Searcharxiv⌕ Search

arXiv subjects

Akash Singh

Publications and source records attributed to Akash Singh.

At least 19 recordsLinked to original sources

ExpertCrowd-SWE: Measuring Frontier Coding Agents on Expert-Authored Engineering Tasks in Unfamiliar Codebases

ExpertCrowd-SWE (EC-SWE) is a benchmark of 100 expert-authored, multi-file software engineering tasks for evaluating autonomous coding agents in large, unfamiliar codebases. Each task is a change request written by a practising engineer against a pinned commit of a real repository, paired with a sealed reference solution and a task-specific grader whose weighted, mostly executable criteria award partial credit; must-pass blockers and pass/fail gates encode what a maintainer would refuse to merge. The tasks span 87 open-source projects and 11 languages. The median ticket is 148 words, while the median reference solution changes 471 lines. Every task image is calibrated before release so that the reference solution scores 1.0 and both the untouched repository and an inert change score 0; nearly half of the tasks also ship alternative, lazy and adversarial control patches that the grader must classify correctly. The final 100 tasks were selected for difficulty against current frontier models. We evaluate seven frontier models, five runs each (3,495 scored runs), in a single platform harness that gives each model its provider-native tool interface. Claude Opus 5.5 (67.3%) and GPT-6 Astra (63.7%) lead without a statistically significant gap, followed by Claude Fable 5.1 (46.5%), Gemini 3.8 Flash (40.0%), and a statistically indistinguishable group of Qwen 3.8 Max, DeepSeek V4.1 Flash and Grok 4.6 (32-34%). No task is fully solved by every model, and on 80 tasks no model earns full credit on all five runs. Scores span 35 points. Depending on the model, agents take a median of 33 to 166 steps per run, with little relation to score. We release 24 tasks with their environments, graders and results, and hold out 76 to limit contamination.

cs.AI↗

Dopant Cluster Engineering for Tunable Room-Temperature Ferromagnetism in 2D Transition Metal Dichalcogenides

Two-dimensional transition metal dichalcogenides (2D-TMDs) offer a promising platform for realizing dilute magnetic semiconductors. However, simultaneously achieving large magnetic anisotropy energy (MAE) and high Curie temperature (T$_\text{c}$) remains a major challenge. In this work, we employ DFT+U to systematically investigate the thermodynamic stability and magnetic properties of 2D-MoSe$_2$ and 2D-WSe$_2$ doped with transition metal atoms. We found that the paired dopant configurations are more stable than previously attributed isolated dopants. Despite their nearly identical crystal structures, 2D-MoSe$_2$ and 2D-WSe$_2$ show contrasting properties upon doping. All paired dopants in 2D-MoSe$_2$ exhibit long-range ferromagnetic coupling (FMC) primarily due to carrier-mediated exchange interactions, whereas those in 2D-WSe$_2$ show weak exchange coupling. Interestingly, 2Mn$_\text{2W}$ shows a transition from less stable antiferromagnetic coupling (AFMC) in the short range to highly stable FMC in the long-range. The combined effect of spin-orbit coupling, exchange interactions, and orbital hybridization gives rise to a large MAE with maximum in-plane values of $-$3.5 meV for 2V$_\text{2Mo}$ and $-$3.2 meV for 2Mn$_\text{2W}$. Monte Carlo simulations estimate T$_\text{c}$ $\simeq$ 440 K for 2Fe$_\text{2Mo}$ and ensure its high-temperature stability in 2D-MoSe$_2$. Our results establish dopant clustering as an effective route to engineer high-temperature ferromagnetism in 2D-TMDs and highlight their potential for nano-spintronic applications.

cond-mat.mtrl-sci↗

Competition and coexistence of baryonic condensates in AdS hardwall model

Cold dense QCD admits several order parameters for baryon-number breaking, among them diquark and dibaryon condensates. We ask whether they exclude one another or coexist, using a five-dimensional AdS hardwall model with full backreaction in which two charged scalars, dual to operators of charges $q_i$ and dimensions $Δ_i$, couple to a common $U(1)_B$ gauge field. With the diquark-like operator fixed at $q_1=2$, $Δ_1=3$, both operators condense only within a narrow window of $Δ_2$, which widens with $q_2$ and closes when the operators coincide. Outside the window, only a single condensate forms. The structure persists in both confined and deconfined phases. In the $μ_B$-$T$ plane, confined matter with both condensates is preferred at low temperature, below a narrow band of deconfined matter with one condensate.

