Systematic optimization of resonance parameters in a transport approach
Carl B. Rosenkvist 1,2,3 Hannah Elfner 3,1,2,4 · 2025
Read Systematic optimization of resonance parameters in a transport approach by Carl B. Rosenkvist 1,2,3 Hannah Elfner 3,1,2,4 (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-05504.txt.
Introduction
Carl Rosenkvist and Hannah Elfner tuned a hadron transport model in two thousand twenty-five with a genetic algorithm, a search method borrowed from evolution. Their model, SMASH, simulates strongly interacting particles, and its resonance settings govern how much strangeness it produces. Hand tuning those many settings is slow and imprecise.
They fit the settings to measured cross-sections for strangeness production at low energies, where data are exclusive and detailed. The tuned model matched experiments far better, above all for pion-proton collisions. The authors present the work as a case for machine learning tools in the careful calibration of transport models.
Catalog record
| CREATOR | Carl B. Rosenkvist 1,2,3 Hannah Elfner 3,1,2,4 |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 29 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-05504
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Systematic optimization of resonance parameters in a transport approach Carl B. Rosenkvist 1,2,3 Hannah Elfner 3,1,2,4 Abstract This study optimizes resonance parameters responsible for strangeness production in the SMASH (Simulating Many Accelerated Strongly-interacting Hadrons) transport model using a genetic algorithm. By fitting resonance parameters to experimental data on exclusive strangeness cross-sections at low energies, we significantly improve the model’s accuracy, especially in pion-proton interactions. Our approach explores how machine learning tools can be used for precise resonance tuning in transport approaches. I Introduction One of the primary goals of heavy-ion collisions is to explore the quantum chromodynamics (QCD) phase diagram, which describes the behavior of strongly interacting matter under extreme conditions of temperature and density. Both theoretical and experimental studies suggest the existence of a phase, known as the quark-gluon plasma (QGP), where quarks and gluons are nearly deconfined Harris and Müller 1996 . This state of matter is expected to emerge at high temperatures and/or densities, providing a unique environment to study the fundamental properties of QCD. Strangeness plays a crucial role in this investigation because it is conserved under the strong interaction and must have been created during the collision. Its presence can offer significant insights into the properties of the quark-gluon plasma (QGP). Various models have been developed to describe strange quark production under different energy and density conditions. At higher energies, event generators such as Pythia/Angantyr Bierlich et al. 2022 ; Bierlich et al. 2018 utilize the Lund string fragmentation model to simulate hadronization, including the production of strange quarks. Recently, the string framework has been extended with rope hadronization Bierlich 2018 , a feature added to these models to address strangeness enhancement. In contrast, models that focus on the thermodynamics of heavy-ion collisions, such as hydrodynamic and statistical hadronization models, describe strange quark production through notably gluon interactions Rafelski and Müller 1982 . These approaches focus on the collective behavior of the medium rather than on individual hadronization processes. At lower energies, where hadronic dynamics dominate, transport approaches provide a more suitable description of the observed phenomena. For example, the Parton-Hadron-String Dynamics (PHSD) approach, which models both nucleon resonances and color-dipoles (strings) Bratkovskaya et al. , incorporates in-medium effects on strangeness production. This is achieved through modifications to the antikaon potential, spectral functions, and reaction cross sections, which depend on baryon density, temperature, and antikaon momentum in the nuclear medium Song et al. 2021 . This approach has emphasized the significance of in-medium effects on strangeness in nuclear matter. Complementing the medium effects, the vacuum properties of resonances, such as pole positions and branching ratios, also play a significant role in strangeness production. In transport models like Ultra-relativistic Quantum Molecular Dynamics (UrQMD), strangeness production at lower energies is included through resonance decay Bass 1998 . These resonances are modeled by their vacuum properties tuned to experimental data. A newer transport model, Simulating Many Accelerated Strongly-interacting Hadrons (SMASH), shares similarities with UrQMD but differs in the selection of resonances and resonance parameters Wergieluk et al. 2024 . At higher energies, both SMASH and UrQMD employ string fragmentation handled by Pythia. However, resonance properties, such as branching ratios, are often poorly constrained by experimental data, leading to