arXiv subjects

Abhishek Sharma

Publications and source records attributed to Abhishek Sharma.

At least 37 records · Page 2Linked to original sources

arXiv2025

Tunable Magnetic and Topological Phases in EuMnXBi$_2$ (X=Mn, Fe, Co, Zn) Pnictides

We present a comprehensive density functional theory (DFT) study of the electronic, magnetic, and topological properties of the layered pnictides EuMnXBi2 (X = Mn, Fe, Co, Zn), focusing in particular on the relatively unexplored Bi-based member of the EuMn2X2 family. Unlike the well-studied As-, Sb-, and P--based analogues, we show that EuMn2Bi2 stabilizes in a C-type antiferromagnetic ground state with a narrow-gap semiconducting character. Inclusion of spin-orbit coupling (SOC) drives a transition from this trivial antiferromagnetic semiconductor to a Weyl semimetal hosting four symmetry-related Weyl points and robust Fermi arc states. Systematic substitution of Mn with Fe, Co, and Zn further reveals a tunable sequence of magnetic ground states: Fe and Co induce ferrimagnetism with semimetallic behavior, while Zn stabilizes a ferromagnetic semimetal with a large net moment. These findings establish Bi-based EuMnXBi2 pnictides as a versatile platform where magnetic exchange interactions and band topology can be engineered through SOC and chemical substitution. The complex interplay of magnetic interactions and topological effects in the proposed bulk and doped pnictides opens a promising avenue to explore a wide range of electronic and magnetic phenomena. In particular, this study demonstrates that EuMn2Bi2 hosts tunable magnetic and topological phases driven by electron correlations, chemical substitution, and spin-orbit coupling.

cond-mat.mtrl-sci↗

arXiv2025

Quantifying the Emergence of Selection Prior to Biological Evolution

Selection is central to biological evolution, yet there has been no general experimental framework for quantifying selection in chemical systems before life. Here we demonstrate that selection in a prebiological chemical system can be directly quantified. Assembly Theory predicts that selection corresponds to a transition from undirected to directed exploration of chemical possibility space, measurable through the amount of Assembly, A, which integrates molecular assembly index with observed copy number. By analysing peptide ensembles produced under diverse polymerisation conditions, we show that undirected reactions explore sequence space almost uniformly, yielding exploration ratios of 0.85-0.95, whereas reactions influenced by evolved proteases generate markedly lower ratios (0.51-0.75) and elevated A, consistent with selective reinforcement of specific assembly pathways. Across multiple environments and amino-acid combinations, the exploration ratio and ensemble assembly A robustly distinguish directed from undirected exploration, establishing a general, experimentally tractable metric for detecting and measuring selection in chemical evolution.

q-bio.MN↗

arXiv2025

Dynamic Lagging for Time-Series Forecasting in E-Commerce Finance: Mitigating Information Loss with A Hybrid ML Architecture

Accurate forecasting in the e-commerce finance domain is particularly challenging due to irregular invoice schedules, payment deferrals, and user-specific behavioral variability. These factors, combined with sparse datasets and short historical windows, limit the effectiveness of conventional time-series methods. While deep learning and Transformer-based models have shown promise in other domains, their performance deteriorates under partial observability and limited historical data. To address these challenges, we propose a hybrid forecasting framework that integrates dynamic lagged feature engineering and adaptive rolling-window representations with classical statistical models and ensemble learners. Our approach explicitly incorporates invoice-level behavioral modeling, structured lag of support data, and custom stability-aware loss functions, enabling robust forecasts in sparse and irregular financial settings. Empirical results demonstrate an approximate 5% reduction in MAPE compared to baseline models, translating into substantial financial savings. Furthermore, the framework enhances forecast stability over quarterly horizons and strengthens feature target correlation by capturing both short- and long-term patterns, leveraging user profile attributes, and simulating upcoming invoice behaviors. These findings underscore the value of combining structured lagging, invoice-level closure modeling, and behavioral insights to advance predictive accuracy in sparse financial time-series forecasting.

