arXiv subjects

Joydeep Ghosh

Publications and source records attributed to Joydeep Ghosh.

At least 19 recordsLinked to original sources

arXiv2026

MILM: Large Language Models for Multimodal Irregular Time Series with Informative Sampling

Multimodal irregular time series (MITS) consist of asynchronous and irregularly sampled observations from heterogeneous numerical and textual channels. In healthcare, for example, patients' electronic health records (EHR) include irregular lab measurements and clinical notes. The irregular timing and channel patterns of observations carry predictive signal alongside the numerical values and textual content. LLMs are natural candidates for processing such heterogeneous data, given their extensive pretrained knowledge spanning textual and numerical domains. We introduce MILM (Multimodal Irregular time series Language Model), which represents MITS as time-ordered triplets in Extensible Markup Language (XML) format and fine-tunes an LLM through a two-stage strategy for MITS classification. The first stage trains on value-redacted MITS to predict from sampling patterns alone, and the second stage trains on full MITS to jointly model sampling patterns and observed values. Our two-stage model (MILM-2S) and its single-stage counterpart (MILM-Direct) achieve the best and second-best average performance on multiple EHR datasets. Further value redaction evaluations confirm that sampling patterns carry predictive signal and that MILM-2S learns to exploit them. In the value pending evaluation we introduce, where some values are unavailable at prediction time, MILM-2S outperforms MILM-Direct by a larger margin compared to standard evaluation. For MILM-2S, preserving the time and channel of value-pending observations as additional sampling information further improves in-hospital mortality prediction.

cs.LG↗

arXiv2026

RRCM: Ranking-Driven Retrieval over Collaborative and Meta Memories for LLM Recommendation

Large Language Models (LLMs) have emerged as a promising paradigm for next-generation recommender systems, offering strong semantic understanding and natural-language reasoning abilities. Despite recent progress, current LLM-based recommenders still face key challenges in constructing decision-relevant contexts from heterogeneous evidence. First, existing methods often rely on fixed context construction strategies: collaborative behavioral evidence and item-side metadata are typically incorporated through predefined prompts, static retrieval pipelines, or handcrafted injection mechanisms, making it difficult to determine what information is truly beneficial for each instance. Second, heterogeneous evidence introduces a severe context-efficiency bottleneck. Rich metadata and collaborative interaction records can quickly overwhelm the context window, while aggressive compression or heuristic filtering may discard fine-grained evidence critical for accurate recommendation. To address these challenges, we propose RRCM, a ranking-driven retrieval-and-reasoning framework over collaborative and metadata memories for LLM-based agentic recommendation. RRCM starts from a lightweight user-history context and learns whether to recommend directly, retrieve collaborative evidence, retrieve item metadata, or interleave both through reasoning. Both memories are represented in natural language and accessed through a unified retrieval interface, enabling flexible evidence acquisition without handcrafted CF injection or fixed retrieval rules. We optimize this memory-reading policy with an outcome-only ranking reward, instantiated using group relative policy optimization, so that retrieval decisions are directly driven by final top-k recommendation quality. Extensive experiments show that RRCM significantly outperforms traditional baselines and diverse LLM-based recommendation approaches.

cs.IR↗

arXiv2026

Goal-Conditioned Supervised Learning for LLM Fine-Tuning

Large language models often require fine-tuning to better align their behavior with user intent at deployment. Existing approaches are commonly divided into online and offline paradigms. Online methods, such as RL-based alignment, can directly optimize outcome quality but typically rely on external reward models and iterative rollouts, making them costly and difficult to deploy in many cases. Offline methods are more efficient, but prevailing approaches such as supervised fine-tuning (SFT) and direct preference optimization (DPO) remain limited: SFT typically collapses graded feedback into binary supervision, while DPO depends on paired preference data that is often unavailable or expensive to construct. In this paper, we propose goal-conditioned supervised learning (GCSL) as an offline fine-tuning framework for LLMs. Our core idea is to treat feedback signals directly as an explicit goal and train the model, purely through supervised learning, to generate responses that achieve that goal. To better exploit graded feedback, we further introduce a novel goal formulation that defines learning as consistently pursuing outcomes above a target quality threshold, rather than imitating samples from a selected high-quality subset. This design mitigates the bounded-learning effect of SFT and classic GCSL by explicitly guiding the model to learn the directional progression of quality. We also propose natural-language goal representations to better leverage the semantic understanding and reasoning capabilities of LLMs. We evaluate our method on three tasks: non-toxic generation, code generation, and LLM for recommendation. Results show that our approach consistently outperforms standard offline fine-tuning baselines while retaining the efficiency, scalability, and simple data requirements of supervised learning.

