betaFeedback

WorksPhysics and Astronomy

Non-Hermitian Schwinger effect in an imaginary magnetic field

Abstract

We investigate a non-Hermitian Schwinger effect by analyzing a one-loop effective Lagrangian of scalar and Dirac quantum electrodynamics (QED) in the presence of a real electric field and a parallel imaginary magnetic field. We show that an imaginary magnetic field produces an additional pole family in the integrand of the proper-time representation. Unlike a real magnetic field, an imaginary magnetic field can generate a nonzero imaginary part of the effective Lagrangian even in the absence of an electric field. The interplay between the two fields leads to nontrivial phenomena, such as a cancellation mechanism between the contributions of two fields. We also point out that when the ratio of two field strengths is rational, some poles among the two families coincide, and a careful treatment of double poles is required. By further examining the complex Landau-level structure of the electric-pole contribution, we find that the field dependence of the vacuum decay rate from each Landau level exhibits a sinusoidal-like behavior, and their sum determines the overall field dependence. Our results clarify the analytic structure of the QED effective Lagrangian in imaginary magnetic backgrounds and provide a basis for investigating its relation to non-Hermitian phenomena in the QED vacuum.

Introduction

The { Schwinger effect} [1, 2, 3, 4], a landmark prediction of quantum electrodynamics (QED), refers to the vacuum instability or the particle-antiparticle pair creation from the vacuum by a strong electric field (see Refs. [5, 6, 7, 8, 9, 10, 11] for reviews). This phenomenon is a cornerstone of { strong-field physics} and continues to drive research in a wide range of fields, including high-intensity laser physics [12, 13]. Despite its long history, the Schwinger mechanism remains a highly active field of study rich with open questions. Non-Hermitian environments occur in nature and can be controllably realized in experiments, but how the Schwinger effect is modified in such environments remains unknown.

Non-Hermitian quantum mechanics has emerged as a powerful framework for describing open systems, dissipative processes, and instability phenomena [14, 15, 16, 17, 18]. A typical example is the Hatano-Nelson model [19, 20], where an imaginary vector potential was introduced through an asymmetric hopping term. Systems with an imaginary vector potential were experimentally found [21, 22]. More generally, the spatial curl and temporal derivative of an imaginary vector potential can be formally interpreted as { imaginary magnetic and electric fields}, respectively (for phenomenological discussion on imaginary fields, see Refs. [23, 24, 25, 26, 27]). Such situations correspond to inhomogeneous and dynamical generalizations of the Hatano-Nelson model and are experimentally accessible. Therefore, such developments raise a fundamental question: how does an imaginary electromagnetic field influence quantum phenomena in relativistic quantum field theory, particularly { vacuum instability}?

To understand this influence, it is instructive to recall how a real magnetic field affects the Schwinger effect [28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79]. While a real magnetic field does not independently produce particles from the vacuum, it can modify the vacuum decay rate induced by an electric field. In the proper-time representation of the effective Lagrangian, a magnetic field introduces hyperbolic structures, such as 1/tanh⁡(eBs)1/\tanh(eBs) for Dirac QED and 1/sinh⁡(eBs)1/\sinh(eBs) for scalar QED, which reflect Landau-level quantization and the corresponding modification of the vacuum spectrum.

In this work, we investigate the Schwinger effect in scalar and Dirac QED in the presence of a real electric field and a parallel imaginary magnetic field. This background is non-Hermitian and modifies the proper-time structure of the effective Lagrangian. As a result, we obtain intuitive formulas of the imaginary part in the manner of Schwinger’s original formula. Using these formulas, we demonstrate that (i) the emergence of a nonzero imaginary part of the effective Lagrangian even at vanishing electric field, which may be referred to as { imaginary magnetic Schwinger effect}; (ii) the nontrivial interplay between the two fields, such as a cancellation mechanism of singularities; and (iii) the role of the complex Landau-level structure induced by an imaginary magnetic field. These findings clarify the analytic structure of the effective Lagrangian in an imaginary magnetic background and its implications for vacuum instability.

The remainder of this paper is organized as follows. In § 1.1, after the standard Schwinger effect is reviewed, we derive the effective Lagrangians in an imaginary magnetic field, their imaginary parts, and complex Landau level representation. § 1.2 presents the numerical results and their physical interpretation. § 1.3 summarizes our findings and outlook. Appendix A provides an additional review of the standard Schwinger effect, while Appendices B and C provide further discussions of the real and complex Landau levels, respectively. Finally, Appendix D discusses the realization of an effective imaginary magnetic field in a Hatano-Nelson lattice.

Schwinger effect in imaginary magnetic fields

In this section, we study the one-loop effective Lagrangians for scalar and Dirac QED in an imaginary magnetic field. These are derived by analytically continuing the proper-time representation known for a real magnetic field. Using the derived effective Lagrangians, we discuss the structure of the imaginary part. Before doing that, we first briefly review the standard Schwinger effect and its modification in the presence of a real magnetic field, where we assume static and uniform electromagnetic fields.

Standard Schwinger effect

We start with the case of parallel electric and magnetic fields. In the natural units (ℏ=c=1\hbar= c = 1), the effective Lagrangians for scalar and Dirac QED are given by [3]

Lscalar(E,B)=116π2∫0∞dss3e−m2s[(eEs)(eBs)sin⁡(eEs−iϵ)sinh⁡(eBs)−1−e2(E2−B2)s26].(1)\mathcal{L}_{\mathrm{scalar}}(E,B)=\frac{1}{16\pi^2}\int_0^\infty\frac{\mathrm{d}s}{s^3}e^{-m^2s}\left[\frac{(eEs)(eBs)}{\sin(eEs-i\epsilon)\sinh(eBs)}-1-\frac{e^2(E^2-B^2)s^2}{6}\right]. \tag*{(1)}

and [2]

LDirac(E,B)=−18π2∫0∞dss3e−m2s[(eEs)(eBs)tan⁡(eEs−iϵ)tanh⁡(eBs)−1+e2(E2−B2)s23].(2)\mathcal{L}_{\mathrm{Dirac}}(E,B)=-\frac{1}{8\pi^2}\int_0^\infty\frac{\mathrm{d}s}{s^3}e^{-m^2s}\left[\frac{(eEs)(eBs)}{\tan(eEs-i\epsilon)\tanh(eBs)}-1+\frac{e^2(E^2-B^2)s^2}{3}\right]. \tag*{(2)}

respectively, where mm is the particle mass, e>0e>0 is the elementary charge, and ss denotes the Schwinger proper-time parameter. The second and third terms represent the vacuum subtraction and charge-renormalization terms, respectively. The electric-field factors sin⁡(eEs−iϵ)\sin(eEs-i\epsilon) and tan⁡(eEs−iϵ)\tan(eEs-i\epsilon) generate a sequence of poles along the positive real proper-time axis at [2]1

sn=nπ∣eE∣,n=1,2,3,…,(3)s_n=\frac{n\pi}{\lvert eE\rvert},\qquad n=1,2,3,\ldots, \tag*{(3)}

where nn labels the poles of the proper-time integrand. The −iϵ-i\epsilon prescription specifies how the integration contour avoids these poles, as illustrated in Figure 1(a). The resulting pole contributions generate the imaginary part of the effective Lagrangian. Since the vacuum persistence probability is ∣eiVTL∣2=e−2VTIm⁡L|e^{iVT\mathcal{L}}|^2=e^{-2VT\operatorname{Im}\mathcal{L}} in a four volume VTVT, the vacuum decay rate (per unit four volume) Γ=−1VTln⁡∣eiVTL∣2\Gamma=-\frac{1}{VT}\ln|e^{iVT\mathcal{L}}|^2 is Γ=2Im⁡L\Gamma=2\operatorname{Im}\mathcal{L} [4]. In contrast, for a real magnetic field, the magnetic poles lie away from the integration contour and therefore do not contribute directly to the imaginary part. The magnetic field instead modifies the residues at the electric poles through the hyperbolic factors in Eqs. (1) and (2), thereby modifying the vacuum decay rate. The detailed discussions of the imaginary parts are provided in Appendix A.

Figure 1. Contour prescription for the proper-time integral. (a) For a real magnetic field, the electric EE poles lie on the real axis and contribute to the imaginary part of effective Lagrangian through the deformed integration contour, while the magnetic BB poles remain on the imaginary axis. (b) After the analytic continuation B→iBIB \to iB_I, the magnetic BIB_I poles move onto the real axis and also contribute to the imaginary part.

Imaginary magnetic field

By performing the analytic continuation of B→iBIB\to iB_I,2 the hyperbolic factors in Eqs. (1) and (2) become

sinh⁡(ieBIs)=isin⁡(eBIs),(4)\sinh(ieB_Is)=i\sin(eB_Is), \tag*{(4)}
tanh⁡(ieBIs)=itan⁡(eBIs).(5)\tanh(ieB_Is)=i\tan(eB_Is). \tag*{(5)}

One can immediately see that the poles originating from the imaginary magnetic field, which lie on the imaginary axis for real magnetic field, are shifted onto the real axis after the analytic continuation as depicted in Figure 1(b). Note that the imaginary units introduced by the analytic continuation in the numerators and denominators in the first terms of Eqs. (1) and (2) cancel each other.

Scalar QED
  1. Scalar QED

In scalar QED, the one-loop effective Lagrangian in an imaginary magnetic field can be written as

Lscalar(E,BI)=116π2∫0∞dss3e−m2s[e2EBIs2sin⁡(eEs−iϵ)sin⁡(eBIs−iϵ)−1−e2(E2+BI2)s26].(6)\mathcal{L}_{\mathrm{scalar}}(E,B_I)=\frac{1}{16\pi^2}\int_0^\infty\frac{\mathrm{d}s}{s^3}e^{-m^2s}\left[\frac{e^2EB_Is^2}{\sin(eEs-i\epsilon)\sin(eB_Is-i\epsilon)}-1-\frac{e^2(E^2+B_I^2)s^2}{6}\right]. \tag*{(6)}

This form can be regarded as an analytic continuation from the real-magnetic-field case, Eq. (1). Thus, the effective Lagrangian is characterized by the product of two cosecant factors.3

Here, we emphasize that although L(E,BI)\mathcal{L}(E,B_I) can be derived by the analytic continuation B→iBIB \to iB_I, in general, its imaginary part Im⁡L(E,BI)\operatorname{Im}\mathcal{L}(E,B_I) cannot be obtained by simply substituting B→iBIB \to iB_I into Im⁡L(E,B)\operatorname{Im}\mathcal{L}(E,B). This is because, under this continuation, the magnetic-field poles move onto the proper-time integration contour. Therefore, to correctly calculate the imaginary part, we need an appropriate pole prescription, which corresponds to introducing the −iϵ-i\epsilon prescription in the Lagrangian.

In the following, to discuss the imaginary part of this Lagrangian, we separately consider the two cases for the field ratio BI/EB_I/E where E≠0E \ne0: (i) a rational and (ii) an irrational ratio. These two cases require different treatments, particularly because the rational ratio can lead to coincident EE and BIB_I poles, giving rise to the double poles.

