Bright Squeezed Vacuum (BSV)

• BSV is a macroscopic quantum state generated via high-gain parametric down-conversion, exhibiting zero mean field and strong intensity fluctuations.
• It displays superbunching and complex multimode behavior, with applications in femtosecond metrology, high-harmonic generation, and strong-field ionization.
• Advanced characterization techniques like single-shot spectral interferometry uncover its unique photon statistics and enable precision in quantum state engineering.

Bright squeezed vacuum (BSV) is the high-gain squeezed-vacuum output of unseeded parametric down-conversion or an unseeded optical parametric amplifier: a macroscopic quantum state with zero mean electric field, large mean photon number, and exceptionally strong quadrature and intensity fluctuations. In contrast to coherent radiation, BSV is vacuum-seeded and retains nonclassical pair correlations at high flux; in contrast to ordinary thermal light, degenerate single-mode BSV can exhibit superbunching with $g^{(2)}(0)\to 3$, while realistic bulk implementations are often multimode in angle and frequency (Kern et al., 22 Sep 2025, Pérez et al., 2014). Over the last decade, BSV has evolved from a subject of macroscopic quantum optics into an ultrafast-light platform relevant to femtosecond metrology, strong-field ionization, high-harmonic generation, nonlinear tunneling, and continuous-variable frequency-domain architectures (Rasputnyi et al., 2024, Liu et al., 8 Apr 2026).

1. Quantum-optical definition and statistical structure

In the single-mode description, squeezed vacuum is generated by the operator

$$S(r)=\exp\!\left[\tfrac{1}{2}\left(r e^{-i\theta} a^2-r e^{i\theta} a^{\dagger 2}\right)\right],$$

acting on $|0\rangle$, so that $|0,r\rangle=S(r)|0\rangle$. For this state the mean field vanishes, $\langle a\rangle=0$, the mean photon number is $\bar n=\sinh^2 r$, and the quadrature variances are reshaped as $\mathrm{Var}(X_\theta)=\tfrac14 e^{\pm 2r}$, with squeezing in one quadrature and antisqueezing in the conjugate quadrature (Kern et al., 22 Sep 2025). In the bright regime, $r$ is large and $\bar n\gg 1$, so individual shots can contain macroscopic photon numbers.

For traveling-wave optical parametric amplification, the two-mode signal–idler description is equally fundamental. A minimal Hamiltonian is

$$H=i\hbar\kappa\left(e^{i\phi}a_s^\dagger a_i^\dagger-e^{-i\phi}a_s a_i\right),$$

which generates Bogoliubov transformations and twin-beam correlations characteristic of high-gain spontaneous parametric down-conversion (Pérez et al., 2014). In bulk, the resulting field is intrinsically multimode in angle and frequency; Schmidt-mode or Bloch–Messiah decompositions make this explicit by reducing the joint amplitude to independent squeezed supermodes. The effective mode number is commonly written as

$$S(r)=\exp\!\left[\tfrac{1}{2}\left(r e^{-i\theta} a^2-r e^{i\theta} a^{\dagger 2}\right)\right],$$0

or, when gain reshapes the mode weights, through gain-dependent modal occupations (Pérez et al., 2014, Sharapova et al., 2018).

BSV differs sharply from both coherent and thermal radiation. A coherent state has Poissonian statistics, $$S(r)=\exp\!\left[\tfrac{1}{2}\left(r e^{-i\theta} a^2-r e^{i\theta} a^{\dagger 2}\right)\right],$$1, and nonzero mean field; a single thermal mode has $$S(r)=\exp\!\left[\tfrac{1}{2}\left(r e^{-i\theta} a^2-r e^{i\theta} a^{\dagger 2}\right)\right],$$2. Degenerate BSV can instead show superbunching, with $$S(r)=\exp\!\left[\tfrac{1}{2}\left(r e^{-i\theta} a^2-r e^{i\theta} a^{\dagger 2}\right)\right],$$3 at the spectral center, as observed near 1040 nm in femtosecond single-shot measurements (Kern et al., 22 Sep 2025). In bright single-mode squeezed vacuum, the photon-number variance is super-Poissonian and the Fano factor satisfies $$S(r)=\exp\!\left[\tfrac{1}{2}\left(r e^{-i\theta} a^2-r e^{i\theta} a^{\dagger 2}\right)\right],$$4 (Liu et al., 8 Apr 2026). For multimode detection, correlation functions are diluted toward unity or toward the thermal limit according to the number of collected modes (Pérez et al., 2014, Iskhakov et al., 2012).

