arXiv subjects

Souvik Das

Publications and source records attributed to Souvik Das.

At least 19 recordsLinked to original sources

arXiv2026

Heavy neutral leptons from light scalar in fixed target and forward search experiments

The observation of neutrino masses strongly motivates $U(1)_{B-L}$ extensions of the Standard Model, in which heavy neutral leptons acquire Majorana masses through spontaneous $U(1)_{B-L}$ symmetry breaking and generate light neutrino masses via the seesaw mechanism. In this framework, the singlet scalar responsible for symmetry breaking mixes with the SM Higgs boson, allowing it to be produced in rare meson decays. We investigate a scenario in which this light scalar promptly decays into a pair of long-lived heavy neutrinos that subsequently decay into visible charged leptons and hadrons through light-heavy neutrino mixing inside the proposed Forward Physics Facility (FPF) at the FCC-hh and the SHiP beam-dump experiment. Taking into account realistic detector geometries, decay probabilities, and visible branching fractions, we estimate the projected sensitivities to the scalar-Higgs mixing angle as a function of the scalar mass and to the light-heavy neutrino mixing as a function of the heavy neutrino mass. We find that FPF and SHiP can significantly extend the discovery reach for both light scalars and long-lived heavy neutrinos beyond existing experimental limits, providing powerful and complementary probes of neutrino-mass generation and hidden-sector physics.

hep-ph↗

arXiv2026

Sparse Robust Optimal Control in Continuous-Time: A Computationally Viable Approach

This article presents a novel, numerically viable algorithm for solving sparse robust optimal control problems in continuous time. We consider a constrained linear noisy system governed by an ordinary differential equation (ODE), with an $L^1$-type objective function in line with the sparse optimal control literature. The resulting optimal control problem is shown to admit a semi-infinite programming (SIP) formulation. Building upon this insight, we develop a new framework that enables the computation of exact solutions -- to our knowledge, the first such achievement in the context of sparse optimal control. We demonstrate that a finite and computationally viable convex optimization problem can be solved to recover, in a lossless manner, both the optimal value and the corresponding optimizers of the original SIP, while also guaranteeing satisfaction of uncountably many constraints. We also show that the parameter-dependent noisy systems and the minimum attention problem fall into our framework and can be solved efficiently via our algorithm. The efficacy of our algorithm is illustrated through a benchmark numerical example.

math.OC↗

arXiv2026

Forward Searches for Heavy Neutrinos and $Z'$ Bosons at FCC-hh

The discovery of neutrino masses strongly motivates extensions of the Standard Model containing heavy neutral leptons and additional gauge interactions. We investigate the prospects for probing these states at the proposed Forward Physics Facility (FPF) of the 100 TeV Future Circular Collider (FCC-hh) within a broad class of anomaly-free chiral $U(1)$ gauge extensions. These models predict a new neutral gauge boson, $Z'$, together with right-handed neutrinos responsible for generating light neutrino masses through the seesaw mechanism. We study long-lived particle signatures arising from both heavy neutrinos and the $Z'$ boson produced in the far-forward region. In particular, we analyze heavy neutrino production from meson decays, visible decays of long-lived $Z'$ bosons produced through meson decays and proton bremsstrahlung, long-lived $Z'$ bosons decaying into heavy-neutrino pairs, and prompt $Z'$ decays yielding long-lived heavy neutrinos. The expected event rates are evaluated for the proposed FPF detector configurations, taking into account realistic detector geometry, decay probabilities, and visible final states. We derive projected sensitivities to the heavy neutrino mass and active-sterile mixing as well as to the $Z'$ mass and gauge coupling for several representative $U(1)$ charge assignments. Our results demonstrate that the FPF at FCC-hh can substantially extend the discovery reach for light long-lived heavy neutrinos and light $Z'$ bosons beyond existing and proposed experiments, providing a powerful and complementary probe of neutrino-mass models and hidden gauge sectors. https://github.com/SouvikPhD/RHN-Detection-with-FASER-2-

