Quantum-Optical Strong-Field Approximation

• Quantum-optical strong-field approximation is a framework that quantizes the driving laser field, enabling calculation of photon-resolved observables beyond semiclassical models.
• It employs coherent-state and path-integral techniques to capture electron–photon entanglement, non-classical light effects, and detailed photon statistics.
• Benchmarking demonstrates that while traditional ATI spectra are recovered, joint electron–photon correlations and metrological enhancements in HHG are uniquely accessible.

Searching arXiv for recent and foundational work on the quantum-optical strong-field approximation and closely related formulations. Quantum-optical strong-field approximation denotes a class of extensions of the strong-field approximation in which the driving laser is treated as a quantized field rather than as a prescribed classical wave. Across the cited literature, the term encompasses coherent-state Schrödinger-picture constructions, path-integral formulations, and Heisenberg-picture perturbative expansions that retain photon statistics, field depletion, and electron–photon entanglement while preserving the Volkov-type logic of strong-field theory (Gonoskov et al., 2016, Varró, 2024, Mao et al., 17 Dec 2025, Lange et al., 16 Dec 2025). In contrast to semiclassical SFA, where observables are built from electron operators in an external field, the quantum-optical formulation promotes the laser mode or modes to dynamical quantum degrees of freedom and thereby makes photon-resolved spectra, squeezing, and joint electron–photon correlations accessible.

1. Definition and relation to conventional strong-field theory

The conventional SFA starts from a decomposition

$H(t)=H_0+V_c+V_d(t),$

with the central assumptions that the initial bound state remains unperturbed until ionization and that, after ionization, the Coulomb potential is neglected so that the continuum electron is described by a Volkov solution. In this setting the lowest-order transition amplitude for a final momentum $\mathbf p$ is

$$M^{(1)}(\mathbf p) =-i\int_{-\infty}^{\infty}dt\, \langle\chi_g(t;\mathbf p)|V_d(t)|\phi_0 e^{-iI_p t}\rangle,$$

or, equivalently, in the usual quasiclassical form with action

$$S(\mathbf p,t)=\frac12\int^t(\mathbf p+\mathbf A(\tau))^2\,d\tau+I_p t$$

(Popov et al., 2016).

Quantum-optical versions keep the atomic binding problem in the strong-field spirit but quantize the radiation from the outset. One representative single-mode Hamiltonian is

$$H=\frac12[p-\hat A]^2+\omega\left(\hat a^\dagger\hat a+\frac12\right),$$

with

$$\hat A(t)=-\frac{\beta}{\sqrt2}\left(\hat a e^{-i\omega t}+\hat a^\dagger e^{i\omega t}\right),$$

while a path-integral ionization benchmark employs

$$H_A=-\frac12\frac{\partial^2}{\partial x^2}-\frac{1}{\sqrt{x^2+2}}, \qquad H_{\rm int}(t)=i f(t)\epsilon_V x(ae^{-i\omega t}-a^\dagger e^{i\omega t})$$

in the interaction picture and dipole gauge (Gonoskov et al., 2016, Mao et al., 17 Dec 2025). The conceptual shift is that the field state is no longer an inert background: photon-number distributions, coherent-state amplitudes, and field quadratures become part of the dynamical output.

A recurring theme is that the quantum-optical extension is not merely a reformulation of standard SFA. It changes the observable content of the theory. Semiclassical SFA can reproduce electron-only quantities, but it does not, by construction, produce photon-number reshaping, squeezing, or electron–photon joint spectra. The quantized-field formulation is introduced precisely to make those observables calculable.

