Extended Furry Picture in QED

• Extended Furry picture is a formulation in QED that replaces the bare Coulomb potential with an effective one, incorporating screening effects.
• It reorganizes perturbation theory around dressed states, reducing higher-order corrections and resolving quasidegeneracies in atomic calculations.
• The approach is applied in bound-state and strong-field QED, bridging rigorous ab initio methods with many-body approximations for high-precision results.

Searching arXiv for recent and foundational papers on the extended Furry picture to ground the article in published work. The extended Furry picture is a formulation of relativistic quantum electrodynamics in which a nontrivial external or effective field is absorbed into the zeroth-order one-particle dynamics, so perturbation theory is organized around dressed states rather than free states or, in bound-state problems, states in the bare nuclear Coulomb field alone. In bound-state QED for highly charged ions and many-electron atoms, this usually means replacing $V_{\rm nuc}$ by an effective potential $V_{\rm eff}=V_{\rm nuc}+V_{\rm scr}$, with $V_{\rm scr}$ a local screening potential; in strong-field QED, it means treating a prescribed classical background field exactly and the remaining interactions perturbatively; and in the mathematical no-pair theory of heavy atoms it is the natural extension of the Furry picture obtained by replacing the bare Coulomb projector with one built from a screened or self-consistent Dirac operator [(Malyshev et al., 2014); (Ginges et al., 2017); (Merz et al., 2020); (Acosta et al., 2023)].

1. Definition, standard form, and scope

In the standard bound-state Furry picture, electrons are treated as moving exactly in the Coulomb field of the nucleus, while electron–electron interaction and coupling to the quantized electromagnetic field are treated perturbatively. For highly charged ions, the corresponding one-electron equation is

$$\left[-i \boldsymbol{\alpha}\cdot \nabla + \beta m + V_{\rm nuc}(\mathbf r)\right]\psi_n(\mathbf r)=\varepsilon_n\psi_n(\mathbf r).$$

The extended Furry picture modifies this starting point by replacing the nuclear potential with an effective potential,

$$V_{\rm eff}(\mathbf r)=V_{\rm nuc}(\mathbf r)+V_{\rm scr}(\mathbf r),$$

so that part of the interelectronic interaction is already incorporated into the zeroth-order Hamiltonian (Malyshev et al., 2014).

This reorganization is not a change of the underlying theory. A recurring misconception is to treat the extended Furry picture as an alternative dynamics rather than as a different splitting between unperturbed and interaction parts. In the atomic literature, the underlying QED interaction is unchanged; what changes is the choice of $H_0$, together with the addition of an explicit counterterm $-V_{\rm scr}$ in perturbation theory to avoid double counting. In strong-field QED, the same structural idea appears with a different physical meaning: a classical external field $A^\mu_{\rm ext}(x)$ is absorbed into the “free” Hamiltonian, and the quantized gauge field is treated perturbatively. For collider applications, the “extended” aspect can mean adapting the Furry picture to the full beam environment, including two non-collinear constant crossed fields or more realistic beam backgrounds (Porto et al., 2013).

The term also has a narrower mathematical usage in relativistic atomic theory. Brown–Ravenhall employ the free Dirac operator for the projector onto the electron space, whereas the Furry picture uses the Dirac operator with the nuclear Coulomb field. An extended Furry picture then amounts to defining the projector with a Dirac operator that contains not only the nuclear field but also a screening or self-consistent mean-field potential, such as a Dirac–Fock or Mittleman-type effective field (Merz et al., 2020).

2. Formal structure in bound-state QED

In many-electron atomic calculations, the effective one-electron Dirac Hamiltonian is written as

$$h^{\rm D}(\mathbf r)=\boldsymbol{\alpha}\cdot\mathbf p+\beta m+V_{\rm eff}(r), \qquad V_{\rm eff}(r)=V_{\rm nucl}(r)+V_{\rm scr}(r).$$

The complete basis of bound and continuum states used in perturbation theory is constructed from eigenstates of this operator, not from eigenstates of the pure Coulomb Hamiltonian. In Li-like hyperfine calculations, this choice partially incorporates the $1s^2$ core already at zeroth order, removes the quasidegeneracy of the $V_{\rm eff}=V_{\rm nuc}+V_{\rm scr}$0 and $V_{\rm eff}=V_{\rm nuc}+V_{\rm scr}$1 configurations already at zeroth order, and improves the zeroth-order radial wave functions of the valence electron (Kosheleva et al., 2020).

