{S}(\mathbb{R}_+^2;\mathbb{C}^4)/\ker\psi_0 E 0 ≃ S ( R + 2 ; C 4 ) / ker ψ 0 . We denote by
H 0 ( 1 ) : = E 0 ‾ ∥ ⋅ ∥ 0 (25) \mathcal{H}_0^{(1)}:=\overline{\mathcal{E}_0}^{\|\cdot\|_0} \tag*{(25)} H 0 ( 1 ) := E 0 ∥ ⋅ ∥ 0 (25)
the completion of the space of functions E 0 \mathcal{E}_0 E 0 with respect to the scalar product ⟨ ⋅ , ⋅ ⟩ 0 \langle\cdot,\cdot\rangle_0 ⟨ ⋅ , ⋅ ⟩ 0 . On account of Proposition 4, the map Ψ 0 \Psi_0 Ψ 0 is injective, and if we write
J 0 : E 0 → V ⊂ F − ( V ) , J 0 ( q ) = Ψ 0 ( q ) Ω . (26) J_0:\mathcal{E}_0\to V\subset\mathcal{F}_-(V),\qquad J_0(q)=\Psi_0(q)\Omega. \tag*{(26)} J 0 : E 0 → V ⊂ F − ( V ) , J 0 ( q ) = Ψ 0 ( q ) Ω. (26)
this is an isometry. Hence, being a bounded operator, there exists a unique continuous extension of J 0 J_0 J 0 from E 0 \mathcal{E}_0 E 0 to the whole H 0 ( 1 ) \mathcal{H}_0^{(1)} H 0 ( 1 ) . The role of Ψ 0 \Psi_0 Ψ 0 is that of a reparameterization of the zero-time Fermionic fields which is the natural starting point to construct a Hamiltonian evolution.
Remark 14 Unlike J 0 J_0 J 0 , the map Ψ 0 : E 0 → A \Psi_0:\mathcal{E}_0\to\mathcal{A} Ψ 0 : E 0 → A is not an isometry, at least when A \mathcal{A} A is equipped with its standard C ∗ C^* C ∗ -norm. At the same time, the vacuum state identifies a seminorm on A \mathcal{A} A by
∥ A ∥ 2 , ω 2 : = ω ( A ∗ A ) = ∥ A Ω ∥ 2 , \lVert A\rVert_{2,\omega}^2:=\omega(A^*A)=\lVert A\Omega\rVert^2, ∥ A ∥ 2 , ω 2 := ω ( A ∗ A ) = ∥ A Ω ∥ 2 ,
which has null space N ω = { A ∈ A : A Ω = 0 } \mathcal{N}_\omega=\{A\in\mathcal{A}:A\Omega=0\} N ω = { A ∈ A : A Ω = 0 } . Proposition 4 gives ∥ Ψ 0 ( q ) ∥ 2 , ω 2 = ∥ q ∥ 0 2 \lVert\Psi_0(q)\rVert_{2,\omega}^2=\lVert q\rVert_0^2 ∥ Ψ 0 ( q ) ∥ 2 , ω 2 = ∥ q ∥ 0 2 . Thus J 0 J_0 J 0 identifies E 0 \mathcal{E}_0 E 0 isometrically with a subspace of the GNS L 2 L^2 L 2 -completion of A / N ω \mathcal{A}/\mathcal{N}_\omega A / N ω . The vacuum state is not faithful on A \mathcal{A} A , so this is not a norm on the operators themselves. More in general, it is possible to realize the Fermionic expectation with a faithful quasi-free state and the non-commutative L p L^p L p -framework, see, e.g. , [5 , 2 ] .
Unfortunately, although E 0 \mathcal{E}_0 E 0 is very useful as a starting point to build the physical Hilbert space, it is too large. In order to understand the reason it is convenient to introduce the following operators: Assume m > 0 m>0 m > 0 and set
h D ( k ) : = i γ E 0 γ E 1 k + m γ E 0 , P ± ( k ) : = 1 2 ( I 2 ± h D ( k ) A ( k ) ) . h_D(k):=i\gamma_E^0\gamma_E^1k+m\gamma_E^0,\qquad P_{\pm}(k):=\frac{1}{2}\left(\mathbb{I}_2\pm\frac{h_D(k)}{A(k)}\right). h D ( k ) := i γ E 0 γ E 1 k + m γ E 0 , P ± ( k ) := 2 1 ( I 2 ± A ( k ) h D ( k ) ) .
It holds that h D ( k ) ∗ = h D ( k ) h_D(k)^*=h_D(k) h D ( k ) ∗ = h D ( k ) , h D ( k ) 2 = A ( k ) 2 I 2 h_D(k)^2=A(k)^2\mathbb{I}_2 h D ( k ) 2 = A ( k ) 2 I 2 , while P ± ( k ) P_{\pm}(k) P ± ( k ) are complementary orthogonal projections of rank one. These functions define continuous multipliers on S ( R ; C 2 ) \mathcal{S}(\mathbb{R};\mathbb{C}^2) S ( R ; C 2 ) and the associated operators, which we denote again with the same symbols, commute with the operator A x A_x A x . We have that
γ E 0 Γ E ( − k ) = 2 A ( k ) P + ( k ) , Γ E ( − k ) = 2 A ( k ) γ E 0 P + ( k ) . (27) \gamma_E^0\Gamma_E(-k)=2A(k)P_+(k),\qquad\Gamma_E(-k)=2A(k)\gamma_E^0P_+(k). \tag*{(27)} γ E 0 Γ E ( − k ) = 2 A ( k ) P + ( k ) , Γ E ( − k ) = 2 A ( k ) γ E 0 P + ( k ) . (27)
Proposition 5. Under the given assumptions, it holds that
Ran Q 1 = S ( R ; C 2 ) , Ran Q 2 = γ E 0 P + S ( R ; C 2 ) , (28) \operatorname{Ran}Q_1=\mathcal{S}(\mathbb{R};\mathbb{C}^2),\qquad\operatorname{Ran}Q_2=\gamma_E^0P_+\mathcal{S}(\mathbb{R};\mathbb{C}^2), \tag*{(28)} Ran Q 1 = S ( R ; C 2 ) , Ran Q 2 = γ E 0 P + S ( R ; C 2 ) , (28) and hence
E 0 = S ( R ; C 2 ) ⊕ γ E 0 P + S ( R ; C 2 ) . (29) \mathcal{E}_0=\mathcal{S}(\mathbb{R};\mathbb{C}^2)\oplus\gamma_E^0P_+\mathcal{S}(\mathbb{R};\mathbb{C}^2). \tag*{(29)} E 0 = S ( R ; C 2 ) ⊕ γ E 0 P + S ( R ; C 2 ) . (29)
Proof To determine the ranges, choose ρ ∈ C c ∞ ( ( 0 , ∞ ) ) \rho\in C^\infty_c((0,\infty)) ρ ∈ C c ∞ (( 0 , ∞ )) with ∫ 0 ∞ ρ ( s ) d s = 1 \int_0^\infty\rho(s)\,\mathrm{d}s=1 ∫ 0 ∞ ρ ( s ) d s = 1 . For q ∈ S ( R ; C 2 ) q\in\mathcal{S}(\mathbb{R};\mathbb{C}^2) q ∈ S ( R ; C 2 ) , set
f q ^ ( k , t ) : = ( A ( k ) + 1 ) e A ( k ) t ρ ( ( A ( k ) + 1 ) t ) q ^ ( k ) . \widehat{f_q}(k,t):=(A(k)+1)e^{A(k)t}\rho((A(k)+1)t)\widehat{q}(k). f q ( k , t ) := ( A ( k ) + 1 ) e A ( k ) t ρ (( A ( k ) + 1 ) t ) q ( k ) .
Since A ( k ) t A(k)t A ( k ) t is bounded on the support of ρ ( ( A ( k ) + 1 ) t ) \rho((A(k)+1)t) ρ (( A ( k ) + 1 ) t ) , we have f q ∈ S ( R + 2 ; C 2 ) f_q\in\mathcal{S}(\mathbb{R}_+^2;\mathbb{C}^2) f q ∈ S ( R + 2 ; C 2 ) , and
Q 1 ( f q ) ^ ( k ) = q ^ ( k ) ∫ 0 ∞ ( A ( k ) + 1 ) ρ ( ( A ( k ) + 1 ) t ) d t = q ^ ( k ) . \widehat{Q_1(f_q)}(k)=\widehat{q}(k)\int_0^\infty(A(k)+1)\rho((A(k)+1)t)\,\mathrm{d}t=\widehat{q}(k). Q 1 ( f q ) ( k ) = q ( k ) ∫ 0 ∞ ( A ( k ) + 1 ) ρ (( A ( k ) + 1 ) t ) d t = q ( k ) .
Thus Q 1 Q_1 Q 1 is surjective. Applying the same argument componentwise to the Laplace transform occurring in Q 2 Q_2 Q 2 , and using (27), yields
Ran Q 2 = Γ E ( − ⋅ ) S ( R ; C 2 ) = 2 A x γ E 0 P + S ( R ; C 2 ) = γ E 0 P + S ( R ; C 2 ) , \operatorname{Ran}Q_2=\Gamma_E(-\cdot)\mathcal{S}(\mathbb{R};\mathbb{C}^2)=2A_x\gamma_E^0P_+\mathcal{S}(\mathbb{R};\mathbb{C}^2)=\gamma_E^0P_+\mathcal{S}(\mathbb{R};\mathbb{C}^2), Ran Q 2 = Γ E ( − ⋅ ) S ( R ; C 2 ) = 2 A x γ E 0 P + S ( R ; C 2 ) = γ E 0 P + S ( R ; C 2 ) ,
because both A x A_x A x and A x − 1 A_x^{-1} A x − 1 preserve the Schwartz space. Equations (28) and (29) follow. □ \square □
Starting from (29) , we can endow S ( R ; C 2 ) ⊕ S ( R ; C 2 ) \mathcal{S}(\mathbb{R};\mathbb{C}^2)\oplus\mathcal{S}(\mathbb{R};\mathbb{C}^2) S ( R ; C 2 ) ⊕ S ( R ; C 2 ) with the inner product ⟨ ⋅ , ⋅ ⟩ 0 \langle\cdot,\cdot\rangle_0 ⟨ ⋅ , ⋅ ⟩ 0 as per (23) . Since, for q ∈ S ( R ; C 4 ) q\in\mathcal{S}(\mathbb{R};\mathbb{C}^4) q ∈ S ( R ; C 4 ) ,
∥ q ∥ 0 2 = ∫ R ∥ q ^ ( k ) ∥ C 4 2 2 k 2 + m 2 d k , (30) \|q\|_0^2=\int_{\mathbb{R}}\frac{\|\widehat{q}(k)\|_{\mathbb{C}^4}^2}{2\sqrt{k^2+m^2}}\,\mathrm{d}k, \tag*{(30)} ∥ q ∥ 0 2 = ∫ R 2 k 2 + m 2 ∥ q ( k ) ∥ C 4 2 d k , (30)
and since m > 0 m>0 m > 0 , this norm is equivalent to that on H − 1 / 2 ( R ; C 4 ) H^{-1/2}(\mathbb{R};\mathbb{C}^4) H − 1/2 ( R ; C 4 ) . Hence, upon completion, we end up with H − 1 / 2 ( R ; C 2 ) ⊕ H − 1 / 2 ( R , C 2 ) H^{-1/2}(\mathbb{R};\mathbb{C}^2)\oplus H^{-1/2}(\mathbb{R},\mathbb{C}^2) H − 1/2 ( R ; C 2 ) ⊕ H − 1/2 ( R , C 2 ) . Observing that E 0 = ( I ⊕ γ E 0 P + ) [ S ( R ; C 2 ) ⊕ S ( R ; C 2 ) ] \mathcal{E}_0=(\mathbb{I}\oplus\gamma_E^0P_+)[\mathcal{S}(\mathbb{R};\mathbb{C}^2)\oplus\mathcal{S}(\mathbb{R};\mathbb{C}^2)] E 0 = ( I ⊕ γ E 0 P + ) [ S ( R ; C 2 ) ⊕ S ( R ; C 2 )] and that I ⊕ γ E 0 P + \mathbb{I}\oplus\gamma_E^0P_+ I ⊕ γ E 0 P + is a bounded operator with closed range, being P + P_+ P + an orthogonal projector, we can conclude that the completion of E 0 \mathcal{E}_0 E 0 with respect to ⟨ ⋅ , ⋅ ⟩ 0 \langle\cdot,\cdot\rangle_0 ⟨ ⋅ , ⋅ ⟩ 0 as per (25) is
H 0 ( 1 ) = H − 1 / 2 ( R ; C 2 ) ⊕ γ E 0 P + H − 1 / 2 ( R , C 2 ) . (31) \mathcal{H}_0^{(1)}=H^{-1/2}(\mathbb{R};\mathbb{C}^2)\oplus\gamma_E^0P_+H^{-1/2}(\mathbb{R},\mathbb{C}^2). \tag*{(31)} H 0 ( 1 ) = H − 1/2 ( R ; C 2 ) ⊕ γ E 0 P + H − 1/2 ( R , C 2 ) . (31)
In view of (26) , the map J 0 J_0 J 0 identifies E 0 \mathcal{E}_0 E 0 isometrically with a subspace of the one-particle Hilbert subspace of F − ( V ) \mathcal{F}_-(V) F − ( V ) . Hence, we can refer to E 0 \mathcal{E}_0 E 0 and, more properly to H 0 ( 1 ) \mathcal{H}_0^{(1)} H 0 ( 1 ) as auxiliary time-zero one-particle spaces. Since γ E 0 \gamma_E^0 γ E 0 is invertible and P + P_+ P + is a rank 1 projection, it turns that, on each fiber, H 0 ( 1 ) \mathcal{H}_0^{(1)} H 0 ( 1 ) has three independent components, while a Dirac spinor in two dimensions has only two. In order to establish which is the redundant one, we use the following criterion. Let f ∈ S ( R ; C 2 ) f\in\mathcal{S}(\mathbb{R};\mathbb{C}^2) f ∈ S ( R ; C 2 ) and denote by q f q_f q f the corresponding element ( f , 0 ) ∈ E 0 (f,0)\in\mathcal{E}_0 ( f , 0 ) ∈ E 0 . Considering ( R q f ) ( t ) (Rq_f)(t) ( R q f ) ( t ) as per (24) and writing the Dirac operator as D E = γ E 0 ( ∂ t + h D ( k ) ) D_E=\gamma_E^0(\partial_t+h_D(k)) D E = γ E 0 ( ∂ t + h D ( k )) in view of (2) , it holds that ( D E ⊕ D E ∗ ) ( R q f ) = 0 (D_E\oplus D_E^*)(Rq_f)=0 ( D E ⊕ D E ∗ ) ( R q f ) = 0 if and only if
( A ( k ) + h D ( k ) ) f ^ ( k ) = 2 A ( k ) P + ( k ) f ^ ( k ) = 0. (A(k)+h_D(k))\widehat{f}(k)=2A(k)P_+(k)\widehat{f}(k)=0. ( A ( k ) + h D ( k )) f ( k ) = 2 A ( k ) P + ( k ) f ( k ) = 0.