2. Schrödinger-picture constructions and the quantum-optical Volkov state

A foundational Schrödinger-picture realization was given through a closed-form solution of the time-dependent Schrödinger equation for a free electron interacting with a quantized coherent laser mode. For an initial state

$\Psi(p,q,0)=\psi_0(p)\psi_c(q),$

with coherent-state field wave function

$\psi_c(q)=\exp[\lambda q^2+g_0 q+c_0],$

the joint electron–field state takes the form

$$\Psi(p,q,t)=C_0\sqrt{M(t)}\,\psi_0(p)\, \exp\!\bigl[a(t)q^2+b(t)p^2+d(t)pq+f(t)p+g(t)q+c_0+c(t)\bigr].$$

This “quantum-optical Volkov” solution is explicitly entangled through the mixed $\mathbf p$0 term and remains quantum-optically nontrivial even in the high-photon-number limit, where simplified coefficients such as

$\mathbf p$1

are obtained (Gonoskov et al., 2016).

A key structural result is that, for any operator $\mathbf p$2 that is independent of the field quadrature $\mathbf p$3,

$\mathbf p$4

When the field is traced out, the electron-sector matrix elements reduce exactly to the usual semiclassical Volkov expressions. This explains why familiar SFA observables can be recovered even when the underlying theory is fully quantized: the semiclassical theory is embedded as the sector insensitive to the explicit photonic coordinate.

A more recent derivation starts from a coherent-state path integral. The full final state is written as

$\mathbf p$5

followed by insertion of coherent-state resolutions of identity and the Feynman path integral over electron phase-space variables. After integrating over photon midpoints one obtains a path-integral expression in which the coherent-state amplitude obeys a Maxwell-equation-like backaction law,

$\mathbf p$6

In the weak-backaction regime $\mathbf p$7, one sets $\mathbf p$8 and $\mathbf p$9, yielding the central ansatz

$$M^{(1)}(\mathbf p) =-i\int_{-\infty}^{\infty}dt\, \langle\chi_g(t;\mathbf p)|V_d(t)|\phi_0 e^{-iI_p t}\rangle,$$0

where the electron evolves under the classical field associated with the coherent-state label $$M^{(1)}(\mathbf p) =-i\int_{-\infty}^{\infty}dt\, \langle\chi_g(t;\mathbf p)|V_d(t)|\phi_0 e^{-iI_p t}\rangle,$$1:

$$M^{(1)}(\mathbf p) =-i\int_{-\infty}^{\infty}dt\, \langle\chi_g(t;\mathbf p)|V_d(t)|\phi_0 e^{-iI_p t}\rangle,$$2

This formulation provides the benchmarked quantum-optical strong-field approximation used for strong ionization by bright squeezed vacuum (Mao et al., 17 Dec 2025).

3. Representations, approximations, and the role of entanglement

The coherent-state path-integral construction rests on a specific hierarchy of approximations. The photon field is expanded in a coherent-state basis; the only dynamical backaction on $$M^{(1)}(\mathbf p) =-i\int_{-\infty}^{\infty}dt\, \langle\chi_g(t;\mathbf p)|V_d(t)|\phi_0 e^{-iI_p t}\rangle,$$3 is of order $$M^{(1)}(\mathbf p) =-i\int_{-\infty}^{\infty}dt\, \langle\chi_g(t;\mathbf p)|V_d(t)|\phi_0 e^{-iI_p t}\rangle,$$4 and is dropped in the strong-pulse weak-backaction regime; different $$M^{(1)}(\mathbf p) =-i\int_{-\infty}^{\infty}dt\, \langle\chi_g(t;\mathbf p)|V_d(t)|\phi_0 e^{-iI_p t}\rangle,$$5 trajectories do not couple directly except through the initial overlap $$M^{(1)}(\mathbf p) =-i\int_{-\infty}^{\infty}dt\, \langle\chi_g(t;\mathbf p)|V_d(t)|\phi_0 e^{-iI_p t}\rangle,$$6; and multi-mode photon correlations beyond the single driving mode are neglected. These assumptions define the operational content of the benchmarked QSFA and distinguish it from diagonal phase-space descriptions (Mao et al., 17 Dec 2025).