When BI/EB_I/E is irrational (i.e., BI/E∉QB_I/E \notin\mathbb{Q}), the imaginary part of Eq. (6) is

Im⁡Lscalar=e2EBI16π2∑n=1∞(−1)n−1nexp⁡(−nπm2∣eE∣)sin⁡(nπBIE)+e2EBI16π2∑k=1∞(−1)k−1kexp⁡(−kπm2∣eBI∣)sin⁡(kπEBI).(7)\operatorname{Im}\mathcal{L}_{\mathrm{scalar}} = \frac{e^2 E B_I}{16\pi^2} \sum_{n=1}^{\infty} \frac{(-1)^{n-1}}{n} \frac{\exp\left(-\frac{n\pi m^2}{\lvert eE\rvert}\right)}{\sin\left(n\pi\frac{B_I}{E}\right)} + \frac{e^2 E B_I}{16\pi^2} \sum_{k=1}^{\infty} \frac{(-1)^{k-1}}{k} \frac{\exp\left(-\frac{k\pi m^2}{\lvert eB_I\rvert}\right)}{\sin\left(k\pi\frac{E}{B_I}\right)} . \tag*{(7)}

The first and second terms arise from the electric-field poles and the imaginary-magnetic-field poles, respectively. Since now we assume an irrational ratio BI/EB_I/E, the electric-field poles sn=nπ/∣eE∣s_n = n\pi/\lvert eE\rvert and the imaginary-magnetic-field poles sk=kπ/∣eBI∣s_k = k\pi/\lvert eB_I\rvert never coincide, and then the expression of the imaginary part can be decomposed into the two simple-pole contributions. In this sense, Eq. (7) may be called the irrational formula or noncoincident-pole formula.

When a field ratio is rational (i.e., BI/E∈QB_I/E \in\mathbb{Q}), the imaginary part of Eq. (6) then becomes

can be interpreted as a parallel configuration, such as the magnetic case ∣eB∣=∣eEI∣\lvert eB\rvert= \lvert eE_I\rvert and the electric case ∣eE∣=∣eBI∣\lvert eE\rvert= \lvert eB_I\rvert. In this case, the effective Lagrangians [81, 82, 83, 84, 85] have 1/sin⁡2(eBs)1/\sin^2(eB s) and 1/sin⁡2(eEs)1/\sin^2(eE s), respectively. Therefore, the Lagrangian (6) with

1/sin⁡(eEs)sin⁡(eBIs)1/\sin(eE s)\sin(eB_I s) can be regarded as a generalized form of the latter.

Im⁡Lscalarrational=e2EBI16π2∑n=1q∤n∞(−1)n−1nexp⁡(−nπm2∣eE∣)sin⁡(nπBIE)+e2EBI16π2∑k=1p∤k∞(−1)k−1kexp⁡(−kπm2∣eBI∣)sin⁡(kπEBI)+116π2∑j=1∞(−1)(p+q)jexp⁡(−πqjm2∣eE∣)[m2∣eE∣qj+(eE)2πq2j2].(8)\operatorname{Im}\mathcal{L}_{\mathrm{scalar}}^{\mathrm{rational}}=\frac{e^2EB_I}{16\pi^2}\sum_{\substack{n=1\\q\nmid n}}^\infty\frac{(-1)^{n-1}}{n}\frac{\exp\left(-\frac{n\pi m^2}{\lvert eE\rvert}\right)}{\sin\left(n\pi\frac{B_I}{E}\right)}+\frac{e^2EB_I}{16\pi^2}\sum_{\substack{k=1\\p\nmid k}}^\infty\frac{(-1)^{k-1}}{k}\frac{\exp\left(-\frac{k\pi m^2}{\lvert eB_I\rvert}\right)}{\sin\left(k\pi\frac{E}{B_I}\right)} +\frac{1}{16\pi^2}\sum_{j=1}^\infty(-1)^{(p+q)j}\exp\left(-\frac{\pi qjm^2}{\lvert eE\rvert}\right)\left[\frac{m^2\lvert eE\rvert}{qj}+\frac{(eE)^2}{\pi q^2j^2}\right]. \tag*{(8)}

The first and second terms are similar to Eq. (7), but the restrictions q∤nq\nmid n and p∤kp\nmid k remove the common poles of electric and imaginary magnetic sectors from the original sums, respectively,4 and p∤kp\nmid k means that kk is not divisible by pp. For common poles, the original pole indices nn and kk can be written as n=qjn=qj and k=pjk=pj, respectively, where j∈N={1,2,3,…}j\in\mathbb{N}=\{1,2,3,\ldots\} is the common-pole index. pp and qq are positive integer indices characterizing the field ratio

BIE=pq,p,q∈N,gcd⁡(p,q)=1,(9)\frac{B_I}{E}=\frac{p}{q},\qquad p,q\in\mathbb{N},\qquad\operatorname{gcd}(p,q)=1, \tag*{(9)}

where gcd⁡\operatorname{gcd} denotes the greatest common divisor.5 The common-pole contributions are contained in the third term, which arises from the double poles and contains the two terms proportional to m2∣eE∣m^2\lvert eE\rvert and (eE)2(eE)^2. Equation (8) may be called the rational formula, which is composed of the first and second terms (i.e., the noncoincident-pole terms) and the last term (the coincident-pole term).

Dirac QED

Similarly, in Dirac QED,

LDirac(E,BI)=−18π2∫0∞dss3e−m2s[e2EBIs2tan⁡(eEs−iϵ)tan⁡(eBIs−iϵ)−1+e2(E2+BI2)s23].(10)\mathcal{L}_{\mathrm{Dirac}}(E,B_I)=-\frac{1}{8\pi^2}\int_0^\infty\frac{ds}{s^3}e^{-m^2s}\left[\frac{e^2EB_Is^2}{\tan(eEs-i\epsilon)\tan(eB_Is-i\epsilon)}-1+\frac{e^2(E^2+B_I^2)s^2}{3}\right]. \tag*{(10)}

Thus, the effective Lagrangian is characterized by the product of two cotangent factors. For BI/E∉QB_I/E\notin\mathbb{Q}, its imaginary part is

Im⁡LDirac=e2EBI8π2∑n=1∞1nexp⁡(−nπm2∣eE∣)tan⁡(nπBIE)+e2EBI8π2∑k=1∞1kexp⁡(−kπm2∣eBI∣)tan⁡(kπEBI).(11)\operatorname{Im}\mathcal{L}_{\mathrm{Dirac}}=\frac{e^2EB_I}{8\pi^2}\sum_{n=1}^\infty\frac{1}{n}\frac{\exp\left(-\frac{n\pi m^2}{\lvert eE\rvert}\right)}{\tan\left(n\pi\frac{B_I}{E}\right)} +\frac{e^2EB_I}{8\pi^2}\sum_{k=1}^\infty\frac{1}{k}\frac{\exp\left(-\frac{k\pi m^2}{\lvert eB_I\rvert}\right)}{\tan\left(k\pi\frac{E}{B_I}\right)}. \tag*{(11)}

For BI/E=p/q∈QB_I/E=p/q\in\mathbb{Q}, we obtain

Im⁡LDiracrational=e2EBI8π2∑n=1q∤n∞1nexp⁡(−nπm2∣eE∣)tan⁡(nπBIE)+e2EBI8π2∑k=1p∤k∞1kexp⁡(−kπm2∣eBI∣)tan⁡(kπEBI)−18π2∑j=1∞exp⁡(−πqjm2∣eE∣)[m2∣eE∣qj+(eE)2πq2j2].(12)\operatorname{Im}\mathcal{L}_{\mathrm{Dirac}}^{\mathrm{rational}}=\frac{e^2EB_I}{8\pi^2}\sum_{\substack{n=1\\q\nmid n}}^\infty\frac{1}{n}\frac{\exp\left(-\frac{n\pi m^2}{\lvert eE\rvert}\right)}{\tan\left(n\pi\frac{B_I}{E}\right)} +\frac{e^2EB_I}{8\pi^2}\sum_{\substack{k=1\\p\nmid k}}^\infty\frac{1}{k}\frac{\exp\left(-\frac{k\pi m^2}{\lvert eB_I\rvert}\right)}{\tan\left(k\pi\frac{E}{B_I}\right)} -\frac{1}{8\pi^2}\sum_{j=1}^\infty\exp\left(-\frac{\pi qjm^2}{\lvert eE\rvert}\right)\left[\frac{m^2\lvert eE\rvert}{qj}+\frac{(eE)^2}{\pi q^2j^2}\right]. \tag*{(12)}
Special cases

Here, we discuss two useful cases. First, in the limit of E→0E\to0, i.e., in pure imaginary magnetic fields, only the second terms in each formula survives:

Im⁡Lscalar(0,BI)=(eBI)216π3∑n=1∞(−1)n−1n2e−nπm2∣eBI∣,(13)\operatorname{Im}\mathcal{L}_{\mathrm{scalar}}(0,B_I)=\frac{(eB_I)^2}{16\pi^3}\sum_{n=1}^\infty\frac{(-1)^{n-1}}{n^2}e^{-\frac{n\pi m^2}{\lvert eB_I\rvert}}, \tag*{(13)}
Im⁡LDirac(0,BI)=(eBI)28π3∑n=1∞1n2e−nπm2∣eBI∣.(14)\operatorname{Im}\mathcal{L}_{\mathrm{Dirac}}(0,B_I)=\frac{(eB_I)^2}{8\pi^3}\sum_{n=1}^\infty\frac{1}{n^2}e^{-\frac{n\pi m^2}{\lvert eB_I\rvert}}. \tag*{(14)}

These expressions are the same functional form as the ordinary pure electric results after ∣eE∣↔∣eBI∣\lvert eE\rvert\leftrightarrow\lvert eB_I\rvert. Second, when the field strengths are equal, ∣eE∣=∣eBI∣\lvert eE\rvert=\lvert eB_I\rvert, we have p=q=1p=q=1. Then, in Eqs. (8) and (12), every pole is a common pole, and the two noncoincidentpole terms vanish. Then, only the third term survives:

Im⁡Lscalar=∣eBI∣16π2[m2Li⁡1(z)+∣eBI∣πLi⁡2(z)](15)\operatorname{Im}\mathcal{L}_{\mathrm{scalar}}=\frac{|eB_I|}{16\pi^2}\left[m^2\operatorname{Li}_1(z)+\frac{|eB_I|}{\pi}\operatorname{Li}_2(z)\right] \tag*{(15)}
Im⁡LDirac=−∣eBI∣8π2[m2Li⁡1(z)+∣eBI∣πLi⁡2(z)](16)\operatorname{Im}\mathcal{L}_{\mathrm{Dirac}}=-\frac{|eB_I|}{8\pi^2}\left[m^2\operatorname{Li}_1(z)+\frac{|eB_I|}{\pi}\operatorname{Li}_2(z)\right] \tag*{(16)}

where we defined z=exp⁡(−πm2/∣eBI∣)z=\exp(-\pi m^2/|eB_I|). Li⁡1(z)=∑j=1∞zjj\operatorname{Li}_1(z)=\sum_{j=1}^{\infty}\frac{z^j}{j} and Li⁡2(z)=∑j=1∞zjj2\operatorname{Li}_2(z)=\sum_{j=1}^{\infty}\frac{z^j}{j^2}, where ∣z∣≤1|z|\le1, are the polylogarithm functions. These formulas are the same as that in Ref. [83], obtained from the effective Lagrangian [81, 82, 83, 84, 85] for a Euclidean-self-dual background. In this sense, ∣eE∣=∣eBI∣|eE|=|eB_I| may be called the Euclidean-self-dual point. Also, there is an exact relation between the two formulas for the scalar and Dirac cases.6

Im⁡LDirac=−2Im⁡Lscalar(∣eE∣=∣eBI∣).(17)\operatorname{Im}\mathcal{L}_{\mathrm{Dirac}}=-2\operatorname{Im}\mathcal{L}_{\mathrm{scalar}}\qquad(|eE|=|eB_I|). \tag*{(17)}

Complex Landau levels

In general, a physical interpretation in terms of Landau levels is useful for understanding the physics in real magnetic fields. For a real magnetic field, the transverse motion of charged particles is quantized into discrete Landau levels, and the vacuum decay rate can be expressed as a sum over these levels (see Appendix B). For an imaginary magnetic field, conventional Landau levels are no longer applicable, but they can be extended to a complex Landau-level spectrum. We here examine the Landau-level structure of the Schwinger effect in the presence of an imaginary magnetic field.

