Searcharxiv⌕ Search

arXiv subjects

Deepak Kumar

Publications and source records attributed to Deepak Kumar.

At least 55 records · Page 3Linked to original sources
arXiv2025

GPT-OSS-20B: A Comprehensive Deployment-Centric Analysis of OpenAI's Open-Weight Mixture of Experts Model

We present a single-GPU (H100, bf16) evaluation of GPT-OSS-20B (Mixture-of-Experts; 20.9B total, approx. 3.61B active) against dense baselines Qwen3-32B and Yi-34B across multiple dimensions. We measure true time-to-first-token (TTFT), full-decode throughput (TPOT), end-to-end latency percentiles, peak VRAM with past key values (PKV) held, and energy via a consistent nvidia-smi-based sampler. At a 2048-token context with 64-token decode, GPT-OSS-20B delivers higher decode throughput and tokens per Joule than dense baselines Qwen3-32B and Yi-34B, while substantially reducing peak VRAM and energy per 1000 generated tokens; its TTFT is higher due to MoE routing overhead. With only 17.3% of parameters active (3.61B of 20.9B), GPT-OSS-20B provides about 31.8% higher decode throughput and 25.8% lower energy per 1000 generated tokens than Qwen3-32B at 2048/64, while using 31.7% less peak VRAM. Normalized by active parameters, GPT-OSS-20B shows markedly stronger per-active-parameter efficiency (APE), underscoring MoE's deployment advantages. We do not evaluate accuracy; this is a deployment-focused study. We release code and consolidated results to enable replication and extension.

cs.AR↗
arXiv2025

Gate Voltage-Controlled Magnetic Anisotropy Effect on Pt-Porphyrin functionalized single-layer graphene

We report a novel approach to engineering large voltage-controlled magnetic anisotropy (VCMA) and enhanced spin-orbit coupling (SOC) at the interface of single-layer graphene (SLG) and NiFe (Py) through non-covalent functionalization with Platinum (II) 5,10,15,20-tetraphenyl porphyrin (Pt-porphyrin). Using chemical vapor deposition (CVD)-grown SLG, we demonstrate that Pt-porphyrin functionalization significantly increases the SOC and enables robust voltage modulation of interfacial magnetic anisotropy, as confirmed by spin-torque ferromagnetic resonance (ST-FMR) measurements. A substantial VCMA coefficient of 375.6 (fJ/(V-m)) is achieved, accompanied by an order-of-magnitude enhancement in spin torque efficiency (θsh) compared to pristine SLG. The resonance field exhibits a clear, reversible shift under applied gate voltage, confirming robust electric-field modulation of interfacial magnetic anisotropy. Raman spectroscopy and X-ray photoelectron spectroscopy (XPS) confirm the structural integrity and effective charge transfer at the functionalized interface. Electrical characterization of back-gated graphene field-effect transistors (GFETs) further reveals tunable electronic properties upon functionalization. Our results establish functionalized graphene/ferromagnet interfaces as a promising platform for low-power, voltage-controlled spintronic devices, paving the way for scalable, energy-efficient memory and logic technologies

cond-mat.mes-hall↗
arXiv2025

Constraints on maximum neutron star mass from proto-neutron star evolution

A proto-neutron star (PNS) gets formed after a successful supernova when the stellar remnant decouples from the ejecta. In this study, we explore a relativistic framework for the finite-temperature $β$-equilibrium limit of equation of state (EOS), constrained via a Bayesian inference methodology. The EOS is constrained by minimal approximations on a few nuclear saturation properties, low-density pure neutron matter constraints from chiral effective field theory, and a neutron star (NS) maximum mass greater than 2.0 $M_{\odot}$. Two sets of EOS derived from the relativistic mean field model for nucleonic and hyperonic matter constrained by a Bayesian inference calculation at the zero temperature limit are used. The thermal adiabatic index ($Γ_{\rm Th}$) is calculated as a function of the baryonic density across several temperatures for both the sets. Our results suggest that the maximum NS mass is of the order of 2.15 $M_\odot$ if hyperons are present. In addition, the present study suggests that an observation of NS with mass larger than $2.2\ M_{\odot}$ can indirectly indicates the absence of hyperons in its core. The deleptonization of hyperonic PNS reduces the stellar maximum mass rendering the PNS exceeding the zero temperature maximum stellar (baryonic) mass limit becomes metastable which is prone to collapse into a black hole while PNS below such a mass threshold evolves to a stable NS.

