GaMMA: Gamma-Ray Modeling & Analysis

• GaMMA is a comprehensive framework that integrates experimental, analytical, and computational approaches to generate, detect, and analyze gamma-ray and spectral data.
• Its supervised NMF method in the gamma_flow package enables real-time spectral decomposition with high classification accuracy (>90%) and reconstruction fidelity (>98%).
• The framework supports a range of applications from high-energy astrophysics and collider physics to spectroscopy and materials science through modular, interpretable pipelines.

GaMMA refers to a set of frameworks, instruments, and computational methods that address the generation, detection, and analysis of gamma-ray or spectral data across multiple disciplines, ranging from high-energy particle physics and astrophysics to analytical spectroscopy and computational modeling. The various Gamma- or GaMMA-derived projects are distinguished by their specialized aims—either experimental (gamma-beam facilities, telescopes, colliders), analytical (matrix-based spectral analysis), or computational (hyperrelativistic fluid simulations). Each instantiation leverages the unique physical properties of gamma photons or the distinct characteristics of one-dimensional spectral data to enable scientific inference or technological application.

1. Supervised Factorization for Spectral Analysis: GaMMA (gamma_flow)

The GaMMA framework, as implemented in the gamma_flow open-source package, provides real-time automated analysis of one-dimensional spectral data—originally for gamma-ray spectra, though the methodology generalizes to other spectroscopic domains (IR, Raman, mass spectrometry, UV-Vis, stellar spectra). The core is a supervised non-negative matrix factorization (NMF) model for dimensionality reduction and interpretable decomposition:

• Formalism: Let $\mathbf{X}\in\mathbb{R}^{n\times m}_{\geq 0}$ represent $n$ spectra (each an $m$-dimensional channel count vector), and let $\mathbf{L}\in\mathbb{R}^{m\times k}_{\geq 0}$ denote the fixed basis whose $k$ columns are the mean spectra for each known training label. The score matrix $\mathbf{S}\in\mathbb{R}^{n\times k}_{\geq 0}$ encodes non-negative contributions of each basis spectrum to each observation, estimated via a row-wise non-negative least-squares (NNLS) optimization:

$$\min_{\mathbf{S}\geq 0}\ \frac{1}{2} \|\mathbf{X} - \mathbf{S}\mathbf{L}^\mathsf{T}\|^2_F + \frac{\lambda}{2}\|\mathbf{S}\|^2_F$$

• Workflow: The typical pipeline is:
  1. Preprocessing: Energy calibration, rebinning to a common channel grid, data aggregation.
  2. Decomposition: Compute $\mathbf{L}$ by averaging spectra per label in the training set. For each test spectrum $\mathbf{x}_i$, solve NNLS to infer $\mathbf{s}_i$.
  3. Denoising: Project input onto physical basis: $n$0.
  4. Classification: Assign label by highest normalized score or multi-label rule with threshold $n$1.
  5. Outlier Detection: Use cosine similarity between input and reconstruction, $n$2, with threshold $n$3.

Table: Key Performance Metrics for GaMMA (gamma_flow) (Rädle et al., 12 Nov 2025)

Metric	Value / Method	Note
Classification accuracy	>90% (single-label), F1 > 0.92 (multi-label)	Demonstrated on test set of 5 isotopes+background
Denoising quality	$n$4, explained variance >98%	High-fidelity reconstruction
Outlier detection	Precision/Recall defined via standard formula	Cosine similarity decision boundary
Inference speed	$n$51 ms/spectrum (CPU), $n$6	No GPU/massive memory required
Generalizability	Any 1D spectra, not just gamma-ray	Requires label-averaged basis per application

Applying supervised NMF with a fixed, physical basis yields interpretable latent axes (each corresponding to a label such as an isotope), offering a transparent alternative to black-box neural network models while maintaining operational efficiency and adaptability (Rädle et al., 12 Nov 2025).