cs.LG↗

arXiv2025

CoLa: Learning to Interactively Collaborate with Large Language Models

LLMs' remarkable ability to tackle a wide range of language tasks opened new opportunities for collaborative human-AI problem solving. LLMs can amplify human capabilities by applying their intuitions and reasoning strategies at scale. We explore whether human guides can be simulated, by generalizing from human demonstrations of guiding an AI system to solve complex language problems. We introduce CoLa, a novel self-guided learning paradigm for training automated $\textit{guides}$ and evaluate it on two QA datasets, a puzzle-solving task, and a constrained text generation task. Our empirical results show that CoLa consistently outperforms competitive approaches across all domains. Moreover, a small-sized trained guide outperforms a strong model like GPT-4 when acting as a guide. We compare the strategies employed by humans and automated guides by conducting a human study on a QA dataset. We show that automated guides outperform humans by adapting their strategies to reasoners' capabilities and conduct qualitative analyses highlighting distinct differences in guiding strategies.

cs.CL↗

arXiv2025

Rapid parameter estimation with the full symphony of compact binary mergers using meshfree approximation

We present a fast Bayesian inference framework to address the growing computational cost of gravitational-wave parameter estimation. The increased cost is driven by improved broadband detector sensitivity, particularly at low frequencies due to advances in detector commissioning, resulting in longer in-band signals and a higher detection rate. Waveform models now incorporate features like higher-order modes, further increasing the complexity of standard inference methods. Our framework employs meshfree likelihood interpolation with radial basis functions to accelerate Bayesian inference using the IMRPhenomXHM waveform model that incorporates higher modes of the gravitational-wave signal. In the initial start-up stage, interpolation nodes are placed within a constant-match metric ellipsoid in the intrinsic parameter space. During sampling, likelihood is evaluated directly using the precomputed interpolants, bypassing the costly steps of on-the-fly waveform generation and overlap-integral computation. We improve efficiency by sampling in a rotated parameter space aligned with the eigenbasis of the metric ellipsoid, where parameters are uncorrelated by construction. This speeds up sampler convergence. This method yields unbiased parameter recovery when applied to 100 simulated neutron-star-black-hole signals (NSBH) in LIGO-Virgo data, while reducing computational cost by up to an order of magnitude for the longest-duration signal. The meshfree framework equally applies to symmetric compact binary systems dominated by the quadrupole mode, supporting parameter estimation across a broad range of sources. Applied to a simulated NSBH signal in Einstein Telescope data, where the effects of Earth's rotation are neglected for simplicity, our method achieves an O(10^4) speed-up, demonstrating its potential use in the third-generation (3G) era.

gr-qc↗

arXiv2025

Exploring molecular assembly as a biosignature using mass spectrometry and machine learning

Molecular assembly offers a promising path to detect life beyond Earth, while minimizing assumptions based on terrestrial life. As mass spectrometers will be central to upcoming Solar System missions, predicting molecular assembly from their data without needing to elucidate unknown structures will be essential for unbiased life detection. An ideal agnostic biosignature must be interpretable and experimentally measurable. Here, we show that molecular assembly, a recently developed approach to measure objects that have been produced by evolution, satisfies both criteria. First, it is interpretable for life detection, as it reflects the assembly of molecules with their bonds as building blocks, in contrast to approaches that discount construction history. Second, it can be determined without structural elucidation, as it can be physically measured by mass spectrometry, a property that distinguishes it from other approaches that use structure-based information measures for molecular complexity. Whilst molecular assembly is directly measurable using mass spectrometry data, there are limits imposed by mission constraints. To address this, we developed a machine learning model that predicts molecular assembly with high accuracy, reducing error by three-fold compared to baseline models. Simulated data shows that even small instrumental inconsistencies can double model error, emphasizing the need for standardization. These results suggest that standardized mass spectrometry databases could enable accurate molecular assembly prediction, without structural elucidation, providing a proof-of-concept for future astrobiology missions.