Since A ( k ) > 0 A(k)>0 A ( k ) > 0 , hence it is invertible, this entails that f ^ ( k ) \widehat{f}(k) f ( k ) must lie in Ran ( P − ) \operatorname{Ran}(P_-) Ran ( P − ) , the orthogonal complement of Ran ( P + ) \operatorname{Ran}(P_+) Ran ( P + ) , namely f ∈ P − [ S ( R ; C 2 ) ] f\in P_-[\mathcal{S}(\mathbb{R};\mathbb{C}^2)] f ∈ P − [ S ( R ; C 2 )] . Observe that, should we have started from a generic element of the form ( 0 , f ′ ) (0,f') ( 0 , f ′ ) with f ′ ∈ S ( R ; C 2 ) f'\in\mathcal{S}(\mathbb{R};\mathbb{C}^2) f ′ ∈ S ( R ; C 2 ) requiring that e t A x f ′ e^{tA_x}f' e t A x f ′ satisfies the adjoint Dirac equation for t ≤ 0 t\leq0 t ≤ 0 , we would have drawn the conclusion that f ′ ∈ γ E 0 P + [ S ( R ; C 2 ) ] f'\in\gamma_E^0P_+[\mathcal{S}(\mathbb{R};\mathbb{C}^2)] f ′ ∈ γ E 0 P + [ S ( R ; C 2 )] . This recovers the restriction already imposed and it justifies once more the factor γ E 0 P + H − 1 / 2 ( R , C 2 ) \gamma_E^0P_+H^{-1/2}(\mathbb{R},\mathbb{C}^2) γ E 0 P + H − 1/2 ( R , C 2 ) in (31) . Furthermore, this also motivates introducing the projection onto the physical degrees of freedom:
P p h y s : E 0 ⟶ E 0 , ( q 1 , q 2 ) ↦ P p h y s ( q 1 , q 2 ) : = ( P − q 1 , q 2 ) . (32) \mathbb{P}_{\mathrm{phys}}:\mathcal{E}_0\longrightarrow\mathcal{E}_0,\qquad(q_1,q_2)\mapsto\mathbb{P}_{\mathrm{phys}}(q_1,q_2):=(P_-q_1,q_2). \tag*{(32)} P phys : E 0 ⟶ E 0 , ( q 1 , q 2 ) ↦ P phys ( q 1 , q 2 ) := ( P − q 1 , q 2 ) . (32)
Having identified P p h y s [ E 0 ] \mathbb{P}_{\mathrm{phys}}[\mathcal{E}_0] P phys [ E 0 ] as the admissible time-zero data, we should ascertain its compatibility with the auxiliary one-particle structure. In particular, we show that it is orthogonal with respect to ⟨ ⋅ , ⋅ ⟩ 0 \langle\cdot,\cdot\rangle_0 ⟨ ⋅ , ⋅ ⟩ 0 and commutes with A x A_x A x . Consequently, it extends to the Hilbert completion and its range defines a closed A x A_x A x -invariant subspace.
Lemma 1 Given P p h y s \mathbb{P}_{\mathrm{phys}} P phys as in Equation (32) , it holds that [ P p h y s , A x ] = 0 [\mathbb{P}_{\mathrm{phys}},A_x]=0 [ P phys , A x ] = 0 and P p h y s \mathbb{P}_{\mathrm{phys}} P phys is an orthogonal projection with respect to the inner product ⟨ ⋅ , ⋅ ⟩ 0 \langle\cdot,\cdot\rangle_0 ⟨ ⋅ , ⋅ ⟩ 0 as in Equation (23) . Hence it extends uniquely to an orthogonal projection on H 0 ( 1 ) \mathcal{H}_0^{(1)} H 0 ( 1 ) and its range and kernel are
E 0 , p h y s : = P p h y s [ E 0 ] = P − S ( R ; C 2 ) ⊕ γ E 0 P + S ( R ; C 2 ) (33) \mathcal{E}_{0,\mathrm{phys}}:=\mathbb{P}_{\mathrm{phys}}[\mathcal{E}_0]=P_{-}\mathcal{S}(\mathbb{R};\mathbb{C}^2)\oplus\gamma_E^0P_{+}\mathcal{S}(\mathbb{R};\mathbb{C}^2) \tag*{(33)} E 0 , phys := P phys [ E 0 ] = P − S ( R ; C 2 ) ⊕ γ E 0 P + S ( R ; C 2 ) (33)
H p h y s ( 1 ) : = P p h y s [ H 0 ( 1 ) ] = P − H − 1 / 2 ( R ; C 2 ) ⊕ γ E 0 P + H − 1 / 2 ( R ; C 2 ) (34) \mathcal{H}_{\mathrm{phys}}^{(1)}:=\mathbb{P}_{\mathrm{phys}}[\mathcal{H}_0^{(1)}]=P_{-}\mathcal{H}^{-1/2}(\mathbb{R};\mathbb{C}^2)\oplus\gamma_E^0P_{+}\mathcal{H}^{-1/2}(\mathbb{R};\mathbb{C}^2) \tag*{(34)} H phys ( 1 ) := P phys [ H 0 ( 1 ) ] = P − H − 1/2 ( R ; C 2 ) ⊕ γ E 0 P + H − 1/2 ( R ; C 2 ) (34)
ker P p h y s = P + S ( R ; C 2 ) ⊕ { 0 } . (35) \ker\mathbb{P}_{\mathrm{phys}}=P_{+}\mathcal{S}(\mathbb{R};\mathbb{C}^2)\oplus\{0\}. \tag*{(35)} ker P phys = P + S ( R ; C 2 ) ⊕ { 0 } . (35)
Proof The identities P − 2 = P − P_-^2=P_- P − 2 = P − and [ P − , A x ] = 0 [P_-,A_x]=0 [ P − , A x ] = 0 imply respectively that P p h y s 2 = P p h y s \mathbb{P}_{\mathrm{phys}}^2=\mathbb{P}_{\mathrm{phys}} P phys 2 = P phys and [ P p h y s , A x ] = 0 [\mathbb{P}_{\mathrm{phys}},A_x]=0 [ P phys , A x ] = 0 . Since P − ∗ = P − P_-^*=P_- P − ∗ = P − and since ( 2 A x ) − 1 (2A_x)^{-1} ( 2 A x ) − 1 commutes with P − P_- P − ,
⟨ P p h y s q , r ⟩ 0 = ⟨ q , P p h y s r ⟩ 0 , q , r ∈ E 0 . \langle\mathbb{P}_{\mathrm{phys}}q,r\rangle_0=\langle q,\mathbb{P}_{\mathrm{phys}}r\rangle_0,\qquad q,r\in\mathcal{E}_0. ⟨ P phys q , r ⟩ 0 = ⟨ q , P phys r ⟩ 0 , q , r ∈ E 0 .
Thus P p h y s \mathbb{P}_{\mathrm{phys}} P phys is an orthogonal projector and, being bounded, it admits a unique extension to H 0 ( 1 ) \mathcal{H}_0^{(1)} H 0 ( 1 ) . Equations (33a), (33b) and (33c) follow immediately from Equation (29) . □ \square □
We conclude this subsection highlighting a structural feature related to P p h y s \mathbb{P}_{\mathrm{phys}} P phys , namely, that the physically admissible time-zero data can be obtained starting from the original space of test functions S ( R + 2 ; C 4 ) \mathcal{S}(\mathbb{R}_+^2;\mathbb{C}^4) S ( R + 2 ; C 4 ) as follows. Recalling Equation (22) , if we set Q p h y s : = P p h y s Q = ( P − Q 1 , Q 2 ) Q_{\mathrm{phys}}:=\mathbb{P}_{\mathrm{phys}}Q=(P_-Q_1,Q_2) Q phys := P phys Q = ( P − Q 1 , Q 2 ) , then
ker Q p h y s = { F = ( f 1 , f 2 ) : P − Q 1 ( f 1 ) = 0 , Q 2 ( f 2 ) = 0 } . (36) \ker Q_{\mathrm{phys}}=\{F=(f_1,f_2):P_-Q_1(f_1)=0,\quad Q_2(f_2)=0\}. \tag*{(36)} ker Q phys = { F = ( f 1 , f 2 ) : P − Q 1 ( f 1 ) = 0 , Q 2 ( f 2 ) = 0 } . (36)
Consequently, being Q p h y s Q_{\mathrm{phys}} Q phys surjective onto E 0 , p h y s \mathcal{E}_{0,\mathrm{phys}} E 0 , phys , the first isomorphism theorem entails that
S ( R + 2 ; C 4 ) ker Q p h y s ≃ E 0 , p h y s . (37) \frac{\mathcal{S}(\mathbb{R}_+^2;\mathbb{C}^4)}{\ker Q_{\mathrm{phys}}}\simeq\mathcal{E}_{0,\mathrm{phys}}. \tag*{(37)} ker Q phys S ( R + 2 ; C 4 ) ≃ E 0 , phys . (37)
Fock spaces and Wick vectors
Having identified the auxiliary and physical one-particle spaces, we investigate the corresponding Fermionic many-particle sectors. Our first goal is to show that the Wick polynomials constructed from the time-zero fields reproduce, when applied to the vacuum, the antisymmetric tensor powers of the one-particle space. This establishes an isometric identification between the Wick sectors and the corresponding Fermionic Fock sectors and it provides the Hilbert space framework used in the construction of the time evolution. To this end, recalling Equation (29) , for n ≥ 1 n\geq1 n ≥ 1 , set
E 0 ∧ n : = ⋀ n E 0 , H 0 ( n ) : = E 0 ∧ n ‾ = ⋀ n H 0 ( 1 ) . (38) \mathcal{E}_0^{\wedge n}:=\bigwedge^n\mathcal{E}_0,\qquad\mathcal{H}_0^{(n)}:=\overline{\mathcal{E}_0^{\wedge n}}=\bigwedge^n\mathcal{H}_0^{(1)}. \tag*{(38)} E 0 ∧ n := ⋀ n E 0 , H 0 ( n ) := E 0 ∧ n = ⋀ n H 0 ( 1 ) . (38)
with the scalar product induced by ⟨ ⋅ , ⋅ ⟩ 0 \langle\cdot,\cdot\rangle_0 ⟨ ⋅ , ⋅ ⟩ 0 and define
W n , 0 Ψ 0 : E 0 ∧ n ⟶ A ≤ 0 W_{n,0}^{\Psi_0}:\mathcal{E}_0^{\wedge n}\longrightarrow\mathcal{A}_{\leq0} W n , 0 Ψ 0 : E 0 ∧ n ⟶ A ≤ 0
q 1 ∧ ⋯ ∧ q n ⟼ : Ψ 0 ( q 1 ) ⋯ Ψ 0 ( q n ) : W n , 0 Ψ 0 ( u ) : = W n , 0 Ψ 0 ( u ) Ω . q_1\wedge\cdots\wedge q_n\longmapsto:\Psi_0(q_1)\cdots\Psi_0(q_n):\qquad W_{n,0}^{\Psi_0}(u):=W_{n,0}^{\Psi_0}(u)\Omega. q 1 ∧ ⋯ ∧ q n ⟼: Ψ 0 ( q 1 ) ⋯ Ψ 0 ( q n ) : W n , 0 Ψ 0 ( u ) := W n , 0 Ψ 0 ( u ) Ω.
Proposition 6 For n ≥ 1 n\geq1 n ≥ 1 and q 1 , … , q n ∈ E 0 q_1,\ldots,q_n\in\mathcal{E}_0 q 1 , … , q n ∈ E 0 ,
W n , 0 Ψ 0 ( q 1 ∧ ⋯ ∧ q n ) = J 0 q 1 ∧ ⋯ ∧ J 0 q n . (39) W_{n,0}^{\Psi_0}(q_1\wedge\cdots\wedge q_n)=J_0q_1\wedge\cdots\wedge J_0q_n. \tag*{(39)} W n , 0 Ψ 0 ( q 1 ∧ ⋯ ∧ q n ) = J 0 q 1 ∧ ⋯ ∧ J 0 q n . (39)
Consequently W n , 0 Ψ 0 W_{n,0}^{\Psi_0} W n , 0 Ψ 0 extends to an isometric embedding of H 0 ( n ) \mathcal{H}_0^{(n)} H 0 ( n ) into F − ( V ) \mathcal{F}_{-}(V) F − ( V ) , and W n , 0 Ψ 0 W_{n,0}^{\Psi_0} W n , 0 Ψ 0 is injective on E 0 ∧ n \mathcal{E}_0^{\wedge n} E 0 ∧ n .
Proof For q ∈ E 0 q\in\mathcal{E}_0 q ∈ E 0 , using Equation (24) , ( R q ) ( s ) : = 1 ( − ∞ , 0 ] ( s ) e s A x q (Rq)(s):=\mathbf{1}_{(-\infty,0]}(s)e^{sA_x}q ( R q ) ( s ) := 1 ( − ∞ , 0 ] ( s ) e s A x q . Proposition 4 gives
Ψ 0 ( q ) = ξ ( R q ) = a ∗ ( J 0 q ) + a ( χ R q ) and ω ( Ψ 0 ( q ) Ψ 0 ( r ) ) = ⟨ χ R q , J 0 r ⟩ V . \Psi_0(q)=\xi(Rq)=a^*(J_0q)+a(\chi Rq)\qquad\text{and}\qquad\omega\bigl(\Psi_0(q)\Psi_0(r)\bigr)=\langle\chi Rq,J_0r\rangle_V. Ψ 0 ( q ) = ξ ( R q ) = a ∗ ( J 0 q ) + a ( χ R q ) and ω ( Ψ 0 ( q ) Ψ 0 ( r ) ) = ⟨ χ R q , J 0 r ⟩ V .