For scalar QED, the corresponding effective squared masses are real

Mℓ2(B)=m2+(2ℓ+1)∣eB∣,ℓ=0,1,2,…(18)M_\ell^2(B)=m^2+(2\ell+1)|eB|,\qquad\ell=0,1,2,\ldots \tag*{(18)}

Using the analytic continuation B→iBIB\to iB_I,7 they become complex,

Mℓ2(BI)=m2+i(2ℓ+1)∣eBI∣.(19)M_\ell^2(B_I)=m^2+i(2\ell+1)|eB_I|. \tag*{(19)}

We refer to this spectrum as complex Landau levels, because the eigenvalues are complex. The imaginary parts of the eigenvalues reflect the resonant nature of the transverse motion, which can be understood in terms of the Gamow states of an inverted harmonic oscillator [87, 88].

Using the complex Landau levels, the electric-pole contribution, i.e., the first term of Eq. (7), can be rewritten as an Abel-regularized Landau-level representation (for a detailed derivation, see Appendix C),

Im⁡Lscalar(E)=lim⁡η→0+Re⁡{ie2EBI8π2∑ℓ=0∞e−(2ℓ+1)η×ln⁡[1+exp⁡(−πMℓ2(BI)∣eE∣)]}.(20)\operatorname{Im}\mathcal{L}_{\mathrm{scalar}}^{(E)}=\lim_{\eta\to0^+}\operatorname{Re}\left\{\frac{ie^2EB_I}{8\pi^2}\sum_{\ell=0}^{\infty}e^{-(2\ell+1)\eta}\times\ln\left[1+\exp\left(-\frac{\pi M_\ell^2(B_I)}{|eE|}\right)\right]\right\}. \tag*{(20)}

Taking the real part is required to obtain the imaginary part of complex quantity. The overall factor ii is regarded as the analytic continuation of the Landau degeneracy factor ∣eB∣/2π|eB|/2\pi. η>0\eta>0 is the Abel regulator. In an imaginary magnetic field, the exponential factors for higher Landau levels no longer provide exponential suppression but instead become oscillatory. Then, the Abel regulator is needed to ensure convergence of the Landau-level sum. By summing all the Landau levels and taking the limit η→0\eta\to0, Eq. (20) agrees with the first term of Eq. (7).

For Dirac QED, the analytically continued effective squared masses are

Mℓ2(BI)=m2+2iℓ∣eBI∣,ℓ=0,1,2,…(21)M_\ell^2(B_I)=m^2+2i\ell|eB_I|,\qquad\ell=0,1,2,\ldots \tag*{(21)}

The difference from the spectrum in scalar QED originates from the spin structure of Dirac field, and, in the nonrelativistic picture, its imaginary shift can be interpreted as the imaginary Zeeman effect.

The electric-pole contribution, i.e., the first term of Eq. (11), can be written as

Im⁡LDirac(E)=lim⁡η→0+Re⁡{−ie2EBI8π2∑ℓ=0∞gℓe−2ℓη×ln⁡[1−exp⁡(−πMℓ2(BI)∣eE∣)]}.(22)\operatorname{Im}\mathcal{L}_{\mathrm{Dirac}}^{(E)}=\lim_{\eta\to0^+}\operatorname{Re}\left\{-\frac{ie^2EB_I}{8\pi^2}\sum_{\ell=0}^{\infty}g_\ell e^{-2\ell\eta}\times\ln\left[1-\exp\left(-\frac{\pi M_\ell^2(B_I)}{|eE|}\right)\right]\right\}. \tag*{(22)}

The overall minus sign and the 1−exp⁡1-\exp form in the logarithmic function reflect Fermi statistics. The degeneracy factor gℓg_\ell comes from the spin structure of Landau levels in Dirac QED. The lowest Landau level ℓ=0\ell=0 has no spin degeneracy, g0=1g_0=1, and higher Landau levels ℓ≥1\ell\ge1 have the spin degeneracy, gℓ=2g_\ell=2.

Note that our current purpose is the comparison of the Landau level structures in real and imaginary magnetic fields. Therefore, in this section, we focused on the Landau-level decompositions of electric poles. Similarly, for the imaginary magnetic poles, we can formally discuss the discrete spectrum depending on the electric field.

In the next section, we numerically investigate the behaviors of individual complex Landau-level contributions and their sum.

Numerical results and discussion

In this section, we present numerical results for the imaginary part of the effective Lagrangian in the presence of an imaginary magnetic field. We focus on four characteristic features: (i) its dependence on BIB_I, (ii) the separate contributions from the electric and imaginary-magnetic poles, (iii) the convergence of the corresponding infinite sums, and (iv) the contributions of the complex Landau levels.

Imaginary part of the effective Lagrangian

In Figure 2(a) and (b), we plot the imaginary part of the one-loop effective Lagrangian on both (E,B)(E,B) and (E,BI)(E,B_I) planes, respectively, where the vertical axis shows the dimensionless quantity Γ/m4\Gamma/m^4, with Γ≡2Im⁡L\Gamma\equiv2\operatorname{Im}\mathcal{L}. In real magnetic fields shown in Figure 2(a), Γ\Gamma vanishes along the E=0E=0 boundary for any value of BB. It is well known that a static real magnetic field does not generate an imaginary part of the effective Lagrangian by itself but modifies the contribution induced by a nonzero electric field. At a fixed nonzero EE, as BB increases, Γ\Gamma is suppressed in scalar QED and enhanced in Dirac QED. This property reflects the different Landau-level structures of the scalar and Dirac fields (see Appendix B for details).

Figure 2. The imaginary part of the effective Lagrangian, Γ=2Im⁡L\Gamma= 2\operatorname{Im}\mathcal{L}, for scalar and Dirac QED as functions of the electric and magnetic fields: (a) real magnetic field BB and (b) imaginary magnetic field BIB_I.

In imaginary magnetic fields, the behavior changes qualitatively, as shown in Figure 2(b). We find that the dependence on EE and BIB_I becomes symmetric under their interchange, and a nonzero BIB_I generates a finite imaginary part even along the E=0E=0 boundary. The ordinary pure-electric result is recovered at BI=0B_I=0. This symmetric behavior is clear from the formulas in Eqs. (7), (8), (11), and (12).

To examine the field dependence more closely, Figure 3 shows Γ\Gamma at fixed electric-field strengths, ∣eE∣/m2=0.5|eE|/m^2=0.5 and 11. The real and imaginary magnetic-field regions are displayed on the positive and negative sides of the horizontal axis, respectively. In the real-field region, the scalar result decreases monotonically as BB increases, whereas the Dirac result increases. In the imaginary-field region, the dependence on BIB_I is considerably more

Imaginary part of the effective Lagrangian for scalar QED

(a)

Imaginary part of the effective Lagrangian for Dirac QED

(b)

Figure 3. The imaginary part of the effective Lagrangian for (a) scalar and (b) Dirac QED as a function of the imaginary (left) and real (right) magnetic fields at fixed ∣eE∣/m2=0.5|eE|/m^2=0.5 (blue) and 11 (red). In the imaginary-field region, the solid lines show the full result, including both EE- and BIB_I-pole contributions, while the dotted lines show the EE-pole contribution alone. A real magnetic field suppresses the scalar result and enhances the Dirac result, whereas the imaginary-field region exhibits nonmonotonic and, for Dirac QED, sign-changing behavior.

intricate: the scalar result increases rapidly, while the Dirac result is nonmonotonic: it is initially negative but becomes positive as BIB_I increases. These differences become more pronounced at larger electric-field strengths. We note that, near BI=0B_I = 0, the full results (solid lines) approach the EE-pole contributions (dotted lines), consistent with the fact that only EE poles contribute for real magnetic fields. As BIB_I increases, the additional BIB_I-pole contributions become significant, leading to increasing deviations.

The qualitative difference between real and imaginary magnetic fields can be understood from the proper-time structure. For a real magnetic field, the BB-dependence is described by hyperbolic functions that introduce no additional poles on the positive real proper-time axis. For an imaginary magnetic field, these functions become trigonometric and generate an additional family of poles, together with the conventional electric poles. The interplay between these two pole families produces the symmetric dependence on EE and BIB_I in Figure 2b, as well as the nonmonotonic and sign-changing behavior shown in Figure 3.

Figure 4 shows the result where the electric and imaginary magnetic fields are varied simultaneously along the line of ∣eE∣=∣eBI∣\lvert eE\rvert= \lvert eB_I\rvert, which corresponds to Euclidean-selfdual field. We find that the scalar result is positive and increases with the common field strength, whereas the Dirac result is negative and decreases. As analytically derived in Eq. (17), the two results satisfy the exact relation ΓDirac=−2Γscalar\Gamma_{\mathrm{Dirac}} = -2\Gamma_{\mathrm{scalar}}.

Figure 4. The imaginary part of the effective Lagrangian for scalar and Dirac QED along the line of ∣eE∣=∣eBI∣\lvert eE\rvert= \lvert eB_I\rvert. The scalar result is positive and increases with the field strength, whereas the Dirac result is negative and satisfies ΓDirac=−2Γscalar\Gamma_{\mathrm{Dirac}} = -2\Gamma_{\mathrm{scalar}}.

The physical interpretation of the imaginary part Γ\Gamma of the effective Lagrangian in an imaginary magnetic field requires some care. For the conventional Hermitian electromagnetic field, Γ\Gamma characterizes the vacuum-persistence probability [4], ∣eiVTL∣2=e−2VTIm⁡L=e−VTΓ\lvert e^{iVT\mathcal{L}}\rvert^2 = e^{-2VT\operatorname{Im}\mathcal{L}} = e^{-VT\Gamma}, where the probabilistic interpretation is based on the unitarity of the time evolution and cannot be directly applied to a non-Hermitian background. In the non-Hermitian case, the sign of Γ\Gamma determines whether the vacuum amplitude decays or grows, and its magnitude determines the strength of this behavior. Thus, Γ\Gamma should not be treated as the probabilistic vacuum decay rate but can be regarded as a rate of vacuum amplitude decay (Γ>0\Gamma> 0) or amplification (Γ<0\Gamma< 0).8

Although in this subsection we found that the total imaginary part appears smooth, each contribution from the electric- or imaginary-magnetic-pole terms exhibits oscillatory and singular structures as functions of BI/EB_I/E because of the trigonometric factors, such as 1/sin⁡(πnBI/E)1/\sin(\pi nB_I/E) and 1/sin⁡(πkE/BI)1/\sin(\pi kE/B_I) in scalar QED and the cotangent factors in Dirac QED. Therefore, we can expect that a cancellation mechanism works. In the next subsection, we examine the two-pole contributions individually and investigate how their singularities are canceled.