nucl-th↗
arXiv2025

Modification of the universal relation between mass, radius and nonradial $f$-mode oscillation in proto-neutron stars

Neutron stars are usually assumed to be cold; however, in certain dynamical astrophysical scenarios such as newly born neutron stars or binary star mergers, the temperature effects play a non-negligible role. We systematically derive the equation of state at finite-temperature within a relativistic mean-field hadronic model applicable to such proto-neutron stars. The equation of state so derived considerably affects the mass-radius curve, thereby affecting the nonradial quadruple $f$-mode oscillation frequencies.} Temperature effectively makes the equation of state stiffer at relatively low and intermediate densities, thereby making the star less compact and flattening the mass-radius curve. The $f$-mode frequency for low and intermediate-mass neutron stars decreases with temperature and thus should be easier to detect. The universal relation (connecting $f$-mode frequency, mass, and radius) changes nonlinearly with temperature. The parameters defining the universal relation [$ωM = a(T) \left(\frac{M}{R}\right) + b(T)$] becomes temperature dependent with the coefficients following a parabolic relation with temperature.

astro-ph.HE↗
arXiv2025

Some Compact Generalization of Bernstein-Type Inequalities Preserved by Modified Smirnov Operator

Let $P(z)$ be a polynomial of degree $n$. In $2004$, Aziz and Rather \cite{aziz2004some} investigated the dependence of \[\bigg|P(Rz)-αP(z)+β\biggl\{\biggl(\frac{R+1}{2}\biggr)^n-|α|\biggr\}P(z)\bigg|, \ \text{for} \ z \in B(\mathbb{D}),\] on $\max_{z\in B(\mathbb{D})}|P(z)|$, for every real and complex number $α, β$ satisfying $|α| \leq 1$, $|β| \leq 1$, and $R \geq 1$. This paper presents a compact generalization of several well-known polynomial inequalities using modified Smirnov operator, demonstrating that the operator preserves inequalities between polynomials.

math.CV↗
arXiv2025

Analyzing the dense matter equation of states in the light of the compact object HESS J1731-347

The recent mass ($0.77 \pm ^{0.20}_{0.17}M_{\odot}$) and radius ($10.4\pm^{0.86}_{0.78} \text{km}$) measurement of HESS J1731-347 made it one of the most fascinating object if it is indeed a neutron star. In this work, we examine the current status of the dense matter equation of states in the context of this compact object being a neutron star. We use three sets of equation of states corresponding to the three classes - neutron stars, strange stars, and hybrid stars and perform Bayesian model selection on them. Our results show that for hadronic models, the EoS is preferred to be stiff at the intermediate densities. This makes the Brueckner-Hartree-Fock approximation and models based on effective interactions deviate from current astrophysical observations on the inclusion of HESS J1731-347. Furthermore, for the strange star family, the equation of states composed of three flavor quarks prefers relatively smaller bag parameters. Analyzing the hybrid family of equation of states consisting of a first-order phase transition revealed preferences for early first-order phase transition. Comparing all the preferred equations of state among each family, it was found that the current astrophysical constraints prefer the hybrid equation of states the most.

astro-ph.HE↗
arXiv2025

Enhancing Contrast and Resolution for Electron-beam Lithography on Insulating Substrates

We report on the effect of ambient gas on the contrast and the resolution of electron beam lithography (EBL) in gaseous environments on insulating substrates. Poly(methyl methacrylate) (PMMA) films were exposed in an environmental scanning electron microscope using a 30 keV electron-beam under 1 mbar pressure of helium, water, nitrogen and argon. We found that the choice of ambient gas results in significant variations in contrast, and the clearing dose increases with the gases molecular weight and proton number, consistent with the increase in scattering cross-section. Significantly higher contrast values are obtained for exposure under helium and are accompanied by improved sensitivity. Despite higher sensitivity, helium exhibited the best resolution with 20-nm half-pitch dense lines and spaces. However, water vapor offered a larger process window, particularly on fused silica substrates. We also demonstrate that higher sensitivity results from effective charge dissipation. Thus, for EBL on insulating substrates, helium and water vapor may be desirable choices for charge dissipation depending on the substrate and process conditions.