In the recursive Wick rule, the contraction terms cancel the action of a ( χ R q 1 ) a(\chi Rq_1) a ( χ R q 1 ) on J 0 q 2 ∧ ⋯ ∧ J 0 q n J_0q_2\wedge\cdots\wedge J_0q_n J 0 q 2 ∧ ⋯ ∧ J 0 q n . More precisely its contraction with the j j j -th factor has coefficient ( − 1 ) j − 1 ω ( Ψ 0 ( q 1 ) Ψ 0 ( q j ) ) (-1)^{j-1}\omega(\Psi_0(q_1)\Psi_0(q_j)) ( − 1 ) j − 1 ω ( Ψ 0 ( q 1 ) Ψ 0 ( q j )) and only a ∗ ( J 0 q 1 ) a^*(J_0q_1) a ∗ ( J 0 q 1 ) remains. Induction on n n n , starting from Ψ 0 ( q ) Ω = J 0 q \Psi_0(q)\Omega=J_0q Ψ 0 ( q ) Ω = J 0 q , yields Equation (39) . Since a ∗ a^* a ∗ is an isometry,
⟨ W n , 0 Ψ 0 ( q 1 ∧ ⋯ ∧ q n ) , W n , 0 Ψ 0 ( r 1 ∧ ⋯ ∧ r n ) ⟩ F − ( V ) = det ( ⟨ q i , r j ⟩ 0 ) i , j = 1 n . \left\langle W_{n,0}^{\Psi_0}(q_1\wedge\cdots\wedge q_n),W_{n,0}^{\Psi_0}(r_1\wedge\cdots\wedge r_n)\right\rangle_{\mathcal{F}_{-}(V)}
=\det\left(\langle q_i,r_j\rangle_0\right)_{i,j=1}^{n}. ⟨ W n , 0 Ψ 0 ( q 1 ∧ ⋯ ∧ q n ) , W n , 0 Ψ 0 ( r 1 ∧ ⋯ ∧ r n ) ⟩ F − ( V ) = det ( ⟨ q i , r j ⟩ 0 ) i , j = 1 n .
This is the scalar product on E 0 ∧ n \mathcal{E}_0^{\wedge n} E 0 ∧ n and it gives the stated extension. To conclude, observe that, given u ∈ E 0 ∧ n u\in\mathcal{E}_0^{\wedge n} u ∈ E 0 ∧ n , W n , 0 Ψ 0 ( u ) = 0 W_{n,0}^{\Psi_0}(u)=0 W n , 0 Ψ 0 ( u ) = 0 implies W n , 0 Ψ 0 ( u ) = 0 W_{n,0}^{\Psi_0}(u)=0 W n , 0 Ψ 0 ( u ) = 0 , hence u = 0 u=0 u = 0 . This establishes injectivity. □ \square □
We now restrict the preceding construction to the physical one-particle subspace identified in Section 3.3. Starting from Equation (33a), we set
E 0 , p h y s ∧ n : = ⋀ n E 0 , p h y s , H p h y s ( n ) : = E 0 , p h y s ∧ n ‾ = ⋀ n H p h y s ( 1 ) (40) \mathcal{E}_{0,\mathrm{phys}}^{\wedge n}:=\bigwedge^n\mathcal{E}_{0,\mathrm{phys}},\qquad\mathcal{H}_{\mathrm{phys}}^{(n)}:=\overline{\mathcal{E}_{0,\mathrm{phys}}^{\wedge n}}=\bigwedge^n\mathcal{H}_{\mathrm{phys}}^{(1)} \tag*{(40)} E 0 , phys ∧ n := ⋀ n E 0 , phys , H phys ( n ) := E 0 , phys ∧ n = ⋀ n H phys ( 1 ) (40)
and let J 0 , p h y s : = J 0 ∣ H p h y s ( 1 ) J_{0,\mathrm{phys}}:=\left.J_0\right|_{\mathcal{H}_{\mathrm{phys}}^{(1)}} J 0 , phys := J 0 ∣ H phys ( 1 ) .
Corollary 1. The restriction
W n , p h y s Ψ 0 : = W n , 0 Ψ 0 ∣ H p h y s ( n ) = J 0 , p h y s ∧ n : H p h y s ( n ) ⟶ F − ( V ) , W_{n,\mathrm{phys}}^{\Psi_0}:=W_{n,0}^{\Psi_0}\big|_{\mathcal{H}_{\mathrm{phys}}^{(n)}}=J_{0,\mathrm{phys}}^{\wedge n}:\mathcal{H}_{\mathrm{phys}}^{(n)}\longrightarrow\mathcal{F}_{-}(V), W n , phys Ψ 0 := W n , 0 Ψ 0 H phys ( n ) = J 0 , phys ∧ n : H phys ( n ) ⟶ F − ( V ) , is an isometric embedding. For q j ∈ E 0 , p h y s q_j\in\mathcal{E}_{0,\mathrm{phys}} q j ∈ E 0 , phys it satisfies
W n , p h y s Ψ 0 ( q 1 ∧ ⋯ ∧ q n ) = W n , 0 Ψ 0 ( q 1 ∧ ⋯ ∧ q n ) Ω . W_{n,\mathrm{phys}}^{\Psi_0}(q_1\wedge\cdots\wedge q_n)=W_{n,0}^{\Psi_0}(q_1\wedge\cdots\wedge q_n)\Omega. W n , phys Ψ 0 ( q 1 ∧ ⋯ ∧ q n ) = W n , 0 Ψ 0 ( q 1 ∧ ⋯ ∧ q n ) Ω. In particular, W n , 0 Ψ 0 W_{n,0}^{\Psi_0} W n , 0 Ψ 0 is injective on E 0 , p h y s ∧ n \mathcal{E}_{0,\mathrm{phys}}^{\wedge n} E 0 , phys ∧ n .
Proof. It suffices to apply Proposition 6 to ⋀ n H p h y s ( 1 ) \bigwedge^n\mathcal{H}_{\mathrm{phys}}^{(1)} ⋀ n H phys ( 1 ) .
To include the vacuum sectors, we set
H 0 ( 0 ) = H p h y s ( 0 ) = C , W 0 , 0 Ψ 0 ( z ) = W 0 , p h y s Ψ 0 ( z ) : = z Ω , \mathcal{H}_0^{(0)}=\mathcal{H}_{\mathrm{phys}}^{(0)}=\mathbb{C},\qquad W_{0,0}^{\Psi_0}(z)=W_{0,\mathrm{phys}}^{\Psi_0}(z):=z\Omega, H 0 ( 0 ) = H phys ( 0 ) = C , W 0 , 0 Ψ 0 ( z ) = W 0 , phys Ψ 0 ( z ) := z Ω ,
so that the auxiliary and physical Fock spaces are
F 0 : = ⨁ n = 0 ∞ H 0 ( n ) = F − ( H 0 ( 1 ) ) , F p h y s : = ⨁ n = 0 ∞ H p h y s ( n ) = F − ( H p h y s ( 1 ) ) . (41) \mathcal{F}_0:=\bigoplus_{n=0}^{\infty}\mathcal{H}_0^{(n)}=\mathcal{F}_{-}\left(\mathcal{H}_0^{(1)}\right),\qquad
\mathcal{F}_{\mathrm{phys}}:=\bigoplus_{n=0}^{\infty}\mathcal{H}_{\mathrm{phys}}^{(n)}=\mathcal{F}_{-}\left(\mathcal{H}_{\mathrm{phys}}^{(1)}\right). \tag*{(41)} F 0 := n = 0 ⨁ ∞ H 0 ( n ) = F − ( H 0 ( 1 ) ) , F phys := n = 0 ⨁ ∞ H phys ( n ) = F − ( H phys ( 1 ) ) . (41)
The preceding propositions also establish the existence of the isometric embeddings
Γ − ( J 0 ) = ⨁ n = 0 ∞ W n , 0 Ψ 0 : F 0 ⟶ F − ( V ) , Γ − ( J 0 , p h y s ) = ⨁ n = 0 ∞ W n , p h y s Ψ 0 : F p h y s ⟶ F − ( V ) . \Gamma_{-}(J_0)=\bigoplus_{n=0}^{\infty}W_{n,0}^{\Psi_0}:\mathcal{F}_0\longrightarrow\mathcal{F}_{-}(V),\qquad
\Gamma_{-}(J_{0,\mathrm{phys}})=\bigoplus_{n=0}^{\infty}W_{n,\mathrm{phys}}^{\Psi_0}:\mathcal{F}_{\mathrm{phys}}\longrightarrow\mathcal{F}_{-}(V). Γ − ( J 0 ) = n = 0 ⨁ ∞ W n , 0 Ψ 0 : F 0 ⟶ F − ( V ) , Γ − ( J 0 , phys ) = n = 0 ⨁ ∞ W n , phys Ψ 0 : F phys ⟶ F − ( V ) .
Time Evolution and Hamiltonian
We now turn our attention to the time evolution of the theory. First of all we show that Euclidean time translations induce a strongly continuous contraction semigroup on the auxiliary and physical one-particle spaces, identifying its positive generator. Subsequently we use the Markovian structure, through the conditional expectation onto the past algebra, to lift this evolution to Wick polynomials and to the corresponding Fock spaces. The resulting Fock Hamiltonian will be shown to coincide with the second quantization of the one-particle counterpart.
The one-particle semigroup
For τ ≥ 0 \tau\geq0 τ ≥ 0 , define the Euclidean time translation
Θ τ : S ( R + 2 ; C 4 ) ⟶ S ( R + 2 ; C 4 ) , ( Θ τ F ) ( t , x ) : = F ( t − τ , x ) . \Theta_\tau:\mathcal{S}(\mathbb{R}_{+}^{2};\mathbb{C}^{4})\longrightarrow\mathcal{S}(\mathbb{R}_{+}^{2};\mathbb{C}^{4}),\qquad
(\Theta_\tau F)(t,x):=F(t-\tau,x). Θ τ : S ( R + 2 ; C 4 ) ⟶ S ( R + 2 ; C 4 ) , ( Θ τ F ) ( t , x ) := F ( t − τ , x ) .
By direct inspection of Equation (21) , for every F = ( f 1 , f 2 ) ∈ S ( R + 2 ; C 4 ) F=(f_1,f_2)\in\mathcal{S}(\mathbb{R}_{+}^{2};\mathbb{C}^{4}) F = ( f 1 , f 2 ) ∈ S ( R + 2 ; C 4 ) ,
Q 1 ( Θ τ f 1 ^ ) ( k ) = e − τ A ( k ) Q 1 ( f 1 ^ ) ( k ) , Q 2 ( Θ τ f 2 ^ ) ( k ) = e − τ A ( k ) Q 2 ( f 2 ^ ) ( k ) . Q_1(\widehat{\Theta_\tau f_1})(k)=e^{-\tau A(k)}Q_1(\widehat{f_1})(k),\qquad
Q_2(\widehat{\Theta_\tau f_2})(k)=e^{-\tau A(k)}Q_2(\widehat{f_2})(k). Q 1 ( Θ τ f 1 ) ( k ) = e − τ A ( k ) Q 1 ( f 1 ) ( k ) , Q 2 ( Θ τ f 2 ) ( k ) = e − τ A ( k ) Q 2 ( f 2 ) ( k ) .
Consequently,
Q ( Θ τ F ) = e − τ A x Q ( F ) , Q(\Theta_\tau F)=e^{-\tau A_x}Q(F), Q ( Θ τ F ) = e − τ A x Q ( F ) ,
and hence the following map is well defined:
T τ : E 0 ⟶ E 0 , T τ Q ( F ) : = Q ( Θ τ F ) = e − τ A x Q ( F ) . (42) T_\tau:\mathcal{E}_0\longrightarrow\mathcal{E}_0,\qquad T_\tau Q(F):=Q(\Theta_\tau F)=e^{-\tau A_x}Q(F). \tag*{(42)} T τ : E 0 ⟶ E 0 , T τ Q ( F ) := Q ( Θ τ F ) = e − τ A x Q ( F ) . (42)
Proposition 7. The family ( T τ ) τ ≥ 0 (T_\tau)_{\tau\geq0} ( T τ ) τ ≥ 0 as in Equation (42) extends uniquely to a strongly continuous self-adjoint contraction semigroup on H 0 ( 1 ) \mathcal{H}_0^{(1)} H 0 ( 1 ) , see Equation (25) . Its positive Hamiltonian is
H 1 , 0 : = A x , D ( H 1 , 0 ) = { q ∈ H 0 ( 1 ) : A x q ∈ H 0 ( 1 ) } , H_{1,0}:=A_x,\qquad D(H_{1,0})=\{q\in\mathcal{H}_0^{(1)}:A_xq\in\mathcal{H}_0^{(1)}\}, H 1 , 0 := A x , D ( H 1 , 0 ) = { q ∈ H 0 ( 1 ) : A x q ∈ H 0 ( 1 ) } , so that T τ = e − τ H 1 , 0 T_\tau=e^{-\tau H_{1,0}} T τ = e − τ H 1 , 0 and its infinitesimal generator is − H 1 , 0 -H_{1,0} − H 1 , 0 . In particular E 0 \mathcal{E}_0 E 0 is a core for H 1 , 0 H_{1,0} H 1 , 0 and for every q ∈ E 0 q\in\mathcal{E}_0 q ∈ E 0 ,
d d τ T τ q ∣ τ = 0 = − A x q . \left.\frac{\mathrm{d}}{\mathrm{d}\tau}T_\tau q\right|_{\tau=0}=-A_xq. d τ d T τ q τ = 0 = − A x q .