Electric- and imaginary-magnetic-pole contributions

Here, we examine the two-pole contributions separately. For scalar QED, the imaginary part can be written schematically as

Im⁡Lscalar=Im⁡L(E)+Im⁡L(BI),(23)\operatorname{Im}\mathcal{L}_{\mathrm{scalar}} = \operatorname{Im}\mathcal{L}^{(E)} + \operatorname{Im}\mathcal{L}^{(B_I)}, \tag*{(23)}

where the first and second terms arise from the electric and imaginary-magnetic poles, respectively.

The scalar-QED results are shown in Figure 5 for truncation orders nmax⁡=kmax⁡=1n_{\max} = k_{\max} = 1, 2, and 3. The electric-pole contribution contains the factor 1/sin⁡(πnBI/E)1/\sin(\pi nB_I/E), so that singularities are located at nBI/E∈ZnB_I/E \in\mathbb{Z}. For n=1n = 1, the poles occur at integer values of BI/EB_I/E, while for n≥2n \ge2, additional poles appear at fractional values of BI/EB_I/E. In contrast, the imaginary-magnetic-pole contribution contains 1/sin⁡(kπE/BI)1/\sin(k\pi E/B_I), and singularities are located at kE/BI∈ZkE/B_I \in\mathbb{Z}. In the left and middle columns of Figure 5, we separately plot the contribution of each pole family as a function of BI/EB_I/E, where we find that Γ\Gamma has singularities and oscillates. With increasing the truncation order, additional singularities appear.

Figure 5. Electric-pole (left), imaginary-magnetic-pole (middle), and total (right) contributions to the imaginary part of the effective Lagrangian in scalar QED as functions of the imaginary magnetic field strength ∣eBI∣/m2|eB_I|/m^2 at fixed ∣eE∣/m2=1|eE|/m^2 = 1. The upper, middle, and lower rows correspond to truncation orders nmax⁡(kmax⁡)=1n_{\max}(k_{\max}) = 1, 2, and 3, respectively.

As shown in the right column of Figure 5, the total contribution exhibits a different behavior due to the interplay between the two pole families. For nmax⁡=kmax⁡=1n_{\max} = k_{\max} = 1, the singularity near BI/E=1B_I/E = 1 is canceled between the contributions from the electric and imaginary magnetic poles, while other singularities near BI/E=1/2B_I/E = 1/2 from the magnetic poles and BI/E=2,3,…B_I/E = 2,3,\ldots from the electric poles remain. Similarly, for nmax⁡=kmax⁡=2n_{\max} = k_{\max} = 2, the singularities near BI/E=1/2B_I/E = 1/2 and 2 are canceled. For nmax⁡=kmax⁡=3n_{\max} = k_{\max} = 3, the singularities near BI/E=1/3B_I/E = 1/3, 2/3, 3/2, and 3 are canceled. Due to this cancellation mechanism, when nmax⁡n_{\max} is sufficiently large, the singular behavior in Γ\Gamma is suppressed, and Γ\Gamma becomes a smooth function.

The corresponding Dirac-QED results are displayed in Figure 6. The two pole families occur at the same field ratios as in scalar QED, but their residues differ because the Dirac effective Lagrangian contains cotangent factors, instead of the cosecant factors of the scalar case. Its cancellation mechanism is similar to the scalar case.

Figure 6. Same as Figure 5, but for Dirac QED.

Convergence of the pole sums

Figure 7 demonstrates the convergence of Γ\Gamma with increasing the truncation order, where we compare the results truncated at nmax⁡=kmax⁡=5n_{\max}=k_{\max}=5 and 100 in the noncoincident-pole formulas in Eqs. (7) and (11). At low truncation order nmax⁡=5n_{\max}=5, the resulting rates exhibit pronounced spike-like structures near the poles of the individual terms. These structures reflect the incomplete cancellation between contributions from the two pole families.

Figure 7. Convergence of Γ=2Im⁡L\Gamma=2\operatorname{Im}\mathcal{L} with increasing pole-sum truncation nmax⁡n_{\max} for (a) scalar and (b) Dirac QED at ∣eE∣/m2=1|eE|/m^2=1. Green and blue points correspond to nmax⁡=kmax⁡=5n_{\max}=k_{\max}=5 and 100, respectively, obtained from the irrational formulas, Eqs. (7) and (11). Red circles denote results at selected rational field ratios in the rational formulas (8) and (12), where nmax⁡n_{\max} is sufficiently large.

Increasing the truncation order modifies the behavior near coincident poles. In particular, the large spikes in low-order truncations are reduced near certain rational field ratios as the corresponding electric- and imaginary-magnetic-pole contributions are combined. For nonsingular field ratios, the partial sums approach the large-nn result.

This convergence behavior is qualitatively similar for scalar and Dirac QED, although their pole contributions differ through the their overall coefficients, sign structure, and trigonometric factors in the respective expressions. The results emphasize that the electric- and imaginary-magnetic-pole sectors must be considered together.

At exactly rational field ratios, BI/E=p/qB_I/E=p/q, the noncoincident-pole formulas in Eqs. (7) and (11) are not applicable because the electric and imaginary-magnetic

poles overlap. Then, the common poles must be excluded from the two simple-pole sums, and their contributions are evaluated from the corresponding double-pole residues, as the third term in the formulas in Eqs. (8) and (12). The red circles in Figure 7 show the results obtained using these formulas. They agree with the limiting values obtained from nearby irrational values of BI/EB_I/E, which confirms the consistency between the rational and irrational formulas.

Complex Landau-level structure

Figure 8 shows the contributions of the lowest three complex Landau levels and their regulated sum, obtained from Eqs. (20) and (22) for scalar and Dirac QED. Here, we fix the Abel regulator at η=0.05\eta=0.05 for illustration. As ℓmax⁡\ell_{\max} increases, the results become closer to the full result. A smaller η\eta requires a larger ℓmax⁡\ell_{\max} to obtain similar results. Each individual level exhibits oscillatory behavior as ∣eBI∣|eB_I| increases, reflecting the phases generated by their complex effective squared masses. Higher Landau levels oscillate more rapidly because their phases are proportional to (2ℓ+1)BI/E(2\ell+1)B_I/E in scalar QED and 2ℓBI/E2\ell B_I/E in Dirac QED.

Figure 8. Landau level decomposition of electric-pole contribution, Γ(E)≡2Im⁡L−Γ(BI)\Gamma^{(E)}\equiv2\operatorname{Im}\mathcal{L}-\Gamma^{(B_I)}, as a function of the imaginary magnetic field strength ∣eBI∣/m2|eB_I|/m^2 for (a) scalar and (b) Dirac QED at fixed ∣eE∣/m2=1|eE|/m^2=1. The Landau-level sum is truncated at ℓmax⁡=50\ell_{\max}=50, and the full result with nmax⁡=1n_{\max}=1 is shown for comparison. Only the electric-pole contribution is included whereas the imaginary-magnetic-pole contribution is not included.

In the two theories, the lowest Landau level (ℓ=0\ell=0) behaves differently. In scalar QED, the effective squared

mass contains an imaginary shift and produces an oscillatory contribution. In Dirac QED, the lowest Landau-level is a real spectrum, M0,Dirac2=m2M_{0,\mathrm{Dirac}}^{2}=m^{2}. Then, its contribution to Γ(E)\Gamma(E) vanishes, as shown by the green dashed curve in Figure 8(b). The mathematical reason is understood from Eqs. (20) and (22): Since Γ(E)\Gamma(E) is obtained by taking the real part of a complex integrand with the overall imaginary unit ii, a real Landau level does not contribute to Γ(E)\Gamma(E). Thus, the nonzero Γ(E)\Gamma(E) is induced by imaginary shifts in the complex Landau-level spectrum.

The blue dashed lines in Figure 8 represent the results obtained by summing over Landau levels up to ℓ=50\ell=50. Although the contribution from each Landau level is a sinusoidal-like oscillation, the sum deforms the functional forms. As the number of Landau levels increases, the sum gradually approaches the full results of Γ(E)\Gamma(E), i.e., the sharp divergences seen in the left column of Figure 5 and Figure 6.

Thus, a decomposition picture in terms of complex Landau levels is formally possible. However, unlike the case of a real magnetic field, the higher Landau levels are not exponentially suppressed. Consequently, an LLL-dominance picture cannot be straightforwardly applied. Nevertheless, the complex Landau-level decomposition may provide a useful qualitative picture for interpreting the full electric-pole result as a sum of individual sinusoidal-like modes.

Conclusion and outlook

In this work, we have investigated the Schwinger effect in the presence of an imaginary magnetic field parallel to a real electric field. We construct the one-loop effective Lagrangians of scalar and Dirac QED, Eqs. (6) and (10), respectively, which are obtained from the analytic continuation B→iBIB\to iB_I. This continuation transforms the hyperbolic factors as a function of real magnetic field into trigonometric functions, which leads to additional poles on the real proper-time axis.

Our main findings are the imaginary parts of the effective Lagrangians, Eqs. (7), (8), (11), and (12). These formulas contain contributions from both the conventional electric-field poles, sn=nπ/∣eE∣s_n=n\pi/|eE|, and the imaginary-magnetic-field poles, sk=kπ/∣eBI∣s_k=k\pi/|eB_I|. Their interplay gives rise to nontrivial field dependence and cancellations between the two contributions. When BI/EB_I/E is rational, some poles from the two families coincide and form double poles. At these field ratios, the separate simple-pole expressions are no longer applicable, and the common poles must be treated carefully.

Natural extensions of the present study are as follows:

  • It would be of interest to study a complex magnetic field, B=BR+iBIB=B_R+iB_I, with both real and imaginary components. Such an extension would allow us to investigate how the proper-time pole structure and complex Landau levels evolve as the magnetic field is continuously varied in the complex plane.

  • It would be interesting to reformulate or rederive our results using other theoretical approaches, including the operator formalism [28], semiclassical approaches such as the worldline instanton method [89], holographic approaches [90], and resurgence methods [91].

  • It would also be interesting to extend the study to the QCD Schwinger effect, where not only quarks [32] but also non-Abelian gauge particles (i.e., gluons) [92, 93] can be produced by a chromoelectric field, and to investigate how an imaginary chromomagnetic field modifies the effective Lagrangian and its pole structure.

  • For experiments, an effective imaginary magnetic field can be realized as a Hatano–Nelson lattice, where a lattice with inhomogeneous nonreciprocal hopping on the two-dimensional plane is needed (see Appendix § 1.2.4). By applying an electric field to the third direction in such lattice systems, we can realize the present situation.

Acknowledgments

A. J. A. is supported by the JAEA Postdoctoral Fellowship Program and partially by the RCNP Collaboration Research Network Program under Project No. COREnet 057. This work is supported by the Japan Society for the Promotion of Science (JSPS) KAKENHI under Grant Nos. JP24K07034 and JP26K17163.

Standard Schwinger effect

For comparison with the imaginary-magnetic-field case, we briefly review the standard Schwinger effect [4] in a constant electric field and its modification by a parallel real magnetic field.

