arXiv subjects

Rahul Ghosh

Publications and source records attributed to Rahul Ghosh.

At least 19 recordsLinked to original sources

arXiv2026

Calibration-Constrained Diagnostics for Evaluating LLM Self-Evaluation

We introduce a reusable evaluation artifact for diagnosing whether LLM self-reported confidence can support confidence-gated act-or-defer decisions. The artifact is designed for evaluation rather than safety certification: it provides a black-box protocol, prompts, judge-reference labels, model outputs, reproduction scripts, and calibration-constrained diagnostic metrics for auditing LLM self-evaluation across math, factual, and commonsense reasoning tasks. Our central finding is that aggregate uncertainty summaries can produce deployment-misleading rankings: a model that appears strongest under standard aggregate calibration metrics can fail under category-level calibration constraints, while a model with worse aggregate calibration can exhibit stronger calibration-constrained behavior in the positive-margin region where high self-selection accuracy and tight calibration coexist. To expose this failure mode, we define PVC and C-PVC as category-level diagnostics over an accuracy-by-calibration surface: PVC counts categories whose self-evaluation accuracy exceeds a threshold, while C-PVC additionally requires category-level calibration error below a tolerance. We aggregate these diagnostics using Volume Under Surface (VUS) and a positive-margin restriction; they are VC/fat-shattering motivated but are empirical evaluation proxies, not literal VC-dimension estimates. Across 13 models and four benchmarks, the artifact shows weak confidence-correctness separation, stable diagnostic rankings under alternative categorizations and judge-label perturbations, and persistent positive-margin failures not captured by ECE or Brier score. Confidence-gated self-evaluation therefore requires calibration-constrained, category-level diagnostics, and we provide a reusable artifact for applying them under black-box access.

cs.AI↗

arXiv2026

PIER: Physics-Informed Environmental Retrieval for Time-Series Modeling

Accurate modeling of environmental systems is fundamental to scientific understanding and decision-making, yet remains challenging because observations are limited and physical dynamics vary across systems. Retrieval-augmented approaches offer a natural path to transfer knowledge across systems, but standard embedding-based retrieval does not guarantee consistency of underlying physical processes, since scenarios with similar embeddings may arise from different underlying mechanisms. We propose Physics-Informed Environmental Retrieval (PIER), a model-agnostic framework that augments embedding-based retrieval with a physics-aware stream that scores candidates by flux-response consistency with the target, using local verifiers trained on physics-derived flux features. A weight adjustment mechanism then learns per-scenario weights that adaptively balance the two retrieval streams based on diagnostic features summarizing physics-stream reliability. Experiments on 356 lakes across the Midwestern United States spanning 41 years show that PIER consistently outperforms baselines for water temperature and dissolved oxygen prediction, and serves as a general augmentation strategy across diverse backbones.

cs.LG↗

arXiv2026

Correlation-Induced Topological Reconstruction in a Periodically Driven Kagome Mott Insulator

We study the interplay between electronic correlations and circularly polarized periodic driving in the Hubbard model on a Kagome lattice. Using Brillouin-Wigner perturbation theory, we derive an effective Floquet Hamiltonian in which the drive renormalizes the bare hopping and generates chiral nearest- and next-nearest-neighbor terms, producing repeated topological transitions and a flat band whose position is continuously tunable across the spectrum. Within slave-rotor mean-field theory, we show that the resulting Mott transition is strongly non-monotonic in the drive amplitude, yielding repeated metal-insulator transitions, and that the spinon excitations inside the Mott phase acquire a band topology distinct from that of the non-interacting Floquet bands. This correlation-driven topological reconstruction produces emergent flat-band spinon insulators inaccessible to either driving or interactions alone. Our results establish periodically driven Kagome systems as a platform for engineering correlated topological flat-band physics out of equilibrium, with proposed realizations in ultracold atomic lattices.

cond-mat.str-el↗

arXiv2026

Is GraphRAG Needed? From Basic RAG to Graph-/Agentic Solutions with Context Optimization

