Event Spectroscopy System

• Event spectroscopy system is an instrumentation paradigm that processes individual, timestamped events rather than integrated frames to reconstruct diverse physical observables.
• It combines precise timing hardware with calibrated inversion techniques and machine learning to accurately recover spectral, energetic, temporal, spatial, or momentum information.
• Applications include fast-ion laser spectroscopy, structured-light multispectral sensing, electron microscopy, and transient astrophysical event observation, demonstrating high temporal resolution and enhanced sensitivity.

An event spectroscopy system is an instrumentation paradigm in which the fundamental datum is an individually timestamped event, or a short event-accumulation window, rather than a conventionally integrated frame. In the published literature, the term spans fast-ion laser spectroscopy based on ion–electron coincidences, structured-light depth and multispectral sensing with event cameras, event-driven electron and photoelectron spectroscopies, neutron time-of-flight systems, fluorescence microscopes that fuse asynchronous and frame-based sensors, and dual-comb spectrometers for single transient events (Udrescu et al., 2023, Geckeler et al., 8 Sep 2025, Auad et al., 13 Oct 2025, Auad et al., 2021). Across these implementations, the common objective is to recover spectral, energetic, temporal, spatial, or momentum information at the level of single events or sparsely sampled intervals, typically by combining precise timing hardware with calibrated inversion, statistical reconstruction, or machine learning.

1. Definition and domain scope

The phrase “event spectroscopy system” is used in at least two distinct senses. In one sense, it denotes an event-resolved measurement architecture in which asynchronous sensor outputs, coincidence packets, or single-particle detections are directly processed into spectral observables. In another, it denotes instrumentation specialized for observing transient astrophysical events, as in the Son Of X-Shooter (SOXS), a single object spectrograph for simultaneous UV–VIS and NIR coverage of astronomical transients (Xian et al., 2019, Santhakumari et al., 2024).

Within event-resolved architectures, the event primitive depends on the physical domain. Event cameras emit tuples such as $e_k=(x_k,t_k,p_k)$ when a log-brightness change crosses threshold; Timepix3 detectors output packetized $(x,y,\mathrm{ToA},\mathrm{ToT})$ hits; delay-line photoemission systems stream $$\{X_i,Y_i,\mathrm{TOF}_i,\mathrm{ENC}_i,\mathrm{trigger\_id}_i\}$$; and fast-beam isotope spectroscopy constructs event vectors $x=(x_i,y_i,x_f,y_f,\Delta t)$ from coincident electron and ion detections (Geckeler et al., 8 Sep 2025, Auad et al., 13 Oct 2025, Udrescu et al., 2023).

Domain	Event primitive	Reconstructed quantity
Fast-isotope laser spectroscopy	$x=(x_i,y_i,x_f,y_f,\Delta t)$ plus $\nu_{\mathrm{laser}}$	$E_0$, $v$, Doppler-corrected $\nu_0$
Structured-light multispectral sensing	$e_k=(x_k,t_k,p_k)$	depth $(x,y,\mathrm{ToA},\mathrm{ToT})$0, multispectral cube $(x,y,\mathrm{ToA},\mathrm{ToT})$1
Event-based STEM/EELS	$(x,y,\mathrm{ToA},\mathrm{ToT})$2 and TDC row stamps	spatially resolved images, $(x,y,\mathrm{ToA},\mathrm{ToT})$3
Multidimensional photoemission	$(x,y,\mathrm{ToA},\mathrm{ToT})$4	$(x,y,\mathrm{ToA},\mathrm{ToT})$5

A direct consequence is that “spectroscopy” in this context is broader than wavelength measurement alone. Depending on the system, the reconstructed observable may be optical frequency, reflectance versus wavelength, electron energy loss, neutron kinetic energy, photoelectron kinetic energy and momentum, or transient state populations.

