Processing math: 100%

Exploring the entanglement of free spin-12, spin-1 and spin-2 fields

  • In this study, we explore the entanglement of free spin-12, spin-1, and spin-2 fields. We start with an example involving Majorana fields in 1+1 and 2+1 dimensions. Subsequently, we perform the Bogoliubov transformation and express the vacuum state with a particle pair state in the configuration space, which is used to calculate the entropy. This clearly demonstrates that the entanglement entropy originates from the particles across the boundary. Finally, we generalize this method to free spin-1 and spin-2 fields. These higher free massless spin fields have well-known complications owing to gauge redundancy. We deal with the redundancy by gauge-fixing in the light-cone gauge. We show that this gauge provides a natural tensor product structure in the Hilbert space, while surrendering explicit Lorentz invariance. We also use the Bogoliubov transformation to calculate the entropy. The area law emerges naturally by this method.
  • The quantum field theory entanglement is widely studied in the literature [1-3]. Recently, the source of entanglement entropy was likewise considered in [4]. It expresses the vacuum state in terms of the superposition of particle pair states in the configuration space by the Bogoliubov transformation, which transforms the original degrees of freedom to actual local degrees of freedom. Interestingly, it tells us that the entanglement entropy originates from the particle pairs across the boundary. The area law [1, 5-7] emerges naturally by this method. In [4], only the free scalar field is discussed. In this study, we generalize it to 1+1 and 2+1 dimensional Majorana (spin-12) field, 2+1 dimensional U(1) gauge (free spin-1) field, and 3+1 dimensional weak gravitational (free massless spin-2) field. Because there is a Hilbert space with a tensor product structure for the spin-12 field, the generalization is straight forward. As for higher spin fields, more treatments are required because of the gauge redundancy.

    The definition of entanglement entropy in gauge theory was discussed in [8, 9]. The definition of entanglement entropy is heavily dependent on the tensor product structure of the Hilbert space. However, because of the gauge redundancy, there is no natural structure as such in the Hilbert space of the gauge field. Some studies addressed the entanglement entropy of the gauge field with various prescriptions [10, 11], which give up the direct tensor product structure and introduced a "center". It is clear that this issue is also present in the gravitational theories [12, 13], as gravitational theories likewise are gauge theories. The tensor product structure problem of gravitational theories is still under discussion [14]. In this study, we deal with the gauge redundancy by the light-cone gauge for both the free U(1) gauge field and the gravitational field (or strictly speaking, a free spin-2 field). We demonstrate that this gauge-fixing provides a natural tensor product structure in the Hilbert spaces belonging to them, while surrendering the explicit Lorentz invariance.

    Imposing the light-cone gauge, we find that the Lagrangian of the U(1) gauge field and gravitational field simplifies significantly and behaves like a massless scalar field. Subsequently, we quantize them and obtain their Hilbert spaces with the tensor product structure. We perform the Bogoliubov transformation of the theories and express the ground state in terms of a local basis of degrees of freedom. Finally, we calculate the second Rényi entropy, which is a good approximation of the entanglement entropy. We find that in such a prescription, the area law is also naturally preserved.

    Our paper is organized as follows. In Section 2, we make a Bogoliubov transformation of the 1+1 dimensional Majorana field and consider its second Rényi entropy. In Section 3, we perform this procedure for the 2+1 dimensional Majorana field. In Section 4, we derive the Lagrangian of the 2+1 dimensional free U(1) gauge field in the light-cone gauge and obtain its Hilbert space with the natural tensor product structure. Then we perform a Bogoliubov transformation and calculate the second Rényi entropy. In Section 5, we generalize our method to the 3+1 dimensional gravitational (free spin-2) field in the light-cone gauge. In Section 6, we provide conclusions of our work.

    The Majorana field ψ is a field sharing the same Lagrangian with the Dirac field, but satisfying the condition ψ=ψ, which means that the anti-particle and particle of the Majorana field are the same [15]. In 1+1 dimensions at t = 0, a Majorana field with γμ=(σy,iσx) has the form [16]

    ψ(x)=dk2π12ω[1ω+k(ω+kiμ)bk+1ωk(ωkiμ)bk]eikx,

    (1)

    where the annihilation and creation operators bk and bp satisfy the anti-commutation relation {bk,bp}=δ(kp). We define the vacuum |0 as the vacuum of this set of operators, i.e. bk|0=0.

    Let us consider the following Bogoliubov transformation [17]

    ψ(x)=dk2π(a1ka2k)eikx,

    (2)

    which transforms the operators (bk,bk) into the new set of operators (a1k,a2k). For the Majorana field, we have the relation

    {ψα(x),ψβ(y)}=δαβδ(xy).

    (3)

    where α,β=0,1 represent the two components of the Majorana field. To satisfy (3), the operators a1k and a2p should have the relation

    {aik,ajp}=δijδ(k+p),

    (4)

    where i,j=1,2. We choose

    a1k=12(ak+ak),

    (5)

    a2k=i12(akak),

    (6)

    where (ak,ak) are new annihilation and creation operators that satisfy the anti-commutation relation {ak,ap}=δ(kp). We can verify that (5) and (6) satisfy the constraint (4). We define the vacuum |Ω, which is annihilated by ak, i.e. ak|Ω=0. Hence, we obtain the Bogoliubov transformation between (ak,ak) and (bk,bk)

    (bkbk)=12ω(ω+kωkω+k+ωkω+k+ωkωkω+k)×(akak).

    (7)

    Certainly, we have

    bk=12ω((ω+kωk)ak+(ω+k+ωk)ak).

    (8)

    Because we have bk|0=0, the new set of operators satisfy

    ((ω+kωk)ak+(ω+k+ωk)ak)|0=0.

    (9)

    The vacuum |0 can be expressed as a state constructed from the new set of operators

    |0=1γekCkakak|Ω

    (10)

    =1γk(1Ckakak)|Ω,

    (11)

    where γ is the normalization factor, |Ω is the vacuum defined by ak|Ω=0 and the coefficient Ck is fixed by

    Ck=ω+k+ωkω+kωk=ω+μk.

    (12)

    Following [4], let us consider a system with a finite number of sites. The inverse Fourier transform of the operator ak is

    ak=NaNeikN,

    (13)

    where N is the site label. The vacuum |Ψ|0 of the operator bk can thus be written as

    |Ψ=1γk(1N,LCkeik(NL)aNaL)|Ω1γ(1kN,LCkfkNLaNaL)|Ω,

    (14)

    where fkNL=eik(NL). To compute the entanglement entropy, the configuration space is divided into two regions A and ˉA, where the sites in each region are labelled by small letters (n) and small letters with bar (ˉn), respectively. The region A we choose can be any subregion of the total system. The state |Ψ is

    |Ψ1γ(1knlCkfknlanalkˉnˉlCkfkˉnˉlaˉnaˉl12kˉnlCk˜fkˉnlaˉnal)|Ω,

    (15)

    where ˜fkˉnl=fkˉnl+fklˉn.

    A reduced density matrix of ρA is constructed by tracing out the degrees of freedom in region ˉA.

    ρA=TrˉA(|ΨΨ|)1γ2[(1knlCkfknlanal)|ΩAAΩ|(1knlCkfknlanal)]ˉAΩ|ΩˉA+|ΩAAΩ|[ˉAΩ|(kˉnˉlCkfkˉnˉlaˉlaˉn)(kˉnˉlCkfkˉnˉlaˉnaˉl)|ΩˉA]+al|ΩAAΩ|al[ˉAΩ|(kˉnlCk˜fkˉnlaˉn)(kˉnlCk˜fkˉnlaˉn)|ΩˉA].

    (16)

    This expression is already quite revealing. We can see that the first and second line actually lead to no entanglement. They contain operators acting only on A or ˉA. The third line explains the large amount of entanglement between regions A and ˉA - entanglement is created when pairs of particles are created, one inside A and the other in ˉA. Moreover, as we shall see below, these particles that are created cannot be separated by too large of a distance, because the amplitude decreases rapidly with separation.

    Tracing out the degrees of freedom in region ˉA gives the following coefficients in ρA

    κ0=ˉAΩ|(kˉnˉlCkfkˉnˉlaˉlaˉn)(kˉnˉlCkfkˉnˉlaˉnaˉl)|ΩˉA=kkCkCk(ˉnˉl(fkˉnˉlfkˉnˉl+fkˉlˉnfkˉnˉl)+ˉn2),ˉAΩ|(kˉnlCk˜fkˉnlaˉn)(kˉnlCk˜fkˉnlaˉn)|ΩˉA=kkˉnllCkCk˜fkˉnl˜fkˉnl.

    (17)

    The reduced density matrix ρA is then

    ρA1γ2[(1+κ0)|ΩΩ||2Ω||Ω2|+|11|+|22|],

    (18)

    where the subscript label A for the vacuum in region A is dropped,

    |2=knlCkfknlanal|Ω,

    (19)

    |11|=kkˉnllCkCk˜fkˉnl˜fkˉnlal|ΩΩ|al.

    (20)

    Let us consider the second Rényi entropy S2. The square of the reduced density matrix is

    ρ2A=1γ4[((1+κ0)2+κ2)|ΩΩ|(1+κ0+κ2)|Ω2|(1+κ0+κ2)|2Ω|+κ1|11|+(1+κ2)|22|].

