Search arXiv⌕ Search

arXiv subjects

Aman Kumar

Publications and source records attributed to Aman Kumar.

At least 19 recordsLinked to original sources

Energy-resolved current response and the crossover between two approximately linear-in-$T$ resistivity regimes in the infinite-$U$ Hubbard model

We study charge transport in the two-dimensional infinite-$U$ Hubbard model with hopping amplitude $t$ using full exact diagonalization (ED) on $N=16$- and $18$-site square clusters with a single doped hole. We first provide a proof, treating degeneracies explicitly, that a constant dc $f_x$ function, defined as the sum of the squared current matrix elements connecting eigenstates at a given energy to eigenstates of nearly the same energy, averaged over degenerate initial states, is necessary and sufficient for strictly linear-in-$T$ resistivity: $ρ_x(T)=αT$ for all $T>0$ with a constant $α$. The numerically extrapolated dc $f_x$ is not exactly constant, but it is approximately flat over a broad region in the interior of the many-body spectrum and varies strongly within a scale $\varepsilon_{f_x}$ of the spectral edges. This scale $\varepsilon_{f_x}$ is related to a characteristic temperature $T_{\mathrm{cr}}$, which quantifies the leading departure of the resistivity from high-temperature linearity upon cooling. Within the studied temperature range $T/t \geq 0.05$, our ED resistivity results reveal a gradual crossover between high- and low-temperature approximately linear-in-$T$ regimes, with characteristic crossover temperatures $T_{\mathrm{cr}}/t\approx0.414$ and $0.467$ for $N=16$ and $18$, respectively. We also find a broad specific-heat peak associated with this gradual crossover, at $T/t \approx 0.70$ and $0.68$ for $N=16$ and $18$, respectively.

cond-mat.str-el↗

The Great Chat in the Sky: Aligning Exotic Astronomical Data to Language with AstroBridge

Astronomical observations include images, spectra, and light curves, while the scientific literature describes these objects in natural language. We present AstroBridge, a framework that connects these observations to an LLM through a literature-assisted caption generation pipeline. We used catalog metadata to match observations with scientific texts that describe the same objects. A vision-language model generates descriptive captions from representations of observations, matched passages, and available metadata, creating observation-caption pairs without manual captioning. We train AstroBridge to predict captions directly from raw observations. Modality-specific frozen encoders extract representations that a shared, trainable fusion network aligns with the embedding space of an LLM. In caption-based evaluation of spectra, AstroBridge achieves 91.5% accuracy in distinguishing quasars from galaxies, compared to 69.0% from the base model prior to fine-tuning. Limitations remain in galaxy morphology, classification of minority supernova classes, and the factual quality of generated captions. We believe that AstroBridge's framework is a step towards the goal of aligning the totality of astronomy's diverse, exotic data through the lingua franca of text.

astro-ph.IM↗

NutriVision: Ingredient-Conditioned Fusion and Prediction for Single-Image Food Nutrition Estimation

Nutrition estimation is a fundamental task in consumer diet tracking, clinical dietetics, chronic disease management, sports and hospital nutrition, and broader food computing systems. The existing approaches have progressed along two largely separate axes, vision models that rely on calibrated RGB-depth captures and ingredient-aware methods that use textual cues but use limited multimodal fusion. We introduce NutriVision, an end-to-end framework that leverages visual geometry and ingredient semantics to estimate calories, mass, fat content, carbohydrates, and protein from a single RGB image and an optional ingredient list. It obtains the unavailable depth modality using DepthAnything-V3 and encodes ingredient descriptions using CLIP. It integrates three complementary mechanisms: (1) an \emph{Ingredient-Conditioned Frequency-Aligned Fusion Module (IC-FAFM)}, which uses textual guidance to reweight and align RGB-depth frequency components; (2) an \emph{Ingredient-Aware Mask-based Prediction Head (IA-MPH)}, whose gating and channel masks are conditioned on food identity; and (3) modality-specific \emph{Internal Semantic Modeling (ISM)} blocks. On the Nutrition5k dataset, NutriVision achieves a mean PMAE of $\mathbf{13.60\pm0.10\%}$, outperforming our IGSMNet implementation by $0.89$ percentage points and OmniFood8k by $2.90$ percentage points (both $p<0.001$). The module-level ablations identify the ingredient-aware prediction head as the primary architectural contributor, improving mean PMAE by $1.50\pm0.17$ percentage points ($p<0.001$). These results demonstrate that ingredient-conditioned prediction and frequency-aware RGB-depth fusion provide measurable gains for single-image nutrient estimation. More broadly, NutriVision offers a practical route toward nutrition-assessment systems that exploit geometric and semantic cues without requiring specialized depth-sensing hardware