Proof. On E 0 \mathcal{E}_0 E 0 , the semigroup property follows immediately from T τ T_\tau T τ being of the form e − τ A x e^{-\tau A_x} e − τ A x . Moreover,
∥ T τ q ∥ 0 2 = ∫ R e − 2 τ A ( k ) 2 A ( k ) ∥ q ^ ( k ) ∥ C 4 2 d k ≤ ∥ q ∥ 0 2 . (43) \lVert T_\tau q\rVert_0^2=\int_{\mathbb{R}}\frac{e^{-2\tau A(k)}}{2A(k)}\lVert\widehat{q}(k)\rVert_{\mathbb{C}^4}^2\,\mathrm{d}k\leq\lVert q\rVert_0^2. \tag*{(43)} ∥ T τ q ∥ 0 2 = ∫ R 2 A ( k ) e − 2 τ A ( k ) ∥ q ( k ) ∥ C 4 2 d k ≤ ∥ q ∥ 0 2 . (43)
Since e − τ A ( k ) → 1 e^{-\tau A(k)}\to1 e − τ A ( k ) → 1 pointwise as τ → 0 \tau\to0 τ → 0 , dominated convergence gives
lim τ ↓ 0 ∥ T τ q − q ∥ 0 2 = lim τ ↓ 0 ∫ R ∣ e − τ A ( k ) − 1 ∣ 2 2 A ( k ) ∥ q ^ ( k ) ∥ C 4 2 d k = 0. \lim_{\tau\downarrow0}\lVert T_\tau q-q\rVert_0^2
=\lim_{\tau\downarrow0}\int_{\mathbb{R}}\frac{\lvert e^{-\tau A(k)}-1\rvert^2}{2A(k)}\lVert\widehat{q}(k)\rVert_{\mathbb{C}^4}^2\,\mathrm{d}k=0. τ ↓ 0 lim ∥ T τ q − q ∥ 0 2 = τ ↓ 0 lim ∫ R 2 A ( k ) ∣ e − τ A ( k ) − 1 ∣ 2 ∥ q ( k ) ∥ C 4 2 d k = 0.
On account of (43) , T τ T_\tau T τ extends uniquely by continuity to a contraction on H 0 ( 1 ) \mathcal{H}_0^{(1)} H 0 ( 1 ) . This property combined with strong continuity on E 0 \mathcal{E}_0 E 0 , which is dense in H 0 ( 1 ) \mathcal{H}_0^{(1)} H 0 ( 1 ) , implies strong continuity of the extended semigroup. Furthermore, it is self-adjoint because its multiplier is real valued. If q ∈ D ( H 1 , 0 ) q \in D(H_{1,0}) q ∈ D ( H 1 , 0 ) , dominated convergence gives
lim τ ↓ 0 ∥ T τ q − q τ + A x q ∥ 0 = 0. \lim_{\tau\downarrow0}\left\Vert\frac{T_\tau q-q}{\tau}+A_xq\right\Vert_0=0. τ ↓ 0 lim τ T τ q − q + A x q 0 = 0.
Conversely, if the difference quotient converges in H 0 ( 1 ) \mathcal{H}_0^{(1)} H 0 ( 1 ) , a pointwise convergent subsequence identifies its limit with − A x q -A_xq − A x q , and hence A x q ∈ H 0 ( 1 ) A_xq \in\mathcal{H}_0^{(1)} A x q ∈ H 0 ( 1 ) . To conclude, since the mass m > 0 m>0 m > 0 , (30) entails that the graph norm of H 1 , 0 = A x H_{1,0}=A_x H 1 , 0 = A x is equivalent to the H 1 / 2 H^{1/2} H 1/2 -norm. Since S ( R ; C 2 ) \mathcal{S}(\mathbb{R};\mathbb{C}^2) S ( R ; C 2 ) is dense in H 1 / 2 ( R ; C 2 ) H^{1/2}(\mathbb{R};\mathbb{C}^2) H 1/2 ( R ; C 2 ) and P + P_+ P + preserves both spaces, E 0 \mathcal{E}_0 E 0 is dense in D ( H 1 , 0 ) D(H_{1,0}) D ( H 1 , 0 ) with respect to the graph norm. Hence E 0 \mathcal{E}_0 E 0 is a core for H 1 , 0 H_{1,0} H 1 , 0 . □
We now restrict the preceding one-particle dynamics to the physical subspace identified in Section 3.3. The commutation of P p h y s P_{\mathrm{phys}} P phys with A x A_x A x ensures that this subspace is preserved under the Euclidean time-evolution semigroup.
Corollary 2 The operators T τ T_\tau T τ preserve both E 0 , p h y s \mathcal{E}_{0,\mathrm{phys}} E 0 , phys and H p h y s ( 1 ) \mathcal{H}_{\mathrm{phys}}^{(1)} H phys ( 1 ) . The restrictions
T τ p h y s : E 0 , p h y s ⟶ E 0 , p h y s , q ↦ T τ p h y s q : = e − τ A x q , T_\tau^{\mathrm{phys}}:\mathcal{E}_{0,\mathrm{phys}}\longrightarrow\mathcal{E}_{0,\mathrm{phys}},\qquad q\mapsto T_\tau^{\mathrm{phys}}q:=e^{-\tau A_x}q, T τ phys : E 0 , phys ⟶ E 0 , phys , q ↦ T τ phys q := e − τ A x q , extend uniquely to a strongly continuous self-adjoint contraction semigroup on H p h y s ( 1 ) \mathcal{H}_{\mathrm{phys}}^{(1)} H phys ( 1 ) . Its positive Hamiltonian is
H 1 , p h y s : = H 1 , 0 ∣ H p h y s ( 1 ) , D ( H 1 , p h y s ) = D ( H 1 , 0 ) ∩ H p h y s ( 1 ) . H_{1,\mathrm{phys}}:=\left.H_{1,0}\right|_{\mathcal{H}_{\mathrm{phys}}^{(1)}},\qquad D(H_{1,\mathrm{phys}})=D(H_{1,0})\cap\mathcal{H}_{\mathrm{phys}}^{(1)}. H 1 , phys := H 1 , 0 ∣ H phys ( 1 ) , D ( H 1 , phys ) = D ( H 1 , 0 ) ∩ H phys ( 1 ) . Equivalently, on the quotient H 0 ( 1 ) / ker P p h y s \mathcal{H}_0^{(1)}/\ker P_{\mathrm{phys}} H 0 ( 1 ) / ker P phys , T τ p h y s [ q ] = [ T τ q ] T_\tau^{\mathrm{phys}}[q]=[T_\tau q] T τ phys [ q ] = [ T τ q ] .
Proof Since [ P p h y s , A x ] = 0 [P_{\mathrm{phys}},A_x]=0 [ P phys , A x ] = 0 , by functional calculus it descends that P p h y s T τ = T τ P p h y s P_{\mathrm{phys}}T_\tau=T_\tau P_{\mathrm{phys}} P phys T τ = T τ P phys . Hence E 0 , p h y s \mathcal{E}_{0,\mathrm{phys}} E 0 , phys and H p h y s ( 1 ) \mathcal{H}_{\mathrm{phys}}^{(1)} H phys ( 1 ) are invariant under the action of T τ T_\tau T τ , while H p h y s ( 1 ) \mathcal{H}_{\mathrm{phys}}^{(1)} H phys ( 1 ) is a reducing subspace. All the claims therefore follow by restriction from Proposition 7. □
Markovian evolution and the Fock Hamiltonian
We now obtain the Fock evolution directly from the Euclidean time translation and conditional expectation. Set E 0 0 = E 0 , p h y s 0 : = C \mathcal{E}_0^0=\mathcal{E}_{0,\mathrm{phys}}^0:=\mathbb{C} E 0 0 = E 0 , phys 0 := C and W 0 , 0 Ψ 0 ( z ) : = z I A W_{0,0}^{\Psi_0}(z):=z\mathbb{I}_{\mathcal{A}} W 0 , 0 Ψ 0 ( z ) := z I A . Define
A 0 a l g : = ⨁ n = 0 + ∞ W n , 0 Ψ 0 ( E 0 ∧ n ) ⊂ A ≤ 0 , (44) \mathcal{A}_0^{\mathrm{alg}}:=\bigoplus_{n=0}^{+\infty}W_{n,0}^{\Psi_0}(\mathcal{E}_0^{\wedge n})\subset\mathcal{A}_{\leq0}, \tag*{(44)} A 0 alg := n = 0 ⨁ + ∞ W n , 0 Ψ 0 ( E 0 ∧ n ) ⊂ A ≤ 0 , (44)
A p h y s a l g : = ⨁ n = 0 + ∞ W n , 0 Ψ 0 ( E 0 , p h y s ∧ n ) ⊂ A ≤ 0 , (45) \mathcal{A}_{\mathrm{phys}}^{\mathrm{alg}}:=\bigoplus_{n=0}^{+\infty}W_{n,0}^{\Psi_0}(\mathcal{E}_{0,\mathrm{phys}}^{\wedge n})\subset\mathcal{A}_{\leq0}, \tag*{(45)} A phys alg := n = 0 ⨁ + ∞ W n , 0 Ψ 0 ( E 0 , phys ∧ n ) ⊂ A ≤ 0 , (45)
where, in both expressions, we consider only the algebraic sum and the algebraic antisymmetric tensor product. They are the unital algebras generated by Ψ 0 ( E 0 ) \Psi_0(\mathcal{E}_0) Ψ 0 ( E 0 ) and Ψ 0 ( E 0 , p h y s ) \Psi_0(\mathcal{E}_{0,\mathrm{phys}}) Ψ 0 ( E 0 , phys ) , respectively. Indeed, the recursive Wick rule allows one to express Wick polynomials in terms of ordinary, non-Wick ordered monomials and, conversely, ordinary, non-Wick ordered monomials in terms of Wick polynomials.
We use the previous algebras to represent the Fermionic fields and their product. In order to obtain this representation we introduce the map
W 0 : ⨁ n = 0 + ∞ E 0 ∧ n ⟶ A 0 a l g , ( u n ) n ⟼ ∑ n W n , 0 Ψ 0 ( u n ) , \mathbb{W}_0:\bigoplus_{n=0}^{+\infty}\mathcal{E}_0^{\wedge n}\longrightarrow\mathcal{A}_0^{\mathrm{alg}},\qquad(u_n)_n\longmapsto\sum_n W_{n,0}^{\Psi_0}(u_n), W 0 : n = 0 ⨁ + ∞ E 0 ∧ n ⟶ A 0 alg , ( u n ) n ⟼ n ∑ W n , 0 Ψ 0 ( u n ) ,
where ( u n ) n ∈ ⨁ n = 0 + ∞ E 0 ∧ n (u_n)_n\in\bigoplus_{n=0}^{+\infty}\mathcal{E}_0^{\wedge n} ( u n ) n ∈ ⨁ n = 0 + ∞ E 0 ∧ n .
Lemma 2 The map W 0 \mathbb{W}_0 W 0 is bijective from ⨁ n = 0 + ∞ E 0 ∧ n \bigoplus_{n=0}^{+\infty}\mathcal{E}_0^{\wedge n} ⨁ n = 0 + ∞ E 0 ∧ n into A 0 a l g \mathcal{A}_0^{\mathrm{alg}} A 0 alg , and its restriction to ⨁ n = 0 + ∞ E 0 , p h y s ∧ n \bigoplus_{n=0}^{+\infty}\mathcal{E}_{0,\mathrm{phys}}^{\wedge n} ⨁ n = 0 + ∞ E 0 , phys ∧ n is bijective onto A p h y s a l g \mathcal{A}_{\mathrm{phys}}^{\mathrm{alg}} A phys alg .
Proof Observe that W 0 \mathbb{W}_0 W 0 is by construction surjective in both instances and, therefore, we need only to prove injectivity. Suppose that ∑ n = 0 N W n , 0 Ψ 0 ( u n ) = 0 \sum_{n=0}^{N}W_{n,0}^{\Psi_0}(u_n)=0 ∑ n = 0 N W n , 0 Ψ 0 ( u n ) = 0 . Applying both sides to Ω \Omega Ω yields ∑ n = 0 N W n , 0 Ψ 0 ( u n ) = 0 \sum_{n=0}^{N}\mathbb{W}_{n,0}^{\Psi_0}(u_n)=0 ∑ n = 0 N W n , 0 Ψ 0 ( u n ) = 0 . These vectors lie in mutually orthogonal particle sectors; their injectivity by Proposition 6 implies u n = 0 u_n=0 u n = 0 for every n n n . This proves the injectivity of W 0 \mathbb{W}_0 W 0 .
We introduce an evolution structure on A 0 a l g \mathcal{A}_0^{\mathrm{alg}} A 0 alg and A p h y s a l g \mathcal{A}_{\mathrm{phys}}^{\mathrm{alg}} A phys alg induced by the combination of the time translation Θ τ \Theta_\tau Θ τ and the conditional expectation ω 0 \omega_0 ω 0 . For τ ≥ 0 \tau\geq0 τ ≥ 0 , we define K τ \mathcal{K}_\tau K τ on elements of the form W n , 0 Ψ 0 ( Q ( F 1 ) ∧ ⋯ ∧ Q ( F n ) ) W_{n,0}^{\Psi_0}(Q(F_1)\wedge\cdots\wedge Q(F_n)) W n , 0 Ψ 0 ( Q ( F 1 ) ∧ ⋯ ∧ Q ( F n )) by
K τ ( W n , 0 Ψ 0 ( Q ( F 1 ) ∧ ⋯ ∧ Q ( F n ) ) ) : = ω 0 ( W n Ψ ( Θ τ F 1 ⊗ ⋯ ⊗ Θ τ F n ) ) , (46) \mathcal{K}_\tau\left(W_{n,0}^{\Psi_0}(Q(F_1)\wedge\cdots\wedge Q(F_n))\right):=\omega_0\left(W_n^\Psi(\Theta_\tau F_1\otimes\cdots\otimes\Theta_\tau F_n)\right), \tag*{(46)} K τ ( W n , 0 Ψ 0 ( Q ( F 1 ) ∧ ⋯ ∧ Q ( F n )) ) := ω 0 ( W n Ψ ( Θ τ F 1 ⊗ ⋯ ⊗ Θ τ F n ) ) , (46)
and we set K τ I A = I A \mathcal{K}_\tau\mathbb{I}_{\mathcal{A}}=\mathbb{I}_{\mathcal{A}} K τ I A = I A .