    (21)

    The coefficients κ1 and κ2 are given by

    κ1|11|=|11|11|=(kkˉnllCkCk˜fkˉnl˜fkˉnlal|ΩΩ|al)×(kkˉpqqCkCk˜fkˉpq˜fkˉpqaq|ΩΩ|aq)=kkkkˉnllˉpqCkCkCkCk˜fkˉnl˜fkˉnl˜fkˉpq˜fkˉplal|ΩΩ|aq,

    κ2=2|2=Ω|(knlCkfknlalan)(knlCkfknlanal)|Ω=kkCkCk(nl(fknlfknl+fklnfknl)+n2).

    S2 is computed by taking the trace of (21)

    S2=lnTrρ2A=ln((1+κ0)2+κ2+|κ1|+(1+κ2)κ2),

    (22)

    where |κ1| is given by

    |κ1|=kkkkˉnˉpllCkCkCkCk˜fkˉnl˜fkˉnl˜fkˉpl˜fkˉpl.

    (23)

    In the 1+1 dimensional Majorana fermion case, the coefficients Ck control the range of the interaction. The factor kCkfknl can be converted to an integral by taking a continuum limit with an IR cut-off ϵ and UV cut-off K

    Fnl=kCkfknlKϵdkω+μkeik(nl)a

    (24)

    F(x)=Kϵdkω+μkeikx.

    (25)

    This is evaluated numerically and shown in Fig. 1.

    Figure 1

    Figure 1.  (color online) Red and blue lines represent the real and imaginary part of the integral (25), respectively. Both the real and imaginary parts of the integral decrease quickly with distance, and the long range contribution is very small. ϵ=0.1 and K = 10 are used in this plot.

    The figure shows that the interaction between sites falls over large distances. The sites near the boundary make the main contribution to the entanglement entropy.

    In 2+1 dimensions at t = 0, the Majorana field with γμ=(σy,iσz,iσx) has the form [16]

    ψ(x)=d2k2π12ω[1ω+ky(ω+kykx+iμ)bk+1ωky(ωkykxiμ)bk]eikx,

    (26)

    where the annihilation and creation operators bk and bp satisfy the anti-commutation relation {bk,bp}=δ(kp). We define the vacuum |0 as the vacuum of this set of operators, i.e. bk|0=0.

    Let us consider the following Bogoliubov transformation

    ψ(x)=d2k2π(a1ka2k)eikx,

    (27)

    which transforms the operators (bk,bk) to the new set of operators (a1k,a2k). For the 2+1 dimensional Majorana field, we have the relation

    {ψα(x),ψβ(y)}=δαβδ(xy),

    (28)

    where α,β=0,1 represent two components of the Majorana field. To satisfy (28), the operator a1k and a2p should have the relation

    {aik,ajp}=δijδ(k+p),

    (29)

    where i,j=1,2. We choose

    a1k=12(ak+ak),

    (30)

    a2k=i12(akak),

    (31)

    where (ak,ak) are new annihilation and creation operators that satisfy the anti-commutation relation {ak,ap}=δ(kp). We can verify that (30) and (31) satisfy the constraint (29). We define the vacuum |Ω, which is annihilated by ak, i.e. ak|Ω=0. Hence, we obtain the Bogoliubov transformation between (ak,ak) and (bk,bk)

    (bkbk)=12ωkxiμkx+iμ(kx+iμωky+iωkykx+iμωkyiωkykx+iμωkyiω+kykx+iμω+ky+iω+ky)(akak).

    (32)

    Certainly, we have

    bk=12ωkxiμkx+iμ((kx+iμωky+iωky)ak(kx+iμωky+iωky)ak).

    (33)

    Because we have bk|0=0, the new set of operators satisfy

    ((kx+iμωky+iωky)ak(kx+iμωky+iωky)ak)|0=0.

    (34)

    The vacuum |0 can be expressed as a state constructed from the new set of operators

    |0=1γekCkakak|Ω

    (35)

    =1γk(1Ckakak)|Ω,

    (36)

    where γ is the normalization factor, |Ω is the vacuum defined by ak|Ω=0, and the coefficient Ck is fixed by

    Ck=kx+iμ+iωipykx+iμiω+ipy.

    (37)

    Following [4], let us consider a system with a finite number of sites. The inverse Fourier transform of the operator ak is

    ak=NaNeikN,

    (38)

    where N is the site label. The vacuum |Ψ|0 of the operator bk can now be written as

    |Ψ1γk(1N,LCkeik(NL)aNaL)|Ω1γ(1kN,LCkfkNLaNaL)|Ω,

    (39)

    where fkNL=eik(NL). For a simplification of notations, we neglect the vector nation and denote n as n. To compute the entanglement entropy, the configuration space is divided into two regions A and ˉA, where the sites in each region are labelled by small letters (n) and small letters with bars (ˉn), respectively. The region A we choose can be any subregion of the total system. The state |Ψ is

    |Ψ1γ(1knlCkfknlanalkˉnˉlCkfkˉnˉlaˉnaˉl12kˉnlCk˜fkˉnlaˉnal)|Ω,

    (40)

    where ˜fkˉnl=fkˉnl+fklˉn.

    Following the same procedure of the case in 1+1 dimensional Majorana field, we can obtain the reduced density matrix and the second Rényi entropy S2, which have the same form as (18) and (22), respectively. The only difference is the expression of Ck.

    In the 2+1 dimensional Majorana fermion case, the coefficients Ck also control the range of the interaction. The factor kCkfknl can be converted to an integral by taking a continuum limit with an IR cut-off ϵ and UV cut-off K

    Fnl=kCkfknlKϵKϵdkxdkykx+iμ+iωikykx+iμiω+ikyeik(nl)a

    (41)

    F(x,y)=KϵKϵdkxdkykx+iμ+iωikykx+iμiω+ikyei(kxx+kyy).

    (42)

    This is evaluated numerically and shown in Fig. 2.

    Figure 2

    Figure 2.  (color online) Plots of integral (42) in 2+1 dimensional Majorana fermion model. The left and right figures represent the real and imaginary parts of the integral, respectively. The amplitude of the surface decreases quickly with distance, and the long range contribution can be neglected. ϵ=0.1 and K = 10 are used in these plots.

    The figures show that the interaction between sites diminishes over large distances. The sites near the boundary contribute most significantly to the entanglement entropy. Hence, an area law is expected in this model.

    In this section, we consider the 2+1 dimensional free U(1) gauge field in the light-cone gauge. To this end, we derive the Lagrangian, and verify that it is the same as a massless scalar field. Subsequently, we perform the Bogoliubov transformation to calculate the entropy.

    We consider the entanglement entropy of 2+1 dimensional Maxwell fields with light-cone gauge-fixing. In this study, we use the Minkowski metric ημν=diag(1,1,1) with the Minkowski coordinate (x0,x1,x2). We also introduce the light-cone coordinate x+,x,x2 with

    x±12(x0±x1).

    (43)

    With regard to the light-cone coordinate, interested readers can read the book [18]. In the light-cone gauge, the vector is defined as

    a±12(a0±a1),

    (44)

    and the metric becomes

    ˆημν=(010100001).

    (45)

    The vector in the light-cone gauge has the properties a+=a and a=a+.

    Following is the calculation of the 2+1 dimensional Maxwell fields. We start with the Lagrangian L=14FμνFμν. Because L is a scalar, μ; ν can be 0, 1, 2 or +, −, 2. We start with the light-cone coordinate. Because of the asymmetry of Fμν, F++=F=F22=0, the Lagrangian becomes

    L=14FμνFμν=14(2F+F++2F+2F+2+2F2F2).

    (46)

    For the light-cone gauge, A+=0 and A=A+=0. By imposing the gauge-fixing, we find that the Lagrangian becomes

    L=12(A)2+A2(+A2+2A).

    (47)

    With the light-cone gauge fixing A+(p)=0, we have A=1p+(p2A2) [18]. We perform the Fourier transformation

    Aμ(x)=d3p(2π)3eipxAμ(p).

    (48)

    The Fourier transformation of L is

    ˜L=12(pA)2pA2(p+A2+p2A)=12(p1p+p2A2)2pA2(p+A2+p21p+p2A2)=12(p2A2)2+pA2pA2+p2A2p2A2=12(p2A2)2+pA2pA2.

    (49)

    When written in the momentum space of Minkowski spacetime, it becomes

    ˜L=12(p2A2)212((p0)2(p1)2)(A2)2=12(p0A2)2+12((p1A2)2+(p2A2)2).

    (50)

    When we perform the inverse Fourier transformation and return to Minkowski coordinates, the expression becomes

    L=12(0A2)212[(1A2)2+(2A2)2].

    (51)

    This is exactly the same Lagrangian as the massless scalar field, and the corresponding Hamiltonian is

    H=12(0A2)2+12[(1A2)2+(2A2)2].

    (52)

    To simplify the notation in the next subsection, we depict A2 as Ay. The Lagrangian and Hamiltonian can be written as

    L=12(tAy)212[(xAy)2+(yAy)2],

    (53)

    H=12(tAy)2+12[(xAy)2+(yA2)2].

    (54)

    Here, there is no gauge redundancy in the Lagrangian (53). The Hilbert space of Ay should have the tensor product structure, so we can consider its entanglement entropy. It behaves like a free scalar field, which coincides with the trivial center case of [11].