hep-th↗

Classical fractons with cosmological fixed points

Classical fractons are Hamiltonian systems that can develop attractors after projection onto configuration or shape variables, although the full phase space admits none. We study a scale-invariant, dipole-conserving two-parameter family of fracton Hamiltonians $H_{α,β}$. By separating coordinates into scale and shape, we obtain autonomous shape dynamics that admit fixed points which leave a purely scale evolution of the form $R(t)\propto |t|^{α/(α-β)}$. The shape fixed points, which determine the distribution of the expanding particles, are central configurations of power-law Riesz potentials. The distinguished model $(α,β)=(-2,1)$ is unique: its scale evolution takes the Einstein-de Sitter form $R(t)\propto |t|^{2/3}$, its fixed-point equation is the equal-mass Newtonian central-configuration, its large-$N$ distribution is a homogeneous ball, and its homothetic trajectories admit a zero-energy Newtonian gravitational dual. The fixed points are locally stable, and simulations at moderate $N$ approach them from random initial data. Large $N$ simulations reveal a richer class of fixed-points: bound clusters of approximately fixed physical size retain internal motion, while their centers approach unequal-mass Newtonian central configurations and preserve large-scale homogeneity. A scale-separation conjecture yields an effective unequal-mass fracton dynamics for the centers and a corresponding zero-energy Newtonian gravitational dual. Trajectories generically exhibit a bidirectional arrow of time: scale and shape complexity grow away from a Janus point, while Boltzmann entropy grows logarithmically. Together, these features reproduce the salient structure of a flat matter-dominated cosmology. In the distinguished fracton model, all these cosmological analogues emerge as attractor properties, making it a toy model for cosmological dynamics without fine-tuning.

cond-mat.stat-mech↗

Lost in State Space: Probing Frozen Mamba Representations

Mamba's recurrent state h_t is, by construction, a compressed summary of every token seen so far. This raises a tempting hypothesis: if we extract token-level outputs y_t at fixed patch boundaries, we obtain semantic sentence summaries for free, with no pooling head, no fine-tuning, and no [CLS] token. We test this hypothesis carefully. Across five benchmarks (SST-2, CoLA, MRPC, STS-B, IMDb), we compare four strategies for extracting frozen sentence representations from a pretrained Mamba-130M backbone under a strict frozen-feature probing protocol, using three random seeds where computationally feasible. The results do not support the hypothesis: patch boundary readouts do not consistently outperform simple mean pooling. We identify and quantify two structural pathologies: severe anisotropy (mean pairwise cosine similarity 0.9999, std 0.000044) and representational collapse in the raw final SSM state (MCC = 0.000 on CoLA across all three seeds, confirmed via confusion matrix). We further propose orthogonal injection, a modified recurrence that constrains new information per

cs.CL↗

Condensate phases of nuclear matter from AdS Hardwall models

This work develops our previous study of confined phases at finite densities in AdS/QCD by systematically exploring the possibility of baryonic condensates. Using phenomenologically motivated boundary conditions in an AdS hardwall model, we show that both baryonic and quark type condensates dominate the phase diagram at low temperatures. We also undertake a careful scan of the parameter space to extract robust conclusions.

hep-th↗

Nanopillar-Driven Antibacterial Surfaces: Elucidating Bactericidal Mechanisms and Engineering Nanostructures for Enhanced Efficacy

Insects like dragonflies and cicadas possess nanoprotusions on their wings that rupture bacterial membranes upon contact, inspiring synthetic antibacterial surfaces mimicking this phenomenon. Designing such biomimetic surfaces requires understanding the mechanical interaction between nanopillars and bacterial membranes. However, the small scales of these interactions pose challenges. Molecular Dynamics simulations offer precise and efficient modeling at these scales. This study presents a coarse-grained membrane model to explore the mechanical responses of gram-positive and gram-negative bacterial membranes to nanopillar arrays. By varying bacterial shapes (spherical and cylindrical), membrane bending rigidity, and loading rates, we identified two distinct failure mechanisms. Low bending rigidity, typical of gram-negative bacteria, leads to tearing near nanopillar tips, contrary to prior assumptions. High bending rigidity, characteristic of gram-positive bacteria, results in puncturing at contact points. Gram-positive bacteria are more resistant, requiring a threefold increase in loading rate for effective piercing. Nanopillar height and spacing also critically impact bactericidal efficacy: greater heights enhance activity beyond a critical threshold, while increased spacing reduces efficacy. This simplified coarse-grained model, representing bacterial membranes with high fidelity, enables cost-effective, full-scale simulations over extended periods. Our findings provide essential insights for optimizing nanopillared surface designs, advancing antibacterial technology through tailored height and spacing configurations.

