Attosecond Quantum Tomography

Updated 5 July 2026

Attosecond quantum tomography is a reconstruction framework that retrieves full quantum state information of electrons and light using attosecond-resolution measurements.
It employs techniques like streaking, homodyne-like reconstruction, and reduced-state tomography to derive phase-space distributions and density matrices.
These methods extract key observables such as coherence, fluctuation dynamics, and sub-cycle phase details, advancing ultrafast strong-field research.

Attosecond quantum tomography denotes the reconstruction, with attosecond temporal resolution, of the full quantum state—or at least its relevant coherent subspace—of electrons and/or light fields in atoms, molecules, solids and liquids (Varillas et al., 16 Apr 2026). In this usage, tomography is not restricted to classical pulse characterization: it targets density operators, phase-space quasi-probabilities such as the Wigner function $W(\alpha)$ and the Husimi $Q$ -function $Q(\alpha)$ , reduced density matrices, and quantum-process information encoded in delay-resolved electron, XUV, and high-harmonic observables (Tzur et al., 23 Nov 2025). Current realizations span sub-cycle Gaussian-state retrieval of intense infrared fields by attosecond streaking, homodyne-like reconstruction of XUV harmonic modes generated with quantum light, reduced-state tomography of trajectory subspaces in high-harmonic generation (HHG), and full density-matrix reconstruction of free-electron and photoelectron wavepackets.

1. Definition and scope

In the standard quantum-optical sense, quantum state tomography reconstructs a full description of a field’s quantum state—typically the density operator $\hat\rho$ , or equivalently a phase-space quasi-probability such as the Wigner function $W(\alpha)$ or the Husimi $Q(\alpha)$ (Tzur et al., 23 Nov 2025). Operationally, one measures a sufficiently rich set of observables—quadrature distributions $P(X_\theta)$ , photon-number statistics, or related marginals—and inverts those data. Attosecond quantum tomography transfers that paradigm to ultrafast strong-field settings, where the relevant observables are delay-resolved photoelectron spectra, harmonic interferograms, trajectory-resolved intensities, or phase-dependent electron sideband populations.

The field is not a single method but a family of reconstructions tied together by attosecond temporal localization. The roadmap formulation is deliberately broad: many attosecond spectroscopies can already be interpreted as tomographic measurements, even if they are not yet formulated in the language of density matrices or Wigner functions (Varillas et al., 16 Apr 2026). This broader view includes reconstruction of electronic wavepackets in the continuum, Gaussian-state parameters of strong infrared modes, XUV harmonic phase-space distributions, and reduced density matrices of effective subspaces such as HHG trajectory manifolds.

A neighboring but distinct line is quantum metrology with attosecond-scale resolution. For example, x-ray Hong–Ou–Mandel interferometry supplies a biphoton source model, an interferometer design, and a forward map from a biphoton amplitude $\Phi$ to coincidence data $R_c(T)$ ; however, it does not itself perform tomography (Volkovich et al., 2019). This distinction is central: attosecond quantum tomography requires an inversion from measured observables to a state description, whereas attosecond quantum metrology may stop at parameter estimation.

2. Quantum objects and observables

The tomographic target depends on the physical platform. In current work, the reconstructed object may be a single effective optical mode, a continuum photoelectron density matrix, an HHG path qubit, or a free-electron state in an energy-sideband basis.

Platform	Reconstructed object	Primary observables
Quantized-field streaking	Single effective mode Gaussian state $\{\alpha,\xi\}$	$Q$ 0, $Q$ 1
Quantum-optical HHG	$Q$ 2, $Q$ 3, attosecond-resolved field variance	Single-shot harmonic intensities, phase-controlled quadratures
HHG trajectory manifold	$Q$ 4 in the $Q$ 5 basis	Path populations, visibility, phase
Free-electron UTEM states	Density matrix $Q$ 6 in the sideband basis $Q$ 7	Phase-dependent sideband populations
Photoelectron wavepackets	Continuous-variable density matrix $Q$ 8	Single XUV–IR delay scan

These objects are not interchangeable. In attosecond streaking with a quantized driving mode, the relevant state parameters are the displacement vector and covariance matrix of a Gaussian single-mode field (Zhou et al., 15 Apr 2026). In quantum-optical HHG driven by bright squeezed vacuum (BSV), the target is the quantum state of selected XUV harmonic modes, reconstructed through $Q$ 9- and $Q(\alpha)$ 0-functions and attosecond-resolved field variance (Tzur et al., 23 Nov 2025). In HHG path-state formalisms, the state is a $Q(\alpha)$ 1 reduced density matrix over short and long trajectories (Marchisio et al., 18 Jun 2026). In ultrafast transmission electron microscopy, the target is the free-electron density matrix in a discrete energy basis (Priebe et al., 2017). In continuum photoionization, the target is a continuous-variable density matrix of the photoelectron energy degree of freedom (Laurell et al., 2024).

This diversity implies that “attosecond quantum tomography” is best understood as a reconstruction framework, not as a unique observable or detector architecture. What unifies the field is the use of attosecond gating or sub-cycle phase control to convert ultrafast coherences into experimentally accessible marginals.