cs.CV↗

Native Geometry vs. Linear Recoverability: How Alignment Changes Platonic Scaling

Do larger models learn increasingly similar representations of the same world? The Platonic Representation Hypothesis (PRH) predicts that they should, but similar can mean native cosine geometry---nearest-neighbour structure induced by cosine similarity before any learned alignment---or linear recoverability: correspondence revealed by a supervised linear map. Using matched Legacy Survey and Hyper Suprime-Cam (HSC) embeddings across five model families, we find that Ridge alignment strengthens the relationship between model size and cross-dataset neighbourhood correspondence relative to native cosine geometry. The effect appears in all five families and persists when both representations are reduced to dimension, so it is not simply that wider embeddings give alignment more dimensions to work with. The learned maps remain far from rotations or uniform scaling, and Sparse Autoencoder (SAE) and Block-Sparse Featuriser (BSF) features give no advantage over the same alignment applied to dense embeddings. Model scale thus improves linear recoverability more clearly than native cosine-neighbourhood correspondence, without establishing convergence toward a shared native metric geometry.

cs.LG↗

Revisiting Non-Adaptive Set Membership Schemes: Removing the Logarithmic Factor

We study non-adaptive bit-probe schemes for static set membership. Each query reads a fixed number of bits and answers by strict majority. We revisit the random bipartite-graph construction of Garg and Radhakrishnan [STACS, 2017]. The main observation is that a slightly stronger expansion property already bounds the number of exceptional elements. This makes a separate probabilistic estimate for exceptional elements unnecessary. A detailed union-bound calculation then gives, for every fixed $t\ge5$, a logarithm-free existence bound of $O_t(m^{1/c}n^{1-1/c})$, where $c=\lfloor(t-1)/2\rfloor$. For odd $t$, this removes the logarithmic factor from the general upper bound of Garg and Radhakrishnan [STACS, 2017].

cs.DS↗

Haerter-Shastry kinetic magnetism and metallicity in the triangular Hubbard model

The fermionic Hubbard model, when combined with frustration, associated with the breaking of particle-hole symmetry, harbors a rich phase diagram. Theoretical findings associated with the nature of magnetism and metallicity across this phase diagram are now being actively investigated in triangular Hubbard cold atom and solid-state (moiré) based emulators. Building on the theoretical work of Haerter and Shastry, we explore the impact of kinetically frustrated magnetism, a phenomenon where antiferromagnetic order emerges without any underlying magnetic interactions, at finite hole density. We numerically study the infinite-$U$ triangular Hubbard model using the density matrix renormalization group algorithm and estimate the stability of the kinetically induced $120^{\circ}$ antiferromagnetic state to hole doping. Beyond the Haerter-Shastry regime, we find an intermediate phase with multimer (involving multiple correlated spins) stripes that eventually gives way to a paramagnet. We also find evidence of gapless charge excitations (metallicity) throughout the phase diagram for finite hole density. We discuss the implications at large, but finite, values of $U/t$, and investigate whether kinetic magnetism and superexchange collaborate or compete.