The distinction between the diagonal and non-diagonal coherent-state representations is central. In the standard $$M^{(1)}(\mathbf p) =-i\int_{-\infty}^{\infty}dt\, \langle\chi_g(t;\mathbf p)|V_d(t)|\phi_0 e^{-iI_p t}\rangle,$$7-representation, one retains only the diagonal Husimi weight

$$M^{(1)}(\mathbf p) =-i\int_{-\infty}^{\infty}dt\, \langle\chi_g(t;\mathbf p)|V_d(t)|\phi_0 e^{-iI_p t}\rangle,$$8

so that

$$M^{(1)}(\mathbf p) =-i\int_{-\infty}^{\infty}dt\, \langle\chi_g(t;\mathbf p)|V_d(t)|\phi_0 e^{-iI_p t}\rangle,$$9

All off-diagonal coherent-state overlaps $$S(\mathbf p,t)=\frac12\int^t(\mathbf p+\mathbf A(\tau))^2\,d\tau+I_p t$$0 with $$S(\mathbf p,t)=\frac12\int^t(\mathbf p+\mathbf A(\tau))^2\,d\tau+I_p t$$1 are discarded, and the photon statistics do not evolve in time under $$S(\mathbf p,t)=\frac12\int^t(\mathbf p+\mathbf A(\tau))^2\,d\tau+I_p t$$2. By contrast, the corresponding $$S(\mathbf p,t)=\frac12\int^t(\mathbf p+\mathbf A(\tau))^2\,d\tau+I_p t$$3-representation keeps the non-diagonal kernel

$$S(\mathbf p,t)=\frac12\int^t(\mathbf p+\mathbf A(\tau))^2\,d\tau+I_p t$$4

with

$$S(\mathbf p,t)=\frac12\int^t(\mathbf p+\mathbf A(\tau))^2\,d\tau+I_p t$$5

Those off-diagonal terms encode the electron–photon entanglement responsible for parity- and coherence-dependent modulations in photon-resolved observables (Mao et al., 17 Dec 2025).

A mathematically related but more drastic approximation is the approximative phase-space description. There one starts from the positive-$$S(\mathbf p,t)=\frac12\int^t(\mathbf p+\mathbf A(\tau))^2\,d\tau+I_p t$$6 representation

$$S(\mathbf p,t)=\frac12\int^t(\mathbf p+\mathbf A(\tau))^2\,d\tau+I_p t$$7

uses the exact relation

$$S(\mathbf p,t)=\frac12\int^t(\mathbf p+\mathbf A(\tau))^2\,d\tau+I_p t$$8

and then approximates the Gaussian in the relative coordinate by a delta function,

$$S(\mathbf p,t)=\frac12\int^t(\mathbf p+\mathbf A(\tau))^2\,d\tau+I_p t$$9

The result is

$$H=\frac12[p-\hat A]^2+\omega\left(\hat a^\dagger\hat a+\frac12\right),$$0

which replaces the initial laser state by the classical stochastic mixture

$$H=\frac12[p-\hat A]^2+\omega\left(\hat a^\dagger\hat a+\frac12\right),$$1

Because $$H=\frac12[p-\hat A]^2+\omega\left(\hat a^\dagger\hat a+\frac12\right),$$2 is positive, this construction cannot produce sub-Poissonian output statistics or quadrature squeezing below vacuum; specifically, it enforces $$H=\frac12[p-\hat A]^2+\omega\left(\hat a^\dagger\hat a+\frac12\right),$$3 and $$H=\frac12[p-\hat A]^2+\omega\left(\hat a^\dagger\hat a+\frac12\right),$$4 (Gothelf et al., 4 Feb 2026).

