arXiv subjects

Suraj Prasad

Publications and source records attributed to Suraj Prasad.

At least 19 recordsLinked to original sources

arXiv2026

Evolution of the underlying event and non-extensive thermodynamics in pp collisions from RHIC to LHC energies

We investigate the geometrical and thermodynamical aspects of particle production in pp collisions over a broad center-of-mass energy, ranging from $\sqrt{s}=62.4$ GeV to 13 TeV using PYTHIA8. We study the azimuthal dependence of charged-particle production relative to the leading particle. This includes the study of the Tsallis--Pareto parameters extracted from the charged-particle transverse-momentum spectra in fixed $\Delta\phi$ intervals relative to the leading charged particle. The simultaneous analysis of charged-particle multiplicity and event topology demonstrates that the underlying event (UE) dominated region can be extended beyond the conventional transverse side region across RHIC and LHC energies. To further disentangle effects from soft and hard particle production mechanisms, the study is performed in different transverse spherocity and charged-particle flattenicity classes. We demonstrate that the UE region extends beyond the conventional transverse region, validating an enlarged angular interval of $40^\circ\lesssim|\Delta\phi|\lesssim140^\circ$ from RHIC to LHC energies. Event-shape selections reveal that jetty events consistently exhibit larger $q$ and lower $T_{\rm s}$ than isotropic events. The $\Delta\phi$ profiles of charged-particle multiplicity and transverse spherocity remain universal in shape across collision energies despite the strong increase in overall event activity, whereas the Tsallis parameters retain a residual, leading-particle-driven energy dependence in the near- and away-side regions.

hep-ph↗

arXiv2026

Constraining $\alpha$-cluster compactness in $^{16}\rm O$ and $^{20}\rm Ne$ at TeV energies using azimuthal anisotropy

Anisotropic flow in ultra-relativistic light-ion collisions is sensitive to the initial geometry of the colliding nuclei. We investigate whether elliptic flow measurements can constrain the parameters of the proposed $\alpha$-clustered nuclear density distributions of $^{16}$O and $^{20}$Ne at LHC energies. Using the hybrid framework IP-Glasma+MUSIC+iSS+UrQMD, we simulate OO and Ne--Ne collisions at $\sqrt{s_{\mathrm{NN}}}=5.36$ TeV for the Woods--Saxon and $\alpha$-clustered configurations with varying cluster compactness. The elliptic flow coefficient $v_2\{2,|\Delta\eta|>1\}$ is calculated in the kinematic acceptances of ALICE, CMS, and ATLAS detectors and is compared with the Run~3 OO and Ne--Ne experimental measurements. It is observed that the final-state elliptic flow is significantly sensitive to the nuclear geometry, especially in OO collisions, where different configurations lead to distinct centrality dependencies and peak positions of $v_{2}$. By performing a systematic variation of the cluster size and inter-cluster separation in $^{16}$O and $^{20}$Ne nuclei, this work attempts to identify the cluster parameter range that provides the best agreement with the experimental data. These results show that the flow observables in TeV-energy light-ion collisions can be used to optimize the nuclear structure parameters of light nuclei.

hep-ph↗

arXiv2026

MAGE-HEP: Monte Carlo Analysis and Graphical Environment for High-Energy Physics

Monte Carlo event generators are central to high-energy physics analysis. However, workflows based on handwritten scripts can be difficult to reuse, modify, and reproduce when multiple Monte Carlo models, tune variations, run variations, and output formats are involved. We present MAGE-HEP, short for Monte Carlo Analysis and Graphical Environment for High-Energy Physics, a Graphical User Interface (GUI) driven workflow environment for reproducible Monte Carlo-based analyses in high-energy physics. MAGE-HEP organizes analysis workflows through a project-study-run hierarchy. The project stores the workspace, the study stores the reusable analysis context, and each run represents a controlled execution of that context. The MAGE-HEP Node API provides the analysis-building layer for defining generator configurations, observables, selections, output rules, and generated C++/ROOT analysis code. A study context can be inspected, reused, or exported as a \texttt{.mcx} context bundle, while the project state can be exported as a portable \texttt{.mgp} bundle. The current beta implementation validates the core idea using a PYTHIA8 and ROOT workflow. It includes background execution, manifest-based run tracking, live ROOT inspection, and particle-table summaries for supported output layouts. This paper describes the architecture, workflow, and current beta implementation of MAGE-HEP.

