At finite temperature, quantum field theory is conventionally split between the Matsubara (imaginary-time) formalism, which yields equilibrium thermodynamics, and real-time formalisms — thermofield dynamics and closed-time-path methods — needed for dynamics and kinetics. The standard derivation of the real-time formalism deforms the Matsubara integration contour in the complex time plane and assumes the vertical connecting pieces decouple. A. A. Abrikosov, Jr. (Modern Physics Letters A 5 (1990) 2183) showed that an interaction switched on adiabatically while obeying Kubo–Martin–Schwinger periodicity and vanishing at real-time infinities necessarily develops operator-valued poles in the strip −β≤Im⁡t≤0-\beta \le \operatorname{Im} t \le 0; their residues obstruct the contour deformation, generate operator insertions T2exp⁡(−βV^)T_2\exp(-\beta\hat V) shifting the thermodynamic potential by ΔΩ=−(1/β)ln⁡tr⁡[T2exp⁡(−βV^)]\Delta\Omega = -(1/\beta)\ln\operatorname{tr}\left[T_2\exp(-\beta\hat V)\right], and imply that a consistent real-time technique must be built on the true interacting thermal vacuum rather than the noninteracting one. The unresolved problem is to turn this observation into a working formalism: operational diagrammatic rules incorporating the pole-residue insertions, a systematic branch-cut prescription separating the equilibrated interacting system from response-to-switching problems, computations of the residue contributions to the thermodynamic potential for concrete interacting theories (including interaction-dominated ground states such as superconducting condensates), and an extension to general singularity structures (multiple or higher-order poles, many-body branch cuts).

Background

Finite-temperature quantum field theory rests on two complementary legs. The Matsubara (imaginary-time) formalism computes equilibrium thermodynamics by treating the partition function Z=tr⁡e−βH^Z=\operatorname{tr} e^{-\beta\hat H}, with β\beta the inverse temperature, as a field theory on a finite imaginary-time interval of length β\beta, along which bosonic fields are periodic and fermionic fields antiperiodic; frequency integrals become sums over discrete Matsubara frequencies, and the thermodynamic potential Ω=−(1/β)ln⁡Z\Omega=-(1/\beta)\ln Z generates equilibrium quantities. Dynamical observables — real-time correlation functions, linear response, transport — instead require a real-time formalism (RTF). Two standard constructions exist: the closed-time-path (Schwinger–Keldysh) contour, and thermofield dynamics (TFD), an operator formulation in which every physical field is accompanied by an unphysical tilde partner so that the doubled Hilbert space carries a temperature-dependent pure state, the thermal vacuum ∣0(β)⟩|0(\beta)\rangle, whose expectation values reproduce thermal averages; for free fields ∣0(β)⟩|0(\beta)\rangle is obtained from the zero-temperature doubled vacuum by a Bogoliubov transformation, a temperature-dependent rotation mixing each mode with its tilde partner.

In the time-path derivation, the real-time formalism is obtained from the Matsubara functional integral by deforming the imaginary-time contour into one running along the real axis (and, in TFD, along a parallel ghost axis at Im⁡t=−β/2\operatorname{Im} t=-\beta/2), with vertical connecting pieces sent to infinity (T→∞T\to\infty). For free fields this is exact and the vertical pieces contribute only a multiplicative renormalization of the statistical sum. Interactions are customarily switched on adiabatically: the interaction V(t)V(t) vanishes at t=±∞t=\pm\infty so that asymptotic states are free.

A. A. Abrikosov, Jr. observed that in a thermal system this picture is self-contradictory. The KMS (Kubo–Martin–Schwinger) condition — the periodicity property that characterizes thermal equilibrium — requires V(t+iβ)=V(t)V(t+i\beta)=V(t); combined with reality on the real axis and vanishing at real-time infinities, VV is analytic on a cylinder in complex time that closes to a sphere once t=±∞t=\pm\infty are added, and the balance of zeros and poles then forces V(t)V(t) to possess singularities — generically two simple poles symmetric about the line Im⁡t=−β/2\operatorname{Im} t=-\beta/2 — inside the strip −β≤Im⁡t≤0-\beta\le\operatorname{Im} t\le 0 through which the contour deformation must pass. Their operator-valued residues W^(tk∗)\hat W(t_k^*) (defined locally by V(t)=W^(t∗)/(t−t∗)+O(1)V(t)=\hat W(t^*)/(t-t^*)+O(1)) are fixed by an integral of VV over the Matsubara segment, and a contour encircling a pole inserts an ordered exponential whose net effect is the insertion of T2exp⁡(−βV^)T_2\exp(-\beta\hat V), where T2T_2 denotes ordering along the imaginary-time segment. The corresponding correction to the thermodynamic potential (per unit volume vv, in the traced convention of eq. (12) of the author's 1993 recap, arXiv:hep-th/9412201) is