cond-mat.str-el↗

Diagnosing Fine-Grained Inconsistency Classification in Financial Disclosure Text

Financial disclosures may contain numerical, temporal, referential, factual, and policy inconsistencies that require different evidence and reasoning to diagnose. We study fine-grained inconsistency classification: given a passage known to contain a conflict, the goal is to identify its type among 11 categories. Using a fixed snapshot of the synthetic SBID-FD benchmark, we compare frozen and fine-tuned encoders, evidence-augmented classifiers, prompted large language models, and LoRA-adapted generative models under a shared evaluation protocol. Task-specific adaptation yields large improvements over frozen representations, and a fine-tuned 300M encoder performs competitively with substantially larger prompted and adapted models. We further study whether localizing the conflicting claims improves classification through matched predicted-span, reference-span, and distractor-span conditions. The results show that automatically extracted evidence provides additional signal but recovers only part of the benefit obtained from reference spans. Per-class and confusion analyses further reveal that some inconsistency types are especially sensitive to localization quality, whereas others remain difficult even when the relevant evidence is supplied. These findings identify evidence localization and fine-grained type discrimination as distinct challenges and show that compact supervised encoders are strong baselines for this task.

cs.CL↗

The Platonic Universe: Do Foundation Models See the Same Sky?

We investigate when foundation models converge towards shared representations, and how this convergence depends on model capacity, training regime, and model architecture. We take a `science-for-AI' approach, using astronomy as an experimental instrument to test the Platonic Representation Hypothesis and its Aristotelian refinement against an external physical reference. The historical success of astrophysics is evidence that a compact, modality-invariant description of galaxy observables exists, and so representation convergence toward reality should be measurable against the physical parameters astronomers already use. Given this framework, we evaluate eleven foundation model families (spanning classification, self-distillation, joint-embedding prediction, autoencoding, vision-language pre-training, and astro-specific architectures from $\mathcal{O}$(10M)${\to}\mathcal{O}$(10B) parameters) on crossmatched JWST, HSC, and Legacy imagery, and DESI spectroscopy. All models are evaluated frozen, with no astronomy-specific fine-tuning. We probe redshift, stellar mass, and sSFR via linear probes, and local (MKNN) and global (CKA) embedding geometry within families, between modalities, and across architectures. We find that physics performance scales predictably with capacity; probe directions align consistently with expected astrophysical correlations and selection effects; and local (not global) embedding alignment tracks physics performance, including between DESI spectra and HSC imagery---modalities that share essentially no low-level statistics. Our results support the ARH over the strict PRH, demonstrate astronomy's value as an experimental framework for neural representation learning, and suggest that astro-foundation models can build on general-purpose pre-trained architectures, capitalizing on the broader open machine learning community's already-spent computational investment.

astro-ph.IM↗

FiMI Banking: A Sovereign Model for Indian Retail Banking

Banks need conversational systems that can answer product questions, assist customers with account-related requests, and operate safely within strict operational and regulatory constraints. General-purpose language models do not reliably meet these requirements. They fall short when a task requires grounded information, correct tool use, or cautious handling of bank-specific sensitive situations. We introduce FiMI Banking, a controlled Indian retail-banking setting. We build it from vetted banking documents, structured ground truth, synthetic customer backgrounds, and banking tools. We evaluate two post-training approaches: preference optimization for response-level behavior, and reinforcement learning with verifiable rewards for multi-turn tool-use tasks. Preference optimization improves safe behavior substantially: out-of-scope refusal rises from 52% to 80%. Reinforcement learning improves edge-case performance from 0.509 to 0.718 and order-sensitive task performance from 0.590 to 0.679, while using 29% fewer generated tokens. These results show that preference optimization and verifiable-reward reinforcement learning address complementary requirements for reliable banking agents.

