Paul Richmond, Prarit Agarwal, Borun Chowdhury
et al.
We present specialized Large Language Models for theoretical High-Energy Physics, obtained as 20 fine-tuned variants of the 8-billion parameter Llama-3.1 model. Each variant was trained on arXiv abstracts (through August 2024) from different combinations of hep-th, hep-ph and gr-qc. For a comparative study, we also trained models on datasets that contained abstracts from disparate fields such as the q-bio and cs categories. All models were fine-tuned using two distinct Low-Rank Adaptation fine-tuning approaches and varying dataset sizes, and outperformed the base model on hep-th abstract completion tasks. We compare performance against leading commercial LLMs (ChatGPT, Claude, Gemini, DeepSeek) and derive insights for further developing specialized language models for High-Energy Theoretical Physics.
Maksym Ovchynnikov, Jean-Loup Tastet, Oleksii Mikulenko
et al.
The increasing interest in Long-Lived Particles (LLPs) has led to numerous proposed experiments in order to search for them. However, the sensitivity estimates published by these experiments tend to rely on disparate assumptions. To ensure an accurate comparison of their potential to find LLPs, a unified estimation of their sensitivity is therefore required. In this contribution, we introduce \texttt{SensCalc}, a \texttt{Mathematica}-based code that uses a semi-analytic approach to calculate the event rate of GeV-scale LLPs, and we present several case studies.
Ezequiel Alvarez, Federico Lamagna, Cesar Miquel
et al.
Current daily paper releases are becoming increasingly large and areas of research are growing in diversity. This makes it harder for scientists to keep up to date with current state of the art and identify relevant work within their lines of interest. The goal of this article is to address this problem using Machine Learning techniques. We model a scientific paper to be built as a combination of different scientific knowledge from diverse topics into a new problem. In light of this, we implement the unsupervised Machine Learning technique of Latent Dirichlet Allocation (LDA) on the corpus of papers in a given field to: i) define and extract underlying topics in the corpus; ii) get the topics weight vector for each paper in the corpus; and iii) get the topics weight vector for new papers. By registering papers preferred by a user, we build a user vector of weights using the information of the vectors of the selected papers. Hence, by performing an inner product between the user vector and each paper in the daily Arxiv release, we can sort the papers according to the user preference on the underlying topics. We have created the website IArxiv.org where users can read sorted daily Arxiv releases (and more) while the algorithm learns each users preference, yielding a more accurate sorting every day. Current IArxiv.org version runs on Arxiv categories astro-ph, gr-qc, hep-ph and hep-th and we plan to extend to others. We propose several new useful and relevant implementations to be additionally developed as well as new Machine Learning techniques beyond LDA to further improve the accuracy of this new tool.
The Deep Underground Neutrino Experiment (DUNE) could be revolutionary for MeV neutrino astrophysics, because of its huge detector volume, unique event reconstruction capabilities, and excellent sensitivity to the $ν_e$ flavor. However, its backgrounds are not yet known. A major background is expected due to muon spallation of argon, which produces unstable isotopes that later beta decay. We present the first comprehensive study of MeV spallation backgrounds in argon, detailing isotope production mechanisms and decay properties, analyzing beta energy and time distributions, and proposing experimental cuts. We show that above a nominal detection threshold of 5-MeV electron energy, the most important backgrounds are --- surprisingly --- due to low-A isotopes, such as Li, Be, and B, even though high-A isotopes near argon are abundantly produced. We show that spallation backgrounds can be powerfully rejected by simple cuts, with clear paths for improvements. We compare these background rates to rates of possible MeV astrophysical neutrino signals in DUNE, including solar neutrinos (detailed in a companion paper [Capozzi et al. arXiv:1808.08232 [hep-ph]]), supernova burst neutrinos, and the diffuse supernova neutrino background. Further, to aid trigger strategies, in the Appendixes we quantify the rates of single and multiple MeV events due to spallation, radiogenic neutron capture, and other backgrounds, including through pileup. Our overall conclusion is that DUNE has high potential for MeV neutrino astrophysics, but reaching this potential requires new experimental initiatives.
Supervised artificial neural networks with the rapidity-mass matrix (RMM) inputs were studied using several Monte Carlo event samples for various pp collision processes. The study shows the usability of this approach for general event classification problems. The proposed standardization of the ANN feature space can simplify searches for signatures of new physics at the LHC when using machine learning techniques. In particular, we illustrate how to improve signal-over-background ratios in searches for new physics, how to filter out Standard Model events for model-agnostic searches, and how to separate gluon and quark jets for Standard Model measurements.
Abstract A new method to measure the mass of the top quark at the LHC is presented [S. Kawabata, Y. Shimizu, Y. Sumino and H. Yokoya, arXiv:1405.2395 [hep-ph] ]. This method uses lepton energy distribution and ideally does not depend on the velocity distribution of the top quark. We perform a simulation analysis of the top quark mass reconstruction using this method at the leading order, taking account of experimental circumstances. We estimate the sensitivity of the mass determination. The results show that this method is viable in realistic experimental conditions and has a possibility to achieve a good accuracy in determining a theoretically well-defined top quark mass by including higher-order corrections.
Quark flavour conserving (QFC) fermionic squark decays, such as ~t_{1,2} -> t neutralino_i, are usually assumed in squark search analyses. Here we study quark flavour violating (QFV) bosonic squark decays, such as ~u_2 -> ~u_1 h^0/Z^0, where the mass eigenstates ~u_{1,2} are mixtures of scharm and stop quarks. We show that the branching ratios of such QFV decays can be very large due to sizable ~c_R - ~t_{R/L} and ~t_R - ~t_L mixing effects despite the very strong constraints on the QFV parameters from B meson data. This can result in remarkable QFV signatures with significant rates at LHC (14 TeV), such as pp -> gluino gluino X -> t c bar{c} bar{c} h^0/Z^0 missing-E_T X and pp -> gluino gluino X -> t t bar{c} bar{c} h^0/Z^0 missing-E_T X. The QFV bosonic squark decays can play an important role in the squark and gluino searches at LHC (14 TeV).
The 'landscape' of fundamental dynamics has changed even for the 'known' matter. The Standard Model has produced at least the leading source of CP violation (CPV) in B decays; the data have not shown CP asymmetries in $D$ transitions. It needs more data and better technologies to understand the underlying forces. Probing three- and four-body final states in $B$ & $D$ & $τ$ decays with better accuracy is crucial about the existence and the features of New Dynamics. Theoretical tools produced about MEP will show even more about HEP in the future. We have to work on the {\em correlations} between different final states on several CKM levels and the connection between known matter and Dark Matter in indirect ways. CPT invariance is usable in $D$ and $τ$ decays.