cond-mat.mes-hall↗
arXiv2025

Axion effects on the non-radial oscillations of neutron stars

The effects of axions on quark matter equation of state (EOS) is studied within the three flavor Nambu--Jona-Lasinio model and its effects on on the non-radial oscillations of neutron stars is investigated. Using such an EOS for quark matter with axions and a EOS for hadronic matter within the relativistic mean field (RMF) theory, we discuss the hadron-quark phase transition (HQPT) using the Gibbs construction. The EOS so obtained is used to investigate the structure of hybrid neutron star (NS)s. It is found that the presence of axions in the core of compact stars stabilizes hybrid NSs in agreement with modern astrophysical constraints. It is further observed that the quadrupolar fundamental modes ($f$-modes) for such hybrid NSs get substantial enhancements both due to a larger quark core in the presence of axions and from the hyperons as compared to a canonical nucleonic neutron stars.

hep-ph↗
arXiv2025

Enhanced Patterned Fluorescence from Polystyrene through Focused Electron Beam Irradiation under Various Gases

We report on a novel method for tuning and enhancing fluorescence from irradiated polystyrene through electron-beam exposure in gaseous environments. We describe the effect of electron dose and ambient gas on the photoluminescence spectra and yield of irradiated PS films on insulating and conductive substrates. PS films were exposed in an environmental scanning electron microscope using a 20 keV electron beam, ambient gas pressures from high vacuum to 3 mbar, and electron doses from 1.8 to 45 mC cm-2. Irradiated PS films were characterized using confocal microscopy, TEM, EDS and FTIR spectroscopy. From emission spectra collected using confocal microscopy we found that the emission wavelength and photon yield of the irradiated film can be tuned by both dose and gas pressure. The emission wavelength blue-shifts with increasing pressure and red-shifts with increasing dose enabling an overall tuning range of 451 - 544 nm. Significant enhancement in the PL intensity, up to 18 times on sapphire substrates under helium when compared to high vacuum, are observed. Overall, the highest PL yield is observed on soda lime glass substrates under argon. Also, the photon-yield on conductive substrates is significantly smaller than that yield from insulating substrates. TEM images revealed e-beam irradiated PS is amorphous in nature and elemental mapping EDS revealed no signs of film oxidation. FTIR spectroscopy revealed that under gaseous environments the decay of aromatic and aliphatic C-H stretches is reduced compared to the high vacuum exposure; in all cases, features associated with the phenyl rings are preserved. Localized e-beam synthesis of fluorophores in PS can be controlled by both dose and by ambient gas pressure. This technique could enable new approaches to photonics where fluorophores with tunable emission properties can be locally introduced by e-beam patterning.

cond-mat.mes-hall↗
arXiv2025

Constraining the neutron star equation of state by including the isoscalar-vector and isovector-vector coupling using the Bayesian analysis

We constrain the nuclear matter equation of state within the relativistic mean field model by including the isoscalar-vector and isovector-vector coupling term at a fundamental level using the Bayesian analysis. We used the nuclear saturation properties and recent astrophysical observations to constrain the dense matter equation of state. We obtained about 20000 sets of equations of states out of sampling about 60 millions sets of equations of states. All 20000 equations of states satisfy nuclear matter saturation properties at saturation densities and produces high mass neutron stars. In our findings, we find that the non-zero value of isoscalar-vector and isovector-vector coupling parameter and negative value of sigma meson self-coupling stiffen the equation of state. Our sets of equations of state produces neutron stars of mass larger than 2.5 M$_{\odot}$ to include the recent gravitational waves observation GW190419.

nucl-th↗
arXiv2025

Use of Metric Learning for the Recognition of Handwritten Digits, and its Application to Increase the Outreach of Voice-based Communication Platforms

