Goldstone theorem, Hugenholtz-Pines theorem, and Ward-Takahashi relation in finite volume Bose-Einstein condensed gases (Q2498784)

A Bose-Einstein condensed system in a box with periodic boundary condition is treated. This system satisfies the Ward-Takahashi relation snd the unitary inequivalent vacua are realized. The Goldstone theorem, which says when a continuous symmetry is broken spontaneously, there exists a gapless mode (Nambu-Goldstone mode), which is necessary to keep the original invariance, follows from the Ward-Takahashi identity. The Hugenholtz-Pines theorem [\textit{N. M. Hugenholtz} and \textit{D. Pines}, Phys. Rev. (2) 116, 489--506 (1959; Zbl 0089.22703)], which relates the chemical potential with the self-energy in a nonperturbative way, is also shown as a consequence of the Ward-Takahashi identity. These were shown at tree level by the authors [\textit{M. Okumura} and \textit{Y. Yamanaka}, Phys. Rev. A 68, 13609 (2003), Physica A 348, 157 (2005)]. In this paper, these results are extended at one-loop level. The Lagrangian density (of the zero temperature case) treated in this paper is \[ {\mathcal L}= \Psi^\dag(x)\Biggl(i\hslash{\partial\over\partial t}+ {\hslash^2\over 2m}\nabla^2+ \mu\Biggr)\Psi(x)- {g\over 2} \Psi^\dag(x)\Psi^\dag(x)\Psi(x), \] where \(m\), \(\mu\), and \(g\) are the mass of a boson, the chemical potential, and the coupling constant, respectively, and \(x\) stands for the space-time coordinate \(({\mathbf x},t)\). It describes a weakly interacting Bose gas in a box with the volume \(V= L^3\). \(\Psi\) satisfies the periodic boundary condition in each direction; \(\Psi(x+ L,y,z,t)= \Psi(x,y,z,t)\), etc.. \({\mathcal L}\) is invariant under the global phase transformation \(\Psi(x)\to e^{i\theta}\Psi(x)\). When a uniform Bose-Einstein condensate is created, the global symmetry phase symmetry is spontaneously broken and the quantum field \(\Psi(x)\) is divided into a classical constant field \(v\) and a quantum field \(\phi(x)\); \(\Psi(x)= \phi(x)+ v\). \(v\) is called th order parameter and \(v^2= n\), is the density of condensed particles and the expectation value of the Heisenberg field with respect to the vacuum \(|\Omega\rangle\); \[ \langle\Omega|\Psi(x)|\Omega\rangle= v,\qquad\langle\Omega|\phi(x)|\Omega\rangle= 0. \] The cannonical conjugate of \(\phi(x)\) is \(\pi(x)= i\hslash\phi^\dag(x)\). The cannonical commutation relations of \(\Phi(x, t)\) and \(\phi^\dag(x',t)\) are described. Introducing an artificial symmetry breaking term \({\mathcal L}_\varepsilon= (\varepsilon\overline\varepsilon) v[\Psi(x)+ \Psi^\dag(x)]\), the total Lagrangian density becomes \({\mathcal L}_{\text{tot}}= {\mathcal L}+{\mathcal L}_\varepsilon\). Then after describing the total Hamiltonian \(H= H_0+ H_{\text{int}}+\text{Const.}\), \(H_0\) and \(H_{\text{int}}\) are unperturbed and perturbed terms, the \(2\times 2\)-matrix propagator for the quantum fields fields \[ G(x- x')= (G_{ij}(x- x')),\;G_{11}(x- x')=-i\langle\Omega|{\mathbf T}[\phi(x)\phi^\dag(x)]|\Omega\rangle, \] etc., \({\mathbf T}\) is the symbol of the time-ordered product (Peano series), is defined. Its Fourier transform is denoted by \(G(p)= (G_{ij}(p))\). Then considering the infinitesimal global phase transformation \(\delta\Psi(x)= i\Psi(x)\), the following Ward-Takahashi relation \[ v= -{(\varepsilon\overline\varepsilon)v\over \hslash}[G_{11}\,(p= 0)- G_{12}\,(p= 0)], \] is derived (\S2.2). Diagonalizing unperturbed Hamiltonian by using the Bogoliubov transformation, the Ward-Takahashi relation is confirmed at tree level (\S2.3). Here the mode with \({\mathbf p}= 0\); \({\mathbf p}={2\pi\hslash\over L}\,(n_x, n_y, n_z)\), is included in the