A distinctive phase property appears in degenerate amplification. Although the mean field is zero, amplification can occur with phase difference $$S(r)=\exp\!\left[\tfrac{1}{2}\left(r e^{-i\theta} a^2-r e^{i\theta} a^{\dagger 2}\right)\right],$$5 or $$S(r)=\exp\!\left[\tfrac{1}{2}\left(r e^{-i\theta} a^2-r e^{i\theta} a^{\dagger 2}\right)\right],$$6 relative to the pump, producing a shot-to-shot $$S(r)=\exp\!\left[\tfrac{1}{2}\left(r e^{-i\theta} a^2-r e^{i\theta} a^{\dagger 2}\right)\right],$$7-phase ambiguity. In the time domain, the average electric field vanishes while fluctuations remain large and oscillate at twice the optical frequency (Kern et al., 22 Sep 2025). This point is central: BSV is not “field-free,” but rather a zero-mean, high-fluctuation quantum field.

2. Generation and mode engineering

BSV is typically generated by high-gain $$S(r)=\exp\!\left[\tfrac{1}{2}\left(r e^{-i\theta} a^2-r e^{i\theta} a^{\dagger 2}\right)\right],$$8 parametric down-conversion or optical parametric amplification in nonlinear crystals, and can also be produced in $$S(r)=\exp\!\left[\tfrac{1}{2}\left(r e^{-i\theta} a^2-r e^{i\theta} a^{\dagger 2}\right)\right],$$9 four-wave-mixing platforms (Sharapova et al., 2019, Tzur et al., 2023). The high-gain regime preserves twin-beam correlations while driving the photon number per mode to macroscopic values, making brightness, modal purity, and phase matching central design parameters.

A major line of work concerns spatial single-mode operation without lossy post-selection. A two-crystal traveling-wave OPA with a macroscopic gap self-selects the spatial mode that overlaps the pump in the second crystal. In a type-I BBO implementation pumped at 355 nm, varying the crystal separation produced nearly single-mode spatial BSV with transverse mode count $|0\rangle$0, while maintaining high brightness and sub-shot-noise signal–idler correlations with $|0\rangle$1 near the single-mode condition (Pérez et al., 2014). The same geometry explains why increasing the gap narrows the accepted angular spectrum: higher-order modes diffract out of pump overlap and are not reamplified.

Spectral and temporal mode engineering has been pursued most explicitly in SU(1,1) interferometers. In a dispersive two-stage configuration, only the chirped frequencies that temporally overlap the pump in the second crystal receive exponential reamplification. Experimentally this narrowed the BSV spectrum from $|0\rangle$2 to $|0\rangle$3 and reduced the effective number of frequency modes from approximately $|0\rangle$4 to $|0\rangle$5 (Lemieux et al., 2016). The same formalism predicts gain-dependent concentration into the leading Schmidt mode and, in nondegenerate overlap conditions, the emergence of two-color BSV with symmetric spectral peaks (Sharapova et al., 2018).

Frequency-domain engineering extends beyond narrowing. Using an accurate high-gain PDC model and nonlinear holography, frequency-bin couplings can be designed directly through the product of pump spectrum and phase-matching response. In periodically poled KTP with apodized poling and multi-Gaussian pumps, this allowed construction of two-dimensional lattice geometries in frequency space and explicit generation of a 4-mode square continuous-variable cluster state whose nullifier variances fall below vacuum noise at optimal phase for $|0\rangle$6 and $|0\rangle$7 (Hurvitz et al., 2023).