hep-ph↗

arXiv2026

OT-DETECT: Optimal Transport-Driven Attack Detection in Cyber-Physical Systems

This letter presents an optimal-transport (OT)-driven, distributionally robust attack detection algorithm, OT-DETECT, for cyber-physical systems (CPS) modeled as partially observed linear stochastic systems. The underlying detection problem is formulated as a minmax optimization problem using 1-Wasserstein ambiguity sets constructed from observer residuals under both the nominal (attack-free) and attacked regimes, and show that the minmax detection problem can be reduced to a finite-dimensional linear program for computing the worst-case distribution (WCD). Off-support residuals are handled via a kernel-smoothed score function that drives a CUSUM procedure for sequential detection. We also establish a non-asymptotic tail bound on the false-positive error of the CUSUM statistic under the nominal (attack-free) condition, under mild assumptions. Numerical illustrations are provided to evaluate the robustness properties of OT-DETECT.

math.OC↗

arXiv2026

Stability and wave dynamics in polytropic Eddington-inspired Born-Infeld gravitating solar plasmas

We investigate the influence of nonlinear gravity corrections, arising from the Eddington-inspired Born-Infeld (EiBI) theory on wave dynamics, stability, and energy transport processes in polytropic, viscous, and turbulent solar plasmas. Analytical and numerical analyses of the Jeans-normalized quadratic dispersion relation demonstrate that both the EiBI gravity parameter $(\chi)$ and the relative polytropic sound speed $(\beta)$ independently regulate oscillation frequencies, growth rates, phase velocities, perturbation energy partitioning, and outward acoustic energy flux. Positive $\chi$ systematically elevates oscillation frequencies, phase velocities, and outward energy flux level by $\sim$10% relative to the Newtonian predictions, while larger $\beta$ enhances them by up to 55%, thereby promoting wave propagation and efficient acoustic transport. Conversely, negative $\chi$ strengthens gravitational binding and increases damping rates by $\sim$40%, particularly for the \textit{g}-modes. Energy partitioning analyses reveal that the EiBI corrections fundamentally restructure the kinetic-electrostatic-gravitational energy balance. While the Newtonian gravity contributes negligibly ($<$4%), nonzero $\chi$ channels up to one-third of oscillation energy into gravitational modes. The modal surface flux calculations further confirm that only the \textit{p}-modes drive outward energy transport (amplification for $\chi>0$, suppression for $\chi<0$). A direct comparative analysis with four years of SDO/HMI Doppler velocity observations demonstrate a robust theoretical agreement for $\chi=3\times10^7$ m$^5$kg$^{-1}$s$^{-2}$, providing the first empirical constraint on the solar EiBI gravity through helioseismology. The findings offer a rigorous framework for advancing our understanding about solar plasma stability, helioseismic signatures, and ambient atmospheric energy transport processes.

astro-ph.SR↗

arXiv2026

On Convergence Analysis of Network-GIANT: An approximate Hessian-based fully distributed optimization algorithm

This paper presents a detailed convergence and performance analysis of a recently developed approximate Newton-type fully distributed optimization method for $L$-smooth, $\mu$-strongly convex local loss functions, called Network-GIANT (inspired by the Federated learning algorithm GIANT possessing mixed linear-quadratic convergence properties). Network-GIANT has been empirically seen to achieve faster linear convergence properties compared to its gradient-based counterparts, and several other existing second order distributed algorithms, while having the same communication complexity (per iteration) as its first order distributed counterparts. We first explicitly characterize a \emph{global linear convergence rate} for Network-GIANT, which can be computed as the spectral radius of a $3 \times 3$ matrix dependent on $L$, $\mu$, and the spectral norm ($\sigma$) of the consensus matrix of the underlying undirected graph. We provide an explicit bound on the step size parameter $\eta$, below which this spectral radius is guaranteed to be less than $1$. Furthermore, we derive a mixed linear-quadratic inequality based upper bound for the optimality gap norm, and provide a rigorous proof of a local asymptotic convergence rate of $1 - \eta \big(1 - \frac{\gamma}{\mu}\big)$ given the Hessian approximation error $\gamma < \mu$, which formally explains the faster convergence rate of Network-GIANT. Numerical experiments are carried out with a reduced CovType dataset for binary logistic regression over a variety of graphs, including heterogeneous data distributions, to illustrate the above theoretical results.