4. Joint electron–photon observables and numerical benchmarking

The distinctive observable of the quantum-optical theory is the joint electron–photon distribution. In the path-integral QSFA, the photoelectron–photon energy spectrum is obtained by projection onto electron energy $$H=\frac12[p-\hat A]^2+\omega\left(\hat a^\dagger\hat a+\frac12\right),$$5 and photon number $$H=\frac12[p-\hat A]^2+\omega\left(\hat a^\dagger\hat a+\frac12\right),$$6,

$$H=\frac12[p-\hat A]^2+\omega\left(\hat a^\dagger\hat a+\frac12\right),$$7

with momentum-resolved analogues obtained by replacing $$H=\frac12[p-\hat A]^2+\omega\left(\hat a^\dagger\hat a+\frac12\right),$$8 with $$H=\frac12[p-\hat A]^2+\omega\left(\hat a^\dagger\hat a+\frac12\right),$$9. The total ATI spectrum,

$$\hat A(t)=-\frac{\beta}{\sqrt2}\left(\hat a e^{-i\omega t}+\hat a^\dagger e^{i\omega t}\right),$$0

is recovered by tracing over photons, but the trace operation erases precisely the structure that distinguishes the quantum-optical theory from the diagonal approximations (Mao et al., 17 Dec 2025).

Benchmarking against the fully quantized time-dependent Schrödinger equation for atomic ionization by bright squeezed vacuum reveals a sharp separation between traced and joint observables. The photon-number distribution

$$\hat A(t)=-\frac{\beta}{\sqrt2}\left(\hat a e^{-i\omega t}+\hat a^\dagger e^{i\omega t}\right),$$1

is reproduced by the $$\hat A(t)=-\frac{\beta}{\sqrt2}\left(\hat a e^{-i\omega t}+\hat a^\dagger e^{i\omega t}\right),$$2-representation formula

$$\hat A(t)=-\frac{\beta}{\sqrt2}\left(\hat a e^{-i\omega t}+\hat a^\dagger e^{i\omega t}\right),$$3

to sub-percent accuracy across all $$\hat A(t)=-\frac{\beta}{\sqrt2}\left(\hat a e^{-i\omega t}+\hat a^\dagger e^{i\omega t}\right),$$4, whereas the $$\hat A(t)=-\frac{\beta}{\sqrt2}\left(\hat a e^{-i\omega t}+\hat a^\dagger e^{i\omega t}\right),$$5-representation gives

$$\hat A(t)=-\frac{\beta}{\sqrt2}\left(\hat a e^{-i\omega t}+\hat a^\dagger e^{i\omega t}\right),$$6

which is time-independent and remains equal to the incident bright-squeezed-vacuum distribution. As a result, the diagonal approximation misses even–odd parity mixing and local modulations present in the full solution. At the same time, the total ATI spectrum is nearly identical under the full TDSE, QSFA in the $$\hat A(t)=-\frac{\beta}{\sqrt2}\left(\hat a e^{-i\omega t}+\hat a^\dagger e^{i\omega t}\right),$$7-representation, and the $$\hat A(t)=-\frac{\beta}{\sqrt2}\left(\hat a e^{-i\omega t}+\hat a^\dagger e^{i\omega t}\right),$$8-representation, even though the joint spectra differ dramatically (Mao et al., 17 Dec 2025).

The numerical regime of validity is correspondingly precise. The benchmarked QSFA is accurate whenever $$\hat A(t)=-\frac{\beta}{\sqrt2}\left(\hat a e^{-i\omega t}+\hat a^\dagger e^{i\omega t}\right),$$9, even at intensities up to $$H_A=-\frac12\frac{\partial^2}{\partial x^2}-\frac{1}{\sqrt{x^2+2}}, \qquad H_{\rm int}(t)=i f(t)\epsilon_V x(ae^{-i\omega t}-a^\dagger e^{i\omega t})$$0 and average photon number $$H_A=-\frac12\frac{\partial^2}{\partial x^2}-\frac{1}{\sqrt{x^2+2}}, \qquad H_{\rm int}(t)=i f(t)\epsilon_V x(ae^{-i\omega t}-a^\dagger e^{i\omega t})$$1–$$H_A=-\frac12\frac{\partial^2}{\partial x^2}-\frac{1}{\sqrt{x^2+2}}, \qquad H_{\rm int}(t)=i f(t)\epsilon_V x(ae^{-i\omega t}-a^\dagger e^{i\omega t})$$2. The $$H_A=-\frac12\frac{\partial^2}{\partial x^2}-\frac{1}{\sqrt{x^2+2}}, \qquad H_{\rm int}(t)=i f(t)\epsilon_V x(ae^{-i\omega t}-a^\dagger e^{i\omega t})$$3-representation succeeds only for photon-traced electron observables and fails for photon-resolved or entanglement-sensitive quantities. This directly corrects a common overinterpretation: agreement on electron-only spectra does not validate a phase-space approximation for photon statistics or joint observables.