The Hamiltonian is then split as

$V_{\rm eff}=V_{\rm nuc}+V_{\rm scr}$2

or equivalently into a zeroth-order part built from $V_{\rm eff}=V_{\rm nuc}+V_{\rm scr}$3 and a residual interaction consisting of the full QED interaction plus the local counterterm $V_{\rm eff}=V_{\rm nuc}+V_{\rm scr}$4. Diagrammatically, electron lines are drawn as double or triple lines, indicating propagation in $V_{\rm eff}=V_{\rm nuc}+V_{\rm scr}$5, while counterterm vertices are denoted by a small circle with a cross, $V_{\rm eff}=V_{\rm nuc}+V_{\rm scr}$6. In the Be-like calculations of Yerokhin, Kozhedub, Shabaev, and collaborators, the explicit counterterm contributions appear already at first and second order, for example through matrix elements $V_{\rm eff}=V_{\rm nuc}+V_{\rm scr}$7, and the rigorous QED contributions up to second order are combined with higher-order electron correlation obtained in the Breit approximation by CI-DFS (Malyshev et al., 2014).

In the heliumlike $V_{\rm eff}=V_{\rm nuc}+V_{\rm scr}$8 transition-energy calculations, the same scheme is implemented with core-Hartree or local Dirac–Fock screening potentials. The zeroth-order one-electron problem is solved in $V_{\rm eff}=V_{\rm nuc}+V_{\rm scr}$9, first- and second-order interelectronic interactions are treated by the two-time Green’s function method, one-electron QED and screened QED corrections are evaluated in the effective field, and third- and higher-order correlation effects are included through the Dirac–Coulomb–Breit Hamiltonian (Malyshev et al., 2018).

For hyperfine structure in heavy neutral atoms and ions, the extended Furry representation is defined by

$V_{\rm scr}$0

Here $V_{\rm scr}$1 is chosen as a core-Hartree or Kohn–Sham potential, and the one-loop self-energy and vacuum-polarization corrections are evaluated rigorously with propagators and wave functions in that effective field. The same formal QED renormalization machinery applies as in the standard Furry picture; only the external potential in $V_{\rm scr}$2 has changed (Ginges et al., 2017).

3. Screening potentials, counterterms, and convergence

The choice of screening potential is central to the extended Furry picture because it determines how much of the many-electron interaction is built into the reference problem. Three local potentials are used in the Be-like ground-state energy calculations: the local Dirac–Fock (LDF) potential, the Kohn–Sham (KS) potential, and the Perdew–Zunger (PZ) potential. All are local and spherically symmetric and are constructed to represent the average field of the electrons in the $V_{\rm scr}$3 ground state (Malyshev et al., 2014).

For the Kohn–Sham construction, the total radial density is

$V_{\rm scr}$4

and the screening potential is

$V_{\rm scr}$5

supplemented by a Latter correction to improve the asymptotic behavior at large $V_{\rm scr}$6 (Malyshev et al., 2014). In hyperfine-structure work on Rb, Cs, Fr, Ba$V_{\rm scr}$7, and Ra$V_{\rm scr}$8, the screening potentials are instead the core-Hartree potential

$V_{\rm scr}$9

and the Kohn–Sham potential built from the total density $$\left[-i \boldsymbol{\alpha}\cdot \nabla + \beta m + V_{\rm nuc}(\mathbf r)\right]\psi_n(\mathbf r)=\varepsilon_n\psi_n(\mathbf r).$$0, again with a Latter correction to enforce the large-$$\left[-i \boldsymbol{\alpha}\cdot \nabla + \beta m + V_{\rm nuc}(\mathbf r)\right]\psi_n(\mathbf r)=\varepsilon_n\psi_n(\mathbf r).$$1 asymptotic form $$\left[-i \boldsymbol{\alpha}\cdot \nabla + \beta m + V_{\rm nuc}(\mathbf r)\right]\psi_n(\mathbf r)=\varepsilon_n\psi_n(\mathbf r).$$2 (Ginges et al., 2017).