math.OC↗

arXiv2025

Algorithmic detection of false data injection attacks in cyber-physical systems

This article introduces an anomaly detection based algorithm (AD-CPS) to detect false data injection attacks that fall under the category of data deception/integrity attacks, but with arbitrary information structure, in cyber-physical systems (CPSs) modeled as stochastic linear time-invariant systems. The core idea of this data-driven algorithm is based on the fact that an honest state (one not compromised by adversaries) generated by the CPS should concentrate near its weighted empirical mean of the immediate past samples. As the first theoretical result, we provide non-asymptotic guarantees on the false positive error incurred by the algorithm for attacks that are 2-step honest, referring to adversaries that act intermittently rather than successively. Moreover, we establish that for adversaries possessing a certain minimum energy, the false negative error incurred by AD-CPS is low. Extensive experiments were conducted on partially observed stochastic LTI systems to demonstrate these properties and to quantitatively compare AD-CPS with an optimal CUSUM-based test.

math.OC↗

arXiv2025

HBNET-GIANT: A communication-efficient accelerated Newton-type fully distributed optimization algorithm

This article presents a second-order fully distributed optimization algorithm, HBNET-GIANT, driven by heavy-ball momentum, for $L$-smooth and $\mu$-strongly convex objective functions. A rigorous convergence analysis is performed, and we demonstrate global linear convergence under certain sufficient conditions. Through extensive numerical experiments, we show that HBNET-GIANT with heavy-ball momentum achieves acceleration, and the corresponding rate of convergence is strictly faster than its non-accelerated version, NETWORK-GIANT. Moreover, we compare HBNET-GIANT with several state-of-the-art algorithms, both momentum-based and without momentum, and report significant performance improvement in convergence to the optimum. We believe that this work lays the groundwork for a broader class of second-order Newton-type algorithms with momentum and motivates further investigation into open problems, including an analytical proof of local acceleration in the fully distributed setting for convex optimization problems.

math.OC↗

arXiv2025

CoNeT-GIANT: A compressed Newton-type fully distributed optimization algorithm

Compression techniques are essential in distributed optimization and learning algorithms with high-dimensional model parameters, particularly in scenarios with tight communication constraints such as limited bandwidth. This article presents a communication-efficient second-order distributed optimization algorithm, termed as CoNet-GIANT, equipped with a compression module, designed to minimize the average of local strongly convex functions. CoNet-GIANT incorporates two consensus-based averaging steps at each node: gradient tracking and approximate Newton-type iterations, inspired by the recently proposed Network-GIANT. Under certain sufficient conditions on the step size, CoNet-GIANT achieves significantly faster linear convergence, comparable to that of its first-order counterparts, both in the compressed and uncompressed settings. CoNet-GIANT is efficient in terms of data usage, communication cost, and run-time, making it a suitable choice for distributed optimization over a wide range of wireless networks. Extensive experiments on synthetic data and the widely used CovType dataset demonstrate its superior performance.

math.OC↗

arXiv2025

Right handed neutrino production from $Z^\prime$ interactions in forward search experiments

We study two general $U(1)$ extensions of the Standard Model (SM) those generate tiny neutrino masses via the seesaw mechanism after general $U(1)$ breaking. These models predict a new neutral gauge boson ($Z'$) and right-handed neutrinos (RHNs), the latter introduced for anomaly cancellation and neutrino mass generation. In both scenarios, left- and right-handed fermions couple differently to the $Z'$, and RHNs mix with light neutrinos, enabling variety of decay modes. Focusing on the high-luminosity LHC (HL-LHC) and the future FASER2 experiment, we explore RHN pair production from $Z'$ decays in two cases: (i) long-lived $Z'$ decays to visible modes and long-lived RHNs, and (ii) short-lived $Z'$ decays to long-lived RHNs, which further decay visibly inside FASER2. We estimate projected limits on the general $U(1)$ gauge coupling, $Z'$ mass, RHN mass, and light-heavy neutrino mixing for various $U(1)$ charge assignments, and compare them with current experimental bounds.

hep-ph↗

arXiv2024

Beyond Discrete Personas: Personality Modeling Through Journal Intensive Conversations