Ultrafast source engineering has also reached the femtosecond regime. A two-crystal type-I BBO source pumped at 515 nm produced single-spatial-mode femtosecond BSV centered at $|0\rangle$8 with average pulse energy $|0\rangle$9; among $|0,r\rangle=S(r)|0\rangle$0 recorded shots, $|0,r\rangle=S(r)|0\rangle$1 single-peak spectra were selected for shot-resolved temporal reconstruction (Kern et al., 22 Sep 2025). At longer wavelength, single-spatiotemporal-mode BSV centered at $|0,r\rangle=S(r)|0\rangle$2 has been generated with $|0,r\rangle=S(r)|0\rangle$3 duration and photon-number distributions extending from $|0,r\rangle=S(r)|0\rangle$4 to $|0,r\rangle=S(r)|0\rangle$5 photons per pulse (Rasputnyi et al., 2024).

3. State characterization and ultrafast field retrieval

BSV characterization spans photon statistics, modal decompositions, quadrature measurements, and, more recently, shot-resolved ultrafast waveform retrieval. Standard pulse diagnostics such as FROG and SPIDER are optimized for coherent pulses with nonzero mean field; they can access averaged BSV properties but not the spectral phase of individual shots. Homodyne tomography probes squeezed quadratures, but is not suited to direct femtosecond field retrieval at high photon number (Kern et al., 22 Sep 2025).

A key advance is single-shot spectral interferometry with a coherent reference. For one shot, the measured interferogram is modeled as

$|0,r\rangle=S(r)|0\rangle$6

so the cross term encodes the spectral phase difference. In the 1040 nm femtosecond experiment, the reference pulse was fully characterized by FROG, Fourier-transform spectral interferometry isolated the sideband, and the BSV group delay $|0,r\rangle=S(r)|0\rangle$7 was retrieved shot by shot (Kern et al., 22 Sep 2025). The reconstructed single-peak shots showed an average pulse duration of $|0,r\rangle=S(r)|0\rangle$8 FWHM with $|0,r\rangle=S(r)|0\rangle$9 standard deviation, compared with a transform-limited duration of $\langle a\rangle=0$0. The group-delay curves were consistent near the central frequency, and the interferograms displayed fixed-frequency nodes arising from the intrinsic $\langle a\rangle=0$1-phase ambiguity. The two phase classes appeared in a $\langle a\rangle=0$2 ratio, corresponding to probability $\langle a\rangle=0$3 for one class (Kern et al., 22 Sep 2025).

Correlation-based diagnostics remain central for modal analysis. In high-gain BSV, spectral covariance reveals signal–idler structure and effective mode number; in the 1040 nm source, the covariance indicated approximately $\langle a\rangle=0$4 spectral modes (Kern et al., 22 Sep 2025). Earlier analog Hanbury Brown–Twiss measurements on high-gain degenerate PDC observed bunching and superbunching up to gains $\langle a\rangle=0$5, corresponding to $\langle a\rangle=0$6 photons per mode, with near-single-mode detection yielding the expected reduction from ideal $\langle a\rangle=0$7 values through $\langle a\rangle=0$8 and $\langle a\rangle=0$9 (Iskhakov et al., 2012).

Phase-space characterization has also moved into the bright, ultrafast regime. By applying a Kerr nonlinearity to single-mode BSV and sampling the output with a single-shot $\bar n=\sinh^2 r$0-to-$\bar n=\sinh^2 r$1 interferometer, the Husimi distribution was observed to transform from an approximately Gaussian profile to an “S”-shaped distribution, interpreted as direct statistical evidence of intensity-dependent nonlinear phase accumulation (Rasputnyi et al., 19 Dec 2025). In that experiment the BSV comprised $\bar n=\sinh^2 r$2 pulses at $\bar n=\sinh^2 r$3 with $\bar n=\sinh^2 r$4 photons per pulse, and the mapping from second-harmonic interferograms to $\bar n=\sinh^2 r$5 provided shot-by-shot Husimi sampling (Rasputnyi et al., 19 Dec 2025).