As advanced RAG variants like GraphRAG and Agentic RAG emerge, one leading question is when and how to use them. Here, we introduce a framework for different RAG scenarios evaluation and comparison on semi-structured knowledge bases, including regular RAG, GraphRAG, Modular RAG and Agentic RAG. We provide implementation for 9 standardized RAG scenarios, and conduct experiments for a comprehensive comparison. These scenarios are designed for real use cases regarding data and domain restrictions, spanning from simple document-based retrieval to advanced features such as hybrid text-graph retrieval, integration with computed or pre-defined domain knowledge graphs, agentic multi-step planning, and agent-graph integration. Besides, we present a novel context engineering method for GraphRAG and Agentic RAG, addressing the context/memory overflow issues, efficiently managing text and graph retrievals with new representations and agentic loop design, leading to 19%-53% reduction on token usage. Moreover, further analysis identifies a retrieval-generation gap where expanded retrieval does not proportionally improve generation quality, suggesting retrieval-oriented metrics overstate advanced retrieval benefits. This work provides data-driven insights on when and how to use them for building production-ready intelligent RAG systems.

cs.CL↗

arXiv2026

Quantifying Consistency in LLM Logical Reasoning via Structural Uncertainty

Large language models can arrive at the same answer through reasoning paths that are unstable, contradictory, or difficult to rank consistently -- a failure mode especially prevalent in multi-step deductive reasoning. Existing methods assess reliability primarily through output dispersion -- measuring how much sampled answers differ -- but this discards a complementary signal: whether the model can consistently rank competing reasoning candidates. We propose structural uncertainty, a consistency-aware framework derived from the stability of self-preference-induced rankings over sampled reasoning solutions. Given a query, we generate multiple candidate solutions and ask the model to judge pairwise preferences among its own outputs. We aggregate self-preferences into ranking distributions via Bradley-Terry modeling with PageRank, and decompose the signal into two entropy-based components: across-trial ranking instability and within-trial candidate ambiguity. Across five LLMs and eight benchmarks, structural signals provide information complementary to answer dispersion: on logical and mathematical reasoning tasks, the combination improves identification of unreliable instances, while on factual retrieval the structural signal collapses toward uniformity, diagnosing a regime boundary where reasoning-level consistency evaluation is uninformative. The two components relate differently to accuracy: within-trial ambiguity correlates positively with correctness -- consistent with settings where multiple plausible solution paths remain competitive -- while across-trial instability correlates negatively, signaling unreliable reasoning. Structural uncertainty is best understood not as a universal confidence estimator, but as a regime-sensitive evaluator of logical reasoning consistency.

cs.AI↗

arXiv2026

Quantum sensing with discrete time crystals in the Lipkin-Meshkov-Glick Model

Quantum phase transitions have been shown to be highly beneficial for quantum sensing, owing to diverging quantum Fisher information close to criticality. In this work we consider a periodically modulated Lipkin-Meshkov-Glick model to show that discrete time crystal (DTC) phase transition in this setup can enable us to achieve quantum-enhanced high-precision sensing of field strength. We employ a detailed finite-size scaling analysis, a time-averaged Inverse Participation Ratio analysis, and mean-field analysis in the thermodynamic limit, to determine the critical properties of this second-order phase transition. Our studies provide a comprehensive understanding of how quantum criticality in DTCs involving long-range interactions can be harnessed for advanced quantum sensing applications.

quant-ph↗

arXiv2026

Stabilizing boundary time crystals through Non-markovian dynamics

We study Boundary time crystals (BTCs) in the presence of non-Markovian dynamics. In contrast to BTCs observed in earlier works in the Markovian regime, we show that non-Markovian dynamics can be highly beneficial for stabilizing BTCs over a wide range of parameter values, even in the presence of intermediate rates of dissipation. Notably, we also observe the emergence of higher-order limit cycles (HO-LCs) for some parameter regimes. We analyze the effect of non-Markovian dynamics on BTCs and HO-LCs using quantum Fisher information, time-averaged magnetization, a measure of non-Markovianity, and a dynamical phase diagram, all of which show complex behaviors with changing non-Markovianity parameters. Our studies can pave the way for stabilizing time crystals in dissipative systems, as well as lead to studies on varied dissipative dynamics on time translational symmetry breaking.