hep-ph↗

arXiv2026

Inferring identified hadron production in $pp$ collisions with physics-informed machine learning at the LHC

Machine learning has become a powerful tool in high-energy collider experiments, which enables the studies based on data-driven approaches to complex reconstruction and regression tasks. The study of identified hadron spectra in pseudorapidity regions beyond detector acceptance, which is limited to mid-rapidity regions, carries important information about particle production, yet remains unmeasured. In this work, we develop a physics-informed neural network, trained on PYTHIA8 $pp$ collisions at $\sqrt{s}=13.6$ TeV, to infer $p_{\rm T}$ spectra of $π^{\pm}$, $K^{\pm}$, $p/\bar{p}$, $Λ/\barΛ$, and $K^{0}_{\mathrm{s}}$ in different rapidity regions. Physics-motivated constraints, including particle yield ratios, spectral shape, and smoothness, are incorporated into the loss function. A staged hyperparameter optimization strategy is used to ensure stability. The model achieves yield uncertainties of ${\sim}1.5\%$, $1.8\%$, and $5.83\%$ in the training, interpolation, and extrapolation regimes, respectively, outperforming XGBoost and LightGBM. It further reproduces key observables such as particle yield ratios, the multiplicity dependence of $\langle p_{\rm T} \rangle$, and kinetic freeze-out parameters, indicating that the model captures the underlying physics and provides reliable predictions beyond the measured phase space.

hep-ph↗

arXiv2026

Nuclear geometry driven symmetry plane correlations in OO and Ne--Ne collisions at the Large Hadron Collider

Symmetry-plane correlations (SPCs) are key observables sensitive to the medium's transport properties and are driven by participant-plane correlations (PPCs) in the nuclear overlap region. This study explores the possibility of nuclear-geometry-driven SPCs in Oxygen--Oxygen (OO) and Neon--Neon (Ne--Ne) collisions at $\sqrt{s_{\rm NN}}=5.36$ TeV using nuclear geometry simulations based on Nuclear Lattice Effective Field Theory (NLEFT) and Projected Generator Coordinate Method (PGCM) configurations. We investigate $\langle \cos[4(ψ_2 - ψ_4)]\rangle_{\rm GE}$ and $\langle \cos[6(ψ_3 - ψ_6)]\rangle_{\rm GE}$ in OO and Ne--Ne collisions at $\sqrt{s_{\rm NN}}=5.36$ TeV using the A Multi-Phase Transport (AMPT) model. We find that Ne--Ne collisions exhibit larger $\langle \cos[4(ψ_2 - ψ_4)]\rangle_{\rm GE}$ values than OO collisions, whereas $\langle \cos[6(ψ_3 - ψ_6)]\rangle_{\rm GE}$ is larger in OO than in Ne--Ne collisions. This behavior indicates a strongly deformed shape of the $^{20}$Ne nucleus and a tetrahedral structure of the $^{16}$O nucleus. We also explore SPCs for events with tip-tip and body-body collision configurations, which further support these findings.

hep-ph↗

arXiv2026

Event Topology Classifiers at the Large Hadron Collider

Event classifiers are the most fundamental observables to probe the event topology of hadronic and nuclear collisions at relativistic energies. Over the last five decades, significant progress has been made to establish suitable event classifiers to probe different physics processes occurring in elementary $e^{+}e^{-}$ to heavy-ion collisions in a broad range of center of mass energies. One of the major motivations to revisit event classifiers at the Large Hadron Collider (LHC) originates from the recent measurements of high multiplicity proton-proton collisions, which have revealed that these small collision systems exhibit features similar to the formation of quark-gluon plasma (QGP), traditionally believed to be only achievable in heavy nucleus-nucleus collisions at ultra-relativistic energies. To pinpoint the origin of these QGP-like phenomena with substantially reduced autocorrelation and selection biases, and to bring all collision systems on equal footing, along with charged-particle multiplicity, lately several event topology classifiers such as transverse sphericity, transverse spherocity, relative transverse activity classifier, and charged-particle flattenicity have been used extensively in experiments as well as in the phenomenological front. In addition, the infrared and collinear safety of event-shape observables makes them ideal for precision studies of jets and heavy-flavors at the LHC. In this review article, we summarise the motivation, scope, and practical use of these event-shape observables. The discussion integrates results and insights from all major LHC experiments, setting the stage for precision investigations for Run 3, Run 4, and future high luminosity upgrades of the LHC.

hep-ph↗

arXiv2026

SynthPID: P&ID digitization from Topology-Preserving Synthetic Data

Automating the digitization of Piping and Instrumentation Diagrams (P&IDs) into structured process graphs would unlock significant value in plant operations, yet progress is bottlenecked by a fundamental data problem: engineering drawings are proprietary, and the entire community shares a single public benchmark of just 12 annotated images. Prior attempts at synthetic augmentation have fallen short because template-based generators scatter symbols at random, producing graphs that bear little resemblance to real process plants and, accordingly, yield only approximately 33% edge detection accuracy under synth-only training. We argue the failure is structural rather than visual and address it by introducing SynthPID, a corpus of 665 synthetic P&IDs whose pipe topology is seeded directly from real drawings. Paired with a patch-based Relationformer adapted for high-resolution diagrams, a model trained on SynthPID alone achieves 63.8 +/- 3.1% edge mAP on PID2Graph OPEN100 without seeing a single real P&ID during training, closing within 8 pp of the real-data oracle. These gains hold up under a controlled comparison against the template-based regime, confirming that generation quality drives performance rather than model choice. A scaling study reveals that gains flatten beyond roughly 400 synthetic images, pointing to seed diversity as the binding constraint.

cs.CV↗

arXiv2026

Thermodynamic and Transport Properties of Quark-Gluon Plasma at Finite Chemical Potential with a DNN framework

The characteristics of a thermal system depend strongly on its response to thermal gradients and the underlying microscopic interactions among constituents. In the present study, we investigate the thermodynamic and transport properties of the quark-gluon plasma (QGP) at finite baryon chemical potential within a deep-learning-assisted quasi-particle model (DLQPM). The temperature ($\mathrm{T}$) and baryon chemical potential ($μ_B$)-dependent thermal masses of quasi-particles are estimated using neural networks trained to reproduce lattice QCD (lQCD) results for the equation of state, obtained via a Taylor-like expansion around vanishing baryon chemical potential. The trained model acts as an effective emulator, enabling us to estimate the thermodynamic and transport properties at finite $μ_B$. We compute the speed of sound, specific heat, viscosity, and conductivity of the deconfined medium. Our findings are in good agreement with available lattice calculations and other phenomenological models. The present study demonstrates that a DNN-based approach provides an efficient framework for studying the properties of the QGP at finite baryon density.

hep-ph↗

arXiv2026

Speech-Synchronized Whiteboard Generation via VLM-Driven Structured Drawing Representations

Creating whiteboard-style educational videos demands precise coordination between freehand illustrations and spoken narration, yet no existing method addresses this multimodal synchronization problem with structured, reproducible drawing representations. We present the first dataset of 24 paired Excalidraw demonstrations with narrated audio, where every drawing element carries millisecond-precision creation timestamps spanning 8 STEM domains. Using this data, we study whether a vision-language model (Qwen2-VL-7B), fine-tuned via LoRA, can predict full stroke sequences synchronized to speech from only 24 demonstrations. Our topic-stratified five-fold evaluation reveals that timestamp conditioning significantly improves temporal alignment over ablated baselines, while the model generalizes across unseen STEM topics. We discuss transferability to real classroom settings and release our dataset and code to support future research in automated educational content generation.

cs.CV↗

arXiv2026

Topological production of charmonia with event-shape engineering in $pp$ collisions at $\sqrt{s} = 13$ TeV using PYTHIA8

The production of heavy quarks (charm and beauty) in high-energy hadronic and nuclear collisions provides an excellent testing ground for the theory of strong interactions and validates models based on quantum chromodynamics (QCD). In this work, prompt and nonprompt production of $\rm{J/}ψ$ in $pp$ collisions at $\sqrt{s}=13$ TeV are studied as a function of transverse spherocity using PYTHIA8. $\rm{J/}ψ$ is reconstructed via its electromagnetic decay to dielectrons and dimuons, in mid- and forward-rapidity, respectively. Transverse spherocity, an event shape observable, is used to distinguish hard QCD events from the softer, isotropic ones. In PYTHIA8, the production of $\rm{J/}ψ$ can be influenced by the average number of multiple parton interactions ($\langle N_{\rm mpi} \rangle$), owing to the underlying events (UE), which have a dominant contribution to particle production at lower transverse momentum. Since transverse spherocity is correlated to $\langle N_{\rm mpi} \rangle$, this can serve as an experimentally accessible tool for event selection to study the underlying QCD processes influencing the prompt and nonprompt $\rm{J/}ψ$ production. This study reveals the correlation between heavy-flavor production dynamics and topological event selection in $pp$ collisions using PYTHIA8, whose relevance awaits experimental validation.

hep-ph↗

arXiv2026

Probing the microscopic origin of prompt and non-prompt $D^{0}$ production through event-shape engineering in proton-proton collisions at the LHC

Heavy-flavour hadrons are produced in the early stages of ultra-relativistic collisions at the LHC via hard partonic interactions and experience the whole system evolution. The study of prompt and non-prompt $D^{0}$ mesons provides an independent avenue to test the theories of quantum chromodynamics and to investigate beauty hadron production. Moreover, the production of both prompt and non-prompt $D^{0}$ is influenced by microscopic processes such as multi-partonic interactions (MPI) and hadronisation through fragmentation. In this study, an attempt is made to understand the production of prompt and non-prompt $D^{0}$ mesons in proton-proton collisions at $\sqrt{s}=13.6$ TeV using the PYTHIA8 event generator, which offers a qualitative description of charm production. The role of the transverse momentum transfer in the hardest partonic scattering ($\hat{p}_{\rm T}$), MPI, and color reconnection is systematically explored. In addition, the charged particle production in different topological regions with respect to the leading $D^{0}$ meson is studied to assess the influence of the $D^{0}$ meson on the event topology and to examine the selection biases arising from the use of charged particle multiplicity as an event classifier.

hep-ph↗

arXiv2025

Event-shape dependence of symmetry plane correlations using the Gaussian estimator in Pb-Pb collisions at the LHC using a multiphase transport model

The study of symmetry plane correlations (SPCs) can be useful in characterizing the direction of the anisotropic emission of produced particles in the final state. The study of SPCs provides an independent method to understand the transport properties of the system formed in heavy-ion collisions. Similar to anisotropic flow coefficients, which are largely influenced by the initial spatial anisotropy, SPCs also depend upon the participant plane correlations measured using the participating nucleons of the collision overlap region. In this paper, SPCs have been studied in Pb-Pb collisions at $\sqrt{s_{\rm NN}}=5.02$ TeV using the event generator AMPT. In addition to their behaviour with the changing centrality of the collision, their event shape dependence has also been studied for the first time, using the event shape classifier transverse spherocity. The Gaussian estimator has been used to evaluate the correlations, and these have been compared to the participant plane correlations defined in an analogous way to the symmetry plane correlations, and a qualitative match has been found between them. These event-shape differentiated symmetry plane correlations can be used to deduce the presence of higher-order anisotropies in the initial energy distribution, thus giving insight into the initial geometry of the colliding system, among other applications like model development and model testing using Bayesian analyses.

nucl-ex↗

arXiv2025

Replication Study: Federated Text-Driven Prompt Generation for Vision-Language Models

Vision-language models like CLIP have demonstrated remarkable zero-shot capabilities, yet their adaptation to federated learning scenarios presents significant challenges, particularly regarding generalization to unseen classes. The original FedTPG paper \cite{Qiu2024} addresses this limitation by introducing a text driven prompt generation network that dynamically creates prompts conditioned on class names, enabling better cross-class generalization in federated settings. In this work, we present a faithful replication study of FedTPG, evaluating the pre-trained model on six diverse vision datasets: Caltech101, Oxford Flowers, FGVC Aircraft, Oxford Pets, Food-101, and DTD. Our evaluation achieves results within 0.2\% of the original paper's reported accuracies, with an average accuracy of 74.58\% on seen (base) classes and 76.00\% on unseen (new) classes, demonstrating a +1.43 percentage point improvement in generalization. These results validate the original paper's core claims: (1) text-driven prompt generation enables superior generalization to unseen classes compared to static prompt learning methods, and (2) federated training of prompt generators maintains high performance across diverse visual domains without sharing private data. Our successful replication confirms the robustness and reproducibility of the FedTPG approach.

cs.CV↗

arXiv2025

Probing the sensitivity of anisotropic flow coefficients to the initial nuclear structure in pO and OO collisions at the LHC

RHIC and LHC have injected $^{16}\rm O$ nuclei in their accelerator complexes with a focus on investigating collectivity and the origin of quark-gluon plasma signatures in small collision systems. The $^{16}\rm O$ nuclei are known to possess clusters of $α$-particles ($^{4}\rm He$) inside the nucleus. This paper attempts to study the clustered-nuclear-geometry dependence of anisotropic flow coefficients such as elliptic flow ($v_2$) and triangular flow ($v_3$), which are sensitive to the nuclear geometry of colliding nuclei. The study is performed in pO and OO collisions at $\sqrt{s_{\rm NN}}=9.61$~TeV and 7~TeV respectively, employing a hybrid model encompassing IP-Glasma + MUSIC + iSS + UrQMD. The results of the clustered nuclear geometry are compared with those of the Woods\,--\,Saxon nuclear profile. Both initial and final state anisotropies are estimated. This study is thus one of its first kind, where the study of anisotropic flow coefficients for pO and OO collisions is presented using a hybrid hydrodynamics model. While the effect of $α$-clustering in pO is found to be small, it is significant for each observable studied in OO collisions. It is also observed that the magnitude of this effect has a noteworthy dependence on the size of the \textsuperscript{4}He.

hep-ph↗

arXiv2025

Machine learning driven identification of heavy flavor decay leptons in proton-proton collisions at the Large Hadron Collider

The study of heavy-flavor hadrons is topical in the era of precision measurements, which is useful to test theories based on pQCD. The heavy-flavor hadrons are produced initially during heavy-ion or hadronic collisions and are one of the best probes to understand the initial stages of the collisions as well as the system evolution. In experiments, the heavy-flavor sectors are studied directly via their decay to different hadrons or di-leptons or via their semi-leptonic decay, which is accompanied by additional neutrinos. However, their measurement in experiments is resource-intensive and requires input from different Monte-Carlo event generators. In this study, we provide an independent method based on Machine Learning algorithms to separate such leptons coming from heavy-flavor semi-leptonic decays. We use PYTHIA8 to generate events for this study, which gives a good qualitative and quantitative description of heavy-flavor production in $pp$ collisions. We use the XGBoost model for this study, which is trained with $pp$ collisions at $\sqrt{s}=13.6$~TeV. We use \DCAXY, \DCAZ~and pseudo-rapidity as the input to the machine. The ML model provides an accuracy of 98\% for heavy-flavor decay electrons and almost 100\% for heavy-flavor decay muons.