cs.LG↗

arXiv2026

Relativistic calculations of electron impact excitation cross-sections of neutral tungsten

Reliable tungsten spectroscopy is essential for modelling and diagnosing tungsten-containing plasmas, particularly in edge/divertor conditions where neutral tungsten (WI) is produced by sputtering and contributes to visible/near-UV emission. In this work, we present fine-structure-resolved EIE cross sections for W I computed using the relativistic distorted-wave (RDW) method. Target states are described by multi-configurational Dirac-Fock (MCDF)wave functions with an extensive configuration-interaction (CI) expansion including valence-valence and core-valence correlation through hole and deeper-core configurations. The resulting energies are bench marked against available theoretical results and the recommended values from the NIST database. Fine-structure-resolved cross sections are reported for excitations from the ground level 5d4 6s2 (5 D0) and six metastable levels 5d4 6s2 (5D1 , 5D2 , 5D3, 5D4 , 3P0) and 5d5 (6S)6s (7S3 ) into excited levels belonging primarily to the 5d4 6s(6 D)6p and 5d5 (6S)6p configurations, over incident electron energies from threshold to 500 eV. With the exception of a limited set of previously studied transitions, most of the reported fine-structure-resolved cross sections are presented here for the first time. The results show a strong dependence on the initial level and indicate that excitation from the metastable state yields the largest cross sections among the states considered, highlighting the importance of metastable populations in W I modelling. In addition, radiative transition probabilities for selected prominent transitions are calculated and compared with existing data. The present dataset is expected to be valuable for collisional-radiative modelling and spectroscopic diagnostics of tungsten plasmas in the 1-50 eV range.

physics.atom-ph↗

arXiv2026

Gyrokinetic simulation of the effect of transient fueling on plasma turbulence in ADITYA-U tokamak

The gradient-driven microturbulence in ADITYA-U tokamak plasmas has been suppressed by injecting short gas puffs. The suppression of microturbulence increases the core temperature and subsequently the energy confinement time following the gas puff. The gas injection modifies the radial density profile, making it relatively flatter near the mid-radius. Global electrostatic gyrokinetic simulations show that this modification to the radial density profile due to gas injection suppresses the existing trapped electron mode (TEM). Simulation results show that the TEM-dominated turbulence suppression reduces the turbulence-driven heat transport, leading to an increase in core temperature. Applying multiple periodic gas-puffs leads to multiple periodic events of TEM suppression, improving the overall energy confinement time, and is used as an active control mechanism to influence microturbulence in ADITYA-U tokamak.

physics.plasm-ph↗

arXiv2026

Effect of static magnetic island on ITG of ADITYA-U tokamak

Magnetic islands play a crucial role in regulating plasma confinement in tokamaks by interacting with micro-instabilities, such as the ion temperature gradient (ITG) mode. This work presents a detailed investigation of the effects of static magnetic islands on ITG instability, relevant to the ADITYA-U tokamak, using the Global Gyrokinetic Code in Cylindrical Coordinates (G2C3), a particle-in-cell (PIC) framework that employs a neural-network-assisted projection scheme. A two-phase simulation strategy is adopted. In the first phase, static magnetic islands with mode numbers (m, n) = (2, 1) and (3, 1) are introduced by perturbing the equilibrium magnetic flux functions. Particle dynamics within these modified topologies result in the flattening of plasma density profiles in the island regions, confirming island formation and its impact on the equilibrium profiles. In the second phase, the flattened profiles serve as new equilibria for linear electrostatic gyrokinetic simulations with adiabatic electrons, enabling the study of the modified ITG behavior. Magnetic islands significantly restructure the ITG mode, producing a spatial redistribution of potential fluctuations within and around the island region. Moreover, as the island width increases, the growth rates of different toroidal ITG modes converge, suggesting a universal stabilization trend. A comparison between the (2,1) and (3,1) islands indicates that higher-q islands lead to a more spatially extended ITG mode structure, reflecting the longer magnetic connection lengths and weaker curvature drive at outer flux surfaces. These results demonstrate the pivotal role of island-induced equilibrium modifications in determining ITG stability and mode structure in tokamak plasmas.

physics.plasm-ph↗

arXiv2026

Sensitivity of External Magnetic Field on the Change in Cross-section of a Toroidal Current