The same caution appears in the HHG-focused approximative phase-space description. There the error in spectral observables is of order $$H_A=-\frac12\frac{\partial^2}{\partial x^2}-\frac{1}{\sqrt{x^2+2}}, \qquad H_{\rm int}(t)=i f(t)\epsilon_V x(ae^{-i\omega t}-a^\dagger e^{i\omega t})$$4 and is usually negligible for strong classical-like drives, but the quadrature-variance error

$$H_A=-\frac12\frac{\partial^2}{\partial x^2}-\frac{1}{\sqrt{x^2+2}}, \qquad H_{\rm int}(t)=i f(t)\epsilon_V x(ae^{-i\omega t}-a^\dagger e^{i\omega t})$$5

accumulates with emitter number or pulse duration, with numerical scaling $$H_A=-\frac12\frac{\partial^2}{\partial x^2}-\frac{1}{\sqrt{x^2+2}}, \qquad H_{\rm int}(t)=i f(t)\epsilon_V x(ae^{-i\omega t}-a^\dagger e^{i\omega t})$$6 and $$H_A=-\frac12\frac{\partial^2}{\partial x^2}-\frac{1}{\sqrt{x^2+2}}, \qquad H_{\rm int}(t)=i f(t)\epsilon_V x(ae^{-i\omega t}-a^\dagger e^{i\omega t})$$7 (Gothelf et al., 4 Feb 2026).

5. High-harmonic generation and quantum-optical emission theory

In high-order harmonic generation, the quantum-optical SFA has been used to connect the strong-field electron dynamics to measurable photon statistics of the transmitted or emitted radiation. A closed-form quantum-optical Volkov solution allows the joint wave function to be expanded in Fock states,

$$H_A=-\frac12\frac{\partial^2}{\partial x^2}-\frac{1}{\sqrt{x^2+2}}, \qquad H_{\rm int}(t)=i f(t)\epsilon_V x(ae^{-i\omega t}-a^\dagger e^{i\omega t})$$8

so that the infrared photon-number probability is

$$H_A=-\frac12\frac{\partial^2}{\partial x^2}-\frac{1}{\sqrt{x^2+2}}, \qquad H_{\rm int}(t)=i f(t)\epsilon_V x(ae^{-i\omega t}-a^\dagger e^{i\omega t})$$9

Without interaction, a coherent input gives the time-independent Poisson law

$\Psi(p,q,0)=\psi_0(p)\psi_c(q),$0

whereas under strong-field interaction each recollision path contributes an amplitude

$\Psi(p,q,0)=\psi_0(p)\psi_c(q),$1

In the high-$\Psi(p,q,0)=\psi_0(p)\psi_c(q),$2 limit this yields

$\Psi(p,q,0)=\psi_0(p)\psi_c(q),$3

For HHG by $\Psi(p,q,0)=\psi_0(p)\psi_c(q),$4 atoms emitting the $\Psi(p,q,0)=\psi_0(p)\psi_c(q),$5th harmonic, energy conservation gives

$\Psi(p,q,0)=\psi_0(p)\psi_c(q),$6

and the reported resolving power is

$\Psi(p,q,0)=\psi_0(p)\psi_c(q),$7

for XUV light around $\Psi(p,q,0)=\psi_0(p)\psi_c(q),$8 (Gonoskov et al., 2016).