cond-mat.soft↗

Confined phases at finite density in the Hardwall model

AdS/QCD models are being extensively studied because they seem to offer an entirely tractable and radically different approach to the phases of QCD whose region of validity is precisely the region of strong interactions. In this context, a missing ingredient has been the absence of a bulk description dual to a confined hadronic phase at finite baryon density. In this work, we discuss a class of backgrounds that fill this lacuna. Applying physically motivated boundary conditions, we obtain phase diagrams which compare reasonably with approaches based on the PNJL model. We also show that such backgrounds can play a similar role in geometries with a dilaton field.

hep-th↗

Reliable machine learning potentials based on artificial neural network for graphene

Graphene is one of the most researched two dimensional (2D) material due to its unique combination of mechanical, thermal and electrical properties. Special 2D structure of graphene enables it to exhibit a wide range of peculiar material properties like high Young's modulus, high specific strength etc. which are critical for myriad of applications including light weight structural materials, multi-functional coating and flexible electronics. It is quite challenging and costly to experimentally investigate graphene/graphene based nanocomposites, computational simulations such as molecular dynamics (MD) simulations are widely adopted for understanding the microscopic origins of their unique properties. However, disparate results were reported from computational studies, especially MD simulations using various empirical inter-atomic potentials. In this work, an artificial neural network based interatomic potential has been developed for graphene to represent the potential energy surface based on first principle calculations. The developed machine learning potential (MLP) facilitates high fidelity MD simulations to approach the accuracy of ab initio methods but with a fraction of computational cost, which allows larger simulation size/length, and thereby enables accelerated discovery/design of graphene-based novel materials. Lattice parameter, coefficient of thermal expansion (CTE), Young's modulus and yield strength are estimated using machine learning accelerated MD simulations (MLMD), which are compared to experimental/first principle calculations from previous literatures. It is demonstrated that MLMD can capture the dominating mechanism governing CTE of graphene, including effects from lattice parameter and out of plane rippling.

physics.comp-ph↗

Revisiting the Thermal AdS partition function

We rewrite the worldsheet torus partition function of the Thermal AdS CFT by isolating the boundary parameters. Using this, we show that the spectrum of the Euclidean BTZ black hole and Lorentzian AdS3 can be extracted -- the latter as a zero temperature limit. A similar procedure recovers the Lorentzian BTZ spectrum proposed in an earlier work. We then use our expression to construct a boundary modular invariant expression as a Poincaré series.

hep-th↗

Phases of a 10-D Holographic hard wall model

In this article, we study the finite temperature properties of a 10-D version of a hardwall model for QCD. Introducing fundamental matter via probe D7-branes and separate cutoffs for the branes and the bulk, we present a detailed exploration of the phases for varying temperature and quark mass. Finite thermodynamic quantities are calculated using the procedure of holographic renormalization and used to characterize the phases. Finally, by fitting glueball and vector meson masses, we show how a unique phase diagram can be isolated.

hep-th↗

Deep set conditioned latent representations for action recognition

In recent years multi-label, multi-class video action recognition has gained significant popularity. While reasoning over temporally connected atomic actions is mundane for intelligent species, standard artificial neural networks (ANN) still struggle to classify them. In the real world, atomic actions often temporally connect to form more complex composite actions. The challenge lies in recognising composite action of varying durations while other distinct composite or atomic actions occur in the background. Drawing upon the success of relational networks, we propose methods that learn to reason over the semantic concept of objects and actions. We empirically show how ANNs benefit from pretraining, relational inductive biases and unordered set-based latent representations. In this paper we propose deep set conditioned I3D (SCI3D), a two stream relational network that employs latent representation of state and visual representation for reasoning over events and actions. They learn to reason about temporally connected actions in order to identify all of them in the video. The proposed method achieves an improvement of around 1.49% mAP in atomic action recognition and 17.57% mAP in composite action recognition, over a I3D-NL baseline, on the CATER dataset.

cs.CV↗

MockingBERT: A Method for Retroactively Adding Resilience to NLP Models

Protecting NLP models against misspellings whether accidental or adversarial has been the object of research interest for the past few years. Existing remediations have typically either compromised accuracy or required full model re-training with each new class of attacks. We propose a novel method of retroactively adding resilience to misspellings to transformer-based NLP models. This robustness can be achieved without the need for re-training of the original NLP model and with only a minimal loss of language understanding performance on inputs without misspellings. Additionally we propose a new efficient approximate method of generating adversarial misspellings, which significantly reduces the cost needed to evaluate a model's resilience to adversarial attacks.