    In the 2+1 dimensional U(1) gauge field with the light-cone gauge, from (53), we have the equation of motion

    Ay=0,

    (55)

    where =2t2+2x2+2y2. We can find that there is only one physical degree of freedom in the 2+1 dimensional U(1) gauge field, and the component with physical freedom Ay satisfies the equation for the massless scalar field. For the non-zero components of the gauge field, we have the solution

    (Ay˙Ay)=d2k2π12ω(1iω)bkeikx+(1iω)bkeikx.

    (56)

    Let us consider a Bogoliubov transformation, which transforms the set of operators (bk,bk) to the following set of operators

    (Ay˙Ay)=d2k2π(a1ka2k)eikx.

    (57)

    With operator a1k and a2k, the commutation relations of Ay and ˙Ay are expressed as below

    [Ay(x,t),˙Ay(y,t)]=d2k2πd2p2π[a1k,a2p]ei(kx+py)=iδ(xy),

    (58)

    [Ay(x,t),Ay(y,t)]=[˙Ay(x,t),˙Ay(y,t)]=0.

    (59)

    We need to have

    [a1k,a2p]=iδ(k+p),

    (60)

    [a1k,a1p]=[a2k,a2p]=0.

    (61)

    We choose

    a1k=12α(ak+ak),

    (62)

    a2k=iα2(akak),

    (63)

    where α is a real parameter. The operator ak and ap have the commutation relations

    [ak,ap]=δ(kp),

    (64)

    [ak,ap]=[ak,ap]=0.

    (65)

    We define ak|Ω=0, where |Ω is the vacuum of the new set of operators. We find that (62) and (63) satisfy (60) and (61). Thus, from (56), (57), (62), and (63), we have the Bogoliubov transformation

    (bkbk)=1i2ω(iω2α+iα2iω2αiα2iω2αiα2iω2α+iα2)(akak),

    (66)

    with ω=k2x+k2y. From the above Bogoliubov transformation, we have

    bk=1i2ω((iω2α+iα2)ak+(iω2αiα2)ak).

    (67)

    Here (ak,ak) are the annihilation and creation operators of the new modes. We also have the vacuum |0, which is annihilated by bk

    bk|0=0.

    (68)

    From (67), we find that the new set of operators satisfy

    ((iω2α+iα2)ak+(iω2αiα2)ak)|0=0.

    (69)

    The vacuum |0 can be expressed in terms of a state constructed from the new set of operators

    |0=1γekCkakak|Ω

    (70)

    1γk(1Ckakak)|Ω,

    (71)

    where γ is the normalization factor, |Ω is the vacuum defined by ak|Ω=0, and the coefficient Ck is fixed by

    Ck=ωαω+α.

    (72)

    We find that the Bogoliubov transformation of the light-cone gauge field is very similar to that of the free scalar field [4].

    Let us consider a system with a finite number of sites. The inverse Fourier transform of the operator ak is

    ak=NaNeikN,

    (73)

    where N is the site label. The vacuum |Ψ|0 of the operator bk can now be written as

    |Ψ1γk(1N,LCkeik(NL)aNaL)|Ω1γ(1kN,LCkfkNLaNaL)|Ω,

    (74)

    where fkNL=eik(NL). For a simplification of notations, we neglect the vector nation and denote n as n. To compute the entanglement entropy, the configuration space is divided into two regions A and ˉA, where the sites in each region are labelled by small letters (n) and small letters with bars (ˉn), respectively. The region A can be chosen as any subregion of the total system. The state |Ψ is

    |Ψ1γ(1knlCkfknlanalkˉnˉlCkfkˉnˉlaˉnaˉl12kˉnlCk˜fkˉnlaˉnal)|Ω,

    (75)

    where ˜fkˉnl=fkˉnl+fklˉn.

    Following the same procedure of previous cases, we can obtain the reduced density matrix and second Rényi entropy S2, which have the same form as (18) and (22), respectively. The only difference is the expression of Ck.

    In 2+1 dimensional U(1) gauge field theory in the light-cone gauge, the coefficients Ck also control the range of the interaction. The factor kCkfknl can be converted to an integral by taking a continuum limit with an IR cut-off ϵ and UV cut-off K

    Fnl=kCkfknlKϵKϵdkxdkyωαω+αeik(nl)a

    (76)

    F(x,y)=KϵKϵdkxdkyωαω+αei(kxx+kyy),

    (77)

    with ω=k2x+k2y. This is evaluated numerically and shown in Fig. 3.

    Figure 3

    Figure 3.  (color online) Plots of integral (77) in 2+1 dimensional U(1) gauge field with light-cone gauge. The left and right figures represent the real and imaginary parts of the integral, respectively. The amplitude of the surface decreases quickly with distance, and the long range contribution can be neglected. ϵ=0.1, K = 10 and α=1 are used in these plots.

    The figures show that the interaction between sites diminishes over large distances. The sites near the boundary contribute most significantly to the entanglement entropy. Hence, an area law is expected in this model.

    In this section, we consider the 3+1 dimensional free spin-2 field theory in the light-cone gauge. To this end, we derive the Lagrangian and find that it is also the same as the massless scalar field. Subsequently, we perform the Bogoliubov transformation to calculate the entropy.

    We consider the entanglement entropy of the 3+1 dimensional weak gravitational field gμν=ημν+hμν with light-cone gauge-fixing. Here, hμν is a spin-2 field. In this study, we use the Minkowski metric ημν=diag(1,1,1,1) with the Minkowski coordinate (x0,x1,x2,x3). We also introduce the light-cone coordinate x+,x,x2,x3 with

    x±12(x0±x1).

    (78)

    In the light-cone gauge, the vector is defined as

    a±12(a0±a1),

    (79)

    and the metric becomes

    ˆημν=ˆημν=(0100100000100001).

    (80)

    The vector in the light-cone gauge has the properties a+=a and a=a+.

    Now we come to the calculation of weak gravitational (spin-2) field. We start with the Ricci scalar R. Because R is a scalar, μ; ν can be 0, 1, 2, 3 or +, −, 2, 3. We begin with the light-cone coordinate. We impose the gauge-fixing h++=h+=h+I=0, where I=2,3. For the components with subscript indexes, the gauge fixing is h=h+=hI=0. For calculating the Ricci scalar R, we use

    Γμαβ=gμν2[gανxβ+gβνxαgαβxν]

    (81)

    and

    Rμν=αΓαμννΓαμα+ΓαβαΓβμνΓαβνΓβμα.

    (82)

    The Christoffel symbols are

    Γ+++=12h++x,

    (83)

    Γ++I=12h+Ix,

    (84)

    Γ+IJ=12hIJx,

    (85)

    Γ++=Γ+=Γ+I=0,

    (86)

    Γ++=12h++x+,

    (87)

    Γ+=12h++x,

    (88)

    Γ=0,

    (89)

    Γ+I=12h++xI,

    (90)

    ΓI=12hI+x,

    (91)

    ΓIJ=12(hI+xJ+hJ+xIhIJx+),

    (92)

    ΓI++=12(2h+Ix+h++xI),

    (93)

    ΓI+=12h+Ix,

    (94)

    ΓI=0,

    (95)

    ΓI+J=12(h+IxJ+hJIx+h+JxI),

    (96)

    ΓIJ=12hJIx,

    (97)

    ΓIJK=12(hJIxK+hKIxJhJKxI).

    (98)

    The Ricci scalar is given by

    R=ˆημνRμν=ˆη+R++ˆη+R++ˆηIJRIJ=2R++RII,

    (99)

    where the repeated I, J, and K are summed, and we take this convention below. We have

    R+=122h++x2+122h+IxIx14hIJx(h+IxJ+hJIx+h+JxI)

    (100)

    and

    \setcounterequation101RII=2hI+xxI+2hIJxJxI12(h+Ix)2hIJx(hIJx+hI+xJ)14[(hJKxI)2(hIKxJhJIxK)2].(101)

    The Ricci scalar can be written as

    \setcounterequation102R=2hx2+hIJxIxJ+12hIJx(hIxJ+hJIx++hJxI)12(hIx)2hIJx(hIJx++hIxJ)14[(hJKxI)2(hIKxJhJIxK)2].(102)

    When expressed in Fourier space and considering the light-cone gauge-fixing, we can obtain [18]

    hI=1p+pJhIJ

    (103)

    and

    h=1p+pIhI=pIpJ(p+)2hIJ.

    (104)

    With the above relations, we have

    R=2pIpJhIJ+12pp+hIJhIJ12pJpKhIJhIK+14[(pIhJK)2(pJhIKpKhJI)2].

    (105)

    Because hIJ is symmetric and traceless, there are only two degrees of freedom h22 and h23. We expand the above expression with I,J,K=2,3, and obtain

    R=2pIpJhIJ+(p+p12p2212p23)[(h22)2+(h23)2]=2pIpJhIJ+12((p0)2(p1)2(p2)2(p3)2)×[(h22)2+(h23)2].

    (106)

    Apart from the total derivative term IJhIJ, the Lagrangian of the gravitational field can be written as

    L=12((thyy)2(hyy)2+(thyz)2(hyz)2),

    (107)

    where =xi+yj+zk. We have replaced h22 and h23 with hyy and hyz respectively.There is no gauge redundancy in the Lagrangian (107). We can expect that the Hilbert space of hyy and hyz should have the tensor product structure, so we can consider the entanglement entropy of this model. Hence, we provide a prescription of the entanglement entropy of a free spin-2 field.

    In the 3+1 dimensional gravitational (spin-2) field with the light-cone gauge, we have two independent degrees of freedom. From (107), we have the equations of motion

    hyy=0,hyz=0.