Due to any toroidal current column, the magnetic field is found to be sensitive as well as insensitive to its cross-sectional area depending on location of subject point, as predicted by numerical approaches [S. Aich, J. Thakkar, and J. Ghosh, Plasma Fusion Res. 17, 2403055 (2022)], and hence the presence of an angle of invariance is found to be present for any toroidal geometry. Present study aims to validate those numerical observations using the measured magnetic field due to Aditya Upgrade tokamak plasma.

physics.plasm-ph↗

arXiv2025

LLM Reasoning for Cold-Start Item Recommendation

Large Language Models (LLMs) have shown significant potential for improving recommendation systems through their inherent reasoning capabilities and extensive knowledge base. Yet, existing studies predominantly address warm-start scenarios with abundant user-item interaction data, leaving the more challenging cold-start scenarios, where sparse interactions hinder traditional collaborative filtering methods, underexplored. To address this limitation, we propose novel reasoning strategies designed for cold-start item recommendations within the Netflix domain. Our method utilizes the advanced reasoning capabilities of LLMs to effectively infer user preferences, particularly for newly introduced or rarely interacted items. We systematically evaluate supervised fine-tuning, reinforcement learning-based fine-tuning, and hybrid approaches that combine both methods to optimize recommendation performance. Extensive experiments on real-world data demonstrate significant improvements in both methodological efficacy and practical performance in cold-start recommendation contexts. Remarkably, our reasoning-based fine-tuned models outperform Netflix's production ranking model by up to 8% in certain cases.

cs.IR↗

arXiv2025

Massively Multimodal Foundation Models: A Framework for Capturing Interactions with Specialized Mixture-of-Experts

Modern applications increasingly involve many heterogeneous input streams, such as clinical sensors, wearable device data, imaging, and text, each with distinct measurement models, sampling rates, and noise characteristics. We define this as massively multimodal setting, where each sensor constitutes a separate modality. As modality counts grow, capturing their complex, time-varying interactions such as delayed physiological cascades between sensors, has becomes essential yet challenging. Mixture-of-Experts (MoE) architectures are naturally suited for this setting since their sparse routing mechanism enables efficient scaling across many modalities. However, existing MoE architectures route tokens based on similarity alone, overlooking the rich temporal dependencies across modalities: this prevents the model from capturing delayed cross-modal effects, leading to suboptimal expert specialization and reduced accuracy. We propose a framework that explicitly quantifies temporal dependencies between modality pairs across multiple discrete time intervals, defined as delays between an event in one input stream and its manifested effect in another, and uses these to guide MoE routing. A interaction-aware router dispatches tokens to specialized experts based on interaction type. This principled routing enables experts to learn generalizable interaction-processing skills. Experiments across healthcare, activity recognition, and affective computing benchmarks demonstrate substantial performance gains and interpretable routing patterns aligned with domain knowledge.

cs.LG↗

arXiv2025

Simulation study of neutral tungsten emissions for fusion applications

The article reports electron-impact excitation cross-sections and rate coefficients for neutral tungsten for three transitions (400.87 nm, 429.46 nm, and 430.21 nm) using the relativistic distorted wave approach within the flexible atomic code. Some of these lines are also observed in tokamak plasma. Cross-sections are computed for incident electron energy up to 30 keV. The energy levels in flexible atomic code were corrected to match the NIST database. The electron impact excitation rate coefficients are also provided.

physics.plasm-ph↗

arXiv2025

PRISM: A MATLAB-Based Application for Structured Probe Data Management and Visualization in Tokamak Diagnostics

The successful operation of tokamak experiments requires accurate documentation, tracking, and visualization of diagnostic instruments, particularly electrical probes. Traditionally, this metadata is maintained manually through handwritten logbooks or semi-digital spreadsheets, leading to inefficiencies and human errors. In response to these challenges, we present PRISM (Probe Registration and Information System for Monitoring)-a MATLAB-based application developed using App Designer. PRISM provides a graphical user interface (GUI) that facilitates structured probe registration, metadata storage, and both 2D and 3D spatial visualization in tokamak geometries. Tested with data from the ADITYA-U tokamak, PRISM helps users enter information accurately, retrieve metadata easily, and visualize probe setups. The tool is built in a flexible way, is not limited to a specific setup, and could potentially support future developments such as digital twins and real-time control systems.

physics.plasm-ph↗

arXiv2025

A Multi-Stage Large Language Model Framework for Extracting Suicide-Related Social Determinants of Health