Initiation, monitoring, and evaluation of development programmes can involve field-based data collection about project activities. This data collection through digital devices may not always be feasible though, for reasons such as unaffordability of smartphones and tablets by field-based cadre, or shortfalls in their training and capacity building. Paper-based data collection has been argued to be more appropriate in several contexts, with automated digitization of the paper forms through OCR (Optical Character Recognition) and OMR (Optical Mark Recognition) techniques. We contribute with providing a large dataset of handwritten digits, and deep learning based models and methods built using this data, that are effective in real-world environments. We demonstrate the deployment of these tools in the context of a maternal and child health and nutrition awareness project, which uses IVR (Interactive Voice Response) systems to provide awareness information to rural women SHG (Self Help Group) members in north India. Paper forms were used to collect phone numbers of the SHG members at scale, which were digitized using the OCR tools developed by us, and used to push almost 4 million phone calls. The data, model, and code have been released in the open-source domain.

cs.AI↗
arXiv2025

Training neural networks without backpropagation using particles

Neural networks are a group of neurons stacked together in multiple layers to mimic the biological neurons in a human brain. Neural networks have been trained using the backpropagation algorithm based on gradient descent strategy for several decades. Several variants have been developed to improve the backpropagation algorithm. The loss function for the neural network is optimized through backpropagation, but several local minima exist in the manifold of the constructed neural network. We obtain several solutions matching the minima. The gradient descent strategy cannot avoid the problem of local minima and gets stuck in the minima due to the initialization. Particle swarm optimization (PSO) was proposed to select the best local minima among the search space of the loss function. The search space is limited to the instantiated particles in the PSO algorithm, and sometimes it cannot select the best solution. In the proposed approach, we overcome the problem of gradient descent and the limitation of the PSO algorithm by training individual neurons separately, capable of collectively solving the problem as a group of neurons forming a network. Our code and data are available at https://github.com/dipkmr/train-nn-wobp/

cs.NE↗
arXiv2025

Generalization of Some Well-Known Polynomial Inequalities for the Modified Smirnov Operator

Let $P(z)$ be a polynomial of degree $n$. In this paper, we consider the modified Smirnov operator, which carries a polynomial $P(z)$ into $\tilde{\mathbb{S}}_a[P](z)=(1+az)P'(z)-naP(z),$ where $a$ is an arbitrary number in $\overline{\mathbb{D}}$. We estimate minimum and maximum moduli of modified Smirnov operator of $P(z)$ on the unit circle with restricted zeros and thereby obtain a generalization of some results of Dewan and Hans \cite{dewan2013some}. This study includes compact generalization of some well-known polynomial inequalities.

math.CV↗
arXiv2025

CP violation in cold dense quark matter and axion effects on the non-radial oscillations of neutron stars

Charge-conjugation and parity violation in strong interaction for cold dense quark matter is studied with axions of quantum chromodynamic within the three flavor Nambu--Jona-Lasinio model that includes the coupling of axions to quarks. We first calculate the effective potential for axions at finite baryon density and zero temperature including the effects of a first order chiral phase transition. Using the equation of state for quark matter with axions and a hadronic matter equation of state in the ambit of a relativistic mean field theory in quantum hadrodynamics, we discuss the hadron-quark phase transition. Inclusion of axions reduces the critical density for chiral transition. We use a Gibbs construct for the hadron-quark phase transition satisfying the constraints of beta equilibrium and charge neutrality as appropriate for the neutron star matter. The equation of state so obtained is used to investigate the structure of hybrid neutron stars. It is found that with the presence of axions, it is possible to have stable hybrid neutron stars having an inner core of quark matter both in pure quark matter phase as well as in a mixed phase with hyperonic matter along with a outer core of hyperonic matter and is in agreement with modern astrophysical constraints. We also discuss the properties of non-radial oscillations of such hybrid neutron stars. It is observed that the quadrupolar fundamental modes ($f$-modes) for such hybrid neutron stars get substantial enhancements both due to a larger quark core in the presence of axions and from the hyperons as compared to a canonical nucleonic neutron stars.

hep-ph↗
arXiv2025

Explainable, Multi-modal Wound Infection Classification from Images Augmented with Generated Captions