    (108)

    The equations of motion are the same as the free massless scalar field. Their solutions are

    hyy(x)=d3k2ω(bkeikx+bkeikx)

    (109)

    and

    hyz(x)=d3k2ω(bkeikx+bkeikx)

    (110)

    respectively, where (bk,bk) and (bk,bk) are the creation and annihilation operators of two physical degrees of freedom.

    Let us consider the mode (bk,bk). For the operator hyy, its canonical momentum is ˙hyy. We obtain the solution

    (hyy˙hyy)=d3k(2π)3212ω(1iω)bkeikx+(1iω)bkeikx.

    (111)

    Let us consider a Bogoliubov transformation, which transforms the set of operators (bk,bk) to the following set of operators

    (hyy˙hyy)=d3k(2π)32(a1ka2k)eikx.

    (112)

    With operator a1k and a2k, the commutation relations of hyy and ˙hyy are expressed as below

    [hyy(x,t),˙hyy(y,t)]=d3k(2π)32d3p(2π)32[a1k,a2p]ei(kx+py)=iδ(xy),

    (113)

    [hyy(x,t),hyy(y,t)]=[˙hyy(x,t),˙hyy(y,t)]=0.

    (114)

    We find that apart from the dimension, the Bogoliubov 3+1 dimensional spin-2 field with the light-cone gauge is the same as that of the 2+1 dimensional U(1) gauge field. Moreover, they are both the same with the free scalar field. The form of the Bogoliubov transformation is the same as (66), with ω=k2x+k2y+k2z.

    We define the vacuum |0 by bk|0=0, and the vacuum |Ω by ak|Ω=0 in the 3+1 dimensional spin-2 field with the light-cone gauge. Because of the Bogoliubov transformation, the vacuum |0 can be expressed by the vacuum |Ω and operator (ak,ap) as

    |0=1γekCkakak|Ω

    (115)

    1γk(1Ckakak)|Ω,

    (116)

    where γ is the normalization factor, and the coefficient Ck is fixed by

    Ck=ωαω+α,

    (117)

    with ω=k2x+k2y+k2z.

    We can consider a system with a finite number of sites. We can divide the system into two parts, A and ˉA. The region A we choose can be any subregion of the total system. Considering the state |Ψ=|0, we can obtain the reduced density matrix ρA and the second Rényi entropy S2, which have the same form of the cases in previous sections. The only difference is the expression of Ck.

    In the 3+1 dimensional spin-2 field with the light-cone gauge, the coefficients Ck also control the range of the interaction. The factor kCkfknl can be converted to an integral by taking a continuum limit with an IR cut-off ϵ and UV cut-off K

    Fnl=kCkfknlKϵKϵKϵdkxdkydkzωαω+αeik(nl)a

    (118)

    F(x,y,z)=KϵKϵKϵdkxdkydkzωαω+αei(kxx+kyy+kzz),

    (119)

    with ω=k2x+k2y+k2z. Numerical evaluation in the special case of x = y = z gives the result shown in Fig. 4.

    Figure 4

    Figure 4.  (color online) Plots of integral (119) in 3+1 dimensional gravity with the light-cone gauge in the special case of x = y = z. Both the real and imaginary parts of the integral decrease quickly with distance, and the long range contribution can be neglected. ϵ=0.1, K = 10 and α=1 are used in this plot.

    The figure shows that the interaction between sites once again diminishes over large distances. The sites near the boundary contribute most significantly to the entanglement entropy. Hence, an area law is expected in this model as well.

    In this study, we explore the entanglement of free spin-12, spin-1 and spin-2 fields. First, we consider the 1+1 dimensional Majorana field, which is just a pair of left and right moving fermions, and the 2+1 dimensional Majorana field. We perform the Bogoliubov transformation of their modes and express the vacuum with a particle pair state in the configuration space. Subsequently, we calculate the second Rényi entropies in the finite systems. Let us emphasize that while a Majorana Weyl fermion is well known to be non-local, a local Hilbert space can be defined when both chiralities are present. This is demonstrated explicitly in the current note. After that, we generalize the method to the 2+1 dimensional free U(1) spin-1 gauge field and the 3+1 dimensional gravitational (free spin-2) field. Because of the gauge redundancy of the higher spin field, there is no Hilbert space with a natural tensor product structure. We take the light-cone gauge for both fields and find that their Lagrangians behave like a free massless scalar field. The light-cone gauge allows simple quantization, while surrendering explicit Lorentz invariance. Nonetheless, it provides a candidate tensor product structure. The definition of entanglement entropy is dependent on both the state and the operator algebra. If the operator algebra is gauge-invariant [10], the corresponding entanglement entropy is likewise gauge-invariant. In this work and in [11], the operator algebras implicitly chosen are not gauge-invariant, such that the corresponding results follow trend. In our past study [11], we explored several different algebras and demonstrated which of those would reproduce the universal log terms found in Casini [10]. It is, however, expected that generic non-gauge invariant algebra choices, such as those considered in the current note, lead to a result that is gauge-dependent. For the U(1) gauge field, the Lagrangian behaves like a scalar field, which coincides with the trivial center case of [11]. As for the gravitational (free spin-2) field, we provide a prescription to observe the tensor product structure of the Hilbert space. Before doing so, there is no prescription of the gravitational (free spin-2) field. After we obtain their Hilbert spaces with a tensor product structure, we calculate the second Rényi entropies. This method can be helpful in dealing with the Hilbert space and entanglement of the perturbative gravitational field, i.e. weak gravitational field. In the non-perturbative regime, the structure of the Hilbert space is still not clear. In all the cases studied, we find that the entropy originates from the particle pairs across the boundary, and the area law emerges naturally.

    We would like to thank Prof. Yong-Shi Wu for critical and meticulous reading of our manuscript. LYH acknowledges the Thousands Young Talents Program.

    [1] Pasquale Calabrese and John Cardy, Journal of Statistical Mechanics: Theory and Experiment, 2004(06): P06002 (2004)
    [2] Pasquale Calabrese and John Cardy, Journal of Physics A: Mathematical and Theoretical, 42(50): 504005 (2009) doi: 10.1088/1751-8113/42/50/504005
    [3] Horacio Casini and Marina Huerta, Journal of Physics A: Mathematical and Theoretical, 42(50): 504007 (2009) doi: 10.1088/1751-8113/42/50/504007
    [4] Arpan Bhattacharyya, Ling-Yan Hung, Pak Hang Chris Lau, and Si-Nong Liu, Entropy, 19(12): 671 (2017) doi: 10.3390/e19120671
    [5] Luca Bombelli, Rabinder K Koul, Joohan Lee, and Rafael D Sorkin, Physical Review D, 34(2): 373 (1986) doi: 10.1103/PhysRevD.34.373
    [6] Mark Srednicki, Physical Review Letters, 71(5): 666 (1993) doi: 10.1103/PhysRevLett.71.666
    [7] Martin B Plenio, Jens Eisert, J Dreissig, and Marcus Cramer, Physical Review Letters, 94(6): 060503 (2005) doi: 10.1103/PhysRevLett.94.060503
    [8] PV Buividovich and MI Polikarpov, Physics Letters B, 670(2): 141-145 (2008) doi: 10.1016/j.physletb.2008.10.032
    [9] Horacio Casini, Marina Huerta, and José Alejandro Rosabal, Physical Review D, 89(8): 085012 (2014) doi: 10.1103/PhysRevD.89.085012
    [10] Horacio Casini and Marina Huerta, Physical Review D, 90(10): 105013 (2014) doi: 10.1103/PhysRevD.90.105013
    [11] Zhi Yang and Ling-Yan Hung, Journal of High Energy Physics, 2018(3): 73 (2018) doi: 10.1007/JHEP03(2018)073
    [12] William Donnelly and Laurent Freidel, Journal of High Energy Physics, 2016(9): 102 (2016) doi: 10.1007/JHEP09(2016)102
    [13] William Donnelly and Steven B Giddings, Physical Review D, 94(10): 104038 (2016) doi: 10.1103/PhysRevD.94.104038
    [14] Daniel Harlow and Daniel Jafferis, (2018), arXiv:1804.01081
    [15] Ettore Majorana, Nuovo Cim, 14(171): 50 (1937)
    [16] Ling-Yan Hung, Michael Smolkin, and Evgeny Sorkin, Journal of High Energy Physics, 2013(12): 22 (2013) doi: 10.1007/JHEP12(2013)022
    [17] K Svozil, Physical Review Letters, 65(26): 3341 (1990) doi: 10.1103/PhysRevLett.65.3341
    [18] Barton Zwiebach, A first course in string theory, (Cambridge University Press, 2004)
  • [1] Pasquale Calabrese and John Cardy, Journal of Statistical Mechanics: Theory and Experiment, 2004(06): P06002 (2004)
    [2] Pasquale Calabrese and John Cardy, Journal of Physics A: Mathematical and Theoretical, 42(50): 504005 (2009) doi: 10.1088/1751-8113/42/50/504005
    [3] Horacio Casini and Marina Huerta, Journal of Physics A: Mathematical and Theoretical, 42(50): 504007 (2009) doi: 10.1088/1751-8113/42/50/504007
    [4] Arpan Bhattacharyya, Ling-Yan Hung, Pak Hang Chris Lau, and Si-Nong Liu, Entropy, 19(12): 671 (2017) doi: 10.3390/e19120671
    [5] Luca Bombelli, Rabinder K Koul, Joohan Lee, and Rafael D Sorkin, Physical Review D, 34(2): 373 (1986) doi: 10.1103/PhysRevD.34.373
    [6] Mark Srednicki, Physical Review Letters, 71(5): 666 (1993) doi: 10.1103/PhysRevLett.71.666
    [7] Martin B Plenio, Jens Eisert, J Dreissig, and Marcus Cramer, Physical Review Letters, 94(6): 060503 (2005) doi: 10.1103/PhysRevLett.94.060503
    [8] PV Buividovich and MI Polikarpov, Physics Letters B, 670(2): 141-145 (2008) doi: 10.1016/j.physletb.2008.10.032
    [9] Horacio Casini, Marina Huerta, and José Alejandro Rosabal, Physical Review D, 89(8): 085012 (2014) doi: 10.1103/PhysRevD.89.085012
    [10] Horacio Casini and Marina Huerta, Physical Review D, 90(10): 105013 (2014) doi: 10.1103/PhysRevD.90.105013
    [11] Zhi Yang and Ling-Yan Hung, Journal of High Energy Physics, 2018(3): 73 (2018) doi: 10.1007/JHEP03(2018)073
    [12] William Donnelly and Laurent Freidel, Journal of High Energy Physics, 2016(9): 102 (2016) doi: 10.1007/JHEP09(2016)102
    [13] William Donnelly and Steven B Giddings, Physical Review D, 94(10): 104038 (2016) doi: 10.1103/PhysRevD.94.104038
    [14] Daniel Harlow and Daniel Jafferis, (2018), arXiv:1804.01081
    [15] Ettore Majorana, Nuovo Cim, 14(171): 50 (1937)
    [16] Ling-Yan Hung, Michael Smolkin, and Evgeny Sorkin, Journal of High Energy Physics, 2013(12): 22 (2013) doi: 10.1007/JHEP12(2013)022
    [17] K Svozil, Physical Review Letters, 65(26): 3341 (1990) doi: 10.1103/PhysRevLett.65.3341
    [18] Barton Zwiebach, A first course in string theory, (Cambridge University Press, 2004)
  • 加载中