2. Instrument architectures

A recurrent architecture combines an asynchronous sensor with actively controlled illumination or excitation. In structured-light event spectroscopy for UAV perception, a Prophesee Gen3 event sensor is paired with either a Sony MP-CL1A RGB laser projector or a broadband light source with six 10 nm-bandpass filters tuned between 650 nm and 850 nm; the projector and sensor operate in a stereo-like geometry with baseline $(x,y,\mathrm{ToA},\mathrm{ToT})$6, and wavelength cycling interleaves depth and spectral acquisition (Geckeler et al., 8 Sep 2025). Closely related fluorescence microscopes split emitted light into two arms: an event-based path for microsecond temporal dynamics and a CMOS or sCMOS path containing a diffractive optical element that encodes spectral content into a diffractogram. One implementation uses a Prophesee EVK3 VGA together with a Thorlabs Zelux CS165MU1; another uses a Prophesee PPS3MVCD DVS with a Thorlabs CS165MU monochrome CIS in a 4F relay (Baird et al., 2024, Baird et al., 19 Mar 2026). A visible-range event-based spectrometer adopts a Czerny–Turner layout with a 50 $(x,y,\mathrm{ToA},\mathrm{ToT})$7m slit, an off-axis parabolic mirror of focal length $(x,y,\mathrm{ToA},\mathrm{ToT})$8 mm, a 600 grooves/mm grating blazed at 500 nm, an achromatic doublet of focal length $(x,y,\mathrm{ToA},\mathrm{ToT})$9 mm, and a beam splitter feeding a Prophesee EVK4 HD and a frame-based reference spectrometer in parallel (Teixeira et al., 8 May 2026).

Particle and atomic implementations emphasize coincidence timing and field-controlled trajectories. In fast-exotic-isotope laser spectroscopy, a beam such as 1 MeV $$\{X_i,Y_i,\mathrm{TOF}_i,\mathrm{ENC}_i,\mathrm{trigger\_id}_i\}$$0 10 keV is decelerated electrostatically to $$\{X_i,Y_i,\mathrm{TOF}_i,\mathrm{ENC}_i,\mathrm{trigger\_id}_i\}$$1 keV, neutralized in a charge-exchange cell, resonantly excited by a collinear laser, and ionized inside the uniform field between two parallel plates; a position-sensitive electron detector above the ionization region and a second position-sensitive ion detector record the correlated products (Udrescu et al., 2023). In continuous-wave atomic photoionization, a reaction microscope with $$\{X_i,Y_i,\mathrm{TOF}_i,\mathrm{ENC}_i,\mathrm{trigger\_id}_i\}$$2 V/cm and $$\{X_i,Y_i,\mathrm{TOF}_i,\mathrm{ENC}_i,\mathrm{trigger\_id}_i\}$$3 G, together with two 80 mm diameter microchannel-plate and delay-line detectors 430 mm from the interaction region, reconstructs ionization times and complete 3D fragment momenta from electron–recoil ion coincidences (Romans et al., 21 Feb 2025). In single-event neutron spectroscopy, a DRACO Petawatt pitcher–catcher source is coupled to a 4×4 mm$$\{X_i,Y_i,\mathrm{TOF}_i,\mathrm{ENC}_i,\mathrm{trigger\_id}_i\}$$4, 500 $$\{X_i,Y_i,\mathrm{TOF}_i,\mathrm{ENC}_i,\mathrm{trigger\_id}_i\}$$5m-thick CVD single-crystal diamond detector at 1.5 m flight path, with a CIVIDEC 2 GHz broadband amplifier and a Tektronix MSO64B oscilloscope (Millán-Callado et al., 24 Jun 2025).