uncertainties in model predictions. This is demonstrated in the UrQMD study Gerhard et al. 2012 , which investigated model stability under variations in resonance mass and widths. Their findings showed that at lower energies, where resonance dynamics dominate, transport models are sensitive to resonance parameters. In SMASH, resonance properties have been tuned to experimental data Steinberg et al. 2019 , but a systematic approach to investigate different parameter sets has not yet been applied. To address this challenge, we propose a more systematic approach by leveraging experimental data on exclusive strangeness cross-sections at low energies to fit the resonance parameters. In this paper, we utilize a genetic algorithm for parameter optimization and apply it within the SMASH transport model to simulate the cross-sections and refine the resonance properties. The remainder of this paper details our simulation set-up, including the use of SMASH and the development of an emulation method. We then describe the application of genetic algorithms to optimize resonance parameters. This is followed by a presentation of our results, where we compare simulation outcomes with experimental data. We discuss the implications of our findings and suggest directions for future research. Finally, we conclude by summarizing the key contributions and significance of our work. II Resonance production in SMASH SMASH is a relativistic hadron transport approach for heavy-ion collisions, incorporating resonances up to a mass of \sim 2\,\text{GeV} as degrees of freedom at lower energies. At higher energies, strings are formed and then fragmented with Pythia . In SMASH, resonances are modeled by a spectral function \mathcal{A} , represented as a relativistic Breit-Wigner distribution. \mathcal{A}(m)=\frac{m^{2}\Gamma(m)}{(m^{2}-m_{0}^{2})^{2}+m^{2}\Gamma(m)^{2}} (1) Where m_{0} is the pole mass and \Gamma(m) is the mass-dependent width. The Manley et al. Manley and Saleski 1992 approach is used to describe the mass-dependent widths as \Gamma_{R\to a+b}(m)=\Gamma^{0}_{R\to a+b}\frac{\rho_{ab}(m)}{\rho_{ab}(m_{0})} (2) where \displaystyle\rho_{ab}(m)=\int dm_{a}dm_{b}\mathcal{A}_{a}(m_{a})\mathcal{A}_{a}(m_{b})\frac{\lvert\vec{p}_{f}\rvert}{m} \displaystyle\crossproduct B_{L}(\lvert\vec{p}_{f}\rvert R)\mathcal{F}_{ab}^{2}(m). (3) B_{L} is the Blatt-Weisskopf function, L is the orbital angular momentum of the decay channel, R is the interaction length and \mathcal{F}_{ab} is a form factor for unstable decay products. The on-shell parameters, branching ratio \Gamma^{0}_{R\to a+b}/\Gamma^{0}_{\text{tot}} , mass m_{0} and width at the pole is taken from the Particle Data Group (PDG) Navas et al. 2024 . III Emulation Transport codes require significant CPU resources, especially for cross-sections that demand extensive statistics, such as those involving strangeness. Therefore, emulation of transport simulation is preferred. Emulation, in this context, means creating a simplified model that approximates the behavior of the full simulation. A straightforward emulation of strangeness production via resonances involves multiplying the resonance cross-section by the respective branching ratio. By summing all possible resonances, one obtains the portion of the strangeness cross-section due to resonances. Since SMASH uses mass-dependent partial widths, the branching ratio must be integrated over the possible mass values along with the spectral function. Strings are not affected by resonance parameters; thus, the string cross-section can be simulated once and then added to the emulation. All of the above considerations lead to the following expression: \displaystyle\sigma_{d+i} \displaystyle\approx\sigma_{\text{string}\to d+i}+\sum_{N^{*}}\sigma_{N^{*}}\left\langle\frac{\Gamma_{N^{*}\to d}}{\Gamma_{N^{*}}}\right\rangle (4) \displaystyle\left\langle\frac{\Gamma_{N^{*}\to d}}{\Gamma_{N^{*}}}\right\rangle \displaystyle=\int_{m_{d}}^{\sqrt{s}-m_{i}}\differential mA(m)\frac{\Gamma_{N^{*}\to d}(m)}{\Gamma_{N^{*}}(m)} (5) where d and i stand for decay products and spectator, respectively. It is important to note if the spectral function is not normalized to the integration interval then the integral in eq. 4 needs to be divided by \mathcal{N}=\int_{m_{d}}^{\sqrt{s}-m_{i}}\differential mA(m) (6) which ensures that the correct mean for the branching ratio is calculated. For pion-proton collisions that form a resonance through a two-to-one reaction, the mass is not sampled for the mass-dependent branching ratio because the mass is set to \sqrt{s} due to kinematics. IV Data Selection & Parameter Bounds To exclude secondary collisions and other interfering strangeness production interactions, exclusive proton-proton and pion-proton cross-sections were used. The data were collected from the same sources as in Steinberg et al. 2019 , which include Balewski et al. 1998 ; Sewerin et al. 1999 ; Kowina et al. 2004 ; El-Samad et al. 2010 ; Abdel-Bary and others COSY-TOF ; Bilger et al. 1998 ; Abdel-Samad and others COSY-TOF ; Baldini 1988 ; Valdau et