Figures(5)

Get Citation
Zhi Yang and Ling-Yan Hung. Exploring the entanglement of free spin-12, spin-1 and spin-2 fields[J]. Chinese Physics C, 2019, 43(5): 053102-1-053102-11. doi: 10.1088/1674-1137/43/5/053102
Zhi Yang and Ling-Yan Hung. Exploring the entanglement of free spin-12, spin-1 and spin-2 fields[J]. Chinese Physics C, 2019, 43(5): 053102-1-053102-11.  doi: 10.1088/1674-1137/43/5/053102 shu
Milestone
Received: 2019-01-30
Article Metric

Article Views(2023)
PDF Downloads(46)
Cited by(0)
Policy on re-use
To reuse of subscription content published by CPC, the users need to request permission from CPC, unless the content was published under an Open Access license which automatically permits that type of reuse.
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Email This Article

Title:
Email:

Exploring the entanglement of free spin-12, spin-1 and spin-2 fields

  • 1. Department of Physics & Center for Field Theory and Particle Physics, Fudan University, 200433 Shanghai, China
  • 2. State Key Laboratory of Surface Physics & Department of Physics, Fudan University, 200433 Shanghai, China

Abstract: In this study, we explore the entanglement of free spin-12, spin-1, and spin-2 fields. We start with an example involving Majorana fields in 1+1 and 2+1 dimensions. Subsequently, we perform the Bogoliubov transformation and express the vacuum state with a particle pair state in the configuration space, which is used to calculate the entropy. This clearly demonstrates that the entanglement entropy originates from the particles across the boundary. Finally, we generalize this method to free spin-1 and spin-2 fields. These higher free massless spin fields have well-known complications owing to gauge redundancy. We deal with the redundancy by gauge-fixing in the light-cone gauge. We show that this gauge provides a natural tensor product structure in the Hilbert space, while surrendering explicit Lorentz invariance. We also use the Bogoliubov transformation to calculate the entropy. The area law emerges naturally by this method.

    HTML

    1.   Introduction
    • The quantum field theory entanglement is widely studied in the literature [1-3]. Recently, the source of entanglement entropy was likewise considered in [4]. It expresses the vacuum state in terms of the superposition of particle pair states in the configuration space by the Bogoliubov transformation, which transforms the original degrees of freedom to actual local degrees of freedom. Interestingly, it tells us that the entanglement entropy originates from the particle pairs across the boundary. The area law [1, 5-7] emerges naturally by this method. In [4], only the free scalar field is discussed. In this study, we generalize it to 1+1 and 2+1 dimensional Majorana (spin-12) field, 2+1 dimensional U(1) gauge (free spin-1) field, and 3+1 dimensional weak gravitational (free massless spin-2) field. Because there is a Hilbert space with a tensor product structure for the spin-12 field, the generalization is straight forward. As for higher spin fields, more treatments are required because of the gauge redundancy.

      The definition of entanglement entropy in gauge theory was discussed in [8, 9]. The definition of entanglement entropy is heavily dependent on the tensor product structure of the Hilbert space. However, because of the gauge redundancy, there is no natural structure as such in the Hilbert space of the gauge field. Some studies addressed the entanglement entropy of the gauge field with various prescriptions [10, 11], which give up the direct tensor product structure and introduced a "center". It is clear that this issue is also present in the gravitational theories [12, 13], as gravitational theories likewise are gauge theories. The tensor product structure problem of gravitational theories is still under discussion [14]. In this study, we deal with the gauge redundancy by the light-cone gauge for both the free U(1) gauge field and the gravitational field (or strictly speaking, a free spin-2 field). We demonstrate that this gauge-fixing provides a natural tensor product structure in the Hilbert spaces belonging to them, while surrendering the explicit Lorentz invariance.

      Imposing the light-cone gauge, we find that the Lagrangian of the U(1) gauge field and gravitational field simplifies significantly and behaves like a massless scalar field. Subsequently, we quantize them and obtain their Hilbert spaces with the tensor product structure. We perform the Bogoliubov transformation of the theories and express the ground state in terms of a local basis of degrees of freedom. Finally, we calculate the second Rényi entropy, which is a good approximation of the entanglement entropy. We find that in such a prescription, the area law is also naturally preserved.

      Our paper is organized as follows. In Section 2, we make a Bogoliubov transformation of the 1+1 dimensional Majorana field and consider its second Rényi entropy. In Section 3, we perform this procedure for the 2+1 dimensional Majorana field. In Section 4, we derive the Lagrangian of the 2+1 dimensional free U(1) gauge field in the light-cone gauge and obtain its Hilbert space with the natural tensor product structure. Then we perform a Bogoliubov transformation and calculate the second Rényi entropy. In Section 5, we generalize our method to the 3+1 dimensional gravitational (free spin-2) field in the light-cone gauge. In Section 6, we provide conclusions of our work.

    2.   1+1 dimensional Majorana fermion - a pair of left and rightmoving fermions
    • The Majorana field ψ is a field sharing the same Lagrangian with the Dirac field, but satisfying the condition ψ=ψ, which means that the anti-particle and particle of the Majorana field are the same [15]. In 1+1 dimensions at t = 0, a Majorana field with γμ=(σy,iσx) has the form [16]

      ψ(x)=dk2π12ω[1ω+k(ω+kiμ)bk+1ωk(ωkiμ)bk]eikx,

      (1)

      where the annihilation and creation operators bk and bp satisfy the anti-commutation relation {bk,bp}=δ(kp). We define the vacuum |0 as the vacuum of this set of operators, i.e. bk|0=0.

      Let us consider the following Bogoliubov transformation [17]

      ψ(x)=dk2π(a1ka2k)eikx,

      (2)

      which transforms the operators (bk,bk) into the new set of operators (a1k,a2k). For the Majorana field, we have the relation

      {ψα(x),ψβ(y)}=δαβδ(xy).

      (3)

      where α,β=0,1 represent the two components of the Majorana field. To satisfy (3), the operators a1k and a2p should have the relation

      {aik,ajp}=δijδ(k+p),

      (4)

      where i,j=1,2. We choose

      a1k=12(ak+ak),

      (5)

      a2k=i12(akak),

      (6)

      where (ak,ak) are new annihilation and creation operators that satisfy the anti-commutation relation {ak,ap}=δ(kp). We can verify that (5) and (6) satisfy the constraint (4). We define the vacuum |Ω, which is annihilated by ak, i.e. ak|Ω=0. Hence, we obtain the Bogoliubov transformation between (ak,ak) and (bk,bk)

      (bkbk)=12ω(ω+kωkω+k+ωkω+k+ωkωkω+k)×(akak).

      (7)

      Certainly, we have

      bk=12ω((ω+kωk)ak+(ω+k+ωk)ak).

      (8)

      Because we have bk|0=0, the new set of operators satisfy

      ((ω+kωk)ak+(ω+k+ωk)ak)|0=0.

      (9)

      The vacuum |0 can be expressed as a state constructed from the new set of operators

      |0=1γekCkakak|Ω

      (10)

      =1γk(1Ckakak)|Ω,

      (11)

      where γ is the normalization factor, |Ω is the vacuum defined by ak|Ω=0 and the coefficient Ck is fixed by

      Ck=ω+k+ωkω+kωk=ω+μk.