Background: Understanding social determinants of health (SDoH) factors contributing to suicide incidents is crucial for early intervention and prevention. However, data-driven approaches to this goal face challenges such as long-tailed factor distributions, analyzing pivotal stressors preceding suicide incidents, and limited model explainability. Methods: We present a multi-stage large language model framework to enhance SDoH factor extraction from unstructured text. Our approach was compared to other state-of-the-art language models (i.e., pre-trained BioBERT and GPT-3.5-turbo) and reasoning models (i.e., DeepSeek-R1). We also evaluated how the model's explanations help people annotate SDoH factors more quickly and accurately. The analysis included both automated comparisons and a pilot user study. Results: We show that our proposed framework demonstrated performance boosts in the overarching task of extracting SDoH factors and in the finer-grained tasks of retrieving relevant context. Additionally, we show that fine-tuning a smaller, task-specific model achieves comparable or better performance with reduced inference costs. The multi-stage design not only enhances extraction but also provides intermediate explanations, improving model explainability. Conclusions: Our approach improves both the accuracy and transparency of extracting suicide-related SDoH from unstructured texts. These advancements have the potential to support early identification of individuals at risk and inform more effective prevention strategies.

cs.CL↗

arXiv2025

Favorable modifications of Scrape-Off Layer (SOL) heat flux width through pulsed fuelling in ADITYA-U Tokamak

Enhancement of the scrape-off layer (SOL) heat flux width has been observed in the ADITYA-U Tokamak following the injection of short fuel gas pulses. A notable reduction in parallel heat flux near the last closed flux surface (LCFS) is observed after each pulse. Comparative analysis indicates that pulsed fuelling is more effective in mitigating heat flux with improved core confinement than continuous gas feeding via real-time density control. Analytical and simulation works are also carried out for validation of experimental results. The analytical model shows that SOL width modification cannot be attributed solely to the decrease of temperature due to gas pulse injection; cross-field plasma diffusion also needs to increase. Simulations with the UEDGE code suggest that an increase in both the cross-field diffusion coefficient and inward pinch velocity is necessary to replicate the experimentally observed broadening of the heat flux SOL width. These findings provide insights into efficient SOL heat flux control strategies for future fusion devices.

physics.plasm-ph↗

arXiv2025

Identification and Characterization of a New Disruption Regime in ADITYA-U Tokamak

Disruptions continue to pose a significant challenge to the stable operation and future design of tokamak reactors. A comprehensive statistical investigation carried out on the ADITYA-U tokamak has led to the observation and characterization of a novel disruption regime. In contrast to the conventional Locked Mode Disruption (LMD), the newly identified disruption exhibits a distinctive two-phase evolution: an initial phase characterized by a steady rise in mode frequency with a nonlinearly saturated amplitude, followed by a sudden frequency collapse accompanied by a pronounced increase in amplitude. This behaviour signifies the onset of the precursor phase on a significantly shorter timescale. Clear empirical thresholds have been identified to distinguish this disruption type from conventional LMD events, including edge safety factor, current decay coefficient, current quench (CQ) time, and CQ rate. The newly identified disruption regime is predominantly governed by the (m/n = 2/1) drift-tearing mode (DTM), which, in contrast to typical disruptions in the ADITYA-U tokamak that involve both m/n = 2/1 and 3/1 modes, consistently manifests as the sole dominant instability. Initiated by core temperature hollowing, the growth of this mode is significantly enhanced by a synergistic interplay between a strongly localized pressure gradient and the pronounced steepening of the current density profile in the vicinity of the mode rational surface.

physics.plasm-ph↗

arXiv2025

OMAC: A Holistic Optimization Framework for LLM-Based Multi-Agent Collaboration

Agents powered by advanced large language models (LLMs) have demonstrated impressive capabilities across diverse complex applications. Recently, Multi-Agent Systems (MAS), wherein multiple agents collaborate and communicate with each other, have exhibited enhanced capabilities in complex tasks, such as high-quality code generation and arithmetic reasoning. However, the development of such systems often relies on handcrafted methods, and the literature on systematic design and optimization of LLM-based MAS remains limited. In this work, we introduce \textbf{OMAC}, a general framework designed for holistic optimization of LLM-based MAS. Specifically, we identify five key optimization dimensions for MAS, encompassing both agent functionality and collaboration structure. Building upon these dimensions, we first propose a general algorithm, utilizing two actors termed the Semantic Initializer and the Contrastive Comparator, to optimize any single dimension. Then, we present an algorithm for joint optimization across multiple dimensions. Extensive experiments demonstrate the superior performance of OMAC on diverse tasks against recent approaches.