Electron-microscopy implementations revolve around Timepix3. A custom scanning unit (SU) based on an FPGA, high-speed DACs, and ADCs generates $$\{X_i,Y_i,\mathrm{TOF}_i,\mathrm{ENC}_i,\mathrm{trigger\_id}_i\}$$6 and $$\{X_i,Y_i,\mathrm{TOF}_i,\mathrm{ENC}_i,\mathrm{trigger\_id}_i\}$$7 deflection waveforms, shares a common reference clock with the Timepix3 readout, and maps each electron timestamp to instantaneous probe position. In related hyperspectral EELS work, four tiled 256 × 256 Timepix3 chips, a SPIDR board with two supplementary TDC lines, and a custom 25 MHz scan engine embed scan timing directly into the event stream (Auad et al., 13 Oct 2025, Auad et al., 2021). Multidimensional photoemission systems use a time-of-flight spectrometer coupled to a delay-line detector, with TDCs collecting X, Y, TOF, and optional encoder channels against a common event clock (Xian et al., 2019). By contrast, SOXS is not asynchronous at the detector-event level; its “event spectroscopy” role is to observe astrophysical transient events using a common path, UV–VIS and NIR spectrographs, an acquisition and guiding camera, ESO NGC controllers, PLC-based control, and an automated Python data reduction pipeline (Santhakumari et al., 2024).

3. Event models and reconstruction formalisms

The mathematical core of event spectroscopy is an inverse problem that maps event primitives into calibrated physical quantities. For event cameras, the canonical triggering relation is

$$\{X_i,Y_i,\mathrm{TOF}_i,\mathrm{ENC}_i,\mathrm{trigger\_id}_i\}$$8

or, equivalently, an event occurs whenever $$\{X_i,Y_i,\mathrm{TOF}_i,\mathrm{ENC}_i,\mathrm{trigger\_id}_i\}$$9 (Geckeler et al., 8 Sep 2025, Baird et al., 19 Mar 2026). In structured-light depth sensing, disparity is converted into range by the pinhole relation

$x=(x_i,y_i,x_f,y_f,\Delta t)$0

while event counts after wavelength cycling are mapped to reflectance through $x=(x_i,y_i,x_f,y_f,\Delta t)$1 (Geckeler et al., 8 Sep 2025).

Fast-beam laser spectroscopy introduces a probabilistic inversion over hidden kinetic energy. The event vector $x=(x_i,y_i,x_f,y_f,\Delta t)$2 is passed through a Mixture Density Network whose output parameterizes

$x=(x_i,y_i,x_f,y_f,\Delta t)$3

with training by negative log-likelihood over reference-atom events. The predicted $x=(x_i,y_i,x_f,y_f,\Delta t)$4 yields $x=(x_i,y_i,x_f,y_f,\Delta t)$5, and the rest-frame transition frequency is recovered from the non-relativistic Doppler relation

$x=(x_i,y_i,x_f,y_f,\Delta t)$6

The same system uses numerical inversion of $x=(x_i,y_i,x_f,y_f,\Delta t)$7 from a SIMION field map when the approximation $x=(x_i,y_i,x_f,y_f,\Delta t)$8 is insufficient (Udrescu et al., 2023).

Time-of-flight formalisms dominate neutron, photoemission, and several electron-spectroscopy systems. For fast neutrons over path $x=(x_i,y_i,x_f,y_f,\Delta t)$9,

$x=(x_i,y_i,x_f,y_f,\Delta t)$0

and detector efficiency $x=(x_i,y_i,x_f,y_f,\Delta t)$1 is incorporated after GEANT4 simulation and background subtraction (Millán-Callado et al., 24 Jun 2025). In multidimensional photoemission, calibrated mappings $x=(x_i,y_i,x_f,y_f,\Delta t)$2, $x=(x_i,y_i,x_f,y_f,\Delta t)$3, and $x=(x_i,y_i,x_f,y_f,\Delta t)$4, together with spherical-timing-aberration and optional space-charge corrections, convert raw event tables into $x=(x_i,y_i,x_f,y_f,\Delta t)$5 hypercubes (Xian et al., 2019). In continuous-wave atomic photoionization, detector coordinates and unknown time-of-flight $x=(x_i,y_i,x_f,y_f,\Delta t)$6 are inverted under homogeneous $x=(x_i,y_i,x_f,y_f,\Delta t)$7 and $x=(x_i,y_i,x_f,y_f,\Delta t)$8 fields to recover $x=(x_i,y_i,x_f,y_f,\Delta t)$9, $\nu_{\mathrm{laser}}$0, and $\nu_{\mathrm{laser}}$1, with the ionization time obtained by minimizing transverse momentum mismatch under the condition $\nu_{\mathrm{laser}}$2 (Romans et al., 21 Feb 2025).