      (12)

      Following [4], let us consider a system with a finite number of sites. The inverse Fourier transform of the operator ak is

      ak=NaNeikN,

      (13)

      where N is the site label. The vacuum |Ψ|0 of the operator bk can thus be written as

      |Ψ=1γk(1N,LCkeik(NL)aNaL)|Ω1γ(1kN,LCkfkNLaNaL)|Ω,

      (14)

      where fkNL=eik(NL). To compute the entanglement entropy, the configuration space is divided into two regions A and ˉA, where the sites in each region are labelled by small letters (n) and small letters with bar (ˉn), respectively. The region A we choose can be any subregion of the total system. The state |Ψ is

      |Ψ1γ(1knlCkfknlanalkˉnˉlCkfkˉnˉlaˉnaˉl12kˉnlCk˜fkˉnlaˉnal)|Ω,

      (15)

      where ˜fkˉnl=fkˉnl+fklˉn.

      A reduced density matrix of ρA is constructed by tracing out the degrees of freedom in region ˉA.

      ρA=TrˉA(|ΨΨ|)1γ2[(1knlCkfknlanal)|ΩAAΩ|(1knlCkfknlanal)]ˉAΩ|ΩˉA+|ΩAAΩ|[ˉAΩ|(kˉnˉlCkfkˉnˉlaˉlaˉn)(kˉnˉlCkfkˉnˉlaˉnaˉl)|ΩˉA]+al|ΩAAΩ|al[ˉAΩ|(kˉnlCk˜fkˉnlaˉn)(kˉnlCk˜fkˉnlaˉn)|ΩˉA].

      (16)

      This expression is already quite revealing. We can see that the first and second line actually lead to no entanglement. They contain operators acting only on A or ˉA. The third line explains the large amount of entanglement between regions A and ˉA - entanglement is created when pairs of particles are created, one inside A and the other in ˉA. Moreover, as we shall see below, these particles that are created cannot be separated by too large of a distance, because the amplitude decreases rapidly with separation.

      Tracing out the degrees of freedom in region ˉA gives the following coefficients in ρA

      κ0=ˉAΩ|(kˉnˉlCkfkˉnˉlaˉlaˉn)(kˉnˉlCkfkˉnˉlaˉnaˉl)|ΩˉA=kkCkCk(ˉnˉl(fkˉnˉlfkˉnˉl+fkˉlˉnfkˉnˉl)+ˉn2),ˉAΩ|(kˉnlCk˜fkˉnlaˉn)(kˉnlCk˜fkˉnlaˉn)|ΩˉA=kkˉnllCkCk˜fkˉnl˜fkˉnl.

      (17)

      The reduced density matrix ρA is then

      ρA1γ2[(1+κ0)|ΩΩ||2Ω||Ω2|+|11|+|22|],

      (18)

      where the subscript label A for the vacuum in region A is dropped,

      |2=knlCkfknlanal|Ω,

      (19)

      |11|=kkˉnllCkCk˜fkˉnl˜fkˉnlal|ΩΩ|al.

      (20)

      Let us consider the second Rényi entropy S2. The square of the reduced density matrix is

      ρ2A=1γ4[((1+κ0)2+κ2)|ΩΩ|(1+κ0+κ2)|Ω2|(1+κ0+κ2)|2Ω|+κ1|11|+(1+κ2)|22|].

      (21)

      The coefficients κ1 and κ2 are given by

      κ1|11|=|11|11|=(kkˉnllCkCk˜fkˉnl˜fkˉnlal|ΩΩ|al)×(kkˉpqqCkCk˜fkˉpq˜fkˉpqaq|ΩΩ|aq)=kkkkˉnllˉpqCkCkCkCk˜fkˉnl˜fkˉnl˜fkˉpq˜fkˉplal|ΩΩ|aq,

      κ2=2|2=Ω|(knlCkfknlalan)(knlCkfknlanal)|Ω=kkCkCk(nl(fknlfknl+fklnfknl)+n2).

      S2 is computed by taking the trace of (21)

      S2=lnTrρ2A=ln((1+κ0)2+κ2+|κ1|+(1+κ2)κ2),

      (22)

      where |κ1| is given by

      |κ1|=kkkkˉnˉpllCkCkCkCk˜fkˉnl˜fkˉnl˜fkˉpl˜fkˉpl.

      (23)

      In the 1+1 dimensional Majorana fermion case, the coefficients Ck control the range of the interaction. The factor kCkfknl can be converted to an integral by taking a continuum limit with an IR cut-off ϵ and UV cut-off K

      Fnl=kCkfknlKϵdkω+μkeik(nl)a

      (24)

      F(x)=Kϵdkω+μkeikx.

      (25)

      This is evaluated numerically and shown in Fig. 1.

      Figure 1.  (color online) Red and blue lines represent the real and imaginary part of the integral (25), respectively. Both the real and imaginary parts of the integral decrease quickly with distance, and the long range contribution is very small. ϵ=0.1 and K = 10 are used in this plot.

      The figure shows that the interaction between sites falls over large distances. The sites near the boundary make the main contribution to the entanglement entropy.

    3.   2+1 dimensional Majorana fermion
    • In 2+1 dimensions at t = 0, the Majorana field with γμ=(σy,iσz,iσx) has the form [16]

      ψ(x)=d2k2π12ω[1ω+ky(ω+kykx+iμ)bk+1ωky(ωkykxiμ)bk]eikx,

      (26)

      where the annihilation and creation operators bk and bp satisfy the anti-commutation relation {bk,bp}=δ(kp). We define the vacuum |0 as the vacuum of this set of operators, i.e. bk|0=0.

      Let us consider the following Bogoliubov transformation

      ψ(x)=d2k2π(a1ka2k)eikx,

      (27)

      which transforms the operators (bk,bk) to the new set of operators (a1k,a2k). For the 2+1 dimensional Majorana field, we have the relation

      {ψα(x),ψβ(y)}=δαβδ(xy),

      (28)

      where α,β=0,1 represent two components of the Majorana field. To satisfy (28), the operator a1k and a2p should have the relation

      {aik,ajp}=δijδ(k+p),

      (29)

      where i,j=1,2. We choose

      a1k=12(ak+ak),

      (30)

      a2k=i12(akak),

      (31)

      where (ak,ak) are new annihilation and creation operators that satisfy the anti-commutation relation {ak,ap}=δ(kp). We can verify that (30) and (31) satisfy the constraint (29). We define the vacuum |Ω, which is annihilated by ak, i.e. ak|Ω=0. Hence, we obtain the Bogoliubov transformation between (ak,ak) and (bk,bk)

      (bkbk)=12ωkxiμkx+iμ(kx+iμωky+iωkykx+iμωkyiωkykx+iμωkyiω+kykx+iμω+ky+iω+ky)(akak).

      (32)

      Certainly, we have

      bk=12ωkxiμkx+iμ((kx+iμωky+iωky)ak(kx+iμωky+iωky)ak).

      (33)

      Because we have bk|0=0, the new set of operators satisfy

      ((kx+iμωky+iωky)ak(kx+iμωky+iωky)ak)|0=0.

      (34)

      The vacuum |0 can be expressed as a state constructed from the new set of operators

      |0=1γekCkakak|Ω

      (35)

      =1γk(1Ckakak)|Ω,

      (36)

      where γ is the normalization factor, |Ω is the vacuum defined by ak|Ω=0, and the coefficient Ck is fixed by

      Ck=kx+iμ+iωipykx+iμiω+ipy.

      (37)

      Following [4], let us consider a system with a finite number of sites. The inverse Fourier transform of the operator ak is

      ak=NaNeikN,

      (38)

      where N is the site label. The vacuum |Ψ|0 of the operator bk can now be written as

      |Ψ1γk(1N,LCkeik(NL)aNaL)|Ω1γ(1kN,LCkfkNLaNaL)|Ω,

      (39)

      where fkNL=eik(NL). For a simplification of notations, we neglect the vector nation and denote n as n. To compute the entanglement entropy, the configuration space is divided into two regions A and ˉA, where the sites in each region are labelled by small letters (n) and small letters with bars (ˉn), respectively. The region A we choose can be any subregion of the total system. The state |Ψ is

      |Ψ1γ(1knlCkfknlanalkˉnˉlCkfkˉnˉlaˉnaˉl12kˉnlCk˜fkˉnlaˉnal)|Ω,

      (40)

      where ˜fkˉnl=fkˉnl+fklˉn.

      Following the same procedure of the case in 1+1 dimensional Majorana field, we can obtain the reduced density matrix and the second Rényi entropy S2, which have the same form as (18) and (22), respectively. The only difference is the expression of Ck.

      In the 2+1 dimensional Majorana fermion case, the coefficients Ck also control the range of the interaction. The factor kCkfknl can be converted to an integral by taking a continuum limit with an IR cut-off ϵ and UV cut-off K

      Fnl=kCkfknlKϵKϵdkxdkykx+iμ+iωikykx+iμiω+ikyeik(nl)a

      (41)

      F(x,y)=KϵKϵdkxdkykx+iμ+iωikykx+iμiω+ikyei(kxx+kyy).

      (42)

      This is evaluated numerically and shown in Fig. 2.

      Figure 2.  (color online) Plots of integral (42) in 2+1 dimensional Majorana fermion model. The left and right figures represent the real and imaginary parts of the integral, respectively. The amplitude of the surface decreases quickly with distance, and the long range contribution can be neglected. ϵ=0.1 and K = 10 are used in these plots.

      The figures show that the interaction between sites diminishes over large distances. The sites near the boundary contribute most significantly to the entanglement entropy. Hence, an area law is expected in this model.