quant-ph↗

arXiv2026

Towards Knowledge Guided Pretraining Approaches for Multimodal Foundation Models: Applications in Remote Sensing

Self-supervised learning has emerged as a powerful paradigm for pretraining foundation models using large-scale data. Existing pretraining approaches predominantly rely on masked reconstruction or next-token prediction strategies, demonstrating strong performance across various downstream tasks, including geoscience applications. However, these approaches do not fully capture the knowledge of causal interplay between different geospatial and environmental variables. To address this limitation, we propose Knowledge Guided Variable-Step Forecasting (KG-VSF), a novel pretraining task that models forecasting as a conditional generation task, where driver variables (e.g., weather) inform the prediction of response variables (e.g., satellite imagery). We demonstrate that pretraining in such a fashion leads to strong embeddings which give enhanced performance when finetuned on downstream tasks where capturing this causality matters such as pixel wise crop type mapping, soil moisture estimation and forecasting, missing image prediction, and future image forecasting when compared to finetuning embeddings from other standard pretraining approaches.

cs.CV↗

arXiv2026

CompAgent: An Agentic Framework for Visual Compliance Verification

Visual compliance verification is a critical yet underexplored problem in computer vision, especially in domains such as media, entertainment, and advertising where content must adhere to complex and evolving policy rules. Existing methods often rely on task-specific deep learning models trained on manually labeled datasets, which are costly to build and limited in generalizability. While recent Multimodal Large Language Models (MLLMs) offer broad real-world knowledge and policy understanding, they struggle to reason over fine-grained visual details and apply structured compliance rules effectively on their own. In this paper, we propose CompAgent, the first agentic framework for visual compliance verification. CompAgent augments MLLMs with a suite of visual tools-such as object detectors, face analyzers, NSFW detectors, and captioning models-and introduces a planning agent that dynamically selects appropriate tools based on the compliance policy. A compliance verification agent then integrates image, tool outputs, and policy context to perform multimodal reasoning. Experiments on public benchmarks show that CompAgent outperforms specialized classifiers, direct MLLM prompting, and curated routing baselines, achieving up to 76% F1 score and a 10% improvement over the state-of-the-art on the UnsafeBench dataset. Our results demonstrate the effectiveness of agentic planning and robust tool-augmented reasoning for scalable, accurate, and adaptable visual compliance verification.

cs.CV↗

arXiv2026

Learning Flow Distributions via Projection-Constrained Diffusion on Manifolds

We present a generative modeling framework for synthesizing physically feasible two-dimensional incompressible flows under arbitrary obstacle geometries and boundary conditions. Whereas existing diffusion-based flow generators either ignore physical constraints, impose soft penalties that do not guarantee feasibility, or specialize to fixed geometries, our approach integrates three complementary components: (1) a boundary-conditioned diffusion model operating on velocity fields; (2) a physics-informed training objective incorporating a divergence penalty; and (3) a projection-constrained reverse diffusion process that enforces exact incompressibility through a geometry-aware Helmholtz-Hodge operator. We derive the method as a discrete approximation to constrained Langevin sampling on the manifold of divergence-free vector fields, providing a connection between modern diffusion models and geometric constraint enforcement in incompressible flow spaces. Experiments on analytic Navier-Stokes data and obstacle-bounded flow configurations demonstrate significantly improved divergence, spectral accuracy, vorticity statistics, and boundary consistency relative to unconstrained, projection-only, and penalty-only baselines. Our formulation unifies soft and hard physical structure within diffusion models and provides a foundation for generative modeling of incompressible fields in robotics, graphics, and scientific computing.

physics.flu-dyn↗

arXiv2025

TelcoAI: Advancing 3GPP Technical Specification Search through Agentic Multi-Modal Retrieval-Augmented Generation

