Time evolution in an external field: The unitarity paradox (Q5934240)

One of the main postulates of the axiomatic relativistic quantum field theory (QFT) in the Wightman approach [\textit{A. S. Wightman}, Phys. Rev. 101, 860-866 (1956; Zbl 0074.22902); \textit{R. F. Streater} and \textit{A. S. Wightman}, PCT, spin and statistics and all that, Amsterdam, Addison-Wesley (1989; Zbl 0704.53058)] is that the Poincaré transformations should be unitary operators in the Hilbert state space [\textit{N. N. Bogoliubov, A. A. Logunov, A. I. Oksal} and \textit{I. T. Todorov}, General principles of quantum field theory, Nauka, Moscow (1987); English transl. Kluwer Academic, Dordrecht (1990; Zbl 0732.46040)]. However, if one considers the quantum electrodynamics in the strong external classical electromagnetic field in the leading order of perturbation theory, the creation and annihilation operators at different moments in time are related by a nonunitary Bogoliubov canonical transformation. Hence it is necessary to use different representations of the canonical commutation relations (CCR) at different moments of time. This suggested that also in the nonperturbative QFT different representations of CCR at different moments of time should be considered, while the time translation is not a unitary operator in the Hilbert space but rather a transformation connecting different representations of CCR [\textit{O. I. Zavyalov} and \textit{V. N. Sushko}, in ``Statistical physics and quantum field theory'', N. N. Bogoliubov (ed.), 411-439, Nauka, Moscow (1973)]. This suggestion is in contradiction with the Wightman axiomatic approach. In the first section of the paper a simple exactly solvable model is considered, and the Hilbert state space and unitary evolution operator are constructed using the Bogoliubov S-matrix approach [\textit{N. N. Bogolyubov}, Dokl. Akad. Nauk SSSR 81, 757-760 (1951; Zbl 0044.23402); \textit{N. N. Bogoliubov} and \textit{D. V. Shirkov}, Introduction to the theory of quantized fields, Interscience, New York (1959; Zbl 0088.21701)]. In the next section the model is studied in the strong external field in the leading order of perturbation theory and the corresponding states are constructed. It is shown that the evolution transformation is a nonunitary canonical transformation. After these two rather technical sections, the apparent contradiction is solved by showing how the representations of CCR which appear to be equivalent in the exact theory are not equivalent in the approximate theory. Therefore unitarity of evolution in the exact theory is not in contradiction with the nonunitarity of the approximate theory.