Infections in Diabetic Foot Ulcers (DFUs) can cause severe complications, including tissue death and limb amputation, highlighting the need for accurate, timely diagnosis. Previous machine learning methods have focused on identifying infections by analyzing wound images alone, without utilizing additional metadata such as medical notes. In this study, we aim to improve infection detection by introducing Synthetic Caption Augmented Retrieval for Wound Infection Detection (SCARWID), a novel deep learning framework that leverages synthetic textual descriptions to augment DFU images. SCARWID consists of two components: (1) Wound-BLIP, a Vision-Language Model (VLM) fine-tuned on GPT-4o-generated descriptions to synthesize consistent captions from images; and (2) an Image-Text Fusion module that uses cross-attention to extract cross-modal embeddings from an image and its corresponding Wound-BLIP caption. Infection status is determined by retrieving the top-k similar items from a labeled support set. To enhance the diversity of training data, we utilized a latent diffusion model to generate additional wound images. As a result, SCARWID outperformed state-of-the-art models, achieving average sensitivity, specificity, and accuracy of 0.85, 0.78, and 0.81, respectively, for wound infection classification. Displaying the generated captions alongside the wound images and infection detection results enhances interpretability and trust, enabling nurses to align SCARWID outputs with their medical knowledge. This is particularly valuable when wound notes are unavailable or when assisting novice nurses who may find it difficult to identify visual attributes of wound infection.

cs.CV↗
arXiv2025

Krutrim LLM: Multilingual Foundational Model for over a Billion People

India is a diverse society with unique challenges in developing AI systems, including linguistic diversity, oral traditions, data accessibility, and scalability. Existing foundation models are primarily trained on English, limiting their effectiveness for India's population. Indic languages comprise only 1 percent of Common Crawl corpora despite India representing 18 percent of the global population, leading to linguistic biases. Thousands of regional languages, dialects, and code mixing create additional representation challenges due to sparse training data. We introduce Krutrim LLM, a 2 trillion token multilingual model designed for India's linguistic landscape. It incorporates the largest known Indic dataset, mitigating data scarcity and ensuring balanced performance across dialects. Krutrim outperforms or matches state-of-the-art models on Indic benchmarks while maintaining competitive English performance. Despite being significantly smaller in training flops, Krutrim LLM matches or exceeds models like LLAMA-2 on 10 out of 16 tasks, with an average score of 0.57 versus 0.55. This evidences Krutrim's flexible multilingual fluency across diverse linguistic contexts. Krutrim is integrated with real-time search to improve factual accuracy in conversational AI applications. This enhances accessibility for over 1 billion users worldwide. Through intentional design choices addressing data imbalances, Krutrim LLM signifies meaningful progress in building ethical, globally representative AI models.

cs.CL↗
arXiv2025

The footprint of nuclear saturation properties on the neutron star $f$ mode oscillation frequencies: a machine learning approach

We investigate the intricate relationships between the non-radial \(f\) mode oscillation frequencies of neutron stars (NS)s and the corresponding nuclear matter equation of state (EOS) using a machine learning (ML) approach within the ambit of the relativistic mean field (RMF) framework for nuclear matter. With two distinct parameterizations of the Walecka model, namely, (1) with non-linear self interactions of the scalar field (NL) and, (2) a density dependent Bayesian model (DDB), we perform a thorough examination of the \(f\) mode frequency in relation to various nuclear saturation properties. The correlations between the \(f\) mode frequencies and nuclear saturation properties reveal, through various analytical and ML methods, the complex nature of NSs and their potential as the cosmic laboratory for studying extreme states of matter. A principal component analysis (PCA) has been performed using mixed datasets from DDB and NL models to discriminate the relative importance of the different components of the EOS on the $f$ mode frequencies. Additionally, a {\it Random forest feature importance} analysis also elucidates the distinct roles of these properties in determining the \(f\) mode frequency across a spectrum of NS masses. Our findings are further supported by symbolic regression searches, yielding high-accuracy relations with strong Pearson coefficients and minimal errors. These relations suggest new methodologies for probing NS core characteristics, such as energy density, pressure, and speed of sound from observations of non-radial \(f\) mode oscillations of NSs.

nucl-th↗