The 3rd Generation Partnership Project (3GPP) produces complex technical specifications essential to global telecommunications, yet their hierarchical structure, dense formatting, and multi-modal content make them difficult to process. While Large Language Models (LLMs) show promise, existing approaches fall short in handling complex queries, visual information, and document interdependencies. We present TelcoAI, an agentic, multi-modal Retrieval-Augmented Generation (RAG) system tailored for 3GPP documentation. TelcoAI introduces section-aware chunking, structured query planning, metadata-guided retrieval, and multi-modal fusion of text and diagrams. Evaluated on multiple benchmarks-including expert-curated queries-our system achieves $87\%$ recall, $83\%$ claim recall, and $92\%$ faithfulness, representing a $16\%$ improvement over state-of-the-art baselines. These results demonstrate the effectiveness of agentic and multi-modal reasoning in technical document understanding, advancing practical solutions for real-world telecommunications research and engineering.

cs.LG↗

arXiv2025

Exactly solvable Stuart-Landau models in arbitrary dimensions

We use Clifford's geometric algebra to extend the Stuart-Landau system to dimensions $D >2$ and give an exact solution of the oscillator equations in the general case. At the supercritical Hopf bifurcation marked by a transition from stable fixed-point dynamics to oscillatory motion, the Jacobian matrix evaluated at the fixed point has $N=\lfloor{D/2}\rfloor$ pairs of complex conjugate eigenvalues which cross the imaginary axis simultaneously. For odd $D$ there is an additional purely real eigenvalue that does the same. Oscillatory dynamics is asymptotically confined to a hypersphere $\mathbb{S}^{D-1}$ and is characterised by extreme multistability, namely the coexistence of an infinite number of limiting orbits each of which has the geometry of a torus $\mathbb{T}^N$ on which the motion is either periodic or quasiperiodic. We also comment on similar Clifford extensions of other limit cycle oscillator systems and their generalisations.

nlin.CD↗

arXiv2025

Learning to Retrieve for Environmental Knowledge Discovery: An Augmentation-Adaptive Self-Supervised Learning Framework

The discovery of environmental knowledge depends on labeled task-specific data, but is often constrained by the high cost of data collection. Existing machine learning approaches usually struggle to generalize in data-sparse or atypical conditions. To this end, we propose an Augmentation-Adaptive Self-Supervised Learning (A$^2$SL) framework, which retrieves relevant observational samples to enhance modeling of the target ecosystem. Specifically, we introduce a multi-level pairwise learning loss to train a scenario encoder that captures varying degrees of similarity among scenarios. These learned similarities drive a retrieval mechanism that supplements a target scenario with relevant data from different locations or time periods. Furthermore, to better handle variable scenarios, particularly under atypical or extreme conditions where traditional models struggle, we design an augmentation-adaptive mechanism that selectively enhances these scenarios through targeted data augmentation. Using freshwater ecosystems as a case study, we evaluate A$^2$SL in modeling water temperature and dissolved oxygen dynamics in real-world lakes. Experimental results show that A$^2$SL significantly improves predictive accuracy and enhances robustness in data-scarce and atypical scenarios. Although this study focuses on freshwater ecosystems, the A$^2$SL framework offers a broadly applicable solution in various scientific domains.

cs.LG↗

arXiv2025

Quantum Phases in a Two-Dimensional Generalized interacting SSH Model

We study interaction-driven quantum phases in a two-dimensional generalized Su-Schrieffer-Heeger (SSH) defined on a square lattice with inequivalent nearest-neighbor hopping, next-nearest-neighbor hopping and a staggered on-site potential. In the non-interacting limit, the model hosts either a quadratic band touching (QBT) at the Brillouin-zone center or symmetry-protected Dirac nodes, depending on microscopic parameters. In the parameter regime with a QBT, our self-consistent Hartree-Fock analysis shows that weak to intermediate interactions can spontaneously break time-reversal symmetry and stabilize a quantum anomalous Hall (QAH) insulating phase with a finite Chern number. Interestingly, this QAH phase is found to weakly break lattice-symmetries, leading to a small but finite nematic bond order. This is in contrast to the standard QAH phase in checkerboard lattice, which preserves all lattice symmetries. Additionally, we find an enhanced bond-nematic Dirac semimetallic (BNDS) phase due to asymmetric hopping, which is thought to be absent in the Hartree-Fock approach. In the parameter regimes where QBT splits into two Dirac nodes, the QAH phase survives up to a finite staggered on-site potential. However, as the staggered potential increases, the QAH phase is suppressed while the BNDS phase grows stronger.