cs.AI↗

What AstroPT knows about galaxies, and what that can teach us about LLMs

Interpretability research increasingly asks when concepts emerge during training and whether linear probes recover real structure, but in language models these claims are hard to validate because language offers little ground-truth ordering of concepts or relationships among them. We propose the use of astronomical ground truth through AstroPT, a transformer trained on millions of galaxy images, as a calibration testbed. AstroPT is an LLM-like model trained within a domain where the difficulty ordering of concepts and the relations among them are known in advance. Probing frozen representations across checkpoints, layers, model sizes, and objective choices, we find that galaxy properties emerge in a fixed order that tracks their known difficulty---quantities written almost directly into the pixels (band magnitude) become decodable early in training and shallow in the network, while multiband/spectra based and inferred quantities (such as redshift and specific star formation rate) emerge later and deeper. This order is invariant to our tested training objectives, and scales in magnitude but not in sequence with capacity. Our linear probe directions further recover the known physical structure among galaxy properties. Our findings suggest that astronomy offers a controlled sandbox for calibrating mechanistic interpretability methods we otherwise apply to LLMs blind.

cs.LG↗

Ascorbic acid mitigates oxidative structural degradation in bovine spermatozoa: a label-free quantitative phase microscopy study

Oxidative stress is a key factor in low fertility outcomes during assisted reproduction technology (ART) and contributes to poor sperm quality. Conventional assessment relies on bright-field microscopy, which lacks the quantitative sensitivity to resolve subcellular structural and biophysical changes without exogenous contrast agents, a requirement that can introduce cytotoxic effects and compromise cell viability. This study uses QPM as a label-free, high-throughput method to demonstrate how ascorbic acid reduces the impact of oxidative structural degradation in Sahiwal bovine spermatozoa. To illustrate this pathology, severe oxidative stress was experimentally induced using hydrogen peroxide (H2O2), and two doses of ascorbic acid (1 mg/ml and 8 mg/ml) were tested to evaluate dose-dependent antioxidant recovery in structurally damaged sperm. QPM was used to quantify structural changes in key biophysical parameters, including dry mass, optical thickness, volume, surface area, sphericity, and surface area-to-volume ratio, as well as intracellular texture parameters. Under acute oxidative stress, bovine spermatozoa exhibited significant reductions in optical thickness and dry mass, alongside measurable changes in intracellular structural organization, collectively indicative of oxidative stress-induced morphological degradation. Co-treatment with ascorbic acid resulted in partial, dose-dependent structural preservation, with the 8 mg/ml formulation demonstrating statistically significant attenuation of these structural changes compared to the 1 mg/ml treatment group. These findings suggest that QPM-derived biophysical parameters may serve as promising, label-free structural indicators for characterizing oxidative damage in bovine spermatozoa. Future studies should incorporate functional validation to determine whether this structural preservation translates to improved outcomes in ART.

physics.optics↗

Ranking Companion: A Visual Analytics Approach to Item-Based Ranking with Hybrid Item Selection

Personalizing item ranking creation is a challenging task, especially when users lack knowledge of data attributes or the ability to express and formalize their attribute preferences. Item-based ranking creation is an approach allowing users to directly externalize preferences through known-item judgments rather than attribute-based scoring. However, a core challenge of item-based ranking is identifying and selecting representative candidate items for externalizing preferences. Existing approaches rely on singular item-selection methods, limiting flexibility and user control. To address this challenge, we present Ranking Companion, a visual analytics approach for item-based ranking that combines model-driven active learning with human-driven item-selection methods. By drawing from six complementary item-selection methods, users can externalize listwise preferences based on selected candidate items, while an iterative machine learning process with a ranking model calculates ranking results, presented to users alongside explanations for interpretation. We evaluated Ranking Companion in a formative user study with 10 participants, in which participants used each item-selection method across three iterations, revealing tradeoffs in perceived ranking quality across accuracy, diversity, novelty, transparency, control, and satisfaction. Ranking Companion contributes a unified interactive item selection space and provides preliminary empirical guidance toward the hybrid use of multiple complementary item-selection methods in personalized item-based ranking creation.