Large Language Models (LLMs) have significantly improved personalized conversational capabilities. However, existing datasets like Persona Chat, Synthetic Persona Chat, and Blended Skill Talk rely on static, predefined personas. This approach often results in dialogues that fail to capture human personalities' fluid and evolving nature. To overcome these limitations, we introduce a novel dataset with around 400,000 dialogues and a framework for generating personalized conversations using long-form journal entries from Reddit. Our approach clusters journal entries for each author and filters them by selecting the most representative cluster, ensuring that the retained entries best reflect the author's personality. We further refine the data by capturing the Big Five personality traits --openness, conscientiousness, extraversion, agreeableness, and neuroticism --ensuring that dialogues authentically reflect an individual's personality. Using Llama 3 70B, we generate high-quality, personality-rich dialogues grounded in these journal entries. Fine-tuning models on this dataset leads to an 11% improvement in capturing personality traits on average, outperforming existing approaches in generating more coherent and personality-driven dialogues.

cs.CL↗

arXiv2024

Solar GES-structure modified with EiBI gravity

In the post-Newtonian era, the Eddington-inspired Born-Infeld (EiBI) theory, considered as an improved modification of the Einsteinian general relativity formalism in the weak field regime (non-relativistic), has enabled us to study the dynamics of dense astroobjects in light of the modified gravitational effects. This EiBI theory imparts a new shape to the usual gravitational Poisson equation through the addition of a cosmological correction factor, termed as the EiBI gravity parameter. A systematic inclusion of this gravity in the basic structure equation could lead to a realistic picture of the existing solar models free from any end-stage singularity. A theoretic model is accordingly proposed to investigate the effect of the EiBI gravity on the Gravito-Electrostatic Sheath (GES) formalism of the equilibrium solar plasma structure. This study shows that the GES-based solar plasma dynamics is noticeably modified against the previously reported Newtonian GES-model studies. An equilibrium bounded solution for the solar self-gravity shows the EiBI-modified solar surface boundary (SSB) to exist at a new helio-centric radial location $\xi = 4$ (on the Jeansean scale). It is found that the EiBI gravity shifts the present SSB outwards by 14.28% relative to the original Newtonian SSB. The EiBI-modified gravity effects on diverse relevant solar parameters, such as the gravito-electrostatic potentials, fields, and Mach numbers, are illustratively analyzed. It is anticipated that our analyses could be applied further to see the solar plasma equilibrium and fluctuation dynamics in realistically modified post-Newtonian gravity environments on both the bounded (interior) and unbounded (exterior) solar plasma scales.

gr-qc↗

arXiv2024

On a probabilistic global optimizer derived from the Walker slice sampling

This article presents a zeroth order probabilistic global optimization algorithm -- SwiftNav -- for (not necessarily convex) functions over a compact domain. A discretization procedure is deployed on the compact domain, starting with a small step-size $h > 0$ and subsequently adaptively refining it in the course of a simulated annealing routine utilizing the Walker slice and the Gibbs sampler, in order to identify a set of global optimizers up to good precision. SwiftNav is parallelizable, which helps with scalability as the dimension of decision variables increases. Several numerical experiments are included here to demonstrate the effectiveness and accuracy of SwiftNav in high-dimensional benchmark optimization problems.

math.OC↗

arXiv2024

Data-driven distributionally robust MPC for systems with multiplicative noise: A semi-infinite semi-definite programming approach

This article introduces a novel distributionally robust model predictive control (DRMPC) algorithm for a specific class of controlled dynamical systems where the disturbance multiplies the state and control variables. These classes of systems arise in mathematical finance, where the paradigm of distributionally robust optimization (DRO) fits perfectly, and this serves as the primary motivation for this work. We recast the optimal control problem (OCP) as a semi-definite program with an infinite number of constraints, making the ensuing optimization problem a \emph{semi-infinite semi-definite program} (SI-SDP). To numerically solve the SI-SDP, we advance an approach for solving convex semi-infinite programs (SIPs) to SI-SDPs and, subsequently, solve the DRMPC problem. A numerical example is provided to show the effectiveness of the algorithm.