The practical rationale is convergence. By placing a substantial mean-field contribution into $$\left[-i \boldsymbol{\alpha}\cdot \nabla + \beta m + V_{\rm nuc}(\mathbf r)\right]\psi_n(\mathbf r)=\varepsilon_n\psi_n(\mathbf r).$$3, first- and second-order interelectronic corrections are reduced, higher-order terms are smaller, and quasidegeneracies present in the pure Coulomb starting point can be removed. The Be-like calculations make this cancellation explicit: for Be-like Ca,

$$\left[-i \boldsymbol{\alpha}\cdot \nabla + \beta m + V_{\rm nuc}(\mathbf r)\right]\psi_n(\mathbf r)=\varepsilon_n\psi_n(\mathbf r).$$4

for LDF, KS, and PZ, respectively, while for Be-like U the final total binding energies are

$$\left[-i \boldsymbol{\alpha}\cdot \nabla + \beta m + V_{\rm nuc}(\mathbf r)\right]\psi_n(\mathbf r)=\varepsilon_n\psi_n(\mathbf r).$$5

again for LDF, KS, and PZ. Individual terms vary strongly with the choice of $$\left[-i \boldsymbol{\alpha}\cdot \nabla + \beta m + V_{\rm nuc}(\mathbf r)\right]\psi_n(\mathbf r)=\varepsilon_n\psi_n(\mathbf r).$$6, but the physical sum is stable once counterterms and radiative corrections are included (Malyshev et al., 2014).

The same logic underlies the use of several starting potentials as an uncertainty diagnostic. In the hyperfine calculations for heavy atoms, CH and KS zeroth-order intervals differ by a few percent, while the relative QED corrections differ by up to about $$\left[-i \boldsymbol{\alpha}\cdot \nabla + \beta m + V_{\rm nuc}(\mathbf r)\right]\psi_n(\mathbf r)=\varepsilon_n\psi_n(\mathbf r).$$7; that spread is used as one component of the uncertainty estimate when rescaling QED corrections onto many-body results. The authors also verify that the relative Bohr–Weisskopf correction is almost insensitive to the choice of atomic potential, with $$\left[-i \boldsymbol{\alpha}\cdot \nabla + \beta m + V_{\rm nuc}(\mathbf r)\right]\psi_n(\mathbf r)=\varepsilon_n\psi_n(\mathbf r).$$8 to within $$\left[-i \boldsymbol{\alpha}\cdot \nabla + \beta m + V_{\rm nuc}(\mathbf r)\right]\psi_n(\mathbf r)=\varepsilon_n\psi_n(\mathbf r).$$9 for all considered systems (Ginges et al., 2017).

4. Principal applications in atomic structure and spectroscopy

The extended Furry picture has become a standard computational framework for precision bound-state QED in highly charged ions. In the Be-like ground-state energy calculation, rigorous QED contributions up to second order of perturbation theory are combined with third- and higher-order electron-correlation contributions obtained within the Breit approximation by the large-scale configuration-interaction Dirac–Fock–Sturm method, and nuclear recoil and nuclear polarization are included. The calculation covers $$V_{\rm eff}(\mathbf r)=V_{\rm nuc}(\mathbf r)+V_{\rm scr}(\mathbf r),$$0–96, and ionization potentials are obtained by subtracting the binding energies of the corresponding lithiumlike ions (Malyshev et al., 2014).

For Li-like hyperfine splitting, the same framework is used to compute rigorous one- and two-photon exchange corrections over the range $$V_{\rm eff}(\mathbf r)=V_{\rm nuc}(\mathbf r)+V_{\rm scr}(\mathbf r),$$1–82, with third and higher orders included in the Breit approximation by recursive perturbation theory. The comparison between Coulomb, core-Hartree, and Kohn–Sham starting potentials shows that all three converge to essentially the same total hyperfine factor $$V_{\rm eff}(\mathbf r)=V_{\rm nuc}(\mathbf r)+V_{\rm scr}(\mathbf r),$$2, but the uncertainty is markedly smaller for the extended-Furry starting points, especially at low $$V_{\rm eff}(\mathbf r)=V_{\rm nuc}(\mathbf r)+V_{\rm scr}(\mathbf r),$$3. For $$V_{\rm eff}(\mathbf r)=V_{\rm nuc}(\mathbf r)+V_{\rm scr}(\mathbf r),$$4, for example, the total values are $$V_{\rm eff}(\mathbf r)=V_{\rm nuc}(\mathbf r)+V_{\rm scr}(\mathbf r),$$5, $$V_{\rm eff}(\mathbf r)=V_{\rm nuc}(\mathbf r)+V_{\rm scr}(\mathbf r),$$6, and $$V_{\rm eff}(\mathbf r)=V_{\rm nuc}(\mathbf r)+V_{\rm scr}(\mathbf r),$$7 for Coulomb, core-Hartree, and Kohn–Sham, respectively (Kosheleva et al., 2020).