Δp=−ΔΩv=1βvln⁡tr⁡[T2exp⁡(−βV^)],\Delta p=-\frac{\Delta\Omega}{v}=\frac{1}{\beta v}\ln\operatorname{tr}\left[T_2\exp(-\beta\hat V)\right],

whose logarithm reproduces the sum of vacuum-loop (Matsubara) contributions. Hence the vertical pieces of the deformed contour do not decouple: an interacting thermal system is described by an interacting thermal vacuum that differs from the free one precisely by the insertions the deformation generates, and the branch structure of the complex-time evolution distinguishes two different physical problems — the response of a free thermal state to switching the interaction on, and the fully equilibrated interacting system.

The author's subsequent Banff talk sharpened the point with examples where the noninteracting thermal vacuum fails qualitatively: temperature-dependent symmetry restoration, superconductivity (where the condensate makes the free quasiparticle basis incomplete), zero modes that defeat the Riemann–Lebesgue argument behind decoupling, and degenerate levels split by an interaction (a spin-1/2 example in which evolution from the degenerate state gives zero magnetization where the correct thermal answer is nonzero).

The question sits at the foundations of real-time thermal perturbation theory. Modern practice largely bypasses it: in the Schwinger–Keldysh closed-time-path formalism the initial state is the full interacting Gibbs density matrix ρ∝e−βH^\rho\propto e^{-\beta\hat H}, so the interacting equilibrium state is built in from the start and the contour-deformation question is never posed; resummation schemes — hard thermal loops (the one-loop diagrams with hard external momenta that dominate hot gauge theories and must be resummed) and two-particle-irreducible effective actions (effective actions written as functionals of full dressed propagators, whose self-consistent skeleton expansion resums the vacuum diagrams) — control the infrared and pinch-singularity problems (divergences of naive real-time loop integrals that arise when the real-time branches of the contour are pinched between poles of thermal propagators) of expansions around the free thermal state. What was never produced is the object the source asks for: a real-time technique whose construction takes the interaction-induced complex-time singularities explicitly into account, with residue insertions, branch prescriptions, and thermodynamic corrections worked out for concrete interacting theories.

Problem Statement

For a thermal quantum system with interaction V^\hat V at inverse temperature β\beta, construct an operational real-time formalism founded on the true interacting thermal vacuum, with the interaction-induced singularities in complex time explicitly incorporated. The scientific objective is the source's single general request — a real-time technique constructed in the true interacting vacuum, taking the interaction into account, whose equilibrium and kinetic predictions — thermodynamic quantities and real-time correlation functions — coincide with those of the interacting Gibbs state e−β(H^0+V^)e^{-\beta(\hat H_0+\hat V)}, where H^0\hat H_0 is the noninteracting Hamiltonian. The construction is to be exhibited in explicit operational form, with the interaction-induced complex-time singularities incorporated, and demonstrated for at least one interacting theory by taking account of those singularities rather than assuming the interacting state as given input. The directions listed below are aspects of this objective discussed in the source and in the author's own 1993 recap; they describe facets of one and the same construction rather than separate tasks. Throughout, T2exp⁡(−βV^)T_2\exp(-\beta\hat V) denotes the exponential of the interaction ordered along the Matsubara imaginary-time segment, and W^(tk∗)\hat W(t_k^*) denotes the operator residue of the interaction's complex-time pole at t=tk∗t=t_k^*, as defined in the Background.

(1) Operational rules. Provide explicit diagrammatic and operator prescriptions showing how the insertions T2exp⁡(−βV^)T_2\exp(-\beta\hat V), generated by the residues W^(tk∗)\hat W(t_k^*) of the interaction's complex-time poles, enter real-time calculations, and how they modify the thermal vacuum of the doubled Hilbert space of thermofield dynamics (TFD), so that real-time expansions are organized around the interacting thermal state rather than the noninteracting one.