A distinct but related formulation derives a quantum-optical strong-field Kramers–Heisenberg formula within nonrelativistic QED. The radiation field is split into strong laser modes treated nonperturbatively and weak harmonic modes treated perturbatively; unitary squeezing and displacement transformations remove the explicit $\Psi(p,q,0)=\psi_0(p)\psi_c(q),$9 and $\psi_c(q)=\exp[\lambda q^2+g_0 q+c_0],$0 terms for the strong modes and introduce the shifted potential

$\psi_c(q)=\exp[\lambda q^2+g_0 q+c_0],$1

For coherent laser input, coherent-state sums over displacement-operator matrix elements generate exact Bessel-function identities, recovering the familiar Jacobi–Anger structure in the large-$\psi_c(q)=\exp[\lambda q^2+g_0 q+c_0],$2 limit. The second iteration of the interaction-picture integral equation yields a HHG amplitude

$\psi_c(q)=\exp[\lambda q^2+g_0 q+c_0],$3

with resonant and non-resonant terms $\psi_c(q)=\exp[\lambda q^2+g_0 q+c_0],$4 and $\psi_c(q)=\exp[\lambda q^2+g_0 q+c_0],$5 built from intermediate bound and continuum states. In the classical limit of large coherent amplitude, negligible depletion, and sharply peaked Poissonian statistics, the standard semiclassical SFA expression is recovered (Varró, 2024).

These HHG formulations show that the quantum-optical theory does not only rephrase the three-step model. It relocates strong-field observables into photon counting, quadrature statistics, and coherent-state trajectory sums, thereby turning the emitted or transmitted light itself into the carrier of attosecond-scale information.

6. Heisenberg-picture developments, metrological consequences, and adjoining refinements

A recent Heisenberg-picture approach reformulates strong-field quantum optics as a controlled perturbation of the time-evolution operator after displacement of the large coherent laser amplitude. For $\psi_c(q)=\exp[\lambda q^2+g_0 q+c_0],$6 identical independent emitters, one writes

$\psi_c(q)=\exp[\lambda q^2+g_0 q+c_0],$7

with

$\psi_c(q)=\exp[\lambda q^2+g_0 q+c_0],$8

and residual coupling

$\psi_c(q)=\exp[\lambda q^2+g_0 q+c_0],$9

Truncating the Dyson expansion at second order in $$\Psi(p,q,t)=C_0\sqrt{M(t)}\,\psi_0(p)\, \exp\!\bigl[a(t)q^2+b(t)p^2+d(t)pq+f(t)p+g(t)q+c_0+c(t)\bigr].$$0 gives

$$\Psi(p,q,t)=C_0\sqrt{M(t)}\,\psi_0(p)\, \exp\!\bigl[a(t)q^2+b(t)p^2+d(t)pq+f(t)p+g(t)q+c_0+c(t)\bigr].$$1

and the driven emitter operator becomes

$$\Psi(p,q,t)=C_0\sqrt{M(t)}\,\psi_0(p)\, \exp\!\bigl[a(t)q^2+b(t)p^2+d(t)pq+f(t)p+g(t)q+c_0+c(t)\bigr].$$2

From this follow closed-form expressions for the HHG spectrum, quadrature squeezing, and photon statistics. The emitted spectrum contains a coherent contribution proportional to $$\Psi(p,q,t)=C_0\sqrt{M(t)}\,\psi_0(p)\, \exp\!\bigl[a(t)q^2+b(t)p^2+d(t)pq+f(t)p+g(t)q+c_0+c(t)\bigr].$$3 and an incoherent one proportional to $$\Psi(p,q,t)=C_0\sqrt{M(t)}\,\psi_0(p)\, \exp\!\bigl[a(t)q^2+b(t)p^2+d(t)pq+f(t)p+g(t)q+c_0+c(t)\bigr].$$4; in the classical limit one recovers

$$\Psi(p,q,t)=C_0\sqrt{M(t)}\,\psi_0(p)\, \exp\!\bigl[a(t)q^2+b(t)p^2+d(t)pq+f(t)p+g(t)q+c_0+c(t)\bigr].$$5