Proposition 8 The maps K τ K_{\tau} K τ as per (46) is well defined on A 0 a l g \mathcal{A}^{\mathrm{alg}}_{0} A 0 alg , it leaves A p h y s a l g \mathcal{A}^{\mathrm{alg}}_{\mathrm{phys}} A phys alg invariant and it forms a linear semigroup which satisfies
K τ W n , 0 Ψ 0 ( u ) = W n , 0 Ψ 0 ( T τ ∧ n u ) , u ∈ E 0 ∧ n , (47) K_{\tau}W^{\Psi_{0}}_{n,0}(u)=W^{\Psi_{0}}_{n,0}\left(T_{\tau}^{\wedge n}u\right),\qquad u\in\mathcal{E}_{0}^{\wedge n}, \tag*{(47)} K τ W n , 0 Ψ 0 ( u ) = W n , 0 Ψ 0 ( T τ ∧ n u ) , u ∈ E 0 ∧ n , (47) where T τ ∧ 0 = I C T_{\tau}^{\wedge0}=\mathbb{I}_{\mathbb{C}} T τ ∧ 0 = I C . Equivalently, on the algebraic direct sum,
W 0 − 1 K τ W 0 = ⨁ n = 0 ∞ T τ ∧ n . (48) \mathbb{W}_{0}^{-1}K_{\tau}\mathbb{W}_{0}=\bigoplus_{n=0}^{\infty}T_{\tau}^{\wedge n}. \tag*{(48)} W 0 − 1 K τ W 0 = n = 0 ⨁ ∞ T τ ∧ n . (48)
Proof. For q j = Q ( F j ) q_j=Q(F_j) q j = Q ( F j ) , Proposition 3, the definition of Ψ 0 \Psi_0 Ψ 0 , and (42) give
ω 0 ( W n ψ 0 ( Θ τ F 1 ⊗ ⋯ ⊗ Θ τ F n ) ) = W n , 0 Ψ 0 ( Θ τ F 1 ⊗ ⋯ ⊗ Θ τ F n ) = W n , 0 Ψ 0 ( Q ( Θ τ F 1 ) ∧ ⋯ ∧ Q ( Θ τ F n ) ) = W n , 0 Ψ 0 ( T τ q 1 ∧ ⋯ ∧ T τ q n ) . \begin{aligned}
\omega_0\left(W_n^{\psi_0}(\Theta_{\tau}F_1\otimes\cdots\otimes\Theta_{\tau}F_n)\right)
&=W_{n,0}^{\Psi_0}(\Theta_{\tau}F_1\otimes\cdots\otimes\Theta_{\tau}F_n)\\
&=W_{n,0}^{\Psi_0}\left(Q(\Theta_{\tau}F_1)\wedge\cdots\wedge Q(\Theta_{\tau}F_n)\right)\\
&=W_{n,0}^{\Psi_0}\left(T_{\tau}q_1\wedge\cdots\wedge T_{\tau}q_n\right).
\end{aligned} ω 0 ( W n ψ 0 ( Θ τ F 1 ⊗ ⋯ ⊗ Θ τ F n ) ) = W n , 0 Ψ 0 ( Θ τ F 1 ⊗ ⋯ ⊗ Θ τ F n ) = W n , 0 Ψ 0 ( Q ( Θ τ F 1 ) ∧ ⋯ ∧ Q ( Θ τ F n ) ) = W n , 0 Ψ 0 ( T τ q 1 ∧ ⋯ ∧ T τ q n ) .
The result depends only on q 1 ∧ ⋯ ∧ q n q_1\wedge\cdots\wedge q_n q 1 ∧ ⋯ ∧ q n , namely it is unchanged when any F j F_j F j is replaced by another representative of q j q_j q j . Furthermore, by linearity it respects all relations in E 0 ∧ n \mathcal{E}_{0}^{\wedge n} E 0 ∧ n . (44) also ensures that the definition is independent of the chosen finite decomposition into Wick monomials. This proves that the map K τ K_{\tau} K τ is well-defined as per (47) . (48) implies K 0 = I K_0=\mathbb{I} K 0 = I and K τ + σ = K τ K σ K_{\tau+\sigma}=K_{\tau}K_{\sigma} K τ + σ = K τ K σ for all τ , σ ≥ 0 \tau,\sigma\geq0 τ , σ ≥ 0 . Finally, T τ T_{\tau} T τ preserves E 0 , p h y s \mathcal{E}_{0,\mathrm{phys}} E 0 , phys by Corollary 2, which implies that A p h y s a l g \mathcal{A}^{\mathrm{alg}}_{\mathrm{phys}} A phys alg is invariant. □
For n ≥ 0 n\geq0 n ≥ 0 , set
T τ , 0 ( 0 ) : = I C , T τ , 0 ( n ) : = T τ ∧ n ( n ≥ 1 ) , (49) T_{\tau,0}^{(0)}:=\mathbb{I}_{\mathbb{C}},\qquad T_{\tau,0}^{(n)}:=T_{\tau}^{\wedge n}\quad(n\geq1), \tag*{(49)} T τ , 0 ( 0 ) := I C , T τ , 0 ( n ) := T τ ∧ n ( n ≥ 1 ) , (49)
and set H 0 , 0 : = 0 H_{0,0}:=0 H 0 , 0 := 0 , while, for n ≥ 1 n\geq1 n ≥ 1 , let H n , 0 H_{n,0} H n , 0 be the closure of
H n , 0 ( q 1 ∧ ⋯ ∧ q n ) : = ∑ r = 1 n q 1 ∧ ⋯ ∧ H 1 , 0 q r ∧ ⋯ ∧ q n , q j ∈ E 0 . (50) H_{n,0}(q_1\wedge\cdots\wedge q_n):=\sum_{r=1}^{n}q_1\wedge\cdots\wedge H_{1,0}q_r\wedge\cdots\wedge q_n,\qquad q_j\in\mathcal{E}_0. \tag*{(50)} H n , 0 ( q 1 ∧ ⋯ ∧ q n ) := r = 1 ∑ n q 1 ∧ ⋯ ∧ H 1 , 0 q r ∧ ⋯ ∧ q n , q j ∈ E 0 . (50)
Proposition 9 With the vacuum norm ∥ a ∥ W : = ∥ a Ω ∥ F − ( V ) \|a\|_{\mathbb{W}}:=\|a\Omega\|_{\mathcal{F}_{-}(V)} ∥ a ∥ W := ∥ a Ω ∥ F − ( V ) , the completion of A 0 a l g \mathcal{A}^{\mathrm{alg}}_0 A 0 alg is canonically isomorphic to F 0 \mathcal{F}_0 F 0 . Under this identification, K τ K_{\tau} K τ extends to the strongly continuous self-adjoint contraction semigroup
T τ , 0 : = Γ − ( T τ ) = ⨁ n = 0 ∞ T τ , 0 ( n ) . (51) T_{\tau,0}:=\Gamma_{-}(T_{\tau})=\bigoplus_{n=0}^{\infty}T_{\tau,0}^{(n)}. \tag*{(51)} T τ , 0 := Γ − ( T τ ) = n = 0 ⨁ ∞ T τ , 0 ( n ) . (51) Its positive Hamiltonian is
T τ , 0 = e − τ H 0 , H 0 : = d Γ ( H 1 , 0 ) = ⨁ n = 0 ∞ H n , 0 , (52) T_{\tau,0}=e^{-\tau H_0},\qquad H_0:=\mathrm{d}\Gamma(H_{1,0})=\bigoplus_{n=0}^{\infty}H_{n,0}, \tag*{(52)} T τ , 0 = e − τ H 0 , H 0 := d Γ ( H 1 , 0 ) = n = 0 ⨁ ∞ H n , 0 , (52) where H n , 0 = d Γ n ( H 1 , 0 ) H_{n,0}=\mathrm{d}\Gamma_n(H_{1,0}) H n , 0 = d Γ n ( H 1 , 0 ) . The set ⨁ n = 0 + ∞ E 0 ∧ n \bigoplus_{n=0}^{+\infty}\mathcal{E}_{0}^{\wedge n} ⨁ n = 0 + ∞ E 0 ∧ n is a core and we have, for q 1 , … , q n ∈ E 0 q_1,\ldots,q_n\in\mathcal{E}_0 q 1 , … , q n ∈ E 0 ,
T τ , 0 W n , 0 Ψ 0 ( q 1 , … , q n ) Ω = W n , 0 Ψ 0 ( T τ q 1 , … , T τ q n ) Ω , (53) T_{\tau,0}W_{n,0}^{\Psi_0}(q_1,\ldots,q_n)\Omega=W_{n,0}^{\Psi_0}(T_{\tau}q_1,\ldots,T_{\tau}q_n)\Omega, \tag*{(53)} T τ , 0 W n , 0 Ψ 0 ( q 1 , … , q n ) Ω = W n , 0 Ψ 0 ( T τ q 1 , … , T τ q n ) Ω , (53) H 0 W n , 0 Ψ 0 ( q 1 , … , q n ) Ω = ∑ r = 1 n W n , 0 Ψ 0 ( q 1 , … , H 1 , 0 q r , … , q n ) Ω . (54) H_0W_{n,0}^{\Psi_0}(q_1,\ldots,q_n)\Omega=\sum_{r=1}^{n}W_{n,0}^{\Psi_0}(q_1,\ldots,H_{1,0}q_r,\ldots,q_n)\Omega. \tag*{(54)} H 0 W n , 0 Ψ 0 ( q 1 , … , q n ) Ω = r = 1 ∑ n W n , 0 Ψ 0 ( q 1 , … , H 1 , 0 q r , … , q n ) Ω. (54)
Proof. For a = W 0 ( ( u n ) n ) a=\mathbb{W}_0((u_n)_n) a = W 0 (( u n ) n ) , Proposition 6 and orthogonality of particle sectors give
∥ a ∥ W 2 = ∑ n ∥ u n ∥ 2 , ∥ K τ a ∥ W 2 = ∑ n ∥ T τ ∧ n u n ∥ 2 ≤ ∥ a ∥ W 2 . \|a\|_{\mathbb{W}}^2=\sum_n\|u_n\|^2,\qquad\|K_{\tau}a\|_{\mathbb{W}}^2=\sum_n\|T_{\tau}^{\wedge n}u_n\|^2\leq\|a\|_{\mathbb{W}}^2. ∥ a ∥ W 2 = n ∑ ∥ u n ∥ 2 , ∥ K τ a ∥ W 2 = n ∑ ∥ T τ ∧ n u n ∥ 2 ≤ ∥ a ∥ W 2 .
Thus W 0 − 1 \mathbb{W}_0^{-1} W 0 − 1 extends to the stated isometric identification, and (48) becomes (51) . More precisely, on Wick vectors in F − ( V ) \mathcal{F}_{-}(V) F − ( V ) ,
( K τ a ) Ω = Γ − ( J 0 ) T τ , 0 Γ − ( J 0 ) − 1 ( a Ω ) , (K_{\tau}a)\Omega=\Gamma_{-}(J_0)T_{\tau,0}\Gamma_{-}(J_0)^{-1}(a\Omega), ( K τ a ) Ω = Γ − ( J 0 ) T τ , 0 Γ − ( J 0 ) − 1 ( a Ω ) ,
with the inverse taken on the range of Γ − ( J 0 ) \Gamma_{-}(J_0) Γ − ( J 0 ) . Each T τ ∧ n T_{\tau}^{\wedge n} T τ ∧ n is a self-adjoint contraction and it is strongly continuous on H 0 ( n ) \mathcal{H}_{0}^{(n)} H 0 ( n ) . To conclude, the contraction bound extends strong continuity from finite-particle vectors to the whole F 0 \mathcal{F}_0 F 0 .
Differentiating T τ ∧ n T_{\tau}^{\wedge n} T τ ∧ n at τ = 0 \tau=0 τ = 0 on E 0 ∧ n \mathcal{E}_{0}^{\wedge n} E 0 ∧ n gives minus the sum in (50) . Since E 0 \mathcal{E}_0 E 0 is a core for H 1 , 0 H_{1,0} H 1 , 0 as per Proposition 7, these finite wedge products form a core for H n , 0 = d Γ n ( H 1 , 0 ) H_{n,0}=\mathrm{d}\Gamma_n(H_{1,0}) H n , 0 = d Γ n ( H 1 , 0 ) . The generator of the direct-sum semigroup is consequently − H 0 -H_0 − H 0 with core ⨁ n = 0 + ∞ E 0 ∧ n \bigoplus_{n=0}^{+\infty}\mathcal{E}_{0}^{\wedge n} ⨁ n = 0 + ∞ E 0 ∧ n . The identities on Wick vectors follow from (39) . □
Define the physical Fock projection by
Π p h y s : = Γ − ( P p h y s ) = ⨁ n = 0 ∞ P p h y s ∧ n , Ran Π p h y s = F p h y s . (55) \Pi_{\mathrm{phys}}:=\Gamma_{-}(\mathbb{P}_{\mathrm{phys}})=\bigoplus_{n=0}^{\infty}\mathbb{P}_{\mathrm{phys}}^{\wedge n},\qquad\operatorname{Ran}\Pi_{\mathrm{phys}}=\mathcal{F}_{\mathrm{phys}}. \tag*{(55)} Π phys := Γ − ( P phys ) = n = 0 ⨁ ∞ P phys ∧ n , Ran Π phys = F phys . (55)
Corollary 3. The restriction of K τ \mathcal{K}_{\tau} K τ to A p h y s a l g \mathcal{A}^{\mathrm{alg}}_{\mathrm{phys}} A phys alg extends in the vacuum norm to
T τ , p h y s : = T τ , 0 ∣ F p h y s = Γ − ( T τ p h y s ) = ⨁ n = 0 ∞ ( T τ p h y s ) ∧ n . (56) \mathbb{T}_{\tau,\mathrm{phys}} := \mathbb{T}_{\tau,0}|_{\mathcal{F}_{\mathrm{phys}}} = \Gamma_{-}\left(\mathbb{T}^{\mathrm{phys}}_{\tau}\right) = \bigoplus_{n=0}^{\infty}\left(\mathbb{T}^{\mathrm{phys}}_{\tau}\right)^{\wedge n}. \tag*{(56)} T τ , phys := T τ , 0 ∣ F phys = Γ − ( T τ phys ) = n = 0 ⨁ ∞ ( T τ phys ) ∧ n . (56) It is a strongly continuous self-adjoint contraction semigroup with
T τ , p h y s = e − τ H p h y s , H p h y s : = d Γ ( H 1 , p h y s ) = H 0 ∣ F p h y s . (57) \mathbb{T}_{\tau,\mathrm{phys}} = e^{-\tau H_{\mathrm{phys}}}, \qquad H_{\mathrm{phys}} := \mathrm{d}\Gamma(H_{1,\mathrm{phys}}) = H_{0}|_{\mathcal{F}_{\mathrm{phys}}}. \tag*{(57)} T τ , phys = e − τ H phys , H phys := d Γ ( H 1 , phys ) = H 0 ∣ F phys . (57) The auxiliary Wick-vector identities above restrict to q j ∈ E 0 , p h y s q_j \in\mathcal{E}_{0,\mathrm{phys}} q j ∈ E 0 , phys with T τ \mathbb{T}_{\tau} T τ , H 1 , 0 H_{1,0} H 1 , 0 , H 0 H_0 H 0 replaced by T τ p h y s \mathbb{T}^{\mathrm{phys}}_{\tau} T τ phys , H 1 , p h y s H_{1,\mathrm{phys}} H 1 , phys , H p h y s H_{\mathrm{phys}} H phys , respectively.