In heavy neutral atoms and alkali-metal-like ions, rigorous one-loop QED corrections to ground-state hyperfine intervals are evaluated in the extended Furry picture with CH and KS potentials, while the full many-body description is completed with the all-orders correlation potential method. The systems $$V_{\rm eff}(\mathbf r)=V_{\rm nuc}(\mathbf r)+V_{\rm scr}(\mathbf r),$$8, $$V_{\rm eff}(\mathbf r)=V_{\rm nuc}(\mathbf r)+V_{\rm scr}(\mathbf r),$$9, $H_0$0, $H_0$1, and $H_0$2 are singled out because of their relevance for parity-violation studies. In this setting, the extended Furry picture is essential because the pure Coulomb starting point is not appropriate for neutral heavy atoms, whose valence states are strongly screened by the core (Ginges et al., 2017).

For heliumlike ions, the approach is applied to the $H_0$3, $H_0$4, $H_0$5, and $H_0$6 x-ray transitions in Ar, Ti, Fe, Cu, and Kr. The calculation combines rigorous QED treatment of first- and second-order effects in the extended Furry picture with higher-order correlation in the Breit approximation, and it treats the quasidegenerate $H_0$7 and $H_0$8 levels through a $H_0$9 effective Hamiltonian. The residual dependence on the choice of effective potential is used, together with estimates of uncalculated two-loop and higher-order screening QED terms, to assign uncertainties at the $-V_{\rm scr}$0–$-V_{\rm scr}$1 eV level for the transition energies (Malyshev et al., 2018).

Across these applications, the extended Furry picture is best understood as a bridge between ab initio QED and many-body methods. Rigorous QED is retained where it is presently tractable—typically first and second order in interelectronic interaction and one-electron radiative corrections—while the reorganized zeroth-order problem and Breit-based many-body treatments absorb much of the remaining correlation.

5. Strong-field, collider, and momentum-space generalizations

In strong-field QED, the Furry picture starts from the Lagrangian

$-V_{\rm scr}$2

so the classical background $-V_{\rm scr}$3 is treated exactly and the quantized photon field $-V_{\rm scr}$4 remains perturbative. The exact fermion states are Volkov solutions, and the corresponding propagators and vertices are background-dressed. At future linear colliders, where the intense beam–beam electromagnetic field can be approximated by a constant crossed field, the same formalism is used to analyze beamstrahlung, coherent pair production, and possible corrections to precision Standard Model processes. Typical quoted values are $-V_{\rm scr}$5 for ILC-1 TeV and $-V_{\rm scr}$6 for CLIC-3 TeV (Porto et al., 2013).

A more recent development reformulates strong-field QED in momentum space. For a pulsed univariate background $-V_{\rm scr}$7, with $-V_{\rm scr}$8, the dressed fermion–fermion–photon vertex is written as

$-V_{\rm scr}$9

with the entire background dependence encoded in the phase integrals $A^\mu_{\rm ext}(x)$0, $A^\mu_{\rm ext}(x)$1, and $A^\mu_{\rm ext}(x)$2. The key Ward-identity relation is

$A^\mu_{\rm ext}(x)$3

and the momentum-space formulation produces the required gauge-restoration term systematically, in contrast to the usual position-space Furry picture, which must be equipped with non-obvious terms to ensure the Ward identity. In this representation, free spinors, the free photon propagator, and the free fermion propagator are used; the external field appears solely in the vertex function (Acosta et al., 2023).

The term “extended Furry picture” also appears in analytic strong-field work on higher-order Volkov processes. Fierz transformations of Volkov spinors provide a systematic simplification of traces in first- and second-order Furry-picture amplitudes such as high-intensity Compton scattering and stimulated Compton scattering. The motivation is explicitly tied to higher-order, resonant processes, in which Volkov propagators can go on shell in the background and produce the resonance behavior of higher-order Furry-picture processes (Hartin, 2016).