B=0 case

For a constant electric field, corresponding to B=0B=0, the imaginary parts of the one-loop effective Lagrangians for scalar and Dirac QED are [4]

Im⁡Lscalar=(eE)216π3∑n=1∞(−1)n−1n2exp⁡(−nπm2∣eE∣)(24)\operatorname{Im}\mathcal{L}_{\mathrm{scalar}}=\frac{(eE)^2}{16\pi^3}\sum_{n=1}^{\infty}\frac{(-1)^{n-1}}{n^2}\exp\left(-\frac{n\pi m^2}{|eE|}\right) \tag*{(24)}

and [4]

Im⁡LDirac=(eE)28π3∑n=1∞1n2exp⁡(−nπm2∣eE∣)(25)\operatorname{Im}\mathcal{L}_{\mathrm{Dirac}}=\frac{(eE)^2}{8\pi^3}\sum_{n=1}^{\infty}\frac{1}{n^2}\exp\left(-\frac{n\pi m^2}{|eE|}\right) \tag*{(25)}

respectively. The Dirac result differs from the scalar one by an overall spin factor of 2 and by the absence of the alternating factor (−1)n−1(-1)^{n-1}.

Figure 9 shows that the vacuum decay rate is exponentially suppressed in the weak-field regime, ∣eE∣≪m2|eE| \ll m^2, by the factor exp⁡(−πm2/∣eE∣)\exp(-\pi m^2/|eE|). As the field approaches the characteristic scale, ∣eE∣∼m2|eE| \sim m^2, this exponential suppression becomes weaker, and the vacuum decay rate increases rapidly. One can find that the Dirac result is larger than the scalar result due to the spin structure. In the weak-field regime, it is larger by a factor of two, because the n=1n=1 contribution dominates. Then, the sums over proper-time poles converge rapidly because higher-nn terms are suppressed by both 1/n21/n^2 and exp⁡(−nπm2/∣eE∣)\exp(-n\pi m^2/|eE|).

Figure 9. Vacuum decay rates in a real electric field for scalar and Dirac QED, where we compare the results using the truncation of proper-time poles, nmax⁡=1n_{\max}=1, 2 and 100. The Dirac case is larger than the scalar case. Both results increase rapidly as the field approaches the critical scale ∣eE∣∼m2|eE| \sim m^2.

B≠0B \ne0 case

A real magnetic field itself does not induce vacuum decay, whereas in the presence of an electric field, it modifies the vacuum decay rate through Landau-level quantization. The resulting imaginary parts of the one-loop effective Lagrangians are

Im⁡Lscalar(B)=e2EB16π2∑n=1∞(−1)n−1nexp⁡(−nπm2∣eE∣)sinh⁡(nπBE).(26)\operatorname{Im}\mathcal{L}_{\mathrm{scalar}}(B)=\frac{e^2EB}{16\pi^2}\sum_{n=1}^{\infty}\frac{(-1)^{n-1}}{n}\frac{\exp\left(-\frac{n\pi m^2}{|eE|}\right)}{\sinh\left(n\pi\frac{B}{E}\right)}. \tag*{(26)}

and [28, 29]

Im⁡LDirac(B)=e2EB8π2∑n=1∞1nexp⁡(−nπm2∣eE∣)tanh⁡(nπBE).(27)\operatorname{Im}\mathcal{L}_{\mathrm{Dirac}}(B)=\frac{e^2EB}{8\pi^2}\sum_{n=1}^{\infty}\frac{1}{n}\frac{\exp\left(-\frac{n\pi m^2}{|eE|}\right)}{\tanh\left(n\pi\frac{B}{E}\right)}. \tag*{(27)}

In the weak magnetic field limit (B→0B \to0), these expressions smoothly reduce to the pure electric-field formulas, Eqs. (24) and (25). In the strong magnetic field B/E≫1B/E \gg1, the vacuum decay rate in scalar QED is suppressed by lim⁡x→∞1/sinh⁡(x)=0\lim_{x\to\infty}1/\sinh(x)=0, whereas in Dirac QED, the hyperbolic factor does not suppress it because of lim⁡x→∞1/tanh⁡(x)=1\lim_{x\to\infty}1/\tanh(x)=1, and the vacuum decay rate linearly increases by the overall factor BB. As shown in Figure 10, for a fixed electric field ∣eE∣/m2=1|eE|/m^2=1, increasing the magnetic-field strength ∣eB∣/m2|eB|/m^2 enhances the vacuum decay rate in Dirac QED while suppressing it in scalar QED.

Figure 10. Vacuum decay rates in parallel real electric and magnetic fields as a function of ∣eB∣|eB|. A magnetic field suppresses the scalar result, whereas it enhances the Dirac result.

B Real Landau levels

Here, we review the formulas of the Schwinger effect rewritten in terms of Landau-level sums.

Scalar QED

We start from the imaginary part of the effective Lagrangian for scalar QED in real electric and magnetic fields,

Im⁡Lscalar(B)=e2∣E∣∣B∣16π2∑n=1∞(−1)n−1ne−nπm2∣eE∣sinh⁡(nπ∣B∣∣E∣).(28)\operatorname{Im}\mathcal{L}_{\mathrm{scalar}}(B)=\frac{e^2|E||B|}{16\pi^2}\sum_{n=1}^{\infty}\frac{(-1)^{n-1}}{n}\frac{e^{-\frac{n\pi m^2}{|eE|}}}{\sinh\left(n\pi\frac{|B|}{|E|}\right)}. \tag*{(28)}

To isolate the contributions from individual Landau levels, we use

e−nπm2∣eE∣sinh⁡(nπ∣B∣∣E∣)=2∑ℓ=0∞e−nπ∣eE∣(m2+(2ℓ+1)∣eB∣).(29)\frac{e^{-\frac{n\pi m^2}{|eE|}}}{\sinh\left(n\pi\frac{|B|}{|E|}\right)}=2\sum_{\ell=0}^{\infty}e^{-\frac{n\pi}{|eE|}\left(m^2+(2\ell+1)|eB|\right)}. \tag*{(29)}

This naturally defines the effective transverse mass of the ℓ\ell th Landau level as mℓ,scalar2≡m2+(2ℓ+1)∣eB∣m_{\ell,\mathrm{scalar}}^2 \equiv m^2+(2\ell+1)|eB|.

Substituting this expansion into the imaginary part gives

Im⁡Lscalar(B)=e2∣E∣∣B∣8π2∑n=1∞∑ℓ=0∞(−1)n−1n(e−πmℓ,scalar2∣eE∣)n.(30)\operatorname{Im}\mathcal{L}_{\mathrm{scalar}}(B)=\frac{e^{2}|E||B|}{8\pi^{2}}\sum_{n=1}^{\infty}\sum_{\ell=0}^{\infty}\frac{(-1)^{n-1}}{n}\left(e^{-\frac{\pi m_{\ell,\mathrm{scalar}}^{2}}{|eE|}}\right)^{n}. \tag*{(30)}

Exchanging the order of the sums and using ∑n=1∞(−1)n−1nxn=ln⁡(1+x)\sum_{n=1}^{\infty}\frac{(-1)^{n-1}}{n}x^{n}=\ln(1+x) for ∣x∣<1|x|<1, we finally obtain

Im⁡Lscalar(B)=e2∣E∣∣B∣8π2∑ℓ=0∞ln⁡[1+e−πmℓ,scalar2∣eE∣].(31)\operatorname{Im}\mathcal{L}_{\mathrm{scalar}}(B)=\frac{e^{2}|E||B|}{8\pi^{2}}\sum_{\ell=0}^{\infty}\ln\left[1+e^{-\frac{\pi m_{\ell,\mathrm{scalar}}^{2}}{|eE|}}\right]. \tag*{(31)}

b. Dirac QED For Dirac QED, we start from the imaginary part [28, 29]

Im⁡LDirac(B)=e2∣E∣∣B∣8π2∑n=1∞1ne−nπm2∣eE∣tanh⁡(nπ∣B∣∣E∣).(32)\operatorname{Im}\mathcal{L}_{\mathrm{Dirac}}(B)=\frac{e^{2}|E||B|}{8\pi^{2}}\sum_{n=1}^{\infty}\frac{1}{n}\frac{e^{-\frac{n\pi m^{2}}{|eE|}}}{\tanh\left(\frac{n\pi|B|}{|E|}\right)}. \tag*{(32)}

The hyperbolic cotangent can be decomposed as

e−nπm2∣eE∣tanh⁡(nπ∣B∣∣E∣)=∑ℓ=0∞(2−δℓ0)e−nπ∣eE∣(m2+2ℓ∣eB∣).(33)\frac{e^{-\frac{n\pi m^{2}}{|eE|}}}{\tanh\left(\frac{n\pi|B|}{|E|}\right)}=\sum_{\ell=0}^{\infty}(2-\delta_{\ell0})e^{-\frac{n\pi}{|eE|}(m^{2}+2\ell|eB|)}. \tag*{(33)}

From this form, we can define the effective transverse mass of the ℓ\ell th Landau level as mℓ,Dirac2≡m2+2ℓ∣eB∣m_{\ell,\mathrm{Dirac}}^{2}\equiv m^{2}+2\ell|eB|.

Substituting this expression gives

Im⁡LDirac(B)=e2∣E∣∣B∣8π2∑ℓ=0∞(2−δℓ0)∑n=1∞1n(e−πmℓ,Dirac2∣eE∣)n.(34)\operatorname{Im}\mathcal{L}_{\mathrm{Dirac}}(B)=\frac{e^{2}|E||B|}{8\pi^{2}}\sum_{\ell=0}^{\infty}(2-\delta_{\ell0})\sum_{n=1}^{\infty}\frac{1}{n}\left(e^{-\frac{\pi m_{\ell,\mathrm{Dirac}}^{2}}{|eE|}}\right)^{n}. \tag*{(34)}

Using ∑n=1∞xnn=−ln⁡(1−x)\sum_{n=1}^{\infty}\frac{x^{n}}{n}=-\ln(1-x) for ∣x∣<1|x|<1, we obtain

Im⁡LDirac(B)=−e2∣E∣∣B∣8π2∑ℓ=0∞(2−δℓ0)ln⁡[1−e−πmℓ,Dirac2∣eE∣].(35)\operatorname{Im}\mathcal{L}_{\mathrm{Dirac}}(B)=-\frac{e^{2}|E||B|}{8\pi^{2}}\sum_{\ell=0}^{\infty}(2-\delta_{\ell0})\ln\left[1-e^{-\frac{\pi m_{\ell,\mathrm{Dirac}}^{2}}{|eE|}}\right]. \tag*{(35)}

Equivalently, separating the lowest Landau level (LLL), ℓ=0\ell=0, from the higher Landau levels,

Im⁡LDirac(B)=−e2∣E∣∣B∣8π2[ln⁡(1−e−πm2∣eE∣)+2∑ℓ=1∞ln⁡(1−e−πmℓ,Dirac2∣eE∣)].(36)\operatorname{Im}\mathcal{L}_{\mathrm{Dirac}}(B)=-\frac{e^{2}|E||B|}{8\pi^{2}}\left[\ln\left(1-e^{-\frac{\pi m^{2}}{|eE|}}\right)+2\sum_{\ell=1}^{\infty}\ln\left(1-e^{-\frac{\pi m_{\ell,\mathrm{Dirac}}^{2}}{|eE|}}\right)\right]. \tag*{(36)}

Figure Figure 11 shows the Landau-level decomposition of vacuum decay rates at fixed ∣eE∣/m2=1|eE|/m^{2}=1. At weak magnetic fields, many Landau levels contribute because the spacing between adjacent levels vanishes as ∣eB∣→0|eB|\to0. With increasing ∣eB∣|eB|, the contributions from higher Landau levels are exponentially suppressed, and the LLL dominates. For scalar QED [Figure 11(a)], the transverse mass of the LLL increases due to m0,scalar2=m2+∣eB∣m_{0,\mathrm{scalar}}^{2}=m^{2}+|eB|, and the vacuum decay rate at large ∣eB∣|eB| is suppressed. In contrast, for Dirac QED [Figure 11(b)], since the LLL does not depend on BB due to m0,Dirac=mm_{0,\mathrm{Dirac}}=m, its contribution to the rate increases linearly in ∣eB∣|eB| through the overall degeneracy factor ∣eB∣/2π|eB|/2\pi.