cs.LG↗

arXiv2025

Chemputer and Chemputation -- A Universal Chemical Compound Synthesis Machine

Chemputation reframes synthesis as the programmable execution of reaction code on a universally re-configurable hardware graph. Here we prove that a chemputer equipped with a finite, but extensible, set of reagents, catalysts and process conditions, together with a chempiler that maps reaction graphs onto hardware, is universal: it can generate any stable, isolable molecule in finite time and in analytically detectable quantity, provided real-time error correction keeps the per-step fidelity above the threshold set by the molecule's assembly index. The proof is constructed by casting the platform as a Chemical Synthesis Turing Machine (CSTM). The CSTM formalism supplies (i) an eight-tuple state definition that unifies reagents, process variables (including catalysts) and tape operations; (ii) the Universal Chemputation Principle; and (iii) a dynamic-error-correction routine ensuring fault tolerant execution. Linking this framework to assembly theory strengthens the definition of a molecule by demanding practical synthesizability and error correction becomes a prerequisite for universality. We validate the abstraction against >100 \c{hi}DL programs executed on a modular chemputer rigs spanning single step to multi-step routes. Mapping each procedure onto CSTM shows that the cumulative number of unit operations grows linearly with synthetic depth. Together, these results elevate chemical synthesis to the status of a general computation: algorithms written in \c{hi}DL are compiled to hardware, executed with closed-loop correction, and produce verifiable molecular outputs. By formalising chemistry in this way, the chemputer offers a path to shareable, executable chemical code, interoperable hardware ecosystems, and ultimately a searchable, provable atlas of chemical space.

cs.ET↗

arXiv2025

Constructing the Molecular Tree of Life using Assembly Theory and Mass Spectrometry

Here we demonstrate the first biochemistry-agnostic approach to map evolutionary relationships at the molecular scale, allowing the construction of phylogenetic models using mass spectrometry (MS) and Assembly Theory (AT) without elucidating molecular identities. AT allows us to estimate the complexity of molecules by deducing the amount of shared information stored within them when . By examining 74 samples from a diverse range of biotic and abiotic sources, we used tandem MS data to detect 24102 analytes (9262 unique) and 59518 molecular fragments (6755 unique). Using this MS dataset, together with AT, we were able to infer the joint assembly spaces (JAS) of samples from molecular analytes. We show how JAS allows agnostic annotation of samples without fingerprinting exact analyte identities, facilitating accurate determination of their biogenicity and taxonomical grouping. Furthermore, we developed an AT-based framework to construct a biochemistry-agnostic phylogenetic tree which is consistent with genome-based models and outperforms other similarity-based algorithms. Finally, we were able to use AT to track colony lineages of a single bacterial species based on phenotypic variation in their molecular composition with high accuracy, which would be challenging to track with genomic data. Our results demonstrate how AT can expand causal molecular inference to non-sequence information without requiring exact molecular identities, thereby opening the possibility to study previously inaccessible biological domains.

q-bio.PE↗

arXiv2025

Finite size effect in Kuramoto oscillators with inertia on simplicial complex

We investigate the finite-size effects on the dynamical evolution of the Kuramoto model with inertia coupled through triadic interactions. Our findings reveal that fluctuations resulting from the finite size drive the system toward a synchronized state at finite coupling, which contrasts with the analytical predictions {in thermodynamic limit} made for the same system. Building on the analytical calculations performed at the thermodynamic limit, we identify the origin of the synchronization transition that arises because of the finite size. We discover a power-law relationship between the network size and the critical coupling at which the first-order transition to synchronization occurs. Additionally, as inertia increases, there is a significant shift in the critical coupling toward higher values, indicating that inertia counteracts the effects caused by finite size.

nlin.AO↗

arXiv2025

Frictional Contact Network in Dense Suspension Flow

Dense particulate suspensions often exhibit a dramatic increase in viscosity in response to external deformation. This shear thickening behavior has been related to a transition from lubricated, unconstrained pairwise motion to a frictional contact network (FCN) at high stresses. Here, we study the characteristics of the FCN formed during shear thickening to investigate the role of constraints, emphasizing the impact of resistance to gear-like rolling. We contrast the FCN formed by sliding friction alone with that formed by particles with sliding and rolling constraints. Particles with sliding constraints only form a highly interconnected network with primary force chains in the compressive direction, which requires orthogonal support from other force chains. However, orthogonal support is not required for mechanical stability when particles have both sliding and rolling constraints. In addition, the force chains appear linear and longer, reducing the jamming volume fraction for rough/faceted particles. Finally, we propose a novel mechanical stability picture for rough/faceted particles with sliding and rolling constraints, which is crucial for understanding the flow behavior of real-life suspensions.