An even more direct generalization is the perturbative treatment of a weak time-dependent field on top of a non-perturbative static background. In the study of dynamically assisted Klein tunneling, the static step potential

$A^\mu_{\rm ext}(x)$4

is taken as the Furry background, while the localized oscillating electric field

$A^\mu_{\rm ext}(x)$5

is treated perturbatively. The retarded Green function is built from the exact scattering states of the step potential, and reflection and transmission probabilities are computed up to second order. A central result is that, even for a subcritical step, a positive-frequency incoming wave can penetrate the negative-frequency region by emitting energy into the oscillating field, with the dynamically assisted Klein condition

$A^\mu_{\rm ext}(x)$6

This is a one-particle realization of the same general principle: non-perturbative background structure handled exactly, weaker time-dependent forcing treated in perturbation theory around that background (Ochiai et al., 14 Apr 2025).

6. Mathematical heavy-atom perspective and asymptotic implications

In the mathematical theory of relativistic heavy atoms, the Furry picture has a sharply defined no-pair meaning. The one-particle reference operator is the Coulomb–Dirac operator with the nuclear field,

$A^\mu_{\rm ext}(x)$7

and the electron Hilbert space is the positive spectral subspace

$A^\mu_{\rm ext}(x)$8

The many-electron no-pair Hamiltonian is then built from $A^\mu_{\rm ext}(x)$9 and the Coulomb repulsion on the antisymmetric tensor product of $$h^{\rm D}(\mathbf r)=\boldsymbol{\alpha}\cdot\mathbf p+\beta m+V_{\rm eff}(r), \qquad V_{\rm eff}(r)=V_{\rm nucl}(r)+V_{\rm scr}(r).$$0 copies of $$h^{\rm D}(\mathbf r)=\boldsymbol{\alpha}\cdot\mathbf p+\beta m+V_{\rm eff}(r), \qquad V_{\rm eff}(r)=V_{\rm nucl}(r)+V_{\rm scr}(r).$$1. In this framework, the strong Scott conjecture is proved: for $$h^{\rm D}(\mathbf r)=\boldsymbol{\alpha}\cdot\mathbf p+\beta m+V_{\rm eff}(r), \qquad V_{\rm eff}(r)=V_{\rm nucl}(r)+V_{\rm scr}(r).$$2 with $$h^{\rm D}(\mathbf r)=\boldsymbol{\alpha}\cdot\mathbf p+\beta m+V_{\rm eff}(r), \qquad V_{\rm eff}(r)=V_{\rm nucl}(r)+V_{\rm scr}(r).$$3 fixed, the rescaled density $$h^{\rm D}(\mathbf r)=\boldsymbol{\alpha}\cdot\mathbf p+\beta m+V_{\rm eff}(r), \qquad V_{\rm eff}(r)=V_{\rm nucl}(r)+V_{\rm scr}(r).$$4 converges weakly on the Scott scale to the density $$h^{\rm D}(\mathbf r)=\boldsymbol{\alpha}\cdot\mathbf p+\beta m+V_{\rm eff}(r), \qquad V_{\rm eff}(r)=V_{\rm nucl}(r)+V_{\rm scr}(r).$$5 of the relativistic Bohr atom (Merz et al., 2020).

The same work gives a precise mathematical interpretation of what an extended Furry picture would mean in this context. One still defines a no-pair Hamiltonian on a positive spectral subspace of a Dirac operator, but the Dirac operator includes not only the nuclear Coulomb field but also a screening or self-consistent mean-field potential, such as a Dirac–Fock operator or Mittleman’s mean field. The paper states that a similar strong Scott result “should be also true when the mean field in the sense of Mittleman [41] is taken into account,” although this is so far known only in Hartree–Fock approximation when the involved projection is given by the Dirac–Fock operator, citing Fournais et al. 14.

This suggests a general structural constraint on asymptotically correct extensions. Near the nucleus, on the Scott scale $$h^{\rm D}(\mathbf r)=\boldsymbol{\alpha}\cdot\mathbf p+\beta m+V_{\rm eff}(r), \qquad V_{\rm eff}(r)=V_{\rm nucl}(r)+V_{\rm scr}(r).$$6, the local behavior is governed by the unscreened Coulomb singularity, and the limiting density is the hydrogenic Bohr-atom density. A plausible implication is that any extended Furry picture intended to remain asymptotically correct for heavy atoms must preserve that local Coulomb-like structure, even if screening or self-consistent potentials modify the Dirac operator at larger radii. In that sense, the heavy-atom Scott-scale theory supplies a rigorous inner-core benchmark for extended Furry constructions: whatever the chosen effective projector, the near-nucleus asymptotics must remain compatible with the Dirac–Coulomb Bohr atom (Merz et al., 2020).