cs.HC↗

GRACE-RAG: Governed Retrieval Architecture for Canonical Evidence Synthesis, Enabling Lightweight Deployment in Closed-Domain Institutional Settings

Retrieval-Augmented Generation (RAG) systems are widely used in institutional question answering settings where responses must be grounded in authoritative documentation (Gao et al., 2023). In entity-dense domains where relevant information is distributed across heterogeneous documents, vector-only retrieval often produces fragmented evidence and increases dependence on inference-time reasoning (Zhao et al., 2024). This paper introduces GRACE-RAG, a retrieval-governed, graph-augmented RAG architecture that externalizes structural reasoning from the generative stage to a structured retrieval layer, resolving structural ambiguity offline, enabling deployment on self-hosted lightweight models calibrated to closed-domain institutional vocabulary. Experiments across three model capacities: Mistral 24B, GPT OSS 120B, and Gemini 2.5 Flash show consistent improvements in completeness, depth, and anticipatory coverage, with overall quality gains of up to 20% under mid-scale models, indicating that retrieval architecture governs structural quality over model scale, reducing computational and latency footprint without dependence on proprietary systems.

cs.IR↗

Knowledge Graphs, the Missing Link in Agentic AI-based Formal Verification

Recent advances in Large Language Models (LLMs) have enabled workflows that generate SystemVerilog Assertions (SVAs) from natural-language specifications, with the potential to accelerate Formal Verification (FV). However, high-quality assertion synthesis remains challenging because specifications are often ambiguous or incomplete and critical micro-architectural details reside in the Register Transfer Level (RTL). Many existing approaches treat the specification and RTL as loosely structured text, which weakens specification-to-RTL grounding and leads to semantic mismatches and frequent syntax failures during formal parsing and elaboration. This work addresses these limitations with a verification-centric Knowledge Graph (KG) constructed from structured Intermediate Representations (IRs) extracted from the specification, RTL, and formal-tool feedback, including syntax diagnostics, Counterexamples (CEXs), and coverage reports. The KG links requirements, design hierarchy, signals, assumptions, and properties to provide traceable, design-grounded context for generation. A multi-agent workflow queries and updates this KG to generate SVAs and to drive three refinement loops: syntax repair guided by tool diagnostics, CEX-guided correction using trace links, and coverage-directed property augmentation. Evaluation across seven benchmark designs indicates that KG-based context retrieval improves specification-to-RTL grounding and consistently produces compilable SVAs with low syntax-repair overhead. The approach achieves formal coverage ranging from 78.5% to 99.4%, though convergence exhibits design dependence with complex temporal and arithmetic reasoning remaining challenging for current LLM capabilities.

cs.AI↗

Melting temperature shifts from quantum fluctuations in generalized Wigner crystals

It is generally believed that quantum fluctuations collaborate with thermal fluctuations, effectively reducing transition temperatures (e.g. for melting of charge order). We show that this is not always the case and that the interplay between quantum and thermal fluctuations can be competitive. We find excellent motivation for addressing this thanks to the discovery of correlated insulating "generalized Wigner crystal" (GWC) states in hetero-bilayer transition metal dichalcogenide (WS$_2$/WSe$_2$) moiré systems [Y. Xu, et al., Nature 587, 214-218 (2020)]. We account for the impact of quantum effects on the melting temperature of GWCs, carrying out finite temperature Lanczos calculations on an extended Hubbard model on the triangular lattice (both with a double-gate screened potential, and the nearest neighbor model) for multiple electron densities. We show that quantum effects capture the shift relative to the classical estimates, which in some cases are more than 50 percent off from the experimental values. Then building on these numerical findings, we provide a qualitative picture that clarifies that while quantum melting of GWC (by increasing the bandwidth) naturally softens the ground state order parameter, it does not always decrease the melting temperature; conversely it can increase it. To do so we employ a finite temperature perturbation theory, treating the kinetic energy perturbatively on top of a classical Wigner crystal. Our predictions should be observable in future experiments where the bandwidth can be tuned.