cond-mat.soft↗

arXiv2025

Symphony of Symmetry Selective Resonances in Fe-MgO-ZnO-MgO-Fe

We propose the perspective of symmetry-selective resonance of the $Δ_1$ states in the Fe/MgO/ZnO/MgO/Fe heterostructures, offering a broad landscape to design magnetic tunnel junctions (MTJs) that yield a towering tunnel magnetoresistance (TMR) up to $3.5\times10^4\%$ with the resistance area (RA) product dipping down to a minimum of $0.05~Ω\cdotμ\text{m}^2$, while maintaining a nearly perfect (99\%) spin polarization. Our predictions are based on the self-consistent coupling of the non-equilibrium Green's function with density functional theory. We also present the charge current, spin current, and TMR with applied voltage of the Fe/MgO(3-layer)/ZnO(3-layer)/MgO(3-layer)/Fe MTJ, which offers a superior performance triad of TMR ($1.3\times10^4\%$), RA ($0.45~Ω\cdotμ\text{m}^2$), and spin polarization (99\%) over a regular Fe/MgO(6-layer)/Fe based MTJ (TMR $\approx 3.4\times10^3\%$, RA $\approx 22~Ω\cdotμ\text{m}^2$). We provide a comprehensive insight integrating the transmission eigenchannel, spectral density, and the band structure of the Fe contacts to establish the role of symmetry-selective resonance in the Fe/MgO/ZnO/MgO/Fe MTJ.

cond-mat.mes-hall↗

arXiv2025

Data-driven stability analysis in a multi-element supercritical Liquid Oxygen-methane combustor

Thermoacoustic instability (TAI) is a pressing problem in rocket combustors. TAI can cause significant damage to a combustor, resulting in mission failure. Therefore, stability analysis is crucial during the design and development phases of a rocket combustor. Stability analysis during the design phase can be substantially aided by the rocket combustor's large eddy simulation (LES). However, the computational cost of LES for full-scale rocket combustors is high. Therefore, using a small set of data from a large eddy simulation of a multi-element full-scale combustor, we investigated the effectiveness and computational needs of many data-driven and physics-driven tools for the classification of the stable and unstable regimes in the current study. Recurrence network analysis (RNA), reservoir computing (RC), and multi-scale permutation entropy (MPEA) analysis are the instruments employed in this study. The regime categorization task is unsuitable for RNA and MPEA, according to the results. With little input data, RC-based metrics may map the stable and unstable regimes and are thought to be computationally inexpensive and straightforward to use. In order to help with the design and development of rocket combustors, the combined LES-RC method to stability analysis is therefore anticipated to result in a notable decrease in processing needs.

physics.flu-dyn↗

arXiv2025

Impact of Fuel Injection Temperature Dynamics on the Stability of Liquid Oxygen-Methane Supercritical Combustion

A crucial factor in the stability of high-pressure rocket-scale combustors is the temperature at which fuel is injected. This study investigates its effect on the stability of supercritical liquid oxygen (LOx)-methane combustion and highlights the impact of shear layer dynamics in cases with lower injection temperatures. The stability features of a rocket-scale combustor operating with multiple injector elements are investigated using a high-fidelity large eddy simulation (LES) framework. The numerical framework combines a flamelet-generated manifold (FGM) combustion model with complex real gas thermodynamics in a scale-resolving simulation setup. It reproduces the non-equilibrium transcritical injection and supercritical combustion characteristics of supercritical methane-oxygen flames. To ascertain the effect of injection temperature on flame and combustor stability, we perform several LES simulations at various methane injection temperatures and produce a stability map. Our analysis shows extremely unstable flame characteristics at lower fuel injection temperatures that are not seen under typical fuel injection circumstances. Below a specific methane injection temperature, LES captures a high-amplitude, self-sustaining instability. It is determined that the combustor becomes unstable below a specific stability boundary temperature. Detailed spectral and dynamic mode decomposition (DMD) analysis of the stable and unstable cases reveals the onset of longitudinal acoustic waves in the combustor. Our thorough investigation pinpoints the instability mechanism, emphasizing that the leading causes of this self-sustaining instability in the combustor are a reduced velocity ratio, fuel buildup, and fuel cut-off occurrences.

physics.flu-dyn↗

arXiv2025

Asymmetric Resonant Ferroelectric Tunnel Junctions for Simultaneous High Tunnel Electroresistance and Low Resistance-Area Product