4. Strong-field, tunneling, and high-harmonic phenomena

BSV has become a driver for extreme nonlinear optics because it combines macroscopic pulse energy with zero mean field and broad intensity fluctuations. In solids, single-spatiotemporal-mode BSV at $\bar n=\sinh^2 r$6 and $\bar n=\sinh^2 r$7 generated nonperturbative high harmonics up to the 7th order. At matched mean intensity of approximately $\bar n=\sinh^2 r$8, BSV enhanced the 4th–7th harmonic yields in x-cut Mg:LiNbO$\bar n=\sinh^2 r$9 by factors between approximately $\mathrm{Var}(X_\theta)=\tfrac14 e^{\pm 2r}$0 and $\mathrm{Var}(X_\theta)=\tfrac14 e^{\pm 2r}$1 relative to coherent light, while the photon-number distribution extended to $\mathrm{Var}(X_\theta)=\tfrac14 e^{\pm 2r}$2 photons per pulse and sampled sub-cycle field excursions on the order of $\mathrm{Var}(X_\theta)=\tfrac14 e^{\pm 2r}$3 (Rasputnyi et al., 2024). The enhancement was traced to the convexity of nonlinear response under heavy-tailed intensity statistics and to access to a broader range of effective Keldysh parameters within a single acquisition.

In atomic strong-field ionization, BSV has been used as a direct tunnel-ionizing driver. Xenon ionization at 1600 nm was demonstrated with near-single-mode BSV of up to $\mathrm{Var}(X_\theta)=\tfrac14 e^{\pm 2r}$4 pulse energy, corresponding to average peak intensities from $\mathrm{Var}(X_\theta)=\tfrac14 e^{\pm 2r}$5 to $\mathrm{Var}(X_\theta)=\tfrac14 e^{\pm 2r}$6, while the measured $\mathrm{Var}(X_\theta)=\tfrac14 e^{\pm 2r}$7 values decreased from $\mathrm{Var}(X_\theta)=\tfrac14 e^{\pm 2r}$8 at $\mathrm{Var}(X_\theta)=\tfrac14 e^{\pm 2r}$9 to $r$0 at $r$1 (Liu et al., 8 Apr 2026). Compared with coherent-state driving, BSV suppressed ATI rings and carpet-like interference structures but selectively enhanced spider-like photoelectron holography. A quantum-light-corrected QTMC model attributed this to a coherence-filter mechanism: trajectories emitted within the same subcycle field fluctuation remain phase-correlated, whereas asynchronous pathways are dephased by field noise (Liu et al., 8 Apr 2026).

Quantum enhancement is even more explicit in nonlinear tunneling. In sodium, a $r$2 BSV beam produced the same tunneling ionization signature as a $r$3 coherent beam, implying a boost factor greater than $r$4 in effective nonlinear efficiency at fixed pulse energy (Jiang et al., 7 Apr 2026). The effect was controlled by tuning $r$5 from $r$6 to $r$7 at constant pulse energy, with the effective intensity scaling linearly as $r$8 (Jiang et al., 7 Apr 2026). The interpretation is fluctuation-driven: the anti-squeezed quadrature supplies rare large field amplitudes to which the ADK tunneling rate is exponentially sensitive.

At metal nanotips, BSV-driven photoemission revealed another distinctive feature: after post-selection on the photon number of each BSV pulse, the electron spectra displayed the familiar strong-field plateau and $r$9 cutoff, whereas spectra averaged over many BSV shots showed no plateau and only broad high-energy tails (Heimerl et al., 28 Mar 2025). For a mean BSV pulse energy of $\bar n\gg 1$0, $\bar n\gg 1$1 shots with average $\bar n\gg 1$2 detected electrons per pulse were recorded, and the cutoff energy increased linearly with the measured BSV photon number. From this scaling the local field enhancement at the tip apex was extracted as $\bar n\gg 1$3, with an inferred mean near-field intensity $\bar n\gg 1$4 (Heimerl et al., 28 Mar 2025). The same work modeled BSV-driven spectra as incoherent mixtures of coherent-drive spectra weighted by the BSV Husimi function.