    4.   Gauge field in light-cone gauge
    • In this section, we consider the 2+1 dimensional free U(1) gauge field in the light-cone gauge. To this end, we derive the Lagrangian, and verify that it is the same as a massless scalar field. Subsequently, we perform the Bogoliubov transformation to calculate the entropy.

    • 4.1.   Model of gauge field in light-cone gauge

    • We consider the entanglement entropy of 2+1 dimensional Maxwell fields with light-cone gauge-fixing. In this study, we use the Minkowski metric ημν=diag(1,1,1) with the Minkowski coordinate (x0,x1,x2). We also introduce the light-cone coordinate x+,x,x2 with

      x±12(x0±x1).

      (43)

      With regard to the light-cone coordinate, interested readers can read the book [18]. In the light-cone gauge, the vector is defined as

      a±12(a0±a1),

      (44)

      and the metric becomes

      ˆημν=(010100001).

      (45)

      The vector in the light-cone gauge has the properties a+=a and a=a+.

      Following is the calculation of the 2+1 dimensional Maxwell fields. We start with the Lagrangian L=14FμνFμν. Because L is a scalar, μ; ν can be 0, 1, 2 or +, −, 2. We start with the light-cone coordinate. Because of the asymmetry of Fμν, F++=F=F22=0, the Lagrangian becomes

      L=14FμνFμν=14(2F+F++2F+2F+2+2F2F2).

      (46)

      For the light-cone gauge, A+=0 and A=A+=0. By imposing the gauge-fixing, we find that the Lagrangian becomes

      L=12(A)2+A2(+A2+2A).

      (47)

      With the light-cone gauge fixing A+(p)=0, we have A=1p+(p2A2) [18]. We perform the Fourier transformation

      Aμ(x)=d3p(2π)3eipxAμ(p).

      (48)

      The Fourier transformation of L is

      ˜L=12(pA)2pA2(p+A2+p2A)=12(p1p+p2A2)2pA2(p+A2+p21p+p2A2)=12(p2A2)2+pA2pA2+p2A2p2A2=12(p2A2)2+pA2pA2.

      (49)

      When written in the momentum space of Minkowski spacetime, it becomes

      ˜L=12(p2A2)212((p0)2(p1)2)(A2)2=12(p0A2)2+12((p1A2)2+(p2A2)2).

      (50)

      When we perform the inverse Fourier transformation and return to Minkowski coordinates, the expression becomes

      L=12(0A2)212[(1A2)2+(2A2)2].

      (51)

      This is exactly the same Lagrangian as the massless scalar field, and the corresponding Hamiltonian is

      H=12(0A2)2+12[(1A2)2+(2A2)2].

      (52)

      To simplify the notation in the next subsection, we depict A2 as Ay. The Lagrangian and Hamiltonian can be written as

      L=12(tAy)212[(xAy)2+(yAy)2],

      (53)

      H=12(tAy)2+12[(xAy)2+(yA2)2].

      (54)

      Here, there is no gauge redundancy in the Lagrangian (53). The Hilbert space of Ay should have the tensor product structure, so we can consider its entanglement entropy. It behaves like a free scalar field, which coincides with the trivial center case of [11].

    • 4.2.   Bogoliubov transformation and entropy of gauge field in light-cone gauge

    • In the 2+1 dimensional U(1) gauge field with the light-cone gauge, from (53), we have the equation of motion

      Ay=0,

      (55)

      where =2t2+2x2+2y2. We can find that there is only one physical degree of freedom in the 2+1 dimensional U(1) gauge field, and the component with physical freedom Ay satisfies the equation for the massless scalar field. For the non-zero components of the gauge field, we have the solution

      (Ay˙Ay)=d2k2π12ω(1iω)bkeikx+(1iω)bkeikx.

      (56)

      Let us consider a Bogoliubov transformation, which transforms the set of operators (bk,bk) to the following set of operators

      (Ay˙Ay)=d2k2π(a1ka2k)eikx.

      (57)

      With operator a1k and a2k, the commutation relations of Ay and ˙Ay are expressed as below

      [Ay(x,t),˙Ay(y,t)]=d2k2πd2p2π[a1k,a2p]ei(kx+py)=iδ(xy),

      (58)

      [Ay(x,t),Ay(y,t)]=[˙Ay(x,t),˙Ay(y,t)]=0.

      (59)

      We need to have

      [a1k,a2p]=iδ(k+p),

      (60)

      [a1k,a1p]=[a2k,a2p]=0.

      (61)

      We choose

      a1k=12α(ak+ak),

      (62)

      a2k=iα2(akak),

      (63)

      where α is a real parameter. The operator ak and ap have the commutation relations

      [ak,ap]=δ(kp),

      (64)

      [ak,ap]=[ak,ap]=0.

      (65)

      We define ak|Ω=0, where |Ω is the vacuum of the new set of operators. We find that (62) and (63) satisfy (60) and (61). Thus, from (56), (57), (62), and (63), we have the Bogoliubov transformation

      (bkbk)=1i2ω(iω2α+iα2iω2αiα2iω2αiα2iω2α+iα2)(akak),

      (66)

      with ω=k2x+k2y. From the above Bogoliubov transformation, we have

      bk=1i2ω((iω2α+iα2)ak+(iω2αiα2)ak).

      (67)

      Here (ak,ak) are the annihilation and creation operators of the new modes. We also have the vacuum |0, which is annihilated by bk

      bk|0=0.

      (68)

      From (67), we find that the new set of operators satisfy

      ((iω2α+iα2)ak+(iω2αiα2)ak)|0=0.

      (69)

      The vacuum |0 can be expressed in terms of a state constructed from the new set of operators

      |0=1γekCkakak|Ω

      (70)

      1γk(1Ckakak)|Ω,

      (71)

      where γ is the normalization factor, |Ω is the vacuum defined by ak|Ω=0, and the coefficient Ck is fixed by

      Ck=ωαω+α.

      (72)

      We find that the Bogoliubov transformation of the light-cone gauge field is very similar to that of the free scalar field [4].

      Let us consider a system with a finite number of sites. The inverse Fourier transform of the operator ak is

      ak=NaNeikN,

      (73)

      where N is the site label. The vacuum |Ψ|0 of the operator bk can now be written as

      |Ψ1γk(1N,LCkeik(NL)aNaL)|Ω1γ(1kN,LCkfkNLaNaL)|Ω,

      (74)

      where fkNL=eik(NL). For a simplification of notations, we neglect the vector nation and denote n as n. To compute the entanglement entropy, the configuration space is divided into two regions A and ˉA, where the sites in each region are labelled by small letters (n) and small letters with bars (ˉn), respectively. The region A can be chosen as any subregion of the total system. The state |Ψ is

      |Ψ1γ(1knlCkfknlanalkˉnˉlCkfkˉnˉlaˉnaˉl12kˉnlCk˜fkˉnlaˉnal)|Ω,

      (75)

      where ˜fkˉnl=fkˉnl+fklˉn.

      Following the same procedure of previous cases, we can obtain the reduced density matrix and second Rényi entropy S2, which have the same form as (18) and (22), respectively. The only difference is the expression of Ck.

      In 2+1 dimensional U(1) gauge field theory in the light-cone gauge, the coefficients Ck also control the range of the interaction. The factor kCkfknl can be converted to an integral by taking a continuum limit with an IR cut-off ϵ and UV cut-off K

      Fnl=kCkfknlKϵKϵdkxdkyωαω+αeik(nl)a

      (76)

      F(x,y)=KϵKϵdkxdkyωαω+αei(kxx+kyy),

      (77)

      with ω=k2x+k2y. This is evaluated numerically and shown in Fig. 3.

      Figure 3.  (color online) Plots of integral (77) in 2+1 dimensional U(1) gauge field with light-cone gauge. The left and right figures represent the real and imaginary parts of the integral, respectively. The amplitude of the surface decreases quickly with distance, and the long range contribution can be neglected. ϵ=0.1, K = 10 and α=1 are used in these plots.

      The figures show that the interaction between sites diminishes over large distances. The sites near the boundary contribute most significantly to the entanglement entropy. Hence, an area law is expected in this model.

    5.   Free spin-2 field in light-cone gauge
    • In this section, we consider the 3+1 dimensional free spin-2 field theory in the light-cone gauge. To this end, we derive the Lagrangian and find that it is also the same as the massless scalar field. Subsequently, we perform the Bogoliubov transformation to calculate the entropy.

    • 5.1.   Model of free spin-2 field in light-cone gauge

    • We consider the entanglement entropy of the 3+1 dimensional weak gravitational field gμν=ημν+hμν with light-cone gauge-fixing. Here, hμν is a spin-2 field. In this study, we use the Minkowski metric ημν=diag(1,1,1,1) with the Minkowski coordinate (x0,x1,x2,x3). We also introduce the light-cone coordinate x+,x,x2,x3 with

      x±12(x0±x1).

      (78)

      In the light-cone gauge, the vector is defined as

      a±12(a0±a1),

      (79)

      and the metric becomes

      ˆημν=ˆημν=(0100100000100001).

      (80)

      The vector in the light-cone gauge has the properties a+=a and a=a+.