cs.MA↗

arXiv2025

Between Linear and Sinusoidal: Rethinking the Time Encoder in Dynamic Graph Learning

Dynamic graph learning is essential for applications involving temporal networks and requires effective modeling of temporal relationships. Seminal attention-based models like TGAT and DyGFormer rely on sinusoidal time encoders to capture temporal dependencies between edge events. Prior work justified sinusoidal encodings because their inner products depend on the time spans between events, which are crucial features for modeling inter-event relations. However, sinusoidal encodings inherently lose temporal information due to their many-to-one nature and therefore require high dimensions. In this paper, we rigorously study a simpler alternative: the linear time encoder, which avoids temporal information loss caused by sinusoidal functions and reduces the need for high-dimensional time encoders. We show that the self-attention mechanism can effectively learn to compute time spans between events from linear time encodings and extract relevant temporal patterns. Through extensive experiments on six dynamic graph datasets, we demonstrate that the linear time encoder improves the performance of TGAT and DyGFormer in most cases. Moreover, the linear time encoder can lead to significant savings in model parameters with minimal performance loss. For example, compared to a 100-dimensional sinusoidal time encoder, TGAT with a 2-dimensional linear time encoder saves 43% of parameters and achieves higher average precision on five datasets. While both encoders can be used simultaneously, our study highlights the often-overlooked advantages of linear time features in modern dynamic graph models. These findings can positively impact the design choices of various dynamic graph learning architectures and eventually benefit temporal network applications such as recommender systems, communication networks, and traffic forecasting.

cs.LG↗

arXiv2025

Effect of convective transport in edge/SOL plasmas of ADITYA-U tokamak

The 2-D edge plasma fluid transport code, UEDGE has been used to simulate the edge region of circular limiter plasmas of ADITYA-U for modelling the measured electron density profile. The limiter geometry of ADITYA-U has been introduced in the UEDGE code, which is primarily developed and used for divertor configuration. The computational mesh defining the limiter geometry is generated by a routine developed in-house, and has successfully been integrated with the UEDGE code to simulate the edge plasma parameters of ADITYA-U. The radial profiles of edge and scrape-off layer (SOL) electron density, ne and temperature are obtained from the simulations and used to model the measured ne profile using Langmuir probe array. It has been found that a convective velocity, vconv. is definitely needed in addition to the constant perpendicular diffusion coefficient, D throughout the edge and SOL regions to model the edge ne profile. The obtained vconv. is inward and radially constant with a value of 1.5 m/s and the radially constant D is ~ 0.2 m 2/s. The value of D ~ 0.2 m2/s is found to be much less than fluctuation induced diffusivities and lies in-between the neoclassical diffusivity and Bohm diffusivity estimated in the edge-SOL region of ADITYA-U tokamak. Furthermore, the transport of radial electron heat flux is found to be maximizing near the limiter tip location in the poloidal plane.

physics.plasm-ph↗

arXiv2025

A multi-purpose reciprocating probe drive system for studying the effect of gas-puffs on edge plasma dynamics in the ADITYA-U tokamak

This article reports the development of a versatile high-speed reciprocating drive system (HRDS) with interchangeable probe heads to characterize the edge plasma region of ADITYA-U tokamak. This reciprocating probe drive system consisting of Langmuir and magnetic probe heads, is designed, fabricated, installed, and operated for studying the extent of fuel/impurity gas propagation and its influence on plasma dynamics in the far-edge region inside the last closed magnetic flux surface (LCFS). The HRDS is driven by a highly accurate, easy-to-control, dynamic, brushless, permanently excited synchronous servo motor operated by a PXI-commanded controller. The system is remotely operated and allows for precise control of the speed, acceleration, and distance traveled of the probe head on a shot-to-shot basis, facilitating seamless control of operations according to experimental requirements. Using this system, consisting of a linear array of Langmuir probes, measurements of plasma density, temperature, potential, and their fluctuations revealed that the fuel gas-puff impact these mean and fluctuating parameters up to three to four cm inside the LCFS. Attaching an array of magnetic probes to this system led to measurements of magnetic fluctuations inside the LCFS. The HRDS system is fully operational and serves as an important diagnostic tool for ADITYA-U tokamak.

physics.plasm-ph↗