transformation, which is essential in finite volume case. The authors say for a finite volume system with a discrete spectrum, the Bogoliubov's prescription without operators the zero-energy mode can not preserve the Ward-Takahashi relations. The Bogoliubov vacuum \(|\Omega_0\rangle\) is explicitely computed in the Appendix. By this computation, \(\langle|\Omega_0\rangle\sim \varepsilon^{1/2}\to 0\) is derived, which shows unitary inequivalence of the Bogoliubov vacuum and \(|0\rangle\), the vacuum annihilated by the annihilation-operators (\S2.3). To derive the Hugenholtz-Pines theorem, the quantum correction \(\delta\mu\) to \(\mu_0\) is introduced by \(\mu= \mu_0+\delta\mu= gv^2- \varepsilon\overline\varepsilon+ \delta\mu\). Then considering the Schwinger-Dyson equation \[ G^{-1}(p)= G^{-1}_0(p)- \Sigma(p)+{\delta\mu\over \hslash} I, \] and apply the Ward-Takahashi relation, the Hugenholtz-Pines theorem \[ {\delta\mu\over\hslash}= \Sigma_{11}\,(p= 0)- \Sigma_{12}\,(p= 0), \] is derived (\S2.5). The Ward-Takahashi relation is keeped at each loop level. By this fact and \(\langle\Omega|\phi(x)|\Omega\rangle= 0\), the validity of the Hugenholtz-Pine theorem at one-loop level is shown in \S2.6. To extend these discussions to finite temperature case, the authors use the thermofield dynamics. In thermofield dynamics, thermal degrees of freedom are introduced by doubling each degree of freedom through the tilde conjugation (\S3.1, cf. \textit{H. Umezawa}, Advanced field theory -- micro, macro, and thermal physics, AIP, New York (1993)]). The total Hamiltonian and Lagrangian take the forms \[ \widehat H= H-\widehat H,\qquad\widehat L= L-\widetilde L. \] The thermal average is expressed as the thermal vacuum expectation value \(_\beta\langle\Omega|A|\Omega\rangle_\beta\) in thermofield dynamics. If the density matrix \(\rho\) is expressed in the form \(f^{a^\dag a}\), we have \[ _\beta\langle\Omega|A|\Omega\rangle_\beta= {\text{tr}[\rho A]\over \text{tr}[\rho]},\quad n= {_\beta\langle\Omega}|a^\dag a|\Omega\rangle_\beta= {f\over 1-f}. \] If \(f= \exp(-\beta e)\), \(\beta\) is the inverse temperature and \(E\) is the energy, respectively. \(_\beta\langle\Omega|A|\Omega\rangle_\beta\) represents the thermal expectation value at finite temperature. Introducing the matrix notation \(\phi^\mu_i\) and \(\overline\phi^\mu_j\), which represent tilde and non-tilde quantum field \(\phi\) and \(\widetilde\phi\), the thermal propagators with four indices and sixteen components are defined by \[ G^{\mu\nu}_{\beta,ij}(x- x')= -i_\beta\langle\Omega |{\mathbf T}[\phi^\mu_i(x) \overline\Phi^\nu_j(x')]|\Omega_\beta. \] Denoting the Fourier transformed thermal propagators by \(G^{\mu\nu}_{ij}(p)\), the Ward-Takahashi relation at finite temperature is \[ v= -{(\varepsilon\overline\varepsilon)v\over \hslash} [G^{11}_{\beta,11}\, (p= 0)- G^{11}_{\beta,12}\,(p= 0)]. \] This is shown in \S3.2 modifying the discussion in zero temperature case. Validity of the Ward-Takahashi relation and the Goldstone theorem at finite temperature case at tree level are shown in \S3.3. Then considering the Schwinger-Dyson equation at finite temperature: \[ G^{-1}_\beta(p)= G^{-1}_{\beta,0}(p)- \Sigma_\beta(p)+ {\delta\mu\over \hslash} I_{4\times 4}, \] the Hugenholtz-Pines theorem at finite temperature \[ {\delta\mu\over\hslash}= \Sigma^{11}_{\beta,11}(p)\,([= 0- \Sigma^{11}_{\beta,12}\,(p= 0), \] is derived in \S3.4. Its validity at one-loop level is hown in \S3.5. Discussions are done by suitable modifycation of the discussions in zero temperature case given in \S2. In the summary of this paper (\S4), the authors emphasize the importance of the study of the zero-energy mode when the volume is finite.