Figure 11. Landau-level decomposition of the vacuum decay rate for (a) scalar and (b) Dirac QED, where we compare the results using truncations of higher Landau levels, ℓmax⁡=0\ell_{\max}=0, 1, 2, and 100.

C Derivation of complex Landau levels

Here, we derive Eqs. (20) and (22) in the complex Landau-level representation of the electric-pole contribution for scalar and Dirac QED in an imaginary magnetic field.

a. Scalar QED. The electric-pole contribution in scalar QED is

Im⁡Lscalar(E)=e2∣E∣∣BI∣16π2∑n=1∞(−1)n−1nexp⁡(−nπm2∣eE∣)sin⁡(nπ∣BI∣∣E∣).(37)\operatorname{Im}\mathcal{L}_{\mathrm{scalar}}^{(E)}=\frac{e^{2}|E||B_{\mathrm{I}}|}{16\pi^{2}}\sum_{n=1}^{\infty}\frac{(-1)^{n-1}}{n}\frac{\exp\left(-\frac{n\pi m^{2}}{|eE|}\right)}{\sin\left(\frac{n\pi|B_{\mathrm{I}}|}{|E|}\right)}. \tag*{(37)}

The cosecant factor can be expressed as 1/sin⁡x=2ie−ix/(1−e−2ix)1/\sin x=2ie^{-ix}/(1-e^{-2ix}). Since the geometric-series expansion of 1/(1−e−2ix)1/(1-e^{-2ix}) is not convergent for real xx, we introduce an Abel regulator η>0\eta>0 and write

1sin⁡(x−iη)=2i∑ℓ=0∞e−(2ℓ+1)ηe−i(2ℓ+1)x.(38)\frac{1}{\sin(x-i\eta)}=2i\sum_{\ell=0}^{\infty}e^{-(2\ell+1)\eta}e^{-i(2\ell+1)x}. \tag*{(38)}

For x∉πZx \notin\pi\mathbb{Z}, the original cosecant is recovered in the limit η→0+\eta\to0^{+}. By substituting this form into the original imaginary part, we obtain

Im⁡Lscalar(E)=lim⁡η→0+Re⁡{ie2∣E∣∣BI∣8π2∑n=1∞∑ℓ=0∞(−1)n−1ne−(2ℓ+1)ηe−nπ∣eE∣(m2+i(2ℓ+1)∣eBI∣)}.(39)\operatorname{Im}\mathcal{L}_{\mathrm{scalar}}^{(E)}=\lim_{\eta\to0^{+}}\operatorname{Re}\left\{\frac{ie^{2}|E||B_I|}{8\pi^{2}}\sum_{n=1}^{\infty}\sum_{\ell=0}^{\infty}\frac{(-1)^{n-1}}{n}e^{-(2\ell+1)\eta}e^{-\frac{n\pi}{|eE|}\left(m^{2}+i(2\ell+1)|eB_I|\right)}\right\}. \tag*{(39)}

Here, we take the real part of the regulated expression to obtain a real-valued quantity for finite η>0\eta>0. The factor of m2+i(2ℓ+1)∣eBI∣≡Mℓ,scalar2(BI)m^{2}+i(2\ell+1)|eB_I|\equiv M_{\ell,\mathrm{scalar}}^{2}(B_I) corresponds to the complex effective squared mass of the ℓ\ell th Landau level for scalar QED. Using ∑n=1∞(−1)n−1nxn=ln⁡(1+x)\sum_{n=1}^{\infty}\frac{(-1)^{n-1}}{n}x^{n}=\ln(1+x) for ∣x∣<1|x|<1, we obtain

Im⁡Lscalar(E)=lim⁡η→0+Re⁡{ie2∣E∣∣BI∣8π2∑ℓ=0∞e−(2ℓ+1)ηln⁡[1+exp⁡(−πMℓ,scalar2(BI)∣eE∣)]}.(40)\operatorname{Im}\mathcal{L}_{\mathrm{scalar}}^{(E)}=\lim_{\eta\to0^{+}}\operatorname{Re}\left\{\frac{ie^{2}|E||B_I|}{8\pi^{2}}\sum_{\ell=0}^{\infty}e^{-(2\ell+1)\eta}\ln\left[1+\exp\left(-\frac{\pi M_{\ell,\mathrm{scalar}}^{2}(B_I)}{|eE|}\right)\right]\right\}. \tag*{(40)}

This is the Abel-regularized complex Landau-level representation for the electric-pole contribution in Eq. (20).

b. Dirac QED. For Dirac QED, the electric-pole contribution is

Im⁡LDirac(E)=e2∣E∣∣BI∣8π2∑n=1∞1nexp⁡(−nπm2∣eE∣)tan⁡(nπ∣BI∣∣E∣).(41)\operatorname{Im}\mathcal{L}_{\mathrm{Dirac}}^{(E)}=\frac{e^{2}|E||B_I|}{8\pi^{2}}\sum_{n=1}^{\infty}\frac{1}{n}\frac{\exp\left(-\frac{n\pi m^{2}}{|eE|}\right)}{\tan\left(n\pi\frac{|B_I|}{|E|}\right)}. \tag*{(41)}

The cotangent factor 1/tan⁡x=i(1+e−2ix)/(1−e−2ix)=i[1+2e−2ix/(1−e−2ix)]1/\tan x=i(1+e^{-2ix})/(1-e^{-2ix})=i[1+2e^{-2ix}/(1-e^{-2ix})] is written as, using an Abel regulator η>0\eta>0,

1tan⁡(x−iη)=i∑ℓ=0∞gℓe−2ℓηe−2iℓx,(42)\frac{1}{\tan(x-i\eta)}=i\sum_{\ell=0}^{\infty}g_{\ell}e^{-2\ell\eta}e^{-2i\ell x}, \tag*{(42)}

where we introduced the degeneracy factor gℓ=2−δℓ0g_{\ell}=2-\delta_{\ell0}: g0=1g_{0}=1 for the lowest Landau level, while gℓ=2g_{\ell}=2 for the higher Landau levels.

Substituting Eq. (42) into Eq. (41) gives

Im⁡LDirac(E)=lim⁡η→0+Re⁡{ie2∣E∣∣BI∣8π2∑n=1∞∑ℓ=0∞gℓne−2ℓηexp⁡[−nπ∣eE∣(m2+2iℓ∣eBI∣)]}.(43)\operatorname{Im}\mathcal{L}_{\mathrm{Dirac}}^{(E)}=\lim_{\eta\to0^{+}}\operatorname{Re}\left\{\frac{ie^{2}|E||B_I|}{8\pi^{2}}\sum_{n=1}^{\infty}\sum_{\ell=0}^{\infty}\frac{g_{\ell}}{n}e^{-2\ell\eta}\exp\left[-\frac{n\pi}{|eE|}\left(m^{2}+2i\ell|eB_I|\right)\right]\right\}. \tag*{(43)}

The corresponding complex effective squared masses are m2+2iℓ∣eBI∣≡Mℓ,Dirac2(BI)m^{2}+2i\ell|eB_I|\equiv M_{\ell,\mathrm{Dirac}}^{2}(B_I). Using ∑n=1∞xnn=−ln⁡(1−x)\sum_{n=1}^{\infty}\frac{x^{n}}{n}=-\ln(1-x) for ∣x∣<1|x|<1, we obtain

Im⁡LDirac(E)=lim⁡η→0+Re⁡{−ie2∣E∣∣BI∣8π2∑ℓ=0∞gℓe−2ℓηln⁡[1−exp⁡(−πMℓ,Dirac2(BI)∣eE∣)]}.(44)\operatorname{Im}\mathcal{L}_{\mathrm{Dirac}}^{(E)}=\lim_{\eta\to0^{+}}\operatorname{Re}\left\{-\frac{ie^{2}|E||B_I|}{8\pi^{2}}\sum_{\ell=0}^{\infty}g_{\ell}e^{-2\ell\eta}\ln\left[1-\exp\left(-\frac{\pi M_{\ell,\mathrm{Dirac}}^{2}(B_I)}{|eE|}\right)\right]\right\}. \tag*{(44)}

This is Eq. (22).

Inhomogeneous Hatano-Nelson lattice

An imaginary magnetic field can be realized theoretically in lattice systems (and also experimentally, if the corresponding Hamiltonian can be implemented). In this Appendix, as an example of such a realization, we consider a two-dimensional lattice system with a nonreciprocal interaction based on the Hatano-Nelson model [19, 20], described by the Hamiltonian

H=∑r∑μ=x,y[tμf(r)cr†cr+aμ^+tμb(r)cr†cr+aμ^].(45)H=\sum_{\mathbf r}\sum_{\mu=x,y}\left[t_{\mu}^{f}(\mathbf r)c_{\mathbf r}^{\dagger}c_{\mathbf r+a\hat{\mu}}+t_{\mu}^{b}(\mathbf r)c_{\mathbf r}^{\dagger}c_{\mathbf r+a\hat{\mu}}\right]. \tag*{(45)}

where cr†c_{\mathbf r}^{\dagger} and crc_{\mathbf r} are the creation and annihilation operators at lattice site r=(x,y)\mathbf r=(x,y), respectively, aa is the lattice spacing, and μ^\hat{\mu} is the unit vector along the μ\mu direction. tμft_{\mu}^{f} and tμbt_{\mu}^{b} are the forward and backward hopping parameters, respectively. In reciprocal systems, tμf=tμbt_{\mu}^{f}=t_{\mu}^{b}.

Here, we consider a spatially dependent nonreciprocity

txf(r)=txb(r)=t,(46)t_{x}^{f}(\mathbf r)=t_{x}^{b}(\mathbf r)=t, \tag*{(46)}
tyf(r)=te−γ(x),tyb(r)=te+γ(x),(47)t_{y}^{f}(\mathbf r)=te^{-\gamma(x)},\quad t_{y}^{b}(\mathbf r)=te^{+\gamma(x)}, \tag*{(47)}

where γ(x)\gamma(x) characterizes the nonreciprocity along the yy direction. In the conventional Hatano-Nelson case, γ\gamma is independent of xx, whereas γ\gamma depends on xx in an inhomogeneous Hatano-Nelson case.

The link variable is defined as

Uμ(r)=exp⁡[ie∫rr+aμ^A(ℓ)⋅dℓ](48)U_{\mu}(\mathbf{r})=\exp\left[ie\int_{\mathbf{r}}^{\mathbf{r}+a\hat{\mu}}\mathbf{A}(\ell)\cdot d\ell\right] \tag*{(48)}

where ee is the particle charge, A=(Ax,Ay)\mathbf{A}=(A_x,A_y) is the gauge potential, aa is the lattice spacing, and ℓ\ell denotes a point along the integration path.