Proof. The relation P p h y s T τ = T τ P p h y s \mathbb{P}_{\mathrm{phys}}\mathbb{T}_{\tau}=\mathbb{T}_{\tau}\mathbb{P}_{\mathrm{phys}} P phys T τ = T τ P phys implies Π p h y s T τ , 0 = T τ , 0 Π p h y s \Pi_{\mathrm{phys}}\mathbb{T}_{\tau,0}=\mathbb{T}_{\tau,0}\Pi_{\mathrm{phys}} Π phys T τ , 0 = T τ , 0 Π phys . Thus F p h y s \mathcal{F}_{\mathrm{phys}} F phys is reducing for the auxiliary semigroup and its generator, proving all the statements by restriction. □ \square □
Appendix
We present the proof of Proposition 3. We also repeat the statement for the sake of readability.
Proposition. For any n ∈ N n \in\mathbb{N} n ∈ N and for any F 1 , … , F n ∈ S ( R + 2 ; C 4 ) F_1,\ldots,F_n \in\mathcal{S}(\mathbb{R}_{+}^{2};\mathbb{C}^{4}) F 1 , … , F n ∈ S ( R + 2 ; C 4 ) ,
ω 0 ( W n ψ ( F 1 ⊗ ⋯ ⊗ F n ) ) = W n ψ 0 ( F 1 ⊗ ⋯ ⊗ F n ) . \omega_0\left(W_n^{\psi}(F_1\otimes\cdots\otimes F_n)\right)=W_n^{\psi_0}(F_1\otimes\cdots\otimes F_n). ω 0 ( W n ψ ( F 1 ⊗ ⋯ ⊗ F n ) ) = W n ψ 0 ( F 1 ⊗ ⋯ ⊗ F n ) .
Proof. We introduce the notations
ψ > ( F ) : = ψ ( F ) − ω 0 ( ψ ( F ) ) = ψ ( F ) − ψ 0 ( F ) , ∀ F ∈ S ( R + 2 ; C 4 ) (58) \psi_{>}(F):=\psi(F)-\omega_0(\psi(F))=\psi(F)-\psi_0(F),\qquad\forall F\in\mathcal{S}(\mathbb{R}_{+}^{2};\mathbb{C}^{4}) \tag*{(58)} ψ > ( F ) := ψ ( F ) − ω 0 ( ψ ( F )) = ψ ( F ) − ψ 0 ( F ) , ∀ F ∈ S ( R + 2 ; C 4 ) (58)
and
W n ψ ( F 1 > ⊗ ⋯ ⊗ F n ) : = : ψ > ( F 1 ) ⋯ ψ ( F n ) : ∀ F 1 , … , F n ∈ S ( R + 2 ; C 4 ) , (59) W_n^{\psi}(F_{1>} \otimes\cdots\otimes F_n):=:\psi_{>}(F_1)\cdots\psi(F_n):\qquad\forall F_1,\ldots,F_n\in\mathcal{S}(\mathbb{R}_{+}^{2};\mathbb{C}^{4}), \tag*{(59)} W n ψ ( F 1 > ⊗ ⋯ ⊗ F n ) :=: ψ > ( F 1 ) ⋯ ψ ( F n ) : ∀ F 1 , … , F n ∈ S ( R + 2 ; C 4 ) , (59)
where F 1 > F_{1>} F 1 > is a convenient notation and it should not be read as an element of S ( R + 2 ; C 4 ) \mathcal{S}(\mathbb{R}_{+}^{2};\mathbb{C}^{4}) S ( R + 2 ; C 4 ) . Observe that, per construction, ω 0 ( ψ > ( F ) ) = 0 \omega_0(\psi_{>}(F))=0 ω 0 ( ψ > ( F )) = 0 . Bearing this property in mind, we prove the statement by induction on the number of fields in the Wick monomials. Starting from n = 1 n=1 n = 1 , Equations (16) and (19), combined with the notations introduced in Equations (18) and (20), yield
ω 0 ( W 1 ψ ( F ) ) = ω 0 ( ψ ( F ) ) = ψ 0 ( F ) = W 1 ψ 0 ( F ) . \omega_0(W_1^{\psi}(F))=\omega_0(\psi(F))=\psi_0(F)=W_1^{\psi_0}(F). ω 0 ( W 1 ψ ( F )) = ω 0 ( ψ ( F )) = ψ 0 ( F ) = W 1 ψ 0 ( F ) .
Following the inductive hypothesis we can assume the statement to be true at order n − 1 n-1 n − 1 in the number of fields. Using (17) which is valid also for n = 2 n=2 n = 2 , in combination with the linearity of ω 0 \omega_0 ω 0 , it holds that
ω 0 ( W n ψ ( F 1 ⊗ ⋯ ⊗ F n ) ) = ω 0 [ ψ ( F 1 ) W n − 1 ψ ( F 2 ⊗ ⋯ ⊗ F n ) ] − ∑ j = 2 n ( − 1 ) j ω ( ψ ( F 1 ) ψ ( F j ) ) ω 0 ( W n − 2 ψ ( F 2 ⊗ ⋯ ⊗ F j ~ ⊗ ⋯ ⊗ F n ) ) = ω 0 [ ψ ( F 1 ) W n − 1 ψ ( F 2 ⊗ ⋯ ⊗ F n ) ] − ∑ j = 2 n ( − 1 ) j ω ( ψ ( F 1 ) ψ ( F j ) ) W n − 2 ψ ( F 2 ⊗ ⋯ ⊗ F j ~ ⊗ ⋯ ⊗ F n ) = ω 0 [ ψ ( F 1 ) W n − 1 ψ ( F 2 ⊗ ⋯ ⊗ F n ) ] − ∑ j = 2 n ( − 1 ) j ω ( ψ 0 ( F 1 ) ψ 0 ( F j ) ) W n − 2 ψ 0 ( F 2 ⊗ ⋯ ⊗ F j ~ ⊗ ⋯ ⊗ F n ) − ∑ j = 2 n ( − 1 ) j ω ( ψ > ( F 1 ) ψ > ( F j ) ) W n − 2 ψ 0 ( F 2 ⊗ ⋯ ⊗ F j ~ ⊗ ⋯ ⊗ F n ) . (60) \begin{aligned}
\omega_0\left(W_n^{\psi}(F_1\otimes\cdots\otimes F_n)\right)
&=\omega_0\left[\psi(F_1)W_{n-1}^{\psi}(F_2\otimes\cdots\otimes F_n)\right]\\
&\quad-\sum_{j=2}^{n}(-1)^j\omega(\psi(F_1)\psi(F_j))\omega_0\left(W_{n-2}^{\psi}(F_2\otimes\cdots\otimes\widetilde{F_j}\otimes\cdots\otimes F_n)\right)\\
&=\omega_0\left[\psi(F_1)W_{n-1}^{\psi}(F_2\otimes\cdots\otimes F_n)\right]\\
&\quad-\sum_{j=2}^{n}(-1)^j\omega(\psi(F_1)\psi(F_j))W_{n-2}^{\psi}(F_2\otimes\cdots\otimes\widetilde{F_j}\otimes\cdots\otimes F_n)\\
&=\omega_0\left[\psi(F_1)W_{n-1}^{\psi}(F_2\otimes\cdots\otimes F_n)\right]\\
&\quad-\sum_{j=2}^{n}(-1)^j\omega(\psi_0(F_1)\psi_0(F_j))W_{n-2}^{\psi_0}(F_2\otimes\cdots\otimes\widetilde{F_j}\otimes\cdots\otimes F_n)\\
&\quad-\sum_{j=2}^{n}(-1)^j\omega(\psi_{>}(F_1)\psi_{>}(F_j))W_{n-2}^{\psi_0}(F_2\otimes\cdots\otimes\widetilde{F_j}\otimes\cdots\otimes F_n). \tag*{(60)}
\end{aligned} ω 0 ( W n ψ ( F 1 ⊗ ⋯ ⊗ F n ) ) = ω 0 [ ψ ( F 1 ) W n − 1 ψ ( F 2 ⊗ ⋯ ⊗ F n ) ] − j = 2 ∑ n ( − 1 ) j ω ( ψ ( F 1 ) ψ ( F j )) ω 0 ( W n − 2 ψ ( F 2 ⊗ ⋯ ⊗ F j ⊗ ⋯ ⊗ F n ) ) = ω 0 [ ψ ( F 1 ) W n − 1 ψ ( F 2 ⊗ ⋯ ⊗ F n ) ] − j = 2 ∑ n ( − 1 ) j ω ( ψ ( F 1 ) ψ ( F j )) W n − 2 ψ ( F 2 ⊗ ⋯ ⊗ F j ⊗ ⋯ ⊗ F n ) = ω 0 [ ψ ( F 1 ) W n − 1 ψ ( F 2 ⊗ ⋯ ⊗ F n ) ] − j = 2 ∑ n ( − 1 ) j ω ( ψ 0 ( F 1 ) ψ 0 ( F j )) W n − 2 ψ 0 ( F 2 ⊗ ⋯ ⊗ F j ⊗ ⋯ ⊗ F n ) − j = 2 ∑ n ( − 1 ) j ω ( ψ > ( F 1 ) ψ > ( F j )) W n − 2 ψ 0 ( F 2 ⊗ ⋯ ⊗ F j ⊗ ⋯ ⊗ F n ) . (60)
where the tilde denotes a removed element. Herein we exploited the inductive hypothesis, that ω 0 \omega_0 ω 0 defines a conditional expectation as per Definition 5, as well as that, as a by-product, of Definition 7
ω ( ψ 0 ( F 1 ) ψ > ( F j ) ) = 0. (61) \omega(\psi_0(F_1)\psi_{>}(F_j))=0. \tag*{(61)} ω ( ψ 0 ( F 1 ) ψ > ( F j )) = 0. (61)
Focusing on the first term, the properties of conditional expectations and (58) entail
ω 0 [ ψ ( F 1 ) W n − 1 ψ ( F 2 ⊗ ⋯ ⊗ F n ) ] = ω 0 [ ψ 0 ( F 1 ) W n − 1 ψ ( F 2 ⊗ ⋯ ⊗ F n ) ] + ω 0 [ ψ > ( F 1 ) W n − 1 ψ ( F 2 ⊗ ⋯ ⊗ F n ) ] = ψ 0 ( F 1 ) W n − 1 ψ 0 ( F 2 ⊗ ⋯ ⊗ F n ) + ω 0 [ ψ > ( F 1 ) W n − 1 ψ ( F 2 ⊗ ⋯ ⊗ F n ) ] . (62) \begin{aligned}
\omega_0\left[\psi(F_1)W_{n-1}^{\psi}(F_2\otimes\cdots\otimes F_n)\right]
&=\omega_0\left[\psi_0(F_1)W_{n-1}^{\psi}(F_2\otimes\cdots\otimes F_n)\right]\\
&\quad+\omega_0\left[\psi_{>}(F_1)W_{n-1}^{\psi}(F_2\otimes\cdots\otimes F_n)\right]\\
&=\psi_0(F_1)W_{n-1}^{\psi_0}(F_2\otimes\cdots\otimes F_n)
+\omega_0\left[\psi_{>}(F_1)W_{n-1}^{\psi}(F_2\otimes\cdots\otimes F_n)\right]. \tag*{(62)}
\end{aligned} ω 0 [ ψ ( F 1 ) W n − 1 ψ ( F 2 ⊗ ⋯ ⊗ F n ) ] = ω 0 [ ψ 0 ( F 1 ) W n − 1 ψ ( F 2 ⊗ ⋯ ⊗ F n ) ] + ω 0 [ ψ > ( F 1 ) W n − 1 ψ ( F 2 ⊗ ⋯ ⊗ F n ) ] = ψ 0 ( F 1 ) W n − 1 ψ 0 ( F 2 ⊗ ⋯ ⊗ F n ) + ω 0 [ ψ > ( F 1 ) W n − 1 ψ ( F 2 ⊗ ⋯ ⊗ F n ) ] . (62)
where, employing the identity ω ( ψ > ( F ) ψ ( F ′ ) ) = ω ( ψ > ( F ) ψ > ( F ′ ) ) \omega(\psi_{>}(F)\psi(F'))=\omega(\psi_{>}(F)\psi_{>}(F')) ω ( ψ > ( F ) ψ ( F ′ )) = ω ( ψ > ( F ) ψ > ( F ′ )) for all F , F ′ ∈ S ( R + 2 ; C 4 ) F,F'\in\mathcal{S}(\mathbb{R}_{+}^{2};\mathbb{C}^{4}) F , F ′ ∈ S ( R + 2 ; C 4 ) ,