Event-based EELS and related microscopy systems exploit explicit synchronization of scan timing with detector timing. In hyperspectral EELS, TDC timestamps $\nu_{\mathrm{laser}}$3 and $\nu_{\mathrm{laser}}$4 determine the row index, while

$\nu_{\mathrm{laser}}$5

assigns the scan column, and the spectrum cube is accumulated as

$\nu_{\mathrm{laser}}$6

For sparse or non-raster scans, reconstruction may use k-nearest-neighbors interpolation or inpainting based on $\nu_{\mathrm{laser}}$7 minimization (Auad et al., 2021, Auad et al., 13 Oct 2025).

Machine learning appears both in spectral inversion and in event triage. Fluorescence fusion systems use CNN classification of diffractograms, with one system reporting a lookup-and-link chain of CNN spectral classification, HAC + GMM alignment, and a Kalman filter, and another using a U-Net with weighted cross-entropy plus generalized Dice loss for two output spectral channels (Baird et al., 19 Mar 2026, Baird et al., 2024). Rare-event force spectroscopy frames the event-selection problem as extreme class imbalance: 1D force–extension traces are rasterized to 224×224 images, passed through a modified ResNet18, optimized with asymmetric focal loss,

$\nu_{\mathrm{laser}}$8

and routed through a dual-threshold triage system $\nu_{\mathrm{laser}}$9 into “Auto-Discard,” “Manual-Review,” or “Auto-Accept” (Rodriguez-Ramos, 8 Jun 2026). In cyclotron-radiation emission spectroscopy, a U-Net performs binary segmentation of sparse time–frequency tracks, after which connected-component labeling, line fitting, and a 4-D SVM veto recover event start frequencies for precise energy spectroscopy (Esfahani et al., 2024).

4. Performance regimes and demonstrated precision

Fast-isotope laser spectroscopy was explicitly designed for extreme beam conditions. Numerical simulations showed kHz-level uncertainties for ion beams produced at temperatures $E_0$0 K, with energy spreads as large as $E_0$1 keV and non-uniform velocity distributions. For $E_0$2, the MDN reduced $E_0$3 from 100 eV to 0.4 eV per event; for $E_0$4, from 10 keV to 77 eV per event. For the 583 nm line of $E_0$5, 100 detected atoms yielded $E_0$6 MHz and $E_0$7 atoms yielded $E_0$8 kHz, while the 255 nm line of $E_0$9 gave $v$0 MHz for $v$1 atoms (Udrescu et al., 2023).

Structured-light event spectroscopy achieved competitive depth and reflectance performance. Across the five scenes “Branch,” “Buddha,” “David,” “Globe,” and “Lamp,” average depth RMSE was $v$2 cm and Chamfer distance $v$3 cm, compared with 0.75/0.88 cm for Kinect V2, 1.09/1.24 cm for PMD Pico-Flexx, 0.65/1.13 cm for Intel RealSense D435, and 0.84/1.2 cm for RealSense in direct-stereo mode, corresponding to up to 60% reduction in RMSE over commercial sensors. Under bright-sun versus low-light scenes, RMSE variation was 5% for the event-spectroscopy method and 56% for RealSense stereo. In 31 Masoala Rainforest RGB-D scenes, mean IoU was 0.365 for depth-only, 0.441 for RGB-only, and 0.497 for RGBD; branch-classification IoU improved from 0.20 to 0.29, and color-chart reconstruction reduced RMSE from 33.2 to 16.6 CIE-$v$4 (Geckeler et al., 8 Sep 2025).