3. Streaking and continuum-state reconstruction

Attosecond streaking provides two distinct tomographic channels. In one direction, it reconstructs properties of the ionizing light field itself. In the other, it reconstructs the quantum phase of the emitted electron wavepacket and hence the phase of bound–continuum transition matrix elements.

For a quantized intense infrared mode prepared in a squeezed coherent state, attosecond streaking maps the light state onto the first two moments of the delay-resolved photoelectron momentum distribution (Zhou et al., 15 Apr 2026). Under the single-mode Gaussian assumption, the mean momentum and variance obey

$Q(\alpha)$ 2

$Q(\alpha)$ 3

For a squeezed coherent state,

$Q(\alpha)$ 4

so the variance is modulated at $Q(\alpha)$ 5, with phase set by the squeezing angle $Q(\alpha)$ 6. The $Q(\alpha)$ 7-oscillation of the mean retrieves the coherent phase $Q(\alpha)$ 8, while the $Q(\alpha)$ 9-oscillation of the variance retrieves the squeezing phase $\hat\rho$ 0 and the relative strengths of coherent and fluctuation contributions. Within that Gaussian single-mode regime, this constitutes a direct inversion from measured streaking moments to $\hat\rho$ 1, i.e. a sub-cycle Gaussian-state tomography of the infrared mode.

A separate streaking result concerns continuum-electron tomography. In “Attosecond streaking enables the measurement of quantum phase” (Yakovlev et al., 2010), the key finding is that the object reconstructed by standard FROG-type retrieval algorithms is not, in general, the XUV envelope but the time-domain electron wavepacket $\hat\rho$ 2. Its Fourier transform gives the momentum-space wavefunction $\hat\rho$ 3, including the phase of the bound–continuum transition matrix element $\hat\rho$ 4. This establishes streaking as a momentum-space quantum tomography technique for the outgoing electron. A complementary all-optical route uses interference of two independent phase-locked attosecond sources to determine recombination-dipole phases and scattering phase shifts over a large energy range, including the dipole phase around the Cooper minimum in argon (Azoury et al., 2018).

The significance of these results is twofold. First, streaking no longer functions only as classical field characterization; it becomes a phase-sensitive reconstruction of either the quantum state of the driving light or the quantum phase of continuum electronic motion. Second, the measurable quantities are low-order moments or spectral phases rather than rare events, which makes the protocols compatible with existing attosecond beamlines.

4. Quantum-optical HHG and XUV-state reconstruction

A decisive expansion of attosecond quantum tomography came from HHG driven by a strong coherent field at $\hat\rho$ 5 nm together with a weak BSV field at $\hat\rho$ 6 nm (Tzur et al., 23 Nov 2025). In this setting, the BSV’s quantum fluctuations perturb four consecutive half-cycles of the strong-field electron trajectories through complex action shifts $\hat\rho$ 7, with the symmetry $\hat\rho$ 8, $\hat\rho$ 9. The XUV dipole then becomes an internal four-slit interferometer: the amplitudes of odd, even, and half-integer harmonics are explicit functions of $W(\alpha)$ 0 and $W(\alpha)$ 1, and measuring four adjacent harmonics allows shot-by-shot inversion to the joint distribution $W(\alpha)$ 2.

This inversion yields two simultaneous reconstructions. On the photonic side, the authors reconstruct Husimi $W(\alpha)$ 3 distributions for individual harmonics and a Wigner function $W(\alpha)$ 4 for the single-photon channel of a half-integer harmonic via a homodyne-like protocol in which the two-color delay plays the role of the local-oscillator phase (Tzur et al., 23 Nov 2025). On the matter side, the same inversion reconstructs the distributions of $W(\alpha)$ 5 and $W(\alpha)$ 6, revealing sub-cycle correlations in tunneling amplitude and continuum phase. The time-domain ensemble of reconstructed harmonic amplitudes also gives the mean attosecond field $W(\alpha)$ 7 and the attosecond-resolved variance

$W(\alpha)$ 8

The measured statistics demonstrate direct transfer of quantum-optical properties from the BSV to the XUV and to the electron wavepacket. The input BSV has $W(\alpha)$ 9; half-integer harmonics show $Q(\alpha)$ 0, while even harmonics show $Q(\alpha)$ 1, consistent with one-BSV-photon and two-BSV-photon pathways, respectively (Tzur et al., 23 Nov 2025). The reconstructed Wigner function of a half-integer harmonic is centered near zero and elongated along one quadrature, while the joint distribution $Q(\alpha)$ 2 reveals correlated redistribution of tunneling-noise fluctuations between neighboring half-cycles. A closely related experimental framing describes the same advance as reconstruction and control of the quantum state of harmonics through homodyne-like tomography, with sub-cycle manipulation of photon statistics by varying the delay between the coherent and BSV driving fields (Tzur et al., 13 Feb 2025).

This body of work is widely taken as the first experimental realization of quantum-state tomography of XUV light with attosecond precision. Its technical novelty lies in using an internal, self-referenced interferometer rather than a conventional external local oscillator, thereby bypassing the present difficulty of phase-locking an XUV local oscillator.