2. Gamma-Ray Astrophysics Missions and Telescopes

Several missions and experimental platforms labeled as “GAMMA” or “GAMMA-LIGHT/400” are designed to probe fundamental questions in astrophysics, dark matter, and cosmic-ray propagation:

• GAMMA-LIGHT covers the 10 MeV–100 GeV energy range, bridging the observational gap left by prior instruments (COMPTEL, AGILE, Fermi-LAT). It features:
  • A high-resolution silicon tracker (41 trays of microstrip detectors) for sub-degree point-spread function ($n$7 at 1 GeV).
  • A CsI calorimeter and anticoincidence systems for full gamma-ray event reconstruction.
  • Energy resolution parameterized as $n$8; effective area up to 2000 cm² at 1 GeV.
  • Sensitivities enable decisive studies of diffuse galactic emission, SNR pion decay, dark matter annihilation/decay, and transients like GRBs (Morselli et al., 2014).
• GAMMA-400 is a next-generation telescope with lateral and top-down detection, unique for its ability to register gamma-ray bursts (GRBs) from lateral directions using a 16 $n$9 CsI(Tl) calorimeter (CC2) and lateral detectors. Salient metrics:
  • Lateral effective area $m$0 m² per side ($m$1 steradian total FoV).
  • Energy resolution $m$210–15% (10–100 MeV), $m$32% (100 GeV).
  • Angular resolution $m$4 (100 MeV), improving to $m$5 (100 GeV).
  • Simulations predict up to 320 GRB detections/yr (lateral mode), facilitating high-statistics prompt and afterglow studies in the 10–100 MeV regime (Leonov et al., 2021).

Such missions generate critical high-energy datasets necessary for resolving the physical origin of cosmic gamma-ray emission and enabling cross-correlation with other spectral bands.

3. Gamma-Gamma and Gamma Factories: Photon Colliders and High-Flux Sources

“Gamma factories” and “gamma-gamma colliders” designate accelerator-based sources for the production of high-intensity, energy-tunable, quasi-monochromatic gamma-ray beams, often with the capability of producing secondary polarized particle beams:

• Gamma Factory (CERN): Utilizes resonant laser excitation of partially stripped ions (PSI) in the LHC to generate gamma rays with photon energies in the 1–400 MeV range. The upscattering process:

$m$6

where $m$7 is the ion Lorentz factor, $m$8 is incident photon energy, and $m$9 the normalized laser strength. With $\mathbf{L}\in\mathbb{R}^{m\times k}_{\geq 0}$0 and $\mathbf{L}\in\mathbb{R}^{m\times k}_{\geq 0}$1 kW, photon fluxes up to $\mathbf{L}\in\mathbb{R}^{m\times k}_{\geq 0}$2 $\mathbf{L}\in\mathbb{R}^{m\times k}_{\geq 0}$3/s are anticipated. Polished secondary beams—polarized $\mathbf{L}\in\mathbb{R}^{m\times k}_{\geq 0}$4, $\mathbf{L}\in\mathbb{R}^{m\times k}_{\geq 0}$5, cold neutrons—are key deliverables for future collider and neutrino-factory concepts.

• SAPPHiRE γγ Higgs Factory: Proposes a pair of 10 GeV recirculating linacs to create 80 GeV electrons, which collide with high-power lasers near the IP, Compton upscattering photons to achieve $\mathbf{L}\in\mathbb{R}^{m\times k}_{\geq 0}$6 GeV suitable for resonant Higgs production. The collider achieves:
  • Peak luminosity $\mathbf{L}\in\mathbb{R}^{m\times k}_{\geq 0}$7
  • $\mathbf{L}\in\mathbb{R}^{m\times k}_{\geq 0}$820,000 Higgs events/year (integrated luminosity $\mathbf{L}\in\mathbb{R}^{m\times k}_{\geq 0}$9120 fb$k$0)
  • Energy spectrum width $k$1 10%
  • Statistical precision: $k$22% ($k$3), $k$45% ($k$5), $k$68% ($k$7), 100 MeV mass scan (Bogacz et al., 2012, Krasny, 2015)

These facilities exploit either Compton-backscattering (electron+laser) or resonant scattering from PSI, offering unprecedented beam intensity and tunability for fundamental physics and applied research.

4. DAΦNE–GAΜMΑ: Storage-Ring Compton Gamma-Ray Source

The DAΦNE–GAΜMΑ project defines a storage-ring–based gamma factory using Compton backscattering of laser photons off a high-current ($k$8 A), low-emittance ($k$9 mm·mrad) electron ring and a %%%%50$n$051%%%% finesse Fabry–Pérot cavity ($\mathbf{S}\in\mathbb{R}^{n\times k}_{\geq 0}$237 kW stored power):