We choose the Landau gauge, A=(0,Ay(x))\mathbf{A}=(0,A_y(x)), with the xx-dependent imaginary gauge potential

Ay(x)=iγ(x)ea.(49)A_y(x)=\frac{i\gamma(x)}{ea}. \tag*{(49)}

Then, the link variables become

Ux(r)=1,Uy(r)=eieaAy(x)=e−γ(x).(50)U_x(\mathbf{r})=1,\qquad U_y(\mathbf{r})=e^{ieaA_y(x)}=e^{-\gamma(x)}. \tag*{(50)}

Thus, the forward and backward link variables, Uy(x)U_y(x) and Uy−1(x)U_y^{-1}(x), correspond to the xx-dependent nonreciprocal hopping parameters in Eq. (47).

The plaquette variable is defined as the product of link variables along the four links:

Pxy(x)=Ux(x,y)Uy(x+a,y)Ux(x,y+a)−1Uy(x,y)−1=Uy(x+a)Uy(x)=e−[γ(x+a)−γ(x)],(51)\begin{aligned} P_{xy}(x)&=U_x(x,y)U_y(x+a,y)U_x(x,y+a)^{-1}U_y(x,y)^{-1}\\ &=\frac{U_y(x+a)}{U_y(x)}\\ &=e^{-[\gamma(x+a)-\gamma(x)]}, \tag*{(51)} \end{aligned}

as shown in Figure 12(a). When we substitute a linear profile γ(x)=eBIax\gamma(x)=eB_Iax, we can find that the plaquette variable is spatially uniform:

Figure 12. Imaginary magnetic field realized on a two-dimensional lattice with inhomogeneous (i.e., xx-dependent) nonreciprocal parameters γ(x)\gamma(x). (a) Plaquette in the imaginary gauge potential A=(0,Ay(x))\mathbf A=(0,A_y(x)) with Ay=iγ(x)eaA_y=\frac{i\gamma(x)}{ea}. (b) Lattice with γ(x)=eBIax\gamma(x)=eB_I ax.

Pxy(x)=e−eBIa2,(52)P_{xy}(x)=e^{-eB_Ia^2}, \tag*{(52)}

Thus, although the profile of γ(x)\gamma(x) is inhomogeneous, each plaquette variable is homogeneous, and the corresponding magnetic field is also homogeneous, as in Figure 12(b).

We note that the extension to a three-dimensional lattice is straightforward. In addition, if one imposes periodic boundary conditions on this system, the xx-direction link variables at the lattice edges in the yy direction must be modified to ensure the homogeneity of plaquette variables.