(2) Response versus equilibrium. Distinguish, within the complex-time construction, calculations that evolve a noninteracting initial thermal state under an adiabatically switched perturbation — which retain a memory of the initial state and can fail qualitatively, as in the spin-1/2 magnetization example of the 1993 recap — from the fully equilibrated interacting system, taking into account that relaxation processes break the analyticity on which the contour-deformation argument rests and generate branch-cut (relaxation-zone) structure in the complex-time plane.

(3) Computations for concrete theories. For specific interacting field theories — including at least one whose noninteracting thermal vacuum is unstable to the interaction, such as a superconducting condensate — compute the residue operators W^(tk∗)\hat W(t_k^*) and their contributions to the thermodynamic-potential correction

Δp=−ΔΩv=1βvln⁡tr⁡[T2exp⁡(−βV^)].\Delta p=-\frac{\Delta\Omega}{v}=\frac{1}{\beta v}\ln\operatorname{tr}\left[T_2\exp(-\beta\hat V)\right].

(4) General singularity structures. Extend the analysis to multiple poles, higher-order poles, and branch cuts induced by many-body (relaxation) effects, and thereby delimit the conditions under which real-time kinetic calculations around the noninteracting thermal vacuum remain valid.

References

  1. A. A. Abrikosov, Jr., 'Real-time formalism of thermofield dynamics and Matsubara approach in quantum statistical physics', Modern Physics Letters A 5 (1990) 2183-2188. DOI: 10.1142/S0217732390002481, https://doi.org/10.1142/S0217732390002481
  2. A. A. Abrikosov (Jr.), 'The Problem of Ground State in Thermo-Field Dynamics', talk given at the 3rd Workshop and Summer School on Thermal Field Theories and their Applications (Banff, Canada, August 15-27, 1993); published in: F. C. Khanna, H. Umezawa, R. Kobes and G. Kunstatter (eds.), 'Thermal Field Theories and their Applications' (World Scientific, Singapore, 1994), pp. 146-154. arXiv:hep-th/9412201, https://arxiv.org/abs/hep-th/9412201
  3. A. A. Abrikosov, 'Interacting thermal vacuum and noninteracting asymptotic states in thermo-field dynamics', Physics of Atomic Nuclei 59 (1996) 352-359 [Yadernaya Fizika 59 (1996) 372-378]. INSPIRE record: https://inspirehep.net/literature/431956
  4. T. Altherr, 'Introduction to thermal field theory', International Journal of Modern Physics A 8 (1993) 5605-5628. DOI: 10.1142/S0217751X93002216, https://doi.org/10.1142/S0217751X93002216; arXiv:hep-ph/9307277, https://arxiv.org/abs/hep-ph/9307277
  5. A. V. Smilga, 'Physics of thermal QCD', Physics Reports 291 (1997) 1-106. DOI: 10.1016/S0370-1573(97)00014-8, https://doi.org/10.1016/S0370-1573(97)00014-8; arXiv:hep-ph/9612347, https://arxiv.org/abs/hep-ph/9612347
  6. J. Berges, 'Introduction to nonequilibrium quantum field theory', 2004. arXiv:hep-ph/0409233, https://arxiv.org/abs/hep-ph/0409233
  7. N. P. Landsman and Ch. G. van Weert, 'Real- and imaginary-time field theory at finite temperature and density', Physics Reports 145 (1987) 141-249. DOI: 10.1016/0370-1573(87)90121-9, https://doi.org/10.1016/0370-1573(87)90121-9
  8. A. Das and P. Kalauni, 'Operator description for thermal quantum field theories on an arbitrary path in the real time formalism', Physical Review D 93 (2016) 125028. DOI: 10.1103/PhysRevD.93.125028, https://doi.org/10.1103/PhysRevD.93.125028
  9. X.-Y. Hou, Z. Huang, H. Guo, Y. He and C.-C. Chien, 'BCS thermal vacuum of fermionic superfluids and its perturbation theory', Scientific Reports 8 (2018) 11995. DOI: 10.1038/s41598-018-30438-1, https://doi.org/10.1038/s41598-018-30438-1
  10. H. Queiroz, 'O operador espalhamento para férmions num campo externo em Thermofield Dynamics' [The scattering operator for fermions in an external field in Thermofield Dynamics], PhD thesis, Instituto de Física Teórica, São Paulo, 2002 (supervisors B. M. P. Escobar and J. de L. Tomazelli). INSPIRE record: https://inspirehep.net/literature/1501092