ω 0 [ ψ > ( F 1 ) W n − 1 ψ ( F 2 ⊗ ⋯ ⊗ F n ) ] = ω 0 [ W n ψ ( F 1 ⊗ F 2 ⊗ ⋯ ⊗ F n ) ] + ∑ j = 2 n ( − 1 ) j ω ( ψ > ( F 1 ) ψ > ( F j ) ) ω 0 [ W n − 2 ψ ( F 2 ⊗ ⋯ ⊗ F j ~ ⊗ ⋯ ⊗ F n ) ] = ω 0 [ W n ψ ( F 1 > ⊗ F 2 ⊗ ⋯ ⊗ F n ) ] + ∑ j = 2 n ( − 1 ) j ω ( ψ > ( F 1 ) ψ > ( F j ) ) W n − 2 ψ 0 ( F 2 ⊗ ⋯ ⊗ F j ~ ⊗ ⋯ ⊗ F n ) . (63) \begin{aligned}
\omega_0\left[\psi_{>}(F_1)W_{n-1}^{\psi}(F_2\otimes\cdots\otimes F_n)\right]
&=\omega_0\left[W_n^{\psi}(F_1\otimes F_2\otimes\cdots\otimes F_n)\right]\\
&\quad+\sum_{j=2}^{n}(-1)^j\omega(\psi_{>}(F_1)\psi_{>}(F_j))\omega_0\left[W_{n-2}^{\psi}(F_2\otimes\cdots\otimes\widetilde{F_j}\otimes\cdots\otimes F_n)\right]\\
&=\omega_0\left[W_n^{\psi}(F_{1>}\otimes F_2\otimes\cdots\otimes F_n)\right]\\
&\quad+\sum_{j=2}^{n}(-1)^j\omega(\psi_{>}(F_1)\psi_{>}(F_j))W_{n-2}^{\psi_0}(F_2\otimes\cdots\otimes\widetilde{F_j}\otimes\cdots\otimes F_n). \tag*{(63)}
\end{aligned} ω 0 [ ψ > ( F 1 ) W n − 1 ψ ( F 2 ⊗ ⋯ ⊗ F n ) ] = ω 0 [ W n ψ ( F 1 ⊗ F 2 ⊗ ⋯ ⊗ F n ) ] + j = 2 ∑ n ( − 1 ) j ω ( ψ > ( F 1 ) ψ > ( F j )) ω 0 [ W n − 2 ψ ( F 2 ⊗ ⋯ ⊗ F j ⊗ ⋯ ⊗ F n ) ] = ω 0 [ W n ψ ( F 1 > ⊗ F 2 ⊗ ⋯ ⊗ F n ) ] + j = 2 ∑ n ( − 1 ) j ω ( ψ > ( F 1 ) ψ > ( F j )) W n − 2 ψ 0 ( F 2 ⊗ ⋯ ⊗ F j ⊗ ⋯ ⊗ F n ) . (63)
::: Inserting Equations (60) and (61) in Equation (58), we end up with
ω 0 ( W n ψ ( F 1 ⊗ ⋯ ⊗ F n ) ) = ψ 0 ( F 1 ) W n − 1 ψ 0 ( F 2 ⊗ ⋯ ⊗ F n ) + ω 0 [ W n ψ ( F 1 > ⊗ F 2 ⊗ ⋯ ⊗ F n ) ] + ∑ j = 2 n ( − 1 ) j ω ( ψ > ( F 1 ) ψ > ( F j ) ) W n − 2 ψ 0 ( F 2 ⊗ ⋯ ⊗ F j ~ ⊗ ⋯ ⊗ F n ) − ∑ j = 2 n ( − 1 ) j ω ( ψ 0 ( F 1 ) ψ 0 ( F j ) ) W n − 2 ψ 0 ( F 2 ⊗ ⋯ ⊗ F j ~ ⊗ ⋯ ⊗ F n ) + ∑ j = 2 n ( − 1 ) j ω ( ψ > ( F 1 ) ψ > ( F j ) ) W n − 2 ψ 0 ( F 2 ⊗ ⋯ ⊗ F j ~ ⊗ ⋯ ⊗ F n ) = ψ 0 ( F 1 ) W n − 1 ψ 0 ( F 2 ⊗ ⋯ ⊗ F n ) + ω 0 [ W n ψ ( F 1 > ⊗ F 2 ⊗ ⋯ ⊗ F n ) ] − ∑ j = 2 n ( − 1 ) j ω ( ψ 0 ( F 1 ) ψ 0 ( F j ) ) W n − 2 ψ 0 ( F 2 ⊗ ⋯ ⊗ F j ~ ⊗ ⋯ ⊗ F n ) = W n ψ 0 ( F 1 ⊗ ⋯ ⊗ F n ) + ω 0 [ W n ψ ( F 1 > ⊗ F 2 ⊗ ⋯ ⊗ F n ) ] . \begin{aligned}
\omega_0\left(W_n^\psi(F_1\otimes\cdots\otimes F_n)\right)
&=\psi_0(F_1)W_{n-1}^{\psi_0}(F_2\otimes\cdots\otimes F_n)+\omega_0\left[W_n^\psi(F_1^{>}\otimes F_2\otimes\cdots\otimes F_n)\right]\\
&\quad+\sum_{j=2}^{n}(-1)^j\omega(\psi_{>}(F_1)\psi_{>}(F_j))W_{n-2}^{\psi_0}(F_2\otimes\cdots\otimes\widetilde{F_j}\otimes\cdots\otimes F_n)\\
&\quad-\sum_{j=2}^{n}(-1)^j\omega(\psi_0(F_1)\psi_0(F_j))W_{n-2}^{\psi_0}(F_2\otimes\cdots\otimes\widetilde{F_j}\otimes\cdots\otimes F_n)\\
&\quad+\sum_{j=2}^{n}(-1)^j\omega(\psi_{>}(F_1)\psi_{>}(F_j))W_{n-2}^{\psi_0}(F_2\otimes\cdots\otimes\widetilde{F_j}\otimes\cdots\otimes F_n)\\
&=\psi_0(F_1)W_{n-1}^{\psi_0}(F_2\otimes\cdots\otimes F_n)+\omega_0\left[W_n^\psi(F_1^{>}\otimes F_2\otimes\cdots\otimes F_n)\right]\\
&\quad-\sum_{j=2}^{n}(-1)^j\omega(\psi_0(F_1)\psi_0(F_j))W_{n-2}^{\psi_0}(F_2\otimes\cdots\otimes\widetilde{F_j}\otimes\cdots\otimes F_n)\\
&=W_n^{\psi_0}(F_1\otimes\cdots\otimes F_n)+\omega_0\left[W_n^\psi(F_1^{>}\otimes F_2\otimes\cdots\otimes F_n)\right].
\end{aligned} ω 0 ( W n ψ ( F 1 ⊗ ⋯ ⊗ F n ) ) = ψ 0 ( F 1 ) W n − 1 ψ 0 ( F 2 ⊗ ⋯ ⊗ F n ) + ω 0 [ W n ψ ( F 1 > ⊗ F 2 ⊗ ⋯ ⊗ F n ) ] + j = 2 ∑ n ( − 1 ) j ω ( ψ > ( F 1 ) ψ > ( F j )) W n − 2 ψ 0 ( F 2 ⊗ ⋯ ⊗ F j ⊗ ⋯ ⊗ F n ) − j = 2 ∑ n ( − 1 ) j ω ( ψ 0 ( F 1 ) ψ 0 ( F j )) W n − 2 ψ 0 ( F 2 ⊗ ⋯ ⊗ F j ⊗ ⋯ ⊗ F n ) + j = 2 ∑ n ( − 1 ) j ω ( ψ > ( F 1 ) ψ > ( F j )) W n − 2 ψ 0 ( F 2 ⊗ ⋯ ⊗ F j ⊗ ⋯ ⊗ F n ) = ψ 0 ( F 1 ) W n − 1 ψ 0 ( F 2 ⊗ ⋯ ⊗ F n ) + ω 0 [ W n ψ ( F 1 > ⊗ F 2 ⊗ ⋯ ⊗ F n ) ] − j = 2 ∑ n ( − 1 ) j ω ( ψ 0 ( F 1 ) ψ 0 ( F j )) W n − 2 ψ 0 ( F 2 ⊗ ⋯ ⊗ F j ⊗ ⋯ ⊗ F n ) = W n ψ 0 ( F 1 ⊗ ⋯ ⊗ F n ) + ω 0 [ W n ψ ( F 1 > ⊗ F 2 ⊗ ⋯ ⊗ F n ) ] .
In order to tame the last term, we observe that
− ω 0 [ W n ψ ( F 1 > ⊗ F 2 ⊗ ⋯ ⊗ F n ) ] = ω 0 [ W n ψ ( F 2 > ⊗ F 1 ⊗ ⋯ ⊗ F n ) ] . -\omega_0\left[W_n^\psi(F_1^{>}\otimes F_2\otimes\cdots\otimes F_n)\right]=\omega_0\left[W_n^\psi(F_2^{>}\otimes F_1\otimes\cdots\otimes F_n)\right]. − ω 0 [ W n ψ ( F 1 > ⊗ F 2 ⊗ ⋯ ⊗ F n ) ] = ω 0 [ W n ψ ( F 2 > ⊗ F 1 ⊗ ⋯ ⊗ F n ) ] .
If we iterate the above procedure, we obtain on account of the Wick rule,
ω 0 [ W n ψ ( F 2 ⊗ F 1 > ⊗ ⋯ ⊗ F n ) ] = ω 0 [ ψ ( F 2 ) W n − 1 ψ ( F 1 > ⊗ F 2 ~ ⊗ ⋯ ⊗ F n ) ] − ω ( ψ ( F 2 ) ψ > ( F 1 ) ) ω 0 ( W n − 2 ψ ( F 3 ⊗ ⋯ ⊗ F n ) ) − ∑ j = 3 n ( − 1 ) j − 1 ω ( ψ ( F 2 ) ψ ( F j ) ) ω 0 ( W n − 2 ψ ( F 1 > ⊗ F 2 ~ ⊗ F 3 ⊗ ⋯ ⊗ F j ~ ⊗ ⋯ ⊗ F n ) ) . (64) \begin{aligned}
\omega_0\left[W_n^\psi(F_2\otimes F_1^{>}\otimes\cdots\otimes F_n)\right]
&=\omega_0\left[\psi(F_2)W_{n-1}^\psi(F_1^{>}\otimes\widetilde{F_2}\otimes\cdots\otimes F_n)\right]\\
&\quad-\omega(\psi(F_2)\psi_{>}(F_1))\omega_0\left(W_{n-2}^\psi(F_3\otimes\cdots\otimes F_n)\right)\\
&\quad-\sum_{j=3}^{n}(-1)^{j-1}\omega(\psi(F_2)\psi(F_j))\omega_0\left(W_{n-2}^\psi(F_1^{>}\otimes\widetilde{F_2}\otimes F_3\otimes\cdots\otimes\widetilde{F_j}\otimes\cdots\otimes F_n)\right). \tag*{(64)}
\end{aligned} ω 0 [ W n ψ ( F 2 ⊗ F 1 > ⊗ ⋯ ⊗ F n ) ] = ω 0 [ ψ ( F 2 ) W n − 1 ψ ( F 1 > ⊗ F 2 ⊗ ⋯ ⊗ F n ) ] − ω ( ψ ( F 2 ) ψ > ( F 1 )) ω 0 ( W n − 2 ψ ( F 3 ⊗ ⋯ ⊗ F n ) ) − j = 3 ∑ n ( − 1 ) j − 1 ω ( ψ ( F 2 ) ψ ( F j )) ω 0 ( W n − 2 ψ ( F 1 > ⊗ F 2 ⊗ F 3 ⊗ ⋯ ⊗ F j ⊗ ⋯ ⊗ F n ) ) . (64)
Using the inductive hypothesis and ω 0 ( ψ > ( F ) ) = 0 \omega_0(\psi_{>}(F))=0 ω 0 ( ψ > ( F )) = 0 for all F ∈ S ( R + 2 ; C 4 ) F\in S(\mathbb{R}_{+}^{2};\mathbb{C}^{4}) F ∈ S ( R + 2 ; C 4 ) ,
ω 0 ( W n − 2 ψ ( F 3 ⊗ ⋯ ⊗ F n ) ) = W n − 2 ψ 0 ( F 3 ⊗ ⋯ ⊗ F n ) , \omega_0\left(W_{n-2}^\psi(F_3\otimes\cdots\otimes F_n)\right)=W_{n-2}^{\psi_0}(F_3\otimes\cdots\otimes F_n), ω 0 ( W n − 2 ψ ( F 3 ⊗ ⋯ ⊗ F n ) ) = W n − 2 ψ 0 ( F 3 ⊗ ⋯ ⊗ F n ) ,
ω 0 ( W n − 2 ψ ( F 1 > ⊗ F 2 ~ ⊗ F 3 ⊗ ⋯ ⊗ F j ~ ⊗ ⋯ ⊗ F n ) ) = W n − 2 ψ 0 ( ω 0 ( ψ > ( F 1 ) ) F 3 ⊗ ⋯ ⊗ F j ~ ⊗ ⋯ ⊗ F n ) = 0. \omega_0\left(W_{n-2}^\psi(F_1^{>}\otimes\widetilde{F_2}\otimes F_3\otimes\cdots\otimes\widetilde{F_j}\otimes\cdots\otimes F_n)\right)=W_{n-2}^{\psi_0}\left(\omega_0(\psi_{>}(F_1))F_3\otimes\cdots\otimes\widetilde{F_j}\otimes\cdots\otimes F_n\right)=0. ω 0 ( W n − 2 ψ ( F 1 > ⊗ F 2 ⊗ F 3 ⊗ ⋯ ⊗ F j ⊗ ⋯ ⊗ F n ) ) = W n − 2 ψ 0 ( ω 0 ( ψ > ( F 1 )) F 3 ⊗ ⋯ ⊗ F j ⊗ ⋯ ⊗ F n ) = 0.