cs.CL↗

Immersed Cantilever Apparatus for Mechanics and Microscopy

We present here a novel cantilever based apparatus to perform translational stress or strain controlled rheology in very soft solids, and obtain simultaneous confocal imaging of the 3 dimensional microstructure. The stress is measured using eddy based sensors. Both the stress and strain are controlled by applying PID control loops on measured quantities and changing position using a micromanipulator. To get rid of surface tension forces, the sample and cantilever are immersed. This enables stress measurement and control down to 3mPa. With this apparatus, we can independently apply shear and normal stress, or strain, with same precision. We demonstrate the technical capability of the setup with steady shear strain or stress experiments on a soft protein gel system. The simultaneous confocal imaging offers insight into the macroscopic breaking observed in an increasing shear strain experiment.

cond-mat.soft↗

Low temperature Magneto-dielectric coupling in nanoscale layered SmFe0.5Co0.5O3 perovskite

In this study, we determined the different physical characteristics of nanoscale layered mixed Fe-Co layers of orthoferrite SmFe0.5Co0.5O3 based on magnetic and dielectric measurements.Magnetization analyzes showed that the system was antiferromagnetic with a magnetic transition around 310 K and two spin-reorientation transitions at around 192 K and 22 K.In this phase, the antisymmetric exchange interaction induced weak ferromagnetism due to canting of the magnetic spins in a similar manner to the Ln-Fe-Co system. Importantly, this nanoscale layered perovskite also exhibited a positive magneto-dielectric effect of around 2.5 % (100 K) and the value decreased on both sides of this temperature.We analyzed the conditions related to the existence of magnetic and dielectric coupling in terms of the magnetic interactions between the cations as well as the spin-lattice interactions in the low temperature range.

cond-mat.mtrl-sci↗

Novel Compliant omnicrawler-wheel transforming module

This paper presents a novel design of a crawler robot which is capable of transforming its chassis from an Omni crawler mode to a large-sized wheel mode using a novel mechanism. The transformation occurs without any additional actuators. Interestingly the robot can transform into a large diameter and small width wheel which enhances its maneuverability like small turning radius and fast/efficient locomotion. This paper contributes on improving the locomotion mode of previously developed hybrid compliant omnicrawler robot CObRaSO. In addition to legged and tracked mechanism, CObRaSO can now display large wheel mode which contributes to its locomotion capabilities. Mechanical design of the robot has been explained in a detailed manner in this paper and also the transforming experiment and torque analysis has been shown clearly

cs.RO↗

Modelling and Path Planning of Snake Robot in cluttered environment

Studying snake robot locomotion in a cluttered environment has been a complicated task because the motion model is discontinuous due to the physical contact with obstacles, and the contact force cannot be determined solely by contact positions. We present a unique mathematical model of the robot interacting with obstacles in which the contact forces are mapped on the basis of a viscous friction model. Also a motion planning strategy has been introduced which helps deriving the simplest path that ensures sufficient number of contacts of the robot with the obstacles required to reach a goal position. Numerical simulations and experimental results are presented to validate the theoretical approach.

cs.RO↗

CObRaSO: Compliant Omni-Direction Bendable Hybrid Rigid and Soft OmniCrawler Module

This paper presents a novel design of an Omnidirectional bendable Omnicrawler module- CObRaSO. Along with the longitudinal crawling and sideways rolling motion, the performance of the OmniCrawler is further enhanced by the introduction of Omnidirectional bending within the module, which is the key contribution of this paper. The Omnidirectional bending is achieved by an arrangement of two independent 1-DOF joints aligned at 90? w.r.t each other. The unique characteristic of this module is its ability to crawl in Omnidirectionally bent configuration which is achieved by a novel design of a 2-DOF roller chain and a backbone of a hybrid structure of a soft-rigid material. This hybrid structure provides compliant pathways for the lug-chain assembly to passively conform with the orientation of the module and crawl in Omnidirectional bent configuration, which makes this module one of its kind. Furthermore, we show that the unique modular design of CObRaSO unveils its versatility by achieving active compliance on an uneven surface, demonstrating its applications in different robotic platforms (an in-pipeline robot, Quadruped and snake robot) and exhibiting hybrid locomotion modes in various configurations of the robots. The mechanism and mobility characteristics of the proposed module have been verified with the aid of simulations and experiments on real robot prototype.

cs.RO↗