Timepix3-based electron spectroscopy established nanosecond timing with high throughput. In the custom scanning-unit architecture, practical temporal uncertainty was measured as $v$5–2.5 ns at 60–200 keV accelerating voltage; quantum efficiency was $v$6 for 60–200 keV electrons; read-out noise was effectively zero; SNR improved by 2–3× over frame-based detectors at equivalent dose; PSF full-width at half-maximum was $v$7m sensor spread; and effective pixel dwell in Lissajous scans reached 10–30 ns. The Timepix3 chip saturated at $v$8 Mhits/s, with $v$9 Mhits/s stated as feasible using four chips tiled or cascaded readouts (Auad et al., 13 Oct 2025). In hyperspectral EELS, fine ToA resolution was 1.5625 ns, TDC stamp resolution 260 ps, the combined temporal resolution budget gave $\nu_0$0 ns, and demonstrated pixel dwell times extended down to 40 ns, the minimum SU limit (Auad et al., 2021).

Single-event neutron spectroscopy demonstrated compact fast-neutron ToF under laser-plasma conditions. Integrated yields above 1 MeV were $\nu_0$1 n/shot for Cu(p,n) and $\nu_0$2 n/shot for Li(p,n), averaged over $\nu_0$3 shots in one day. With source burst width $\nu_0$4 ns and flight path $\nu_0$5 m, the energy resolution satisfied $\nu_0$6 at 10 MeV and $\nu_0$7 at higher energies (Millán-Callado et al., 24 Jun 2025).

Fluorescence and visible-range event spectrometers prioritized microsecond dynamics. Asynchronous-spectral fusion fluorescence microscopy reported $\nu_0$8m spatial resolution over a 0.5 mm field of view, effective temporal resolution down to 100 $\nu_0$9s, separation of fluorophores with emission peaks 23 nm apart, 96% pixel-level classification accuracy, localization error $e_k=(x_k,t_k,p_k)$0 pixel, and temporal jitter $e_k=(x_k,t_k,p_k)$1s (Baird et al., 19 Mar 2026). A widefield fluorescence system using EBIS and CIS fusion achieved temporal resolution of $e_k=(x_k,t_k,p_k)$2s, minimum $e_k=(x_k,t_k,p_k)$3s, spatial resolution $e_k=(x_k,t_k,p_k)$4m, spectral resolution $e_k=(x_k,t_k,p_k)$5 nm for fluorophores with 88% spectral overlap, and tracking throughput of $e_k=(x_k,t_k,p_k)$6 localization events per second (Baird et al., 2024). The Czerny–Turner event-based spectrometer covered a 234 nm visible bandwidth with spectral resolution of approximately 0.18 nm per pixel, reconstructed spectra at probing rates up to tens of kilohertz, reached an effective 60 kHz probing rate at $e_k=(x_k,t_k,p_k)$7 kHz and 80 kHz at $e_k=(x_k,t_k,p_k)$8 kHz, and in a microfluidic dye experiment increased temporal sampling from $e_k=(x_k,t_k,p_k)$9 frame-based samples to $(x,y,\mathrm{ToA},\mathrm{ToT})$00 event-based spectral slices over a 16 ms transition, with peak-to-baseline contrast $(x,y,\mathrm{ToA},\mathrm{ToT})$01 higher and dye-front timing resolved with $(x,y,\mathrm{ToA},\mathrm{ToT})$02 ms precision (Teixeira et al., 8 May 2026).