Using once more ω 0 ( ψ > ( F ) ) = 0 \omega_0(\psi_{>}(F))=0 ω 0 ( ψ > ( F )) = 0 , ω ( ψ > ( F ) ψ ( F ′ ) ) = ω ( ψ > ( F ) ψ > ( F ′ ) ) \omega(\psi_{>}(F)\psi(F'))=\omega(\psi_{>}(F)\psi_{>}(F')) ω ( ψ > ( F ) ψ ( F ′ )) = ω ( ψ > ( F ) ψ > ( F ′ )) for all F , F ′ ∈ S ( R + 2 ; C 4 ) F,F'\in S(\mathbb{R}_{+}^{2};\mathbb{C}^{4}) F , F ′ ∈ S ( R + 2 ; C 4 ) , (64) reads
ω 0 [ W n ψ ( F 2 ⊗ F 1 > ⊗ ⋯ ⊗ F n ) ] = ω 0 [ ψ ( F 2 ) W n − 1 ψ ( F 1 > ⊗ F 2 ~ ⊗ ⋯ ⊗ F n ) ] − ω ( ψ ( F 2 ) ψ > ( F 1 ) ) W n − 2 ψ 0 ( F 3 ⊗ ⋯ ⊗ F n ) = ω 0 [ ψ > ( F 2 ) W n − 1 ψ ( F 1 > ⊗ F 2 ~ ⊗ F 3 ⊗ ⋯ ⊗ F n ) ] − ω ( ψ > ( F 2 ) ψ > ( F 1 ) ) W n − 2 ψ 0 ( F 3 ⊗ ⋯ ⊗ F n ) . (65) \begin{aligned}
\omega_0\left[W_n^\psi(F_2\otimes F_1^{>}\otimes\cdots\otimes F_n)\right]
&=\omega_0\left[\psi(F_2)W_{n-1}^\psi(F_1^{>}\otimes\widetilde{F_2}\otimes\cdots\otimes F_n)\right]\\
&\quad-\omega(\psi(F_2)\psi_{>}(F_1))W_{n-2}^{\psi_0}(F_3\otimes\cdots\otimes F_n)\\
&=\omega_0\left[\psi_{>}(F_2)W_{n-1}^\psi(F_1^{>}\otimes\widetilde{F_2}\otimes F_3\otimes\cdots\otimes F_n)\right]\\
&\quad-\omega(\psi_{>}(F_2)\psi_{>}(F_1))W_{n-2}^{\psi_0}(F_3\otimes\cdots\otimes F_n). \tag*{(65)}
\end{aligned} ω 0 [ W n ψ ( F 2 ⊗ F 1 > ⊗ ⋯ ⊗ F n ) ] = ω 0 [ ψ ( F 2 ) W n − 1 ψ ( F 1 > ⊗ F 2 ⊗ ⋯ ⊗ F n ) ] − ω ( ψ ( F 2 ) ψ > ( F 1 )) W n − 2 ψ 0 ( F 3 ⊗ ⋯ ⊗ F n ) = ω 0 [ ψ > ( F 2 ) W n − 1 ψ ( F 1 > ⊗ F 2 ⊗ F 3 ⊗ ⋯ ⊗ F n ) ] − ω ( ψ > ( F 2 ) ψ > ( F 1 )) W n − 2 ψ 0 ( F 3 ⊗ ⋯ ⊗ F n ) . (65)
where we used implicitly the following chain of identities:
ω 0 [ ψ ( F 2 ) W n − 1 ψ ( F 1 > ⊗ F 2 ~ ⊗ ⋯ ⊗ F n ) ] = ω 0 [ ψ > ( F 2 ) W n − 1 ψ ( F 1 > ⊗ F 2 ~ ⊗ ⋯ ⊗ F n ) ] + ω 0 [ ψ 0 ( F 2 ) W n − 1 ψ ( F 1 > ⊗ F 2 ~ ⊗ ⋯ ⊗ F n ) ] = ω 0 [ ψ > ( F 2 ) W n − 1 ψ ( F 1 > ⊗ F 2 ~ ⊗ ⋯ ⊗ F n ) ] + ψ 0 ( F 2 ) ω 0 [ W n − 1 ψ ( F 1 > ⊗ F 2 ~ ⊗ ⋯ ⊗ F n ) ] = ω 0 [ ψ > ( F 2 ) W n − 1 ψ ( F 1 > ⊗ F 2 ~ ⊗ ⋯ ⊗ F n ) ] + ψ 0 ( F 2 ) W n − 1 ψ 0 ( F 1 > ⊗ F 2 ~ ⊗ ⋯ ⊗ F n ) = ω 0 [ ψ > ( F 2 ) W n − 1 ψ ( F 1 > ⊗ F 2 ~ ⊗ ⋯ ⊗ F n ) ] , \begin{aligned}
\omega_0\left[\psi(F_2)W_{n-1}^\psi(F_1^{>}\otimes\widetilde{F_2}\otimes\cdots\otimes F_n)\right]
&=\omega_0\left[\psi_{>}(F_2)W_{n-1}^\psi(F_1^{>}\otimes\widetilde{F_2}\otimes\cdots\otimes F_n)\right]\\
&\quad+\omega_0\left[\psi_0(F_2)W_{n-1}^\psi(F_1^{>}\otimes\widetilde{F_2}\otimes\cdots\otimes F_n)\right]\\
&=\omega_0\left[\psi_{>}(F_2)W_{n-1}^\psi(F_1^{>}\otimes\widetilde{F_2}\otimes\cdots\otimes F_n)\right]\\
&\quad+\psi_0(F_2)\omega_0\left[W_{n-1}^\psi(F_1^{>}\otimes\widetilde{F_2}\otimes\cdots\otimes F_n)\right]\\
&=\omega_0\left[\psi_{>}(F_2)W_{n-1}^\psi(F_1^{>}\otimes\widetilde{F_2}\otimes\cdots\otimes F_n)\right]\\
&\quad+\psi_0(F_2)W_{n-1}^{\psi_0}(F_1^{>}\otimes\widetilde{F_2}\otimes\cdots\otimes F_n)\\
&=\omega_0\left[\psi_{>}(F_2)W_{n-1}^\psi(F_1^{>}\otimes\widetilde{F_2}\otimes\cdots\otimes F_n)\right],
\end{aligned} ω 0 [ ψ ( F 2 ) W n − 1 ψ ( F 1 > ⊗ F 2 ⊗ ⋯ ⊗ F n ) ] = ω 0 [ ψ > ( F 2 ) W n − 1 ψ ( F 1 > ⊗ F 2 ⊗ ⋯ ⊗ F n ) ] + ω 0 [ ψ 0 ( F 2 ) W n − 1 ψ ( F 1 > ⊗ F 2 ⊗ ⋯ ⊗ F n ) ] = ω 0 [ ψ > ( F 2 ) W n − 1 ψ ( F 1 > ⊗ F 2 ⊗ ⋯ ⊗ F n ) ] + ψ 0 ( F 2 ) ω 0 [ W n − 1 ψ ( F 1 > ⊗ F 2 ⊗ ⋯ ⊗ F n ) ] = ω 0 [ ψ > ( F 2 ) W n − 1 ψ ( F 1 > ⊗ F 2 ⊗ ⋯ ⊗ F n ) ] + ψ 0 ( F 2 ) W n − 1 ψ 0 ( F 1 > ⊗ F 2 ⊗ ⋯ ⊗ F n ) = ω 0 [ ψ > ( F 2 ) W n − 1 ψ ( F 1 > ⊗ F 2 ⊗ ⋯ ⊗ F n ) ] ,
where W n − 1 ψ 0 ( F 1 > ⊗ F 2 ~ ⊗ ⋯ ⊗ F n ) W_{n-1}^{\psi_0}(F_1^{>}\otimes\widetilde{F_2}\otimes\cdots\otimes F_n) W n − 1 ψ 0 ( F 1 > ⊗ F 2 ⊗ ⋯ ⊗ F n ) vanishes as a consequence of ψ 0 ( F 1 > ) = 0 \psi_0(F_1^{>})=0 ψ 0 ( F 1 > ) = 0 . Focusing on the first term on the right hand side of (65) , we can tame it using the chain of identities
ω 0 [ W n ψ ( F 2 ⊗ F 1 > ⊗ ⋯ ⊗ F n ) ] = ω 0 [ W n ψ ( F 2 > ⊗ F 1 > ⊗ ⋯ ⊗ F n ) ] + ω ( ψ > ( F 2 ) ψ > ( F 1 ) ) ω 0 ( W n − 2 ψ ( F 3 ⊗ ⋯ ⊗ F n ) ) + ∑ j = 3 n ( − 1 ) j − 1 ω ( ψ > ( F 2 ) ψ ( F j ) ) ω 0 ( W n − 2 ψ ( F 1 > ⊗ F 2 ~ ⊗ ⋯ ⊗ F j ~ ⊗ ⋯ ⊗ F n ) ) − ω ( ψ > ( F 2 ) ψ > ( F 1 ) ) W n − 2 ψ 0 ( F 3 ⊗ ⋯ ⊗ F n ) = ω 0 [ W n ψ ( F 2 > ⊗ F 1 > ⊗ F 3 ⊗ ⋯ ⊗ F n ) ] , \begin{aligned}
\omega_0\left[W_n^\psi(F_2\otimes F_1^{>}\otimes\cdots\otimes F_n)\right]
&=\omega_0\left[W_n^\psi(F_2^{>}\otimes F_1^{>}\otimes\cdots\otimes F_n)\right]\\
&\quad+\omega(\psi_{>}(F_2)\psi_{>}(F_1))\omega_0\left(W_{n-2}^\psi(F_3\otimes\cdots\otimes F_n)\right)\\
&\quad+\sum_{j=3}^{n}(-1)^{j-1}\omega(\psi_{>}(F_2)\psi(F_j))\omega_0\left(W_{n-2}^\psi(F_1^{>}\otimes\widetilde{F_2}\otimes\cdots\otimes\widetilde{F_j}\otimes\cdots\otimes F_n)\right)\\
&\quad-\omega(\psi_{>}(F_2)\psi_{>}(F_1))W_{n-2}^{\psi_0}(F_3\otimes\cdots\otimes F_n)\\
&=\omega_0\left[W_n^\psi(F_2^{>}\otimes F_1^{>}\otimes F_3\otimes\cdots\otimes F_n)\right],
\end{aligned} ω 0 [ W n ψ ( F 2 ⊗ F 1 > ⊗ ⋯ ⊗ F n ) ] = ω 0 [ W n ψ ( F 2 > ⊗ F 1 > ⊗ ⋯ ⊗ F n ) ] + ω ( ψ > ( F 2 ) ψ > ( F 1 )) ω 0 ( W n − 2 ψ ( F 3 ⊗ ⋯ ⊗ F n ) ) + j = 3 ∑ n ( − 1 ) j − 1 ω ( ψ > ( F 2 ) ψ ( F j )) ω 0 ( W n − 2 ψ ( F 1 > ⊗ F 2 ⊗ ⋯ ⊗ F j ⊗ ⋯ ⊗ F n ) ) − ω ( ψ > ( F 2 ) ψ > ( F 1 )) W n − 2 ψ 0 ( F 3 ⊗ ⋯ ⊗ F n ) = ω 0 [ W n ψ ( F 2 > ⊗ F 1 > ⊗ F 3 ⊗ ⋯ ⊗ F n ) ] ,
where in the last step we used that, on account of the inductive hypothesis,
ω ( ψ > ( F 2 ) ψ > ( F 1 ) ) ω 0 ( W n − 2 ψ ( F 3 ⊗ ⋯ ⊗ F n ) ) = ω ( ψ > ( F 2 ) ψ > ( F 1 ) ) W n − 2 ψ 0 ( F 3 ⊗ ⋯ ⊗ F n ) , \omega(\psi_{>}(F_2)\psi_{>}(F_1))\omega_0\left(W_{n-2}^\psi(F_3\otimes\cdots\otimes F_n)\right)=\omega(\psi_{>}(F_2)\psi_{>}(F_1))W_{n-2}^{\psi_0}(F_3\otimes\cdots\otimes F_n), ω ( ψ > ( F 2 ) ψ > ( F 1 )) ω 0 ( W n − 2 ψ ( F 3 ⊗ ⋯ ⊗ F n ) ) = ω ( ψ > ( F 2 ) ψ > ( F 1 )) W n − 2 ψ 0 ( F 3 ⊗ ⋯ ⊗ F n ) ,
while all the terms associated to the sum over j j j are vanishing using the inductive hypothesis in combination with ω 0 ( ψ > ( F 1 ) ) = 0 \omega_0(\psi_{>}(F_1))=0 ω 0 ( ψ > ( F 1 )) = 0 . To summarize,
ω 0 [ W n ψ ( F 1 > ⊗ F 2 ⊗ ⋯ ⊗ F n ) ] = − ω 0 [ W n ψ ( F 2 > ⊗ F 1 > ⊗ ⋯ ⊗ F n ) ] . \omega_0\left[W_n^\psi(F_1^{>}\otimes F_2\otimes\cdots\otimes F_n)\right]=-\omega_0\left[W_n^\psi(F_2^{>}\otimes F_1^{>}\otimes\cdots\otimes F_n)\right]. ω 0 [ W n ψ ( F 1 > ⊗ F 2 ⊗ ⋯ ⊗ F n ) ] = − ω 0 [ W n ψ ( F 2 > ⊗ F 1 > ⊗ ⋯ ⊗ F n ) ] .
Iterating this procedure we end up with
ω 0 [ W n ψ ( F 1 > ⊗ F 2 ⊗ ⋯ ⊗ F n ) ] = ( − 1 ) n − 1 ω 0 [ W n ψ ( F n > ⊗ ⋯ ⊗ F 1 ) ] = 0. \omega_0\left[W_n^\psi(F_1^{>}\otimes F_2\otimes\cdots\otimes F_n)\right]=(-1)^{n-1}\omega_0\left[W_n^\psi(F_n^{>}\otimes\cdots\otimes F_1)\right]=0. ω 0 [ W n ψ ( F 1 > ⊗ F 2 ⊗ ⋯ ⊗ F n ) ] = ( − 1 ) n − 1 ω 0 [ W n ψ ( F n > ⊗ ⋯ ⊗ F 1 ) ] = 0.
where in the last equality we used the defining properties of the zero-time Fermionic Wick polynomials. In turn this yields
ω 0 [ W n ψ ( F 1 ⊗ F 2 ⊗ ⋯ ⊗ F n ) ] = 0 , \omega_0[W_n^\psi(F_1 \otimes F_2 \otimes\cdots\otimes F_n)] = 0, ω 0 [ W n ψ ( F 1 ⊗ F 2 ⊗ ⋯ ⊗ F n )] = 0 ,
which concludes the proof. ✷
Acknowledgments
CD and PR acknowledge the support of the INFN Sezione di Pavia and of Gruppo Nazionale di Fisica Matematica, part of INdAM. PF acknowledges the support of the INFN Sezione di Genova and of Gruppo Nazionale di Fisica Matematica, part of INdAM. Part of this work has appeared in the Master thesis of PF submitted on the 19/09/25 as a partial fulfillment of the requirement to obtain a Master’s degree in Physics at the University of Pavia.