For large $$\Psi(p,q,t)=C_0\sqrt{M(t)}\,\psi_0(p)\, \exp\!\bigl[a(t)q^2+b(t)p^2+d(t)pq+f(t)p+g(t)q+c_0+c(t)\bigr].$$6, the degree of squeezing increases with emitter number, whereas $$\Psi(p,q,t)=C_0\sqrt{M(t)}\,\psi_0(p)\, \exp\!\bigl[a(t)q^2+b(t)p^2+d(t)pq+f(t)p+g(t)q+c_0+c(t)\bigr].$$7 because the leading $$\Psi(p,q,t)=C_0\sqrt{M(t)}\,\psi_0(p)\, \exp\!\bigl[a(t)q^2+b(t)p^2+d(t)pq+f(t)p+g(t)q+c_0+c(t)\bigr].$$8 terms dominate numerator and denominator equally (Lange et al., 16 Dec 2025).

The metrological implications of quantized strong fields have also been analyzed within a quantum-optical SFA for strong-field photoelectron holography driven by squeezed light. In the displaced-squeezed frame the electron action becomes operator-valued through

$$\Psi(p,q,t)=C_0\sqrt{M(t)}\,\psi_0(p)\, \exp\!\bigl[a(t)q^2+b(t)p^2+d(t)pq+f(t)p+g(t)q+c_0+c(t)\bigr].$$9

and, to first order in photon-number fluctuations,

$\mathbf p$00

The associated random phase $\mathbf p$01 produces the dephasing factor

$\mathbf p$02

Amplitude squeezing reduces $\mathbf p$03 and stabilizes the action; phase squeezing increases $\mathbf p$04 and causes a rapid collapse of interference visibility. The dephasing exponent scales quartically with wavelength,

$\mathbf p$05

and the resulting photoelectron distribution can be used to evaluate a Classical Fisher Information concentrated in a “dark-port” tunneling tail. On that basis an “Attosecond Quantum Tomography” protocol has been proposed as an in-situ, reference-free reconstruction of the driving field’s Wigner distribution (Khurelbaatar et al., 14 Feb 2026).

Alongside these photonic generalizations, several developments refine the electron side of SFA by improving Coulomb and rescattering physics. High-order Coulomb-corrected SFA replaces the continuum Volkov state by an eikonal Coulomb–Volkov wave function with phases $\mathbf p$06, yielding Coulomb-induced momentum shifts such as $\mathbf p$07 in the tunneling limit and $\mathbf p$08 in the nonadiabatic regime (Klaiber et al., 2016). A Faddeev-like reformulation alternates Coulomb and laser Green’s functions and was reported to converge rapidly toward TDSE results, unlike the divergent standard series in the binding potential (Popov et al., 2016). A velocity-gauge Volterra integro-differential approach incorporates initial-state depletion, discrete-state couplings, and continuum–continuum phases $\mathbf p$09, improving ATI spectra relative to Keldysh–Faisal–Reiss theory (Rodríguez, 2016). This suggests that the long-term development of quantum-optical SFA is likely to involve simultaneous control of photonic quantization, Coulomb corrections, and rescattering structure rather than a modification of only one sector.

In its most precise contemporary form, the quantum-optical strong-field approximation is therefore a controlled reduction of the fully quantized light–matter problem, not a single fixed formula. Its defining criterion is the retention of the field as a quantum dynamical subsystem. When only photon-traced electron spectra are required, diagonal phase-space descriptions may suffice; when photon-resolved observables, squeezing, or joint electron–photon correlations are sought, off-diagonal coherent-state structure or fully quantized propagation becomes essential.