cond-mat.str-el↗

Agentic AI-based Coverage Closure for Formal Verification

Coverage closure is a critical requirement in Integrated Chip (IC) development process and key metric for verification sign-off. However, traditional exhaustive approaches often fail to achieve full coverage within project timelines. This study presents an agentic AI-driven workflow that utilizes Large Language Model (LLM)-enabled Generative AI (GenAI) to automate coverage analysis for formal verification, identify coverage gaps, and generate the required formal properties. The framework accelerates verification efficiency by systematically addressing coverage holes. Benchmarking open-source and internal designs reveals a measurable increase in coverage metrics, with improvements correlated to the complexity of the design. Comparative analysis validates the effectiveness of this approach. These results highlight the potential of agentic AI-based techniques to improve formal verification productivity and support comprehensive coverage closure.

cs.AI↗

Type Iax supernovae as a source of iron-rich silicate dust

We model the formation of dust in the ejecta of Type Iax supernovae (SNe), which is a low-luminosity subclass of Type Ia SNe. A non-equilibrium chemical kinetic approach is adopted to trace the synthesis of molecules, molecular clusters, and dust grains in the ejecta of thermonuclear SNe. We find that Type Iax SNe provide conditions conducive to the formation of several O-rich dust species in the ejecta. Particularly, iron-rich silicates of chemical type FeSiO3, Fe2SiO4, and MgFeSiO4 are found to form in abundance, suggesting that the ejecta of low-luminosity thermonuclear SNe can be a site where a large fraction of iron is locked up in dust, unlike other stellar sources. The final mass of dust formed in the ejecta ranges between 10^{-5} and 10^{-4} Msun, where most of the dust forms between 1000 and 2000 days post-explosion. Apart from Fe-rich silicates, Mg-silicates, and silicon carbide are also formed in the ejecta of Type Iax SNe. When compared to the dust budget of typical Type Ia SNe, we find that the expected dust-to-ejecta mass ratio is 1 or 2 orders of magnitude larger in Type Iax SNe. We conclude that the ejecta of typical Type Ia SNe form a negligible amount of dust, in agreement with observation, while the low-luminosity subclass Type Iax SNe are potential producers of iron-rich silicates.

astro-ph.SR↗

Saarthi for AGI: Towards Domain-Specific General Intelligence for Formal Verification

Saarthi is an agentic AI framework that uses multi-agent collaboration to perform end-to-end formal verification. Even though the framework provides a complete flow from specification to coverage closure, with around 40% efficacy, there are several challenges that need to be addressed to make it more robust and reliable. Artificial General Intelligence (AGI) is still a distant goal, and current Large Language Model (LLM)-based agents are prone to hallucinations and making mistakes, especially when dealing with complex tasks such as formal verification. However, with the right enhancements and improvements, we believe that Saarthi can be a significant step towards achieving domain-specific general intelligence for formal verification. Especially for problems that require Short Term, Short Context (STSC) capabilities, such as formal verification, Saarthi can be a powerful tool to assist verification engineers in their work. In this paper, we present two key enhancements to the Saarthi framework: (1) a structured rulebook and specification grammar to improve the accuracy and controllability of SystemVerilog Assertion (SVA) generation, and (2) integration of advanced Retrieval Augmented Generation (RAG) techniques, such as GraphRAG, to provide agents with access to technical knowledge and best practices for iterative refinement and improvement of outputs. We also benchmark these enhancements for the overall Saarthi framework using challenging test cases from NVIDIA's CVDP benchmark targeting formal verification. Our benchmark results stand out with a 70% improvement in the accuracy of generated assertions, and a 50% reduction in the number of iterations required to achieve coverage closure.

cs.AI↗