BSV can also act as a weak, nonclassical control field in bichromatic strong-field configurations. Simulations with a strong coherent $\bar n\gg 1$5 field and a weak BSV $\bar n\gg 1$6 field found photoelectron asymmetries exceeding those obtained with classical fields of comparable intensity by orders of magnitude; the enhancement originated at the tunneling step, where anisotropic BSV fluctuations modify the instantaneous field amplitude, while leaving continuum dynamics essentially unchanged (Singh et al., 14 Apr 2026). In resonant multiphoton ionization, a fully quantum treatment predicted an anomalous Autler–Townes splitting whose magnitude grows with ATI order under BSV, with $\bar n\gg 1$7, unlike the ATI-order-independent coherent case (Zhang et al., 5 Jun 2026).

5. Metrology, entanglement, and quantum-state engineering

BSV is relevant to metrology not only because of its nonclassical statistics, but also because it can carry ultrafast structure inaccessible to weak entangled-photon sources. The shot-resolved $\bar n\gg 1$8 pulse retrieval at 1040 nm established that individual BSV realizations have consistent group delay and can therefore serve as femtosecond probes for attosecond sub-cycle metrology, even though the absolute phase is random up to a $\bar n\gg 1$9 flip (Kern et al., 22 Sep 2025). In a different regime, high-gain BSV used for two-photon spectroscopy demonstrated simultaneous high resolution in time and frequency, and in space and transverse wavevector. Measured correlation widths in the collinear degenerate configuration gave $$H=i\hbar\kappa\left(e^{i\phi}a_s^\dagger a_i^\dagger-e^{-i\phi}a_s a_i\right),$$0, $$H=i\hbar\kappa\left(e^{i\phi}a_s^\dagger a_i^\dagger-e^{-i\phi}a_s a_i\right),$$1, $$H=i\hbar\kappa\left(e^{i\phi}a_s^\dagger a_i^\dagger-e^{-i\phi}a_s a_i\right),$$2, and $$H=i\hbar\kappa\left(e^{i\phi}a_s^\dagger a_i^\dagger-e^{-i\phi}a_s a_i\right),$$3, with products $$H=i\hbar\kappa\left(e^{i\phi}a_s^\dagger a_i^\dagger-e^{-i\phi}a_s a_i\right),$$4 and $$H=i\hbar\kappa\left(e^{i\phi}a_s^\dagger a_i^\dagger-e^{-i\phi}a_s a_i\right),$$5, both far below $$H=i\hbar\kappa\left(e^{i\phi}a_s^\dagger a_i^\dagger-e^{-i\phi}a_s a_i\right),$$6, demonstrating violation of the classical Fourier or Mancini-type constraint (Cutipa et al., 2021).

At the same time, BSV does not provide a universal nonlinear-spectroscopy advantage. For nonresonant molecular two-photon absorption driven by broadband squeezed vacuum, the high-gain bright regime yields no rate advantage over a quasi-monochromatic coherent pulse of the same duration and mean photon number when the final-state linewidth is much narrower than the optical bandwidth. In the opposite broad-line regime, the TPA rate becomes proportional to $$H=i\hbar\kappa\left(e^{i\phi}a_s^\dagger a_i^\dagger-e^{-i\phi}a_s a_i\right),$$7, and with optimal dispersion compensation one obtains the idealized Type-0/I result $$H=i\hbar\kappa\left(e^{i\phi}a_s^\dagger a_i^\dagger-e^{-i\phi}a_s a_i\right),$$8, where $$H=i\hbar\kappa\left(e^{i\phi}a_s^\dagger a_i^\dagger-e^{-i\phi}a_s a_i\right),$$9 is the mean photon number per temporal mode (Raymer et al., 2022). This distinction has become a standard caution against treating “quantum enhancement” as bandwidth-independent.

Macroscopic entanglement in BSV has been analyzed most explicitly for polarization Bell states. For the four-mode macroscopic triplet and singlet states generated in high-gain PDC, feasible entanglement witnesses are given by Stokes-operator variance combinations such as

$$S(r)=\exp\!\left[\tfrac{1}{2}\left(r e^{-i\theta} a^2-r e^{i\theta} a^{\dagger 2}\right)\right],$$00

for separable states, with negative expectation values of the corresponding witness operators signaling entanglement (Iskhakov et al., 2012). For these states, the effective Schmidt number is

$$S(r)=\exp\!\left[\tfrac{1}{2}\left(r e^{-i\theta} a^2-r e^{i\theta} a^{\dagger 2}\right)\right],$$01

in the bright limit, the negativity is $$S(r)=\exp\!\left[\tfrac{1}{2}\left(r e^{-i\theta} a^2-r e^{i\theta} a^{\dagger 2}\right)\right],$$02, and the logarithmic negativity is $$S(r)=\exp\!\left[\tfrac{1}{2}\left(r e^{-i\theta} a^2-r e^{i\theta} a^{\dagger 2}\right)\right],$$03 (Iskhakov et al., 2012). The quadratic scaling with mean photon number formalizes the idea of macroscopic entanglement in BSV.

Frequency-domain continuous-variable processing pushes BSV toward quantum computing architectures. Nonlinear holography in PPKTP has been used to realize graph-structured frequency-bin couplings, including a square cluster-state adjacency matrix with edge weights $$S(r)=\exp\!\left[\tfrac{1}{2}\left(r e^{-i\theta} a^2-r e^{i\theta} a^{\dagger 2}\right)\right],$$04, and the resulting nullifier variances drop below vacuum at the optimized homodyne phase (Hurvitz et al., 2023). More speculatively, BSV has been proposed as a multiplexable route toward high-purity single-photon generation: by engineering Schmidt modes, distributing brightness across many frequency bins, and using photon-number-resolving detectors with switching, the delivery probability can scale as $$S(r)=\exp\!\left[\tfrac{1}{2}\left(r e^{-i\theta} a^2-r e^{i\theta} a^{\dagger 2}\right)\right],$$05, where $$S(r)=\exp\!\left[\tfrac{1}{2}\left(r e^{-i\theta} a^2-r e^{i\theta} a^{\dagger 2}\right)\right],$$06 is the number of usable bins (Yang et al., 5 Jan 2026).

Finally, BSV is a starting point for deterministic bright non-Gaussian optics. Kerr interaction in fused silica converts ultrafast single-mode BSV into a non-Gaussian state with an “S”-shaped Husimi distribution, and numerical analysis shows that phase-squeezed components of the resulting mixture can develop Wigner negativity that is considerably more loss-tolerant than the coherent-state case (Rasputnyi et al., 19 Dec 2025). This positions BSV as a bridge between Gaussian high-flux resources and genuinely non-Gaussian continuous-variable states.

6. Limitations, misconceptions, and propagation effects

A recurrent misconception is that because BSV has $$S(r)=\exp\!\left[\tfrac{1}{2}\left(r e^{-i\theta} a^2-r e^{i\theta} a^{\dagger 2}\right)\right],$$07, it cannot drive strong-field dynamics. Experiments and simulations now show the opposite: BSV can generate tunneling, rescattering, and attosecond signatures, but the relevant observables may require shot-resolved conditioning or phase-insensitive readout because the field is realized as a fluctuating ensemble rather than a deterministic carrier (Heimerl et al., 28 Mar 2025, Liu et al., 8 Apr 2026). What zero mean field precludes is not strong-field physics itself, but simple single-shot interpretation in terms of a fixed classical waveform. In degenerate femtosecond BSV, the intrinsic $$S(r)=\exp\!\left[\tfrac{1}{2}\left(r e^{-i\theta} a^2-r e^{i\theta} a^{\dagger 2}\right)\right],$$08 phase ambiguity additionally prevents deterministic carrier-envelope phase control across shots (Kern et al., 22 Sep 2025).

Modal purity and loss are the second major limitation. Many BSV signatures weaken rapidly under multimode averaging: $$S(r)=\exp\!\left[\tfrac{1}{2}\left(r e^{-i\theta} a^2-r e^{i\theta} a^{\dagger 2}\right)\right],$$09 drops toward $$S(r)=\exp\!\left[\tfrac{1}{2}\left(r e^{-i\theta} a^2-r e^{i\theta} a^{\dagger 2}\right)\right],$$10, HOM or homodyne mode matching becomes difficult, and post-selected statistics become sensitive to detection bandwidth and geometry (Pérez et al., 2014, Sharapova et al., 2018). In single-shot temporal retrieval, only $$S(r)=\exp\!\left[\tfrac{1}{2}\left(r e^{-i\theta} a^2-r e^{i\theta} a^{\dagger 2}\right)\right],$$11 out of $$S(r)=\exp\!\left[\tfrac{1}{2}\left(r e^{-i\theta} a^2-r e^{i\theta} a^{\dagger 2}\right)\right],$$12 recorded shots were retained for analysis because single-peak spectra near 1040 nm were required for reliable overlap with the coherent reference, so the reconstructed femtosecond pulses characterize a selected subensemble rather than the full BSV ensemble (Kern et al., 22 Sep 2025).

Loss-induced mixedness is especially severe for ultrafast bright states. For Kerr-processed BSV, the purity of a single-mode Gaussian state under linear loss can be written as

$$S(r)=\exp\!\left[\tfrac{1}{2}\left(r e^{-i\theta} a^2-r e^{i\theta} a^{\dagger 2}\right)\right],$$13

where $$S(r)=\exp\!\left[\tfrac{1}{2}\left(r e^{-i\theta} a^2-r e^{i\theta} a^{\dagger 2}\right)\right],$$14 and $$S(r)=\exp\!\left[\tfrac{1}{2}\left(r e^{-i\theta} a^2-r e^{i\theta} a^{\dagger 2}\right)\right],$$15 are transmission and reflection, and $$S(r)=\exp\!\left[\tfrac{1}{2}\left(r e^{-i\theta} a^2-r e^{i\theta} a^{\dagger 2}\right)\right],$$16 is the initial mean photon number (Rasputnyi et al., 19 Dec 2025). With $$S(r)=\exp\!\left[\tfrac{1}{2}\left(r e^{-i\theta} a^2-r e^{i\theta} a^{\dagger 2}\right)\right],$$17 and $$S(r)=\exp\!\left[\tfrac{1}{2}\left(r e^{-i\theta} a^2-r e^{i\theta} a^{\dagger 2}\right)\right],$$18, the reported purity is $$S(r)=\exp\!\left[\tfrac{1}{2}\left(r e^{-i\theta} a^2-r e^{i\theta} a^{\dagger 2}\right)\right],$$19, explaining why Wigner-function negativity is not directly observed for the full bright mixed state even when individual squeezed-coherent components would remain non-Gaussian (Rasputnyi et al., 19 Dec 2025).

Propagation through nonlinear media imposes an additional macroscopic ceiling. A fully quantized propagation model for BSV-driven HHG in argon found that infrared photon losses, ground-state depletion, and ionization strongly decohere the driver, limiting the BSV-preserving medium length to roughly $$S(r)=\exp\!\left[\tfrac{1}{2}\left(r e^{-i\theta} a^2-r e^{i\theta} a^{\dagger 2}\right)\right],$$20 under typical conditions, where $$S(r)=\exp\!\left[\tfrac{1}{2}\left(r e^{-i\theta} a^2-r e^{i\theta} a^{\dagger 2}\right)\right],$$21 is the absorption length (Rivera-Dean et al., 23 Sep 2025). In the plateau region, the macroscopic harmonic yield is then reduced by more than two orders of magnitude relative to coherent driving, although plateau harmonics remain detectable at the level of order $$S(r)=\exp\!\left[\tfrac{1}{2}\left(r e^{-i\theta} a^2-r e^{i\theta} a^{\dagger 2}\right)\right],$$22 photons per pulse and beyond-cutoff harmonics persist at much lower absolute yield (Rivera-Dean et al., 23 Sep 2025). The implication is not that BSV is unusable in propagation-dominated strong-field physics, but that gas density, propagation length, and acceptable driver decoherence must be optimized jointly.

The most stable outlook therefore lies in applications that exploit, rather than suppress, BSV’s defining features: high flux without coherent displacement, strong higher-order moments, tunable mode structure, and shot-resolved fluctuations. Current work indicates that the most productive directions are single- or few-mode source engineering, low-loss interferometric shaping, per-shot optical or electronic tagging, and observables that depend on intensity correlations, conditional spectra, or relative timing rather than deterministic absolute phase (Kern et al., 22 Sep 2025, Yang et al., 5 Jan 2026).