      Now we come to the calculation of weak gravitational (spin-2) field. We start with the Ricci scalar R. Because R is a scalar, μ; ν can be 0, 1, 2, 3 or +, −, 2, 3. We begin with the light-cone coordinate. We impose the gauge-fixing h++=h+=h+I=0, where I=2,3. For the components with subscript indexes, the gauge fixing is h=h+=hI=0. For calculating the Ricci scalar R, we use

      Γμαβ=gμν2[gανxβ+gβνxαgαβxν]

      (81)

      and

      Rμν=αΓαμννΓαμα+ΓαβαΓβμνΓαβνΓβμα.

      (82)

      The Christoffel symbols are

      Γ+++=12h++x,

      (83)

      Γ++I=12h+Ix,

      (84)

      Γ+IJ=12hIJx,

      (85)

      Γ++=Γ+=Γ+I=0,

      (86)

      Γ++=12h++x+,

      (87)

      Γ+=12h++x,

      (88)

      Γ=0,

      (89)

      Γ+I=12h++xI,

      (90)

      ΓI=12hI+x,

      (91)

      ΓIJ=12(hI+xJ+hJ+xIhIJx+),

      (92)

      ΓI++=12(2h+Ix+h++xI),

      (93)

      ΓI+=12h+Ix,

      (94)

      ΓI=0,

      (95)

      ΓI+J=12(h+IxJ+hJIx+h+JxI),

      (96)

      ΓIJ=12hJIx,

      (97)

      ΓIJK=12(hJIxK+hKIxJhJKxI).

      (98)

      The Ricci scalar is given by

      R=ˆημνRμν=ˆη+R++ˆη+R++ˆηIJRIJ=2R++RII,

      (99)

      where the repeated I, J, and K are summed, and we take this convention below. We have

      R+=122h++x2+122h+IxIx14hIJx(h+IxJ+hJIx+h+JxI)

      (100)

      and

      \setcounterequation101RII=2hI+xxI+2hIJxJxI12(h+Ix)2hIJx(hIJx+hI+xJ)14[(hJKxI)2(hIKxJhJIxK)2].(101)

      The Ricci scalar can be written as

      \setcounterequation102R=2hx2+hIJxIxJ+12hIJx(hIxJ+hJIx++hJxI)12(hIx)2hIJx(hIJx++hIxJ)14[(hJKxI)2(hIKxJhJIxK)2].(102)

      When expressed in Fourier space and considering the light-cone gauge-fixing, we can obtain [18]

      hI=1p+pJhIJ

      (103)

      and

      h=1p+pIhI=pIpJ(p+)2hIJ.

      (104)

      With the above relations, we have

      R=2pIpJhIJ+12pp+hIJhIJ12pJpKhIJhIK+14[(pIhJK)2(pJhIKpKhJI)2].

      (105)

      Because hIJ is symmetric and traceless, there are only two degrees of freedom h22 and h23. We expand the above expression with I,J,K=2,3, and obtain

      R=2pIpJhIJ+(p+p12p2212p23)[(h22)2+(h23)2]=2pIpJhIJ+12((p0)2(p1)2(p2)2(p3)2)×[(h22)2+(h23)2].

      (106)

      Apart from the total derivative term IJhIJ, the Lagrangian of the gravitational field can be written as

      L=12((thyy)2(hyy)2+(thyz)2(hyz)2),

      (107)

      where =xi+yj+zk. We have replaced h22 and h23 with hyy and hyz respectively.There is no gauge redundancy in the Lagrangian (107). We can expect that the Hilbert space of hyy and hyz should have the tensor product structure, so we can consider the entanglement entropy of this model. Hence, we provide a prescription of the entanglement entropy of a free spin-2 field.

    • 5.2.   Bogoliubov transformation and entropy of free spin-2 field in light-cone gauge

    • In the 3+1 dimensional gravitational (spin-2) field with the light-cone gauge, we have two independent degrees of freedom. From (107), we have the equations of motion

      hyy=0,hyz=0.

      (108)

      The equations of motion are the same as the free massless scalar field. Their solutions are

      hyy(x)=d3k2ω(bkeikx+bkeikx)

      (109)

      and

      hyz(x)=d3k2ω(bkeikx+bkeikx)

      (110)

      respectively, where (bk,bk) and (bk,bk) are the creation and annihilation operators of two physical degrees of freedom.

      Let us consider the mode (bk,bk). For the operator hyy, its canonical momentum is ˙hyy. We obtain the solution

      (hyy˙hyy)=d3k(2π)3212ω(1iω)bkeikx+(1iω)bkeikx.

      (111)

      Let us consider a Bogoliubov transformation, which transforms the set of operators (bk,bk) to the following set of operators

      (hyy˙hyy)=d3k(2π)32(a1ka2k)eikx.

      (112)

      With operator a1k and a2k, the commutation relations of hyy and ˙hyy are expressed as below

      [hyy(x,t),˙hyy(y,t)]=d3k(2π)32d3p(2π)32[a1k,a2p]ei(kx+py)=iδ(xy),

      (113)

      [hyy(x,t),hyy(y,t)]=[˙hyy(x,t),˙hyy(y,t)]=0.

      (114)

      We find that apart from the dimension, the Bogoliubov 3+1 dimensional spin-2 field with the light-cone gauge is the same as that of the 2+1 dimensional U(1) gauge field. Moreover, they are both the same with the free scalar field. The form of the Bogoliubov transformation is the same as (66), with ω=k2x+k2y+k2z.

      We define the vacuum |0 by bk|0=0, and the vacuum |Ω by ak|Ω=0 in the 3+1 dimensional spin-2 field with the light-cone gauge. Because of the Bogoliubov transformation, the vacuum |0 can be expressed by the vacuum |Ω and operator (ak,ap) as

      |0=1γekCkakak|Ω

      (115)

      1γk(1Ckakak)|Ω,

      (116)

      where γ is the normalization factor, and the coefficient Ck is fixed by

      Ck=ωαω+α,

      (117)

      with ω=k2x+k2y+k2z.

      We can consider a system with a finite number of sites. We can divide the system into two parts, A and ˉA. The region A we choose can be any subregion of the total system. Considering the state |Ψ=|0, we can obtain the reduced density matrix ρA and the second Rényi entropy S2, which have the same form of the cases in previous sections. The only difference is the expression of Ck.

      In the 3+1 dimensional spin-2 field with the light-cone gauge, the coefficients Ck also control the range of the interaction. The factor kCkfknl can be converted to an integral by taking a continuum limit with an IR cut-off ϵ and UV cut-off K

      Fnl=kCkfknlKϵKϵKϵdkxdkydkzωαω+αeik(nl)a

      (118)

      F(x,y,z)=KϵKϵKϵdkxdkydkzωαω+αei(kxx+kyy+kzz),

      (119)

      with ω=k2x+k2y+k2z. Numerical evaluation in the special case of x = y = z gives the result shown in Fig. 4.

      Figure 4.  (color online) Plots of integral (119) in 3+1 dimensional gravity with the light-cone gauge in the special case of x = y = z. Both the real and imaginary parts of the integral decrease quickly with distance, and the long range contribution can be neglected. ϵ=0.1, K = 10 and α=1 are used in this plot.

      The figure shows that the interaction between sites once again diminishes over large distances. The sites near the boundary contribute most significantly to the entanglement entropy. Hence, an area law is expected in this model as well.

    6.   Conclusions
    • In this study, we explore the entanglement of free spin-12, spin-1 and spin-2 fields. First, we consider the 1+1 dimensional Majorana field, which is just a pair of left and right moving fermions, and the 2+1 dimensional Majorana field. We perform the Bogoliubov transformation of their modes and express the vacuum with a particle pair state in the configuration space. Subsequently, we calculate the second Rényi entropies in the finite systems. Let us emphasize that while a Majorana Weyl fermion is well known to be non-local, a local Hilbert space can be defined when both chiralities are present. This is demonstrated explicitly in the current note. After that, we generalize the method to the 2+1 dimensional free U(1) spin-1 gauge field and the 3+1 dimensional gravitational (free spin-2) field. Because of the gauge redundancy of the higher spin field, there is no Hilbert space with a natural tensor product structure. We take the light-cone gauge for both fields and find that their Lagrangians behave like a free massless scalar field. The light-cone gauge allows simple quantization, while surrendering explicit Lorentz invariance. Nonetheless, it provides a candidate tensor product structure. The definition of entanglement entropy is dependent on both the state and the operator algebra. If the operator algebra is gauge-invariant [10], the corresponding entanglement entropy is likewise gauge-invariant. In this work and in [11], the operator algebras implicitly chosen are not gauge-invariant, such that the corresponding results follow trend. In our past study [11], we explored several different algebras and demonstrated which of those would reproduce the universal log terms found in Casini [10]. It is, however, expected that generic non-gauge invariant algebra choices, such as those considered in the current note, lead to a result that is gauge-dependent. For the U(1) gauge field, the Lagrangian behaves like a scalar field, which coincides with the trivial center case of [11]. As for the gravitational (free spin-2) field, we provide a prescription to observe the tensor product structure of the Hilbert space. Before doing so, there is no prescription of the gravitational (free spin-2) field. After we obtain their Hilbert spaces with a tensor product structure, we calculate the second Rényi entropies. This method can be helpful in dealing with the Hilbert space and entanglement of the perturbative gravitational field, i.e. weak gravitational field. In the non-perturbative regime, the structure of the Hilbert space is still not clear. In all the cases studied, we find that the entropy originates from the particle pairs across the boundary, and the area law emerges naturally.

      We would like to thank Prof. Yong-Shi Wu for critical and meticulous reading of our manuscript. LYH acknowledges the Thousands Young Talents Program.

Reference (18)

目录

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return