5. Reduced-state, photoelectron, and free-electron tomography

Attosecond quantum tomography also encompasses reduced-state reconstructions that do not target a full optical mode. In HHG, the dominant short and long trajectories define an experimentally addressable two-level subsystem: the Attosecond Path Qubit (APQ) (Marchisio et al., 18 Jun 2026). Its normalized density matrix,

$Q(\alpha)$ 3

encodes path populations and coherence. Classical dephasing from ensemble averaging and quantum decoherence from tracing out unobserved variables, such as transverse momentum, are separated explicitly. Visibility $Q(\alpha)$ 4, predictability $Q(\alpha)$ 5, and purity $Q(\alpha)$ 6 provide a tomographic language for trajectory-based coherence loss. This does not yet amount to full HHG-state tomography, but it supplies a reduced-state framework with operational reconstruction targets.

For continuum photoelectrons, a more direct density-matrix reconstruction is achieved by the rainbow-KRAKEN protocol (Laurell et al., 2024). It reconstructs the continuous-variable density matrix of a photoelectron in a single time delay scan by combining a broadband infrared probe scanned in time with a narrowband infrared reference fixed to the XUV pulse. The protocol is illustrated for a Fano resonance in He and mixed states in Ar arising from spin–orbit splitting, and it yields excellent fidelities and near-perfect estimation of the purity. Because the reconstructed object is $Q(\alpha)$ 7, this work squarely places attosecond photoionization in the domain of continuous-variable quantum-state tomography.

Free-electron quantum optics provides the most explicit density-matrix reconstructions. SQUIRRELS—“Spectral Quantum Interference for the Regularized Reconstruction of free-Electron States”—reconstructs the density matrix of attosecond-structured free-electron ensembles from phase-dependent sideband populations produced by a known optical unitary (Priebe et al., 2017). From the reconstructed $Q(\alpha)$ 8, the authors obtain Wigner functions and temporal profiles, including an attosecond pulse train with rms width $Q(\alpha)$ 9 as and FWHM $P(X_\theta)$ 0 as. A later reanalysis introduced maximum likelihood estimation, Bayesian inversion, and deep learning for the same class of data, obtaining pulse-durations of about $P(X_\theta)$ 1 as and predicting a degree of coherence of $P(X_\theta)$ 2 per cent for radiations and excitations produced by these electrons (Jeng et al., 25 Aug 2025).

Model-based reconstructions of electronic coherences also appear in core-level dynamics. In nitric oxide, an isolated attosecond soft x-ray pulse excites a coherent superposition of core-excited states near the oxygen K-edge, and a circularly polarized infrared field acts as a clock to time-resolve Auger–Meitner decay (Li et al., 2021). The observed revival in the Auger–Meitner yield around $P(X_\theta)$ 3 fs provides direct evidence of coherence between core-excited states. Although this is not presented as full density-matrix tomography, it is a reconstruction of populations and coherences in a reduced core-excited subspace from attosecond-resolved electron observables.

6. Limitations, misconceptions, and outlook

A recurrent misconception is that every attosecond characterization protocol is already a full quantum tomography. The current literature is more restrictive. In streaking-based quantum-noise retrieval, only the first two moments of a single effective mode are reconstructed; that is complete only under the single-mode Gaussian assumption (Zhou et al., 15 Apr 2026). In AQI-based proposals, the reconstructed AQT-distribution is Wigner-like, but the authors stress that the approach does not constitute a genuine quantum state tomography method because the effective local oscillator is not independent of the probed state, and the two-color delay changes the state itself (Rivera-Dean et al., 2 Nov 2025). In x-ray Hong–Ou–Mandel metrology, the source and forward model are tomographically relevant, but the scheme provides building blocks rather than an actual reconstruction (Volkovich et al., 2019).

Another important limitation is modality. Some protocols reconstruct photonic states, others reconstruct electronic states, and others reconstruct reduced manifolds or process parameters. The roadmap therefore treats attosecond quantum tomography as a programme rather than a single experiment type, emphasizing the need for high-flux, multi-colour, polarization-controllable attosecond sources, integrated quantum-optical detection, and robust inversion algorithms grounded in realistic Hamiltonians (Varillas et al., 16 Apr 2026). This suggests that future progress will depend as much on detector and source engineering as on formal inversion schemes.

Within those constraints, the field is converging on several concrete directions. The quantized-field streaking framework explicitly points to higher moments and multi-mode extensions as routes toward full non-Gaussian state reconstruction (Zhou et al., 15 Apr 2026). APQ methods indicate how mode selection and conditioning can separate classical dephasing from intrinsic decoherence in HHG (Marchisio et al., 18 Jun 2026). Quantum-optical HHG with BSV has already shown that both electrons and XUV harmonics can carry reconstructable nonclassical correlations on sub-cycle timescales (Tzur et al., 23 Nov 2025). Taken together, these results support a precise conclusion: attosecond quantum tomography is no longer merely a conceptual analogy to homodyne detection, but an emerging set of experimentally grounded reconstruction protocols for ultrafast quantum states of light and matter.