References

References

  1. [1]F. Sauter, Uber das Verhalten eines Elektrons im homogenen elektrischen Feld nach der relativistischen Theorie, Z. Phys. 69, 742 (1931).DOI
  2. [2]W. Heisenberg and H. Euler, Consequences of Dirac’s theory of positrons, Z. Phys. 98, 714 (1936), arXiv:physics/0605038.
  3. [3]V. Weisskopf, Über die elektrodynamik des vakuums auf grund der quantentheorie des elektrons, Kong. Dans. Vid. Selsk. Math.-fys. Medd. XIV (1936), reprinted in Quantum Electrodynamics, edited by J. Schwinger, Dover, New York (1958).
  4. [4]J. S. Schwinger, On gauge invariance and vacuum polarization, Phys. Rev. 82, 664 (1951).
  5. [5]G. V. Dunne, Heisenberg-Euler effective Lagrangians: Basics and extensions, in From fields to strings: Circumnavigating theoretical physics. Ian Kogan memorial collection (3 volume set), edited by M. Shifman, A. Vainshtein, and J. Wheater (2004) pp. 445–522, arXiv:hep-th/0406216.DOI
  6. [6]R. Ruffini, G. Vereshchagin, and S.-S. Xue, Electron-positron pairs in physics and astrophysics: from heavy nuclei to black holes, Phys. Rept. 487, 1 (2010), arXiv:0910.0974 [astro-ph.HE].DOI
  7. [7]F. Gelis and N. Tanji, Schwinger mechanism revisited, Prog. Part. Nucl. Phys. 87, 1 (2016), arXiv:1510.05451 [hep-ph].DOI
  8. [8]P. Copinger and S. Pu, Chirality production with mass effects — Schwinger pair production and the axial Ward identity, Int. J. Mod. Phys. A 35, 2030015 (2020), arXiv:2008.03635 [hep-ph].DOI
  9. [9]A. Fedotov, A. Ilderton, F. Karbstein, B. King, D. Seipt, H. Taya, and G. Torgrimsson, Advances in QED with intense background fields, Phys. Rept. 1010, 1 (2023), arXiv:2203.00019 [hep-ph].DOI
  10. [10]K. Hattori, K. Itakura, and S. Ozaki, Strong-field physics in QED and QCD: From fundamentals to applications, Prog. Part. Nucl. Phys. 133, 104068 (2023), arXiv:2305.03865 [hep-ph].DOI
  11. [11]H. Taya, The strong-field effect in QCD and nuclear physics (2026) arXiv:2603.07847 [hep-ph].
  12. [12]V. Yakimenko et al., Prospect of Studying Nonperturbative QED with Beam-Beam Collisions, Phys. Rev. Lett. 122, 190404 (2019), arXiv:1807.09271 [physics.plasm-ph].DOI
  13. [13]C. N. Danson et al., Petawatt and exawatt class lasers worldwide, High Power Laser Science and Engineering 7, e54 (2019).
  14. [14]C. M. Bender, Making sense of non-Hermitian Hamiltonians, Rept. Prog. Phys. 70, 947 (2007), arXiv:hep-th/0703096.DOI
  15. [15]R. El-Ganainy, K. G. Makris, M. Khajavikhan, Z. H. Musslimani, S. Rotter, and D. N. Christodoulides, Non-Hermitian physics and PT symmetry, Nature Phys. 14, 11 (2018).
  16. [16]E. J. Bergqoltz, J. C. Budich, and F. K. Kunst, Exceptional topology of non-Hermitian systems, Rev. Mod. Phys. 93, 015005 (2021), arXiv:1912.10048 [cond-mat.mes-hall].
  17. [17]Y. Ashida, Z. Gong, and M. Ueda, Non-Hermitian physics, Adv. Phys. 69, 249 (2021), arXiv:2006.01837 [cond-mat.mes-hall].DOI
  18. [18]K. Ding, C. F. Fang, and G. Ma, Non-Hermitian topology and exceptional-point geometries, Nature Rev. Phys. 4, 745 (2022), arXiv:2204.11601 [quant-ph].DOI
  19. [19]N. Hatano and D. R. Nelson, Localization transitions in non-Hermitian quantum mechanics, Phys. Rev. Lett. 77, 570 (1996), arXiv:cond-mat/9603165.DOI
  20. [20]N. Hatano and D. R. Nelson, Vortex pinning and non-Hermitian quantum mechanics, Phys. Rev. B 56, 8651 (1997), arXiv:cond-mat/9705290.DOI
  21. [21]S. Weidemann, M. Kremer, T. Helbig, T. Hofmann, A. Stegmaier, M. Greiter, R. Thomale, and A. Szameit, Topological funneling of light, Science 368, 311 (2020).DOI
  22. [22]Y. G. N. Liu, Y. Wei, O. Hemmatyar, G. G. Pyrialakos, P. S. Jung, D. N. Christodoulides, and M. Khajavikhan, Complex skin modes in non-hermitian coupled laser arrays, Light: Science & Applications 11, 336 (2022).DOI
  23. [23]T. Ozawa and T. Hayata, Two-dimensional lattice with an imaginary magnetic field, Phys. Rev. B 109, 085113 (2024), arXiv:2307.14635 [cond-mat.mes-hall].DOI
  24. [24]Z. Wang, D. Liao, Q. Mo, Y. Ma, C. L. Ju, J. Li, and S. Zhang, Tunable landau quantization via imaginary electric fields, Sci. Bull. 71, 4112 (2026).DOI
  25. [25]A. Montag and T. Ozawa, Non-Hermitian Landau Levels, Phys. Rev. B 114, 245128 (2026), arXiv:2605.23613 [quant-ph].arxiv.org/abs/2605.23613
  26. [26]A. J. Arifi and K. Suzuki, Hadronic exceptional points, (2026), arXiv:2606.31697 [hep-ph].arxiv.org/abs/2606.31697
  27. [27]S. Alminas, D. Burba, and G. Juzeliūnas, Imaginary magnetic fields for ultracold atoms, (2026), arXiv:2610.06793 [cond-mat.quant-gas].arxiv.org/abs/2610.06793
  28. [28]A. I. Nikishov, Pair production by a constant external field, Zh. Eksp. Teor. Fiz. 57, 1210 (1969).
  29. [29]F. V. Bunkin and I. I. Tugov, The possibility of electron-positron pair production in vacuum when laser radiation is focussed, Dokl. Akad. Nauk Ser. Fiz. 187, 541 (1969).
  30. [30]M. S. Marinov and V. S. Popov, Pair production in electromagnetic field (case of arbitrary spin), Yad. Fiz. 15, 1271 (1972).
  31. [31]J. K. Daugherty and I. Lerche, Theory of pair production in strong electric and magnetic fields and its applicability to pulsars, Phys. Rev. D 14, 340 (1976).DOI
  32. [32]M. Claudson, A. Yildiz, and P. H. Cox, Vacuum behavior in quantum chromodynamics. II, Phys. Rev. D 22, 2022 (1980).
  33. [33]H. Suganuma and T. Tatsumi, On the behavior of symmetry and phase transitions in a strong electromagnetic field, Annals Phys. 208, 470 (1991).
  34. [34]B. V. Ivanov, On vacuum polarization in constant electromagnetic fields, Phys. Lett. B 282, 228 (1992).DOI
  35. [35]H. Suganuma and T. Tatsumi, Chiral symmetry and quark - anti-quark pair creation in a strong color electromagnetic field, Prog. Theor. Phys. 90, 379 (1993).
  36. [36]S. R. Valluri, D. R. Lamm, and W. J. Mielniczuk, Applications of the representation of the Heisenberg-Euler Lagrangian by means of special functions, Can. J. Phys. 71, 389 (1993).DOI
  37. [37]R. Soldati and L. Sorbo, Effective action for Dirac spinors in the presence of general uniform electromagnetic fields, Phys. Lett. B 426, 82 (1998), arXiv:hep-th/9802167.DOI
  38. [38]S. P. Gavrilov, D. M. Gitman, and A. E. Goncalves, QED in external field with space-time uniform invariants: Exact solutions, J. Math. Phys. 39, 3547 (1998), arXiv:hep-th/0004132.
  39. [39]Q.-g. Lin, Electron-positron pair creation in vacuum by an electromagnetic field in 3 + 1-dimensions and lower dimensions, J. Phys. G 25, 17 (1999), arXiv:hep-th/9810037.
  40. [40]M. N. Hounkonnou and M. Naciri, Production of Dirac particles in vacuum by external fields in d + 1 (d = 3,2,1) dimensions, J. Phys. G 26, 1849 (2000).DOI
  41. [41]Y. M. Cho and D. G. Pak, Effective action: A Convergent series of QED, Phys. Rev. Lett. 86, 1947 (2001), arXiv:hep-th/0006057.DOI
  42. [42]W. S. Bae, Y. M. Cho, and D. G. Pak, Electric magnetic duality in QED effective action, Phys. Rev. D 64, 017303 (2001), arXiv:hep-th/0011196.DOI
  43. [43]U. D. Jentschura, H. Gies, S. R. Valluri, D. R. Lamm, and E. J. Weniger, QED effective action revisited, Can. J. Phys. 80, 267 (2002), arXiv:hep-th/0107135.DOI
  44. [44]A. I. Nikishov, Vector boson in constant electromagnetic field, J. Exp. Theor. Phys. 93, 197 (2001), arXiv:hep-th/0104019.DOI
  45. [45]S. I. Kruglov, Pair production and vacuum polarization of arbitrary spin particles with EDM and AMM, Annals Phys. 293, 228 (2001), arXiv:hep-th/0110061.DOI
  46. [46]S. I. Kruglov, Pair production and vacuum polarization of vector particles with electric dipole moments and anomalous magnetic moments, Eur. Phys. J. C 22, 89 (2001), arXiv:hep-ph/0110100.DOI
  47. [47]S. P. Kim and D. N. Page, Schwinger pair production in electric and magnetic fields, Phys. Rev. D 73, 065020 (2006), arXiv:hep-th/0301132.DOI
  48. [48]S. P. Kim, H. K. Lee, and Y. Yoon, Effective action of scalar QED in electric field backgrounds, Phys. Rev. D 78, 105013 (2008), arXiv:0807.2696 [hep-th].DOI
  49. [49]N. Tanji, Dynamical view of pair creation in uniform electric and magnetic fields, Annals Phys. 324, 1691 (2009), arXiv:0810.4429 [hep-ph].DOI
  50. [50]A. Iwazaki, Pair Creation in Electric Flux Tube and Chiral Anomaly, Phys. Rev. C 80, 052202 (2009), arXiv:0908.4466 [hep-ph].DOI
  51. [51]K. Fukushima, D. E. Kharzeev, and H. J. Warringa, Real-time dynamics of the Chiral Magnetic Effect, Phys. Rev. Lett. 104, 212001 (2010), arXiv:1002.2495 [hep-ph].DOI
  52. [52]N. Tanji and K. Itakura, Schwinger mechanism enhanced by the Nielsen–Olesen instability, Phys. Lett. B 713, 117 (2012), arXiv:1111.6772 [hep-ph].DOI
  53. [53]H. J. Warringa, Dynamics of the Chiral Magnetic Effect in a weak magnetic field, Phys. Rev. D 86, 085029 (2012), arXiv:1205.5679 [hep-th].DOI
  54. [54]S. Bolognesi, F. Kiefer, and E. Rabinovici, Comments on Critical Electric and Magnetic Fields from Holography, JHEP 01, 174, arXiv:1210.4170 [hep-th].
  55. [55]Y. Sato and K. Yoshida, Holographic description of the Schwinger effect in electric and magnetic fields, JHEP 04, 111, arXiv:1303.0112 [hep-th].DOI
  56. [56]K. Hashimoto, T. Oka, and A. Sonoda, Electromagnetic instability in holographic QCD, JHEP 06, 001, arXiv:1412.4254 [hep-th].arxiv.org/abs/1412.4254
  57. [57]S. Ozaki, T. Arai, K. Hattori, and K. Itakura, Euler-Heisenberg-Weiss action for QCD+QED, Phys. Rev. D 92, 016002 (2015), arXiv:1504.07532 [hep-ph].DOI
  58. [58]M. Ghodrati, Schwinger Effect and Entanglement Entropy in Confining Geometries, Phys. Rev. D 92, 065015 (2015), arXiv:1506.08557 [hep-th].DOI
  59. [59]P. Copinger and K. Fukushima, Spatially assisted Schwinger mechanism and magnetic catalysis, Phys. Rev. Lett. 117, 081603 (2016), [Erratum: Phys.Rev.Lett. 118, 099903 (2017)], arXiv:1605.05957 [hep-th].arxiv.org/abs/1605.05957
  60. [60]E. Bavarsad, S. P. Kim, C. Stahl, and S.-S. Xue, Effect of a magnetic field on Schwinger mechanism inside Sitter spacetime, Phys. Rev. D 97, 025017 (2018), arXiv:1707.03975 [hep-th].DOI
  61. [61]P. Copinger, K. Fukushima, and S. Pu, Axial Ward identity and the Schwinger mechanism – Applications to the real-time chiral magnetic effect and condensates, Phys. Rev. Lett. 121, 261602 (2018), arXiv:1807.04416 [hep- th].DOI
  62. [62]M. Korwar and A. M. Thalapillil, Finite temperature Schwinger pair production in coexistent electric and magnetic fields, Phys. Rev. D 98, 076016 (2018), arXiv:1808.01295 [hep-th].DOI
  63. [63]X.-L. Sheng, R.-H. Fang, Q. Wang, and D. H. Rischke, Wigner function and pair production in parallel electric and magnetic fields, Phys. Rev. D 99, 056004 (2019), arXiv:1812.01146 [hep-ph].DOI
  64. [64]D. Karabali, S. Kürkçüoğlu, and V. P. Nair, Magnetic Field and Curvature Effects on Pair Production I: Scalars and Spinors, Phys. Rev. D 100, 065005 (2019), arXiv:1904.11687 [hep-th].DOI
  65. [65]D. Karabali, S. Kürkçüoğlu, and V. P. Nair, Magnetic Field and Curvature Effects on Pair Production II: Vectors and Implications for Chromodynamics, Phys. Rev. D 100, 065006 (2019), arXiv:1905.12391 [hep-th].DOI
  66. [66]Z.-R. Zhu, D.-f. Hou, and X. Chen, Potential analysis of holographic Schwinger effect in the magnetized background, Eur. Phys. J. C 80, 550 (2020), arXiv:1912.05806 [hep-ph].DOI
  67. [67]K. Hattori, K. Itakura, and S. Ozaki, Note on all-order Landau-level structures of the Heisenberg-Euler effective actions for QED and QCD, (2020), arXiv:2001.06131 [hep-ph].arxiv.org/abs/2001.06131
  68. [68]H. Taya, Dynamically assisted Schwinger mechanism and chirality production in parallel electromagnetic field, Phys. Rev. Res. 2, 023257 (2020), arXiv:2003.08948 [hep-ph].DOI
  69. [69]S. Bhattacharya, S. Chakrabortty, H. Hoshino, and S. Kaushal, Background magnetic field and quantum correlations in the Schwinger effect, Phys. Lett. B 811, 135875 (2020), arXiv:2005.12866 [hep-th].DOI
  70. [70]P. Copinger and P. Morales, Schwinger pair production in SL(2,C) topologically nontrivial fields via non-Abelian worldline instantons, Phys. Rev. D 103, 036004 (2021), arXiv:2011.12526 [hep-th].DOI
  71. [71]M. S. Ali, S. Bhattacharya, S. Chakrabortty, and S. Kaushal, Fermionic Bell violation in the presence of background electromagnetic fields in the cosmological de Sitter spacetime, Phys. Rev. D 104, 125012 (2021), arXiv:2102.11745 [hep-th].arxiv.org/abs/2102.11745
  72. [72]S. Kaushal, Schwinger effect and a uniformly accelerated observer, Eur. Phys. J. C 82, 872 (2022), arXiv:2201.03906 [hep-th].DOI
  73. [73]P. Copinger and Y. Hidaka, Angular momentum inheritance from the Schwinger effect in (chromo)electromagnetic fields, PTEP 2023, 023B08 (2023), arXiv:2203.10917 [hep-ph].DOI
  74. [74]P. Copinger, K. Hattori, and D.-L. Yang, Euler-Heisenberg Lagrangian under an axial gauge field, Phys. Rev. D 107, 056016 (2023), arXiv:2208.12913 [hep-th].DOI
  75. [75]C. Yu, Evolution of chirality in a multiphoton pair production process, Phys. Rev. D 108, 116009 (2023), arXiv:2309.03570 [hep-ph].DOI
  76. [76]S. Kaushal, Fermionic entanglement in the presence of background electric and magnetic fields, Eur. Phys. J. C 85, 75 (2025), arXiv:2408.15309 [hep-th].DOI
  77. [77]G. V. Dunne, Heisenberg-Euler and the quantum dilogarithm, Phys. Rev. D 113, 085018 (2026), arXiv:2512.14915 [hep-th].DOI
  78. [78]D. Gupta and A. M. Thalapillil, Resurgence in scalar and spinor QED: the Euler–Heisenberg Lagrangian in parallel field backgrounds, Eur. Phys. J. C 86, 750 (2026), arXiv:2512.22775 [hep-th].DOI
  79. [79]S. Tahery and Q. Chang, Holographic Schwinger effect in a step dilaton background, JHEP 06, 181, arXiv:2604.17743 [hep-th].DOI
  80. [80]A. A. Belavin, A. M. Polyakov, A. S. Schwartz, and Y. S. Tyupkin, Pseudoparticle Solutions of the Yang-Mills Equations, Phys. Lett. B 59, 85 (1975).
  81. [81]G. V. Dunne and C. Schubert, Closed-form two-loop Euler-Heisenberg Lagrangian in a selfdual background, Phys. Lett. B 526, 55 (2002), arXiv:hep-th/0111134.DOI
  82. [82]G. V. Dunne and C. Schubert, Two-loop self-dual Euler-Heisenberg Lagrangians. 1. Real part and helicity amplitudes, JHEP 08, 053, arXiv:hep-th/0205004.
  83. [83]G. V. Dunne and C. Schubert, Two-loop self-dual Euler-Heisenberg Lagrangians. 2. Imaginary part and Borel analysis, JHEP 06, 042, arXiv:hep-th/0205005.DOI
  84. [84]G. V. Dunne, H. Gies, and C. Schubert, Zero modes, beta functions and IR/UV interplay in higher loop QED, JHEP 11, 032, arXiv:hep-th/0210240.DOI
  85. [85]G. V. Dunne, Two-loop diagrammatics in a self-dual background, JHEP 02, 013, arXiv:hep-th/0311167.
  86. [86]A. D’Adda and P. Di Vecchia, Supersymmetry and Instantons, Phys. Lett. B 73, 162 (1978).
  87. [87]G. Barton, Quantum mechanics of the inverted oscillator potential, Annals Phys. 166, 322 (1986).DOI
  88. [88]V. Subramanyan, S. S. Hegde, S. Vishveshwara, and B. Bradlyn, Physics of the Inverted Harmonic Oscillator: From the lowest Landau level to event horizons, Annals Phys. 435, 168470 (2021), arXiv:2012.09875 [cond-mat.mes-hall].DOI
  89. [89]I. K. Affleck, O. Alvarez, and N. S. Manton, Pair Production at Strong Coupling in Weak External Fields, Nucl. Phys. B 197, 509 (1982).DOI
  90. [90]G. W. Semenoff and K. Zarembo, Holographic Schwinger Effect, Phys. Rev. Lett. 107, 171601 (2011), arXiv:1109.2920 [hep-th].DOI
  91. [91]A. Florio, Schwinger pair production from Padé–Borel reconstruction, Phys. Rev. D 101, 013007 (2020), arXiv:1911.03489 [hep-th].DOI
  92. [92]I. A. Batalin, S. G. Matinyan, and G. K. Savvidy, Vacuum Polarization by a Source-Free Gauge Field, Sov. J. Nucl. Phys. 26, 214 (1977).
  93. [93]A. Yildiz and P. H. Cox, Vacuum Behavior in Quantum Chromodynamics, Phys. Rev. D 21, 1095 (1980).DOI

Paper details

vlnc:W_2xrdww67
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents