arXiv subjects

Suman Deb

Publications and source records attributed to Suman Deb.

At least 19 recordsLinked to original sources

arXiv2026

AgentRadio: Passive Awareness for Long-Horizon Multi-Agent Collaboration

Understanding large codebases is a long-horizon task for Large Language Model (LLM) agents: answering a single question can require building and running the software, tracing execution across files, and synthesizing evidence over tens of minutes. On SWE-Atlas QnA, a benchmark of long-horizon questions over production repositories, a single Claude Code agent (Opus 4.6) resolves only 32.3% of tasks. Dividing the work among agents with clean contexts mitigates this limitation. However, the subtasks of code comprehension are interdependent. One agent's findings can rewrite another's task, so agents must coordinate during execution, not only at phase boundaries. Existing multi-agent systems support such exchange only between phases, through staged handoffs or synchronized rounds. Communication and work remain mutually exclusive. A discovery made mid-execution cannot be shared until the next boundary. We present AgentRadio, an asynchronous message-passing layer that equips coding-agent harnesses with three primitives: threads, messages, and waiting for mentions. The last runs as a background task, surfacing teammates' messages without interrupting foreground work, so each agent remains passively aware of its peers and folds new findings into its ongoing task. Under a five-phase protocol of division of labor and negotiation, four agents organized by AgentRadio resolve 62.1% of tasks, 29.8 points above a single agent and above Claude Code with the newer Opus 4.8 (57.2%). Rubric-level analysis shows the gain growing with task difficulty, consistent with mid-course correction as the underlying mechanism. Our code is available at https://github.com/Coral-Protocol/AgentRadio.

cs.MA↗

arXiv2026

Beyond Rule-Based Workflows: An Information-Flow-Orchestrated Multi-Agents Paradigm via Agent-to-Agent Communication from CORAL

Most existing Large Language Model (LLM)-based Multi-Agent Systems (MAS) rely on predefined workflows, where human engineers enumerate task states in advance and specify routing rules and contextual injections accordingly. Such workflow-driven designs are essentially rule-based decision trees, which suffer from two fundamental limitations: they require substantial manual effort to anticipate and encode possible task states, and they cannot exhaustively cover the state space of complex real-world tasks. To address these issues, we propose an Information-Flow-Orchestrated Multi-Agent Paradigm via Agent-to-Agent (A2A) Communication from CORAL, in which a dedicated information flow orchestrator continuously monitors task progress and dynamically coordinates other agents through the A2A toolkit using natural language, without relying on predefined workflows. We evaluate our approach on the general-purpose benchmark GAIA, using the representative workflow-based MAS OWL as the baseline while controlling for agent roles and underlying models. Under the pass@1 setting, our method achieves 63.64% accuracy, outperforming OWL's 55.15% by 8.49 percentage points with comparable token consumption. Further case-level analysis shows that our paradigm enables more flexible task monitoring and more robust handling of edge cases. Our implementation is publicly available at: https://github.com/Coral-Protocol/Beyond-Rule-Based-Workflows

cs.AI↗

arXiv2025

Efficient and Encrypted Inference using Binarized Neural Networks within In-Memory Computing Architectures

Binarized Neural Networks (BNNs) are a class of deep neural networks designed to utilize minimal computational resources, which drives their popularity across various applications. Recent studies highlight the potential of mapping BNN model parameters onto emerging non-volatile memory technologies, specifically using crossbar architectures, resulting in improved inference performance compared to traditional CMOS implementations. However, the common practice of protecting model parameters from theft attacks by storing them in an encrypted format and decrypting them at runtime introduces significant computational overhead, thus undermining the core principles of in-memory computing, which aim to integrate computation and storage. This paper presents a robust strategy for protecting BNN model parameters, particularly within in-memory computing frameworks. Our method utilizes a secret key derived from a physical unclonable function to transform model parameters prior to storage in the crossbar. Subsequently, the inference operations are performed on the encrypted weights, achieving a very special case of Fully Homomorphic Encryption (FHE) with minimal runtime overhead. Our analysis reveals that inference conducted without the secret key results in drastically diminished performance, with accuracy falling below 15%. These results validate the effectiveness of our protection strategy in securing BNNs within in-memory computing architectures while preserving computational efficiency.

cs.CR↗

arXiv2025

Anemoi: A Semi-Centralized Multi-agent System Based on Agent-to-Agent Communication MCP server from Coral Protocol

Recent advances in generalist multi-agent systems (MAS) have largely followed a context-engineering plus centralized paradigm, where a planner agent coordinates multiple worker agents through unidirectional prompt passing. While effective under strong planner models, this design suffers from two critical limitations: (1) strong dependency on the planner's capability, which leads to degraded performance when a smaller LLM powers the planner; and (2) limited inter-agent communication, where collaboration relies on prompt concatenation rather than genuine refinement through structured discussions. To address these challenges, we propose Anemoi, a semi-centralized MAS built on the Agent-to-Agent (A2A) communication MCP server from Coral Protocol. Unlike traditional designs, Anemoi enables structured and direct inter-agent collaboration, allowing all agents to monitor progress, assess results, identify bottlenecks, and propose refinements in real time. This paradigm reduces reliance on a single planner, supports adaptive plan updates, and minimizes redundant context passing, resulting in more scalable execution. Evaluated on the GAIA benchmark, Anemoi achieved 52.73% accuracy with a small LLM (GPT-4.1-mini) as the planner, surpassing the strongest open-source baseline OWL (43.63%) by +9.09% under identical LLM settings. Our implementation is publicly available at https://github.com/Coral-Protocol/Anemoi.

cs.MA↗

arXiv2025

Securing the Internet of Medical Things (IoMT): Real-World Attack Taxonomy and Practical Security Measures

The Internet of Medical Things (IoMT) has the potential to radically improve healthcare by enabling real-time monitoring, remote diagnostics, and AI-driven decision making. However, the connectivity, embedded intelligence, and inclusion of a wide variety of novel sensors expose medical devices to severe cybersecurity threats, compromising patient safety and data privacy. In addition, many devices also have direct capacity - individually or in conjunction with other IoMT devices - to perform actions on the patient, such as delivering an electrical stimulus, administering a drug, or activating a motor, which can potentially be life-threatening. We provide a taxonomy of potential attacks targeting IoMT, presenting attack surfaces, vulnerabilities, and mitigation strategies across all layers of the IoMT architecture. It answers key questions such as: What makes IoMT security different from traditional IT security? What are the cybersecurity threats to medical devices? How can engineers design secure IoMT systems and protect hospital networks from cyberattacks? By analyzing historical cyber incidents, we highlight critical security gaps and propose practical security guidelines for medical device engineers and security professionals. This work bridges the gap between research and implementation, equipping healthcare stakeholders with actionable insights to build resilient and privacy-preserving IoMT ecosystems. Finally, we present the latest standardization and compliance frameworks, that IoMT security designers should be aware of.

cs.CR↗

arXiv2025

Coral Protocol: Open Infrastructure Connecting The Internet of Agents

Coral Protocol is an open and decentralized collaboration infrastructure that enables communication, coordination, trust and payments for The Internet of Agents. It addresses the growing need for interoperability in a world where organizations are deploying multiple specialized AI agents that must work together across domains and vendors. As a foundational platform for multi-agent AI ecosystems, Coral establishes a common language and coordination framework allowing any agent to participate in complex workflows with others. Its design emphasizes broad compatibility, security, and vendor neutrality, ensuring that agent interactions are efficient and trustworthy. In particular, Coral introduces standardized messaging formats for agent communication, a modular coordination mechanism for orchestrating multi-agent tasks, and secure team formation capabilities for dynamically assembling trusted groups of agents. Together, these innovations position Coral Protocol as a cornerstone of the emerging "Internet of Agents," unlocking new levels of automation, collective intelligence, and business value through open agent collaboration.

cs.MA↗

arXiv2024

A Comprehensive Study of Quantum Arithmetic Circuits

In recent decades, the field of quantum computing has experienced remarkable progress. This progress is marked by the superior performance of many quantum algorithms compared to their classical counterparts, with Shor's algorithm serving as a prominent illustration. Quantum arithmetic circuits, which are the fundamental building blocks in numerous quantum algorithms, have attracted much attention. Despite extensive exploration of various designs in the existing literature, researchers remain keen on developing novel designs and improving existing ones. In this review article, we aim to provide a systematically organized and easily comprehensible overview of the current state-of-the-art in quantum arithmetic circuits. Specifically, this study covers fundamental operations such as addition, subtraction, multiplication, division and modular exponentiation. We delve into the detailed quantum implementations of these prominent designs and evaluate their efficiency considering various objectives. We also discuss potential applications of presented arithmetic circuits and suggest future research directions.

quant-ph↗

arXiv2024

Optimal Toffoli-Depth Quantum Adder

Efficient quantum arithmetic circuits are commonly found in numerous quantum algorithms of practical significance. Till date, the logarithmic-depth quantum adders includes a constant coefficient k >= 2 while achieving the Toffoli-Depth of klog n + O(1). In this work, 160 alternative compositions of the carry-propagation structure are comprehensively explored to determine the optimal depth structure for a quantum adder. By extensively studying these structures, it is shown that an exact Toffoli-Depth of log n + O(1) is achievable. This presents a reduction of Toffoli-Depth by almost 50% compared to the best known quantum adder circuits presented till date. We demonstrate a further possible design by incorporating a different expansion of propagate and generate forms, as well as an extension of the modular framework. Our paper elaborates on these designs, supported by detailed theoretical analyses and simulation-based studies, firmly substantiating our claims of optimality. The results also mirror similar improvements, recently reported in classical adder circuit complexity.

quant-ph↗

arXiv2023

$J/\psi$ and $\psi$(2S) polarization in proton-proton collisions at energies available at the CERN Large Hadron Collider using PYTHIA8

The production mechanisms of charmonium states in both hadronic and heavy-ion collisions hold great significance for investigating the hot and dense QCD matter. Studying charmonium polarization in ultra-relativistic collisions can also provide insights into the underlying production mechanisms. With this motivation, we explore the $J/\psi$ and $\psi$(2S) polarization in proton+proton collisions at $\sqrt{s}$ = 7, 8, and 13 TeV using a pQCD-inspired Monte-Carlo event generator called PYTHIA8. This work considers reconstructed quarkonia through their dimuons decay channel in the ALICE forward rapidity acceptance range of $2.5 < y_{\mu \mu} < 4$. Further, we calculate the polarization parameters $\lambda_{\theta}$, $\lambda_{\phi}$, $\lambda_{\theta \phi}$ from the polar and azimuthal angular distributions of the dimuons in helicity and Collins-Soper frames. This study presents a comprehensive measurement of the polarization parameters as a function of transverse momentum, charged-particle multiplicity, and rapidity at the LHC energies. Our findings of charmonium polarization are in qualitative agreement with the corresponding experimental data.

hep-ph↗

arXiv2023

Characterizing nuclear modification effects in high-energy O-O collisions at energies available at the CERN Large Hadron Collider: A transport model perspective

The present work focuses on Oxygen-Oxygen (O-O) collisions, which are planned at the CERN Large Hadron Collider. Oxygen, being a doubly magic number nucleus, has some very unique features. This study attempts to probe the exotic state of QCD matter in O-O collisions. Additionally, the role of different nuclear density profiles in governing the final state dynamics in ultra-relativistic nuclear collisions is also explored. Using a multi-phase transport model, we obtain the nuclear modification factor ($\textit R_{\textit {AA}}$) for all charged hadrons and identified particles for O-O collisions at $\sqrt{s_{\rm{NN}}}$ = 7 TeV. Furthermore, we investigate the behavior of $\textit R_{\textit {AA}}$ as a function of transverse momentum ($\textit{p}_{\rm{T}}$) for three centralities (most central, mid-central, and peripheral) considering both $\alpha$-cluster and Woods-Saxon nuclear density profiles. We also extend this work to study the rapidity dependence of $\textit R_{\textit {AA}}$ for all charged hadrons. To better understand our findings of O-O collisions, the results are confronted with the available data of $\textit R_{\textit {AA}}$ for Pb-Pb collisions. The present study sheds light on particle production mechanisms, emphasizing factors influencing particle yield from pre-collision to post-collision stages in the context of O-O collisions.

hep-ph↗

arXiv2023

Energy flow in ultra-high energy cosmic ray interactions as a probe of thermalization: a potential solution to the muon puzzle

Signatures of the formation of a strongly interacting thermalized matter of partons have been observed in nucleus-nucleus, proton-nucleus, and high-multiplicity proton-proton collisions at LHC energies. Strangeness enhancement in such ultra-relativistic heavy-ion collisions is considered to be a consequence of this thermalized phase, known as quark-gluon plasma (QGP). Simultaneously, proper modeling of hadronic energy fraction in interactions of ultra-high energy cosmic rays (UHECR) has been proposed as a solution for the muon puzzle, an unexpected excess of muons in air showers. These interactions have center-of-mass collision energies of the order of energies attained at the LHC or even higher, indicating that the possibility of a thermalized partonic state cannot be overlooked in UHECR-air interactions. This work investigates the hadronic energy fraction and strangeness enhancement to explore QGP-like phenomena in UHECR-air interactions using various high-energy hadronic models. A core-corona system with a thermalized core undergoing statistical hadronization is considered through the EPOS LHC model. In contrast, PYTHIA 8, QGSJET II-04, and SYBILL 2.3d consider string fragmentation without thermalization. We have found that EPOS LHC gives a better description of strangeness enhancement as compared to other models. We conclude that adequately treating all the relevant effects and further retuning the models is necessary to explain the observed effects.

hep-ph↗

arXiv2022

Dynamics of Hot QCD Matter -- Current Status and Developments

The discovery and characterization of hot and dense QCD matter, known as Quark Gluon Plasma (QGP), remains the most international collaborative effort and synergy between theorists and experimentalists in modern nuclear physics to date. The experimentalists around the world not only collect an unprecedented amount of data in heavy-ion collisions, at Relativistic Heavy Ion Collider (RHIC), at Brookhaven National Laboratory (BNL) in New York, USA, and the Large Hadron Collider (LHC), at CERN in Geneva, Switzerland but also analyze these data to unravel the mystery of this new phase of matter that filled a few microseconds old universe, just after the Big Bang. In the meantime, advancements in theoretical works and computing capability extend our wisdom about the hot-dense QCD matter and its dynamics through mathematical equations. The exchange of ideas between experimentalists and theoreticians is crucial for the progress of our knowledge. The motivation of this first conference named "HOT QCD Matter 2022" is to bring the community together to have a discourse on this topic. In this article, there are 36 sections discussing various topics in the field of relativistic heavy-ion collisions and related phenomena that cover a snapshot of the current experimental observations and theoretical progress. This article begins with the theoretical overview of relativistic spin-hydrodynamics in the presence of the external magnetic field, followed by the Lattice QCD results on heavy quarks in QGP, and finally, it ends with an overview of experiment results.

nucl-th↗

arXiv2022

Multiplicity, transverse momentum and pseudorapidity dependence of open-heavy flavored hadron production in proton+proton collisions at $\sqrt{s}$= 13 TeV using PYTHIA8

Recently, with the upgradation of LHC, it is realized that study of heavy-flavored hadrons, namely $\Lambda_{c}^{+}$ and $\rm{D}^{0}$ in hadronic collisions, could reveal the possibility of thermalization of charm quarks. With this motivation, we study the production dynamics of these hadrons using a pQCD-inspired Monte Carlo event generator called PYTHIA8 in proton+proton collisions at $\sqrt{s}$ = 13 TeV. The production dynamics of these hadrons are analyzed through charged-particle multiplicity, transverse momentum, and pseudorapidity. With the help of the established PYTHIA8 tunes to mimic the behavior of flow-like features, we investigated the variation of effective temperature and degree of non-extensivity using thermodynamically consistent non-extensive Tsallis statistics. We further attempted to establish a connection between the initial state and final state by estimating the correlation between the number of multi-partonic interactions ($\rm{n}_{MPI}$) with the Knudsen number.

hep-ph↗

arXiv2021

Extension of Glauber-like model for Proton-Proton collisions using anisotropic and inhomogeneous density profile

Results from proton-proton ($pp$) collisions have routinely been used as a baseline to analyze and understand the production of QCD matter expected to be produced in nuclear collisions. But recent studies of small systems formed in $pp$ collisions at the LHC energies hint at the possibility of producing medium with collective behavior. Therefore, results from $pp$ collisions required more careful investigation to understand whether QCD matter is produced in high multiplicity $pp$ collisions. With this motivation, the Glauber model traditionally used to study the heavy-ion collision dynamics at high energy is applied here to understand the dynamics of $pp$ collisions. We have used anisotropic and inhomogeneous quark/gluon-based proton density profile, a realistic picture obtained from deep inelastic scattering results and this model explains the charged particle multiplicity distribution of $pp$ collisions at LHC energies very well. Collision geometric properties like impact parameter and mean number of binary collisions ($\langle N_{coll} \rangle$), mean number of participants ($\langle N_{part} \rangle$) at different multiplicities are determined for $pp$ collisions. We further used these collision geometric properties to estimate average charged-particle pseudorapidity density ($\langle dN_{ch}/d\eta \rangle$) and found it to be comparable with the experimental results. Knowing $\langle N_{coll} \rangle$, we have obtained nuclear modification-like factor ($R_{pp}$) in $pp$ collisions which has not been done before to the best of our knowledge.

hep-ph↗

arXiv2021

Implementation of machine learning techniques to predict impact parameter and transverse spherocity in heavy-ion collisions at the LHC

Machine learning techniques have been quite popular recently in the high-energy physics community and have led to numerous developments in this field. In heavy-ion collisions, one of the crucial observables, the impact parameter, plays an important role in the final-state particle production. This being extremely small (i.e. of the order of a few fermi), it is almost impossible to measure impact parameter in experiments. In this work, we implement the ML-based regression technique via Gradient Boosting Decision Trees (GBDT) to obtain a prediction of impact parameter in Pb-Pb collisions at $\sqrt{s_{NN}}$ = 5.02 TeV using A Multi-Phase Transport (AMPT) model. After its successful implementation in small collision systems, transverse spherocity, an event shape observable, holds an opportunity to reveal more about the particle production in heavy-ion collisions as well. In the absence of any experimental exploration in this direction at the LHC yet, we suggest an ML-based regression method to estimate centrality-wise transverse spherocity distributions in Pb-Pb collisions at $\sqrt{s_{NN}}$ = 5.02 TeV by training the model with minimum bias collision data. Throughout this work, we have used a few final state observables as the input to the ML-model, which could be easily made available from collision data. Our method seems to work quite well as we see a good agreement between the simulated true values and the predicted values from the ML-model.

hep-ph↗

arXiv2021

Charmonium suppression in ultra-relativistic proton-proton collisions at LHC energies: A hint for QGP in small systems

Proton-proton ($pp$) collision has been considered as a baseline to study the system produced in relativistic heavy-ion (AA) collisions with the basic assumption that no thermal medium is formed in $pp$ collisions. This warrants a cautious analysis of the system produced in $pp$ collisions at relativistic energies.In this work we investigate the charmonium suppression in $pp$ collisions at $\sqrt{s} = 5.02, 7$ and $13$ TeV energies. Further, charmonium suppression has been studied for various event multiplicities and transverse momenta by including the mechanisms of color screening, gluonic dissociation, collisional damping along with regeneration due to correlated $c\bar c$ pairs. Here we obtain a net suppression of charmonia at high-multiplicity events indicating the possibility of the formation of quark-gluon plasma in $pp$ collisions.

hep-ph↗

arXiv2021

Hadron gas in the presence of a magnetic field using non-extensive statistics: A transition from diamagnetic to paramagnetic system

Non-central heavy-ion collisions at ultra-relativistic energies are unique in producing magnetic fields of the largest strength in the laboratory. Such fields being produced at the early stages of the collision could affect the properties of Quantum Chromodynamics (QCD) matter formed in the relativistic heavy-ion collisions. The transient magnetic field leaves its reminiscence, which in principle, can affect the thermodynamic and transport properties of the final state dynamics of the system. In this work, we study the thermodynamic properties of a hadron gas in the presence of an external static magnetic field using a thermodynamically consistent non-extensive Tsallis distribution function. Various thermodynamical observables such as energy density ($\epsilon$), entropy density ($s$), pressure ($P$) and speed of sound ($c_{\rm s}$) are studied. Investigation of magnetization ($M$) is also performed and this analysis reveals an interplay of diamagnetic and paramagnetic nature of the system in the presence of a magnetic field of varying strength. Further, to understand the system dynamics under equilibrium and non-equilibrium conditions, the effect of the non-extensive parameter ($q$) on the above observables is also studied.

hep-ph↗

arXiv2021

Estimation of Impact Parameter and Transverse Spherocity in heavy-ion collisions at the LHC energies using Machine Learning

Recently, machine learning (ML) techniques have led to a range of numerous developments in the field of nuclear and high-energy physics. In heavy-ion collisions, the impact parameter of a collision is one of the crucial observables which has a significant impact on the final state particle production. However, calculation of such a quantity is nearly impossible in experiments as the length scale ranges in the level of a few fermi. In this work, we implement the ML-based regression technique via Boosted Decision Tree (BDT) to obtain a prediction of impact parameter in Pb-Pb collisions at $\sqrt{s_{\rm NN}}$ = 5.02 TeV using A Multi-Phase Transport (AMPT) model. In addition, we predict an event shape observable, transverse spherocity in Pb-Pb collisions at $\sqrt{s_{\rm NN}}$ = 2.76 and 5.02 TeV using AMPT and PYTHIA8 based on Angantyr model. After a successful implementation in small collision systems, the use of transverse spherocity in heavy-ion collisions has potential to reveal new results from heavy-ion collisions where the production of a QGP medium is already established. We predict the centrality dependent spherocity distributions from the training of minimum bias simulated data and it was found that the predictions from BDT based ML technique match with true simulated data. In the absence of experimental measurements, we propose to implement Machine learning based regression technique to obtain transverse spherocity from the known final state observables in heavy-ion collisions.

hep-ph↗