Ferroelectric tunnel junctions offer potential for non-volatile memory with low power, fast switching, and scalability, but their performance is limited by a high resistance-area product and a low tunnel electroresistance ratio. To address these challenges, we propose a doped HfO2-based, silicon-compatible asymmetric resonant ferroelectric tunnel junction design with a quantum well embedded between two ferroelectric layers, replacing the conventional metal-ferroelectric-metal structure. Using a self-consistent coupling of the non-equilibrium Green's function method with a Preisach-based model, we demonstrate that the quantum well enhances resonant tunneling effects, leading to a simultaneous reduction in the resistance-area product and a boost in the tunnel electroresistance ratio. The low-resistance state becomes more robust, while the high-resistance state is suppressed, improving readout speed and reducing power usage. We observed that incorporating a 2 nm quantum well significantly enhances the tunnel electroresistance ratio, achieving a peak value of approximately 6.15 x 10^4 percent, while simultaneously minimizing the resistance-area product to 47.1 Ohm-cm^2 at 0.175 V. Additionally, the device exhibits negative differential resistance, further enhancing its functionality. Our results confirm that this design enables scalable, energy-efficient, and high-performance non-volatile memory, making it a strong candidate for future memory technologies.

cond-mat.mes-hall↗

arXiv2025

Skyrmion on magnetic tunnel junction: weaving quantum transport with micro-magnetism

Over the last two decades, non-trivial magnetic textures, especially the magnetic skyrmion family, have been extensively explored out of fundamental interest and for diverse possible applications. Given the possible technological and scientific ramifications of skyrmion-texture on magnetic tunnel junctions (Sk-MTJs), in this work, we present a non-equilibrium Green's function (NEGF) based description of Sk-MTJs for both Néel and Bloch textures to capture the spin/charge current across different voltages, temperatures, and sizes. We predict the emergence of a textured spin current from the uniform layer of the Sk-MTJs, along with a radially varying, asymmetric voltage dependence of spin torque. We discuss the identification of Néel-type and Bloch-type skyrmions, including their helicity, based on local measurements of the spin current. We describe the tunneling magnetoresistance (TMR) roll-off in Sk-MTJs with lower cross-sectional areas and higher temperatures based on transmission spectra analysis. Additionally, considering the significant implications of MTJ scaling, we uncover the scaling effects in all skyrmion(AS)-MTJs, where both contacts host skyrmions. We demonstrate the impact of scaling on TMR, spin, and charge currents, providing a pathway for the Sk/AS-MTJs design optimization with miniaturization. We also introduce a computationally efficient, analytically grounded coupled spatio-eigen framework of NEGF, alleviating the sine qua non of the 3D-NEGF for systems that lack translational invariance in the transverse direction, such as Sk/AS-MTJs.

cond-mat.mes-hall↗

arXiv2025

Neural Finite-State Machines for Surgical Phase Recognition

Surgical phase recognition (SPR) is crucial for applications in workflow optimization, performance evaluation, and real-time intervention guidance. However, current deep learning models often struggle with fragmented predictions, failing to capture the sequential nature of surgical workflows. We propose the Neural Finite-State Machine (NFSM), a novel approach that enforces temporal coherence by integrating classical state-transition priors with modern neural networks. NFSM leverages learnable global state embeddings as unique phase identifiers and dynamic transition tables to model phase-to-phase progressions. Additionally, a future phase forecasting mechanism employs repeated frame padding to anticipate upcoming transitions. Implemented as a plug-and-play module, NFSM can be integrated into existing SPR pipelines without changing their core architectures. We demonstrate state-of-the-art performance across multiple benchmarks, including a significant improvement on the BernBypass70 dataset - raising video-level accuracy by 0.9 points and phase-level precision, recall, F1-score, and mAP by 3.8, 3.1, 3.3, and 4.1, respectively. Ablation studies confirm each component's effectiveness and the module's adaptability to various architectures. By unifying finite-state principles with deep learning, NFSM offers a robust path toward consistent, long-term surgical video analysis.

eess.IV↗

arXiv2025

Quantifying the Complexity of Materials with Assembly Theory

Quantifying the evolution and complexity of materials is of importance