Machine-learning-based event reconstruction in spectroscopy-like sparse domains emphasized efficiency rather than direct sensor precision. In cyclotron radiation emission spectroscopy, the CNN+SVM chain improved event reconstruction efficiency by 24.1% relative to the point-clustering baseline, while maintaining comparable accuracy in track-parameter reconstruction; at matched false-positive rate, the ML branch reconstructed 355 true positives against 285 for the baseline (Esfahani et al., 2024). In automated force-spectroscopy triage under 1.34% positive prevalence, the modified ResNet18 reached overall accuracy 0.9196 and recall 0.9231, while automatically discarding 880 traces and reducing manual curation workload by over 90% (Rodriguez-Ramos, 8 Jun 2026).

5. Scientific uses and disciplinary impact

In nuclear and atomic physics, event spectroscopy directly addresses short lifetimes, hot beams, and low yields. Event-by-event Doppler correction on highly energetic beams makes in-flight isotope-shift and hyperfine studies possible for lifetimes $(x,y,\mathrm{ToA},\mathrm{ToT})$03 ms, including short-lived drip-line nuclei such as $(x,y,\mathrm{ToA},\mathrm{ToT})$04, and the stated applications include nuclear-structure models, astrophysical reaction rates for the r-process, and searches for new force carriers via King-plot nonlinearity (Udrescu et al., 2023). Laser-driven neutron ToF spectroscopy is positioned for fast neutron-induced reaction studies, especially where short-lived isotopes or high instantaneous flux are required (Millán-Callado et al., 24 Jun 2025). Continuous-wave 3D momentum spectroscopy in atomic photoionization recovers nanosecond ionization timing together with full fragment momentum vectors, thereby accessing incoherent population dynamics and, prospectively, coherent atomic dynamics on the nanosecond timescale (Romans et al., 21 Feb 2025). Cyclotron-radiation emission spectroscopy applies event reconstruction to precise $(x,y,\mathrm{ToA},\mathrm{ToT})$05-decay energy spectra in the Project 8 neutrino-mass program (Esfahani et al., 2024).

In electron microscopy and condensed-matter spectroscopy, the main effect is to decouple scanning speed from frame-based detector limitations. Event-based hyperspectral EELS reconstructs hyperspectral images at SU-limited dwell time, allowing temporal response on the scale of tens of nanoseconds rather than milliseconds and enabling in situ observation of calcite decomposition into CaO and CO$(x,y,\mathrm{ToA},\mathrm{ToT})$06 under beam irradiation (Auad et al., 2021). The custom scanning-unit/Timepix3 platform extends this logic to hyperspectral EELS, cathodoluminescence excitation spectroscopy, electron energy-gain spectroscopy, photon-induced near-field electron microscopy, and nanosecond nanothermometry, including a proof-of-concept $(x,y,\mathrm{ToA},\mathrm{ToT})$07 map in h-BN acquired in $(x,y,\mathrm{ToA},\mathrm{ToT})$08 s at 10–30 ns effective dwell (Auad et al., 13 Oct 2025). In photoemission, end-to-end event workflows make billion-count single-electron data streams tractable for energy–momentum band mapping and database integration (Xian et al., 2019).

In imaging, machine vision, and microscopy, event spectroscopy frequently fuses temporal and spectral channels that are difficult to obtain simultaneously with a single conventional sensor. Structured-light event spectroscopy adds wavelength-tunable active illumination to event-camera depth sensing, thereby producing depth and multispectral reflectance from a single sensor and supporting material differentiation between leaves and branches in forest environments (Geckeler et al., 8 Sep 2025). Fluorescence implementations similarly use sensor fusion to combine microsecond event timing with diffractive spectral encoding, enabling spectrally resolved tracking without scanning or filter switching and motivating applications in live-animal fluorescence microscopy, fast multiparameter flow cytometry, microfluidic assays, sub-millisecond calcium or voltage imaging, and ultrafast reaction dynamics (Baird et al., 19 Mar 2026, Baird et al., 2024). The visible Czerny–Turner event spectrometer validates the same high-temporal-resolution principle in a microscope-coupled absorption experiment (Teixeira et al., 8 May 2026).