hep-ph↗

arXiv2025

Federated Cross-Modal Style-Aware Prompt Generation

Prompt learning has propelled vision-language models like CLIP to excel in diverse tasks, making them ideal for federated learning due to computational efficiency. However, conventional approaches that rely solely on final-layer features miss out on rich multi-scale visual cues and domain-specific style variations in decentralized client data. To bridge this gap, we introduce FedCSAP (Federated Cross-Modal Style-Aware Prompt Generation). Our framework harnesses low, mid, and high-level features from CLIP's vision encoder alongside client-specific style indicators derived from batch-level statistics. By merging intricate visual details with textual context, FedCSAP produces robust, context-aware prompt tokens that are both distinct and non-redundant, thereby boosting generalization across seen and unseen classes. Operating within a federated learning paradigm, our approach ensures data privacy through local training and global aggregation, adeptly handling non-IID class distributions and diverse domain-specific styles. Comprehensive experiments on multiple image classification datasets confirm that FedCSAP outperforms existing federated prompt learning methods in both accuracy and overall generalization.

cs.CV↗

arXiv2025

Higher order flow coefficients -- A Messenger of QCD medium formed in heavy-ion collisions at the Large Hadron Collider

Anisotropic flow and fluctuations are sensitive observables of the initial state effects in heavy ion collisions and are characterized by the medium properties and final state interactions. Using event-shape observables, one can constrain the probability distributions of anisotropic flow coefficients, thus reducing the linear and nonlinear contributions in the measured higher-order harmonics. In this paper, we use transverse spherocity as an event shape observable to study the flow coefficients and elliptic flow fluctuations. Transverse spherocity is found to have a strong correlation with elliptic flow and its fluctuations. We exploit this feature of transverse spherocity to remove the contribution to elliptic flow from higher-order harmonics. The study is performed in Pb--Pb collisions at $\sqrt{s_{\rm NN}}=5.02$ TeV using a multi-phase transport model. The multi-particle Q-cumulant method estimates the anisotropic flow coefficients, which reduces the non-flow contributions. We observe a stronger system response to the flow coefficients for the events with smaller values of elliptic flow.

nucl-th↗

arXiv2025

Role of clustered nuclear geometry in particle production through p-C and p-O collisions at the Large Hadron Collider

Long-range multi-particle correlations in heavy-ion collisions have shown conclusive evidence of the hydrodynamic behavior of strongly interacting matter and are associated with the final-state azimuthal momentum anisotropy. In small collision systems, azimuthal anisotropy can be influenced by the hadronization mechanism and residual jet-like correlations. Thus, one of the motives of the planned p--O and O--O collisions at the LHC and RHIC is to understand the origin of small system collectivity. As the anisotropic flow coefficients ($v_n$) are sensitive to the initial-state effects including nuclear shape, deformation, and charge density profiles, studies involving $^{12}$C and $^{16}$O nuclei are transpiring due to the presence of exotic $α$ ($^{4}$He) clusters in such nuclei. In this study, for the first time, we investigate the effects of nuclear $α$--clusters on the azimuthal anisotropy of the final-state hadrons in p--C and p--O collisions at $\sqrt{s_{\rm NN}}= 9.9$~TeV within a multi-phase transport model framework. We report the transverse momentum ($p_{\rm T}$) and pseudorapidity ($η$) spectra, participant eccentricity ($ε_2$) and triangularity ($ε_3$), and estimate the elliptic flow ($v_2$) and triangular flow ($v_3$) of the final-state hadrons using the two-particle cumulant method. These results are compared with a model-independent Sum of Gaussians (SOG) type nuclear density profile for $^{12}$C and $^{16}$O nuclei.

nucl-th↗