• Photon Energy: Tunable via electron energy (250–900 MeV) and laser wavelength (0.5–10 μm), yielding $\mathbf{S}\in\mathbb{R}^{n\times k}_{\geq 0}$3 in 2–9 MeV range.
• Key Source Metrics (benchmark case):
  • $\mathbf{S}\in\mathbb{R}^{n\times k}_{\geq 0}$4 ph/s, $\mathbf{S}\in\mathbb{R}^{n\times k}_{\geq 0}$5, spectral density $\mathbf{S}\in\mathbb{R}^{n\times k}_{\geq 0}$6 ph/s/eV.
  • Minimal perturbation to e-beam (energy spread/emittance) due to separation of collision and damping timescales.
  • Rapid energy tuning and compact layouts facilitate deployment in nuclear physics, radiology, and materials science.
• Comparison: Demonstrates flux and bandwidth competitive with or surpassing other sources (Duke HIGS, ELI-NP, Mega-Ray, IRIDE), at MHz repetition and continuous tunability (Alesini et al., 2014).

5. Computational Methods: GAMMA for Relativistic Blastwave Modeling

The GAMMA code represents a modern approach for simulating relativistic hydrodynamics and associated non-thermal emission, particularly in the context of gamma-ray burst afterglows:

• Scheme: Implements ALE (arbitrary Lagrangian–Eulerian) SRHD on a moving mesh—advecting along the dominant fluid direction to avoid mesh entanglement, maximizing local resolution at shocks, and enabling efficient large-scale blast wave evolution.
• Microphysics: Includes in situ shock detection, injection of power-law electron distributions ($\mathbf{S}\in\mathbb{R}^{n\times k}_{\geq 0}$7 for $\mathbf{S}\in\mathbb{R}^{n\times k}_{\geq 0}$8), local radiative cooling via synchrotron and inverse-Compton processes, and per-zone broadband synchrotron spectra.
• Astrophysical Impact: Demonstrates that the local treatment of synchrotron cooling results in a critical frequency shift (cooling break) by a factor $\mathbf{S}\in\mathbb{R}^{n\times k}_{\geq 0}$9 above predictions from global, spatially averaged models. The package provides validated, high-performance light-curve synthesis from early relativistic to late Newtonian regimes (Ayache et al., 2021).

6. Scientific and Industrial Applications

The ensemble of GaMMA-derived technologies and methods underpins a broad array of scientific and technological endeavors:

• High-energy astrophysics: Studies of cosmic-ray acceleration, non-thermal processes in supernova remnants, dark matter indirect detection, and gamma-ray burst phenomenology.
• Collider physics: Precision Higgs property measurements, electroweak quartic coupling studies, and searches for new physics via dedicated $$\min_{\mathbf{S}\geq 0}\ \frac{1}{2} \|\mathbf{X} - \mathbf{S}\mathbf{L}^\mathsf{T}\|^2_F + \frac{\lambda}{2}\|\mathbf{S}\|^2_F$$0 and $$\min_{\mathbf{S}\geq 0}\ \frac{1}{2} \|\mathbf{X} - \mathbf{S}\mathbf{L}^\mathsf{T}\|^2_F + \frac{\lambda}{2}\|\mathbf{S}\|^2_F$$1 collisions.
• Nuclear and materials science: Exploitation of high-brilliance, quasi-monochromatic gamma-ray sources for isotope production, waste transmutation, spectroscopy, neutron radiography, industrial tomography, and non-destructive materials testing.
• Spectroscopy and analytical chemistry: Real-time, robust analysis of multi-component spectra for research and industrial quality control across diverse modalities.

In all manifestations, GaMMA solutions strive for high performance, physical interpretability, and flexibility—leveraging real-time computation, high photon flux, and/or modular hybrid numerical strategies as appropriate for the application domain.

---

Key References:

• Supervised NMF for spectral analysis: "GAMMA_FLOW: Guided Analysis of Multi-label spectra by MAtrix Factorization for Lightweight Operational Workflows" (Rädle et al., 12 Nov 2025)
• High-energy astrophysics missions: "GAMMA-LIGHT: High-Energy Astrophysics above 10 MeV" (Morselli et al., 2014); "Capabilities of the GAMMA-400 gamma-ray telescope..." (Leonov et al., 2021)
• Gamma factories and colliders: "The Gamma Factory proposal for CERN" (Krasny, 2015), "SAPPHiRE: a Small Gamma-Gamma Higgs Factory" (Bogacz et al., 2012)
• Storage-ring Compton sources: "Daφne gamma-rays factory" (Alesini et al., 2014)
• Computational GRB modeling: "GAMMA: a new method for modeling relativistic hydrodynamics and non-thermal emission on a moving mesh" (Ayache et al., 2021)