In astronomy, SOXS extends the term toward transient-event operations rather than asynchronous-event sensing. It is a specialized facility for observing transient events with simultaneous spectral coverage in UV–VIS (350–850 nm) and NIR (800–2000 nm), average resolving power $(x,y,\mathrm{ToA},\mathrm{ToT})$09 for a 1 arcsec slit in UV–VIS and $(x,y,\mathrm{ToA},\mathrm{ToT})$10 in NIR, a flexible scheduler, and a fully automated Python pipeline that produces 1D spectra and quality-control products (Santhakumari et al., 2024). This usage underscores that “event spectroscopy” may refer either to event-resolved sensing or to spectroscopy of transient external phenomena.

6. Limitations, design constraints, and future directions

The principal limitations are domain-specific but structurally similar: sensor nonidealities, synchronization error, throughput saturation, and the need for strong calibration. Structured-light event spectroscopy is currently constrained by projector power and spectral hardware. The portable prototype is limited to RGB because of the commercial projector spectrum, while true multispectral operation would require custom multispectral LEDs or miniaturized filter wheels; under direct sunlight or at long ranges, projection power may become insufficient; and full UAV deployment still requires further mechanical miniaturization and vibration isolation (Geckeler et al., 8 Sep 2025). In asynchronous-spectral fusion microscopy, event sensors provide microsecond timing but lack intrinsic spectral discrimination, which is why a diffractive spectral path and synchronization procedure are required in the first place (Baird et al., 19 Mar 2026).

Timepix3-based systems face timing and bandwidth constraints. Continuous-gun jitter, energy-dependent timewalk, scanning-coil inductance, amplifier bandwidth, and data bandwidth above 40 Mhits/s are identified as the main technical hurdles, and proposed remedies include pre-emphasis or inverse-filter coil drive, electric deflectors, Timepix4 with 200 ps timestamps, thinner sensors or alternative high-$(x,y,\mathrm{ToA},\mathrm{ToT})$11 materials, FPGA-based real-time compression, and GPU-accelerated deconvolution or inpainting (Auad et al., 13 Oct 2025). Event-based hyperspectral EELS adds pixel dead time of $(x,y,\mathrm{ToA},\mathrm{ToT})$12 ns, global throughput up to $(x,y,\mathrm{ToA},\mathrm{ToT})$13 Mevents/s, and column-tree saturation as practical limits on usable current (Auad et al., 2021).

Source technology can also dominate the constraint set. The DRACO PW neutron system relies on a single plasma mirror for contrast enhancement, which limits repetition rate to $(x,y,\mathrm{ToA},\mathrm{ToT})$14 shot/10 s; the forward-looking route is higher-repetition-rate PW-class lasers, automated multi-shot target delivery, and improved ion-acceleration efficiency (Millán-Callado et al., 24 Jun 2025). In fast-beam laser spectroscopy, the stated outcome is already kHz-level precision on beams too hot and too short-lived for cooling or trapping; the paper further states that event-by-event Doppler correction can recover intrinsic transition frequencies with kHz, and ultimately Hz, precision (Udrescu et al., 2023).

A plausible implication is that future event spectroscopy will increasingly depend on co-design of sensor physics, timing distribution, and computational inversion. The literature already points in that direction: variational fusion of asynchronous and spectral modalities in fluorescence microscopy, dual-threshold human-in-the-loop triage for rare force-spectroscopy events, plug-and-play modules for EELS, CL, EEGS, and PINEM in one microscope platform, and autonomous scheduling or quick-look classification loops in transient astronomy (Baird et al., 19 Mar 2026, Rodriguez-Ramos, 8 Jun 2026, Auad et al., 13 Oct 2025, Santhakumari et al., 2024). Across these variants, the defining trend is not a single detector technology but an operational principle: sparse event capture, explicit timing, and post hoc reconstruction of observables that conventional frame-based instrumentation cannot recover with equivalent temporal efficiency.