cond-mat.str-el↗

arXiv2025

RAG for Geoscience: What We Expect, Gaps and Opportunities

Retrieval-Augmented Generation (RAG) enhances language models by combining retrieval with generation. However, its current workflow remains largely text-centric, limiting its applicability in geoscience. Many geoscientific tasks are inherently evidence-hungry. Typical examples involve imputing missing observations using analog scenes, retrieving equations and parameters to calibrate models, geolocating field photos based on visual cues, or surfacing historical case studies to support policy analyses. A simple ``retrieve-then-generate'' pipeline is insufficient for these needs. We envision Geo-RAG, a next-generation paradigm that reimagines RAG as a modular retrieve $\rightarrow$ reason $\rightarrow$ generate $\rightarrow$ verify loop. Geo-RAG supports four core capabilities: (i) retrieval of multi-modal Earth data; (ii) reasoning under physical and domain constraints; (iii) generation of science-grade artifacts; and (iv) verification of generated hypotheses against numerical models, ground measurements, and expert assessments. This shift opens new opportunities for more trustworthy and transparent geoscience workflows.

cs.ET↗

arXiv2025

Generating discrete time crystals through optimal control

In this work we use optimal control to generate Discrete Time Crystals (DTC) in generic many-body quantum systems. We define appropriate cost functions, which, when optimized, result in the formation of DTCs. This hitherto unexplored method represents DTCs as an optimization problem, and allows us to find non-trivial realistic periodic control pulses and parameter regimes which result in spontaneous breaking of time-translational symmetry in quantum systems. We exemplify our approach using many-body quantum systems in the presence, as well as absence of dissipation. We also discuss possible experimental realization of the control protocol for generating DTCs.

quant-ph↗

arXiv2025

Electric and thermoelectric response for Weyl and multi-Weyl semimetals in planar Hall configurations including the effects of strain

We investigate the response tensors in planar Hall (or planar thermal Hall) configurations such that a three-dimensional Weyl or multi-Weyl semimetal is subjected to the influence of an electric field $\mathbf E $ (or temperature gradient $\nabla_{\mathbf r } T$) and an effective magnetic field $\mathbf B^{\rm tot}$, which are oriented at a generic angle with respect to each other. The effective magnetic field consists of two parts: (a) an actual/physical magnetic field $\mathbf B $, and (b) an emergent magnetic field $\mathbf B_5 $ which quantifies the elastic deformations of the sample. $\mathbf B_5 $ is an axial pseudomagnetic field because it couples to conjugate nodal points with opposite chiralities with opposite signs. We study the interplay of the orientations of these two components of $\mathbf B^{\rm tot}$ with respect to the direction of the electric field (or temperature gradient) and elucidate how it affects the characteristics involving the chirality of the node. Additionally, we show that the magnitude and sharpness of the conductivity tensor profiles strongly depend on the value of the topological charge at the node in question.

cond-mat.mes-hall↗

arXiv2025

Hierarchically Disentangled Recurrent Network for Factorizing System Dynamics of Multi-scale Systems: An application on Hydrological Systems

We present a framework for modeling multi-scale processes, and study its performance in the context of streamflow forecasting in hydrology. Specifically, we propose a novel hierarchical recurrent neural architecture that factorizes the system dynamics at multiple temporal scales and captures their interactions. This framework consists of an inverse and a forward model. The inverse model is used to empirically resolve the system's temporal modes from data (physical model simulations, observed data, or a combination of them from the past), and these states are then used in the forward model to predict streamflow. Experiments on several catchments from the National Weather Service North Central River Forecast Center show that FHNN outperforms standard baselines, including physics-based models and transformer-based approaches. The model demonstrates particular effectiveness in catchments with low runoff ratios and colder climates. We further validate FHNN on the CAMELS (Catchment Attributes and MEteorology for Large-sample Studies), which is a widely used continental-scale hydrology benchmark dataset, confirming consistent performance improvements for 1-7 day streamflow forecasts across diverse hydrological conditions. Additionally, we show that FHNN can maintain accuracy even with limited training data through effective pre-training strategies and training global models.

cs.LG↗