wrinkle.uniri.hr

---

Summary

One of the most important open questions in physics is the description of gravity at the quantum scale. A possible discovery of quantum gravity (QG) would revolutionise science and technology and have an impact on society comparable to the discovery of quantum mechanics in the early 20th century. The basic problem of the search for QG is the lack of experimental measurements.

The focus of the proposed research is to measure the impact of phenomenological models of QG, based on the possible deviations from Lorentz symmetry (LIV), using very high-energy gamma-rays from astrophysical sources. The main goals of the research are (i) to develop an innovative method for data analysis based on simulations and machine learning and (ii) to compile a catalogue of astrophysical gamma-ray sources suitable for LIV research. An additional goal of the project will be preparation of applications for competitive research projects.

Team members from the University of Rijeka will work on the development of an analysis method based on artificial neural networks, using computer simulations derived from observations performed with the Fermi-LAT, CTAO-LST and MAGIC telescopes and on the properties of the future SWGO detector. Based on the same observations, we will establish a catalog of potentially usable gamma-ray sources. External collaborators are experts in the theory and phenomenology of LIV. With their help, we will determine the specific LIV models that will be tested and incorporate them in the analysis and the catalogue.

The project results will be published in relevant scientific journals in the field and at international scientific conferences.

---

Objectives

• Objective 1: Develop a data analysis method for the search for LIV effects, which includes the simultaneous effects predicted by theoretical models. The method is based on artificial intelligence and machine learning. Artificial neural networks (ANN) will be trained on computer simulations: the gamma-ray emission from astrophysical sources and the background radiation will be simulated in the part of the spectrum covered by the CTAO-LST, MAGIC and SWGO detectors, and the potential LIV effects predicted by various phenomenological models will be applied to the simulated gamma rays, creating statistical ensembles. Since the simulations have to cover a multidimensional parameter space, AI will also determine which parts of the parameter space need to be simulated, shortening the data generation and reducing the required processing time and storage. In the final step, AI will be used for the data analysis. Once fully developed, the method will be easily adaptable to other research that uses data from astronomical observations.
• Objective 2: Compile a catalogue of sources suitable for LIV research. The catalogue will be organised as a multidimensional parameter space, with parameters divided into three basic categories: (i) source properties (position, distance, mean flux or flux range, emission spectrum, highest energy at which the emission was detected, etc.), (ii) detector properties (location and visible part of the sky, energy range, sensitivity, etc.), and (iii) LIV properties (existing bounds on individual LIV effects and the fundamental LIV model from which each effect derives, which also accounts for the effects that occur simultaneously).
• Additional objective: Applications to calls for competitive research projects (HRZZ, ERC, etc.), with at least one application submitted during the project.

---

Team

• Tomislav Terzić (University of Rijeka) – principal investigator; leads the development of ANN as a LIV analysis method
• Filip Reščić (University of Rijeka) – doctoral student; leads the design of the catalogue of sources suitable for LIV research
• José Manuel Carmona (University of Zaragoza, Spain) – identifies which LIV effects appear simultaneously and determines their magnitudes, so that they can be built into the simulations used for ANN training, with emphasis on the changes in gamma-ray absorption under the influence of LIV; co-supervisor of Filip Reščić
• Giulia Gubitosi (University of Naples Federico II, Italy) – phenomenological modelling of various aspects of LIV, with emphasis on the influence on the gamma-ray propagation speed, and the influence of the choice of cosmological model on the measurement of LIV-induced delays of gamma-ray signals
• Christian Pfeifer (University of Rijeka) – calculation of the possible solutions of various LIV models and their parametrisation, so that they can be considered in the data analysis

---

Infrastructure and data

The research is mostly carried out at the Faculty of Physics of the University of Rijeka. Simulated data are generated and neural networks trained on the supercomputer Bura and the computing resources of the University Computing Centre in Zagreb (Srce). Gamma-ray observation data are obtained through membership in the international scientific collaborations CTAO-LST and MAGIC; data from the Fermi-LAT detector are publicly available.

---

Results and publications

Publications and other results are listed in the CRORIS project page.

---

News and outreach

• 20 May 2026: Tomislav Terzić, public talk “Nabori u prostor-vremenu” (Wrinkles in space-time) at the Pint of Science 2026 festival, Klub Palach, Rijeka.

---

AstroQG is funded by the University of Rijeka / NextGenerationEU research grant uniri-iz-25-119.
UNIRI web page of the project: https://portal.uniri.hr/ProjectsPU2025/Details/0187