math.OC↗

arXiv2024

Mitigating Clickbait: An Approach to Spoiler Generation Using Multitask Learning

This study introduces 'clickbait spoiling', a novel technique designed to detect, categorize, and generate spoilers as succinct text responses, countering the curiosity induced by clickbait content. By leveraging a multi-task learning framework, our model's generalization capabilities are significantly enhanced, effectively addressing the pervasive issue of clickbait. The crux of our research lies in generating appropriate spoilers, be it a phrase, an extended passage, or multiple, depending on the spoiler type required. Our methodology integrates two crucial techniques: a refined spoiler categorization method and a modified version of the Question Answering (QA) mechanism, incorporated within a multi-task learning paradigm for optimized spoiler extraction from context. Notably, we have included fine-tuning methods for models capable of handling longer sequences to accommodate the generation of extended spoilers. This research highlights the potential of sophisticated text processing techniques in tackling the omnipresent issue of clickbait, promising an enhanced user experience in the digital realm.

cs.CL↗

arXiv2024

Entropy Guided Extrapolative Decoding to Improve Factuality in Large Language Models

Large language models (LLMs) exhibit impressive natural language capabilities but suffer from hallucination -- generating content ungrounded in the realities of training data. Recent work has focused on decoding techniques to improve factuality during inference by leveraging LLMs' hierarchical representation of factual knowledge, manipulating the predicted distributions at inference time. Current state-of-the-art approaches refine decoding by contrasting early-exit distributions from a lower layer with the final layer to exploit information related to factuality within the model forward procedure. However, such methods often assume the final layer is the most reliable and the lower layer selection process depends on it. In this work, we first propose extrapolation of critical token probabilities beyond the last layer for more accurate contrasting. We additionally employ layer-wise entropy-guided lower layer selection, decoupling the selection process from the final layer. Experiments demonstrate strong performance - surpassing state-of-the-art on multiple different datasets by large margins. Analyses show different kinds of prompts respond to different selection strategies.

cs.CL↗

arXiv2024

A theoretic analysis of magnetoactive GES-based turbulent solar plasma instability

A recently reported gravito-electrostatic sheath (GES) model is procedurally applied to study the turbumagnetoactive helioseismic oscillation features on the entire bi-fluidic solar plasma system. The bounded solar interior plasma (SIP, internally self-gravitating) and the unbounded solar wind plasma (SWP, externally point-gravitating) are coupled through the interfacial diffused solar surface boundary (SSB) due to an exact gravito-electrostatic interplay. A numerical platform on the developed theoretic formalism reveals the evolution of both dispersive and non-dispersive features of the modified GES mode fluctuations in new parametric windows. Different colourspectral profiles exhibit important features of the GES-based SIP-SWP perturbations elaborately. It is illustratively shown that the thermostatistical GES stability depends mainly on the radial distance, magnetic field, equilibrium plasma density, and plasma temperature. We see that their dispersive features are more pertinently pronounced in the self-gravitational domains (SIP) than the electrostatic ones (SWP). Besides, different characteristic parameters with accelerating (or decelerating) and stabilizing (or destabilizing) effects influencing the entire solar plasma stability are illustratively portrayed. We speculate that, in the SIP, the long-wave (gravitational-like) helioseismic fluctuations become highly dispersive showing more propagatory nature than the shorter ones (acoustic-like). The short waves show more propagatory propensity than the longer ones in the SSB and SWP regime. The reliability of our proposed investigation is bolstered along with the tentative applicability and future scope in light of the current solar observational scenarios, such as SOHO, STEREO, SDO, PSP, and SolO.

astro-ph.SR↗

arXiv2024

Improving Dialog Safety using Socially Aware Contrastive Learning

State-of-the-art conversational AI systems raise concerns due to their potential risks of generating unsafe, toxic, unethical, or dangerous content. Previous works have developed datasets to teach conversational agents the appropriate social paradigms to respond effectively to specifically designed hazardous content. However, models trained on these adversarial datasets still struggle to recognize subtle unsafe situations that appear naturally in conversations or introduce an inappropriate response in a casual context. To understand the extent of this problem, we study prosociality in both adversarial and casual dialog contexts and audit the response quality of general-purpose language models in terms of propensity to produce unsafe content. We propose a dual-step fine-tuning process to address these issues using a socially aware n-pair contrastive loss. Subsequently, we train a base model that integrates prosocial behavior by leveraging datasets like Moral Integrity Corpus (MIC) and ProsocialDialog. Experimental results on several dialog datasets demonstrate the effectiveness of our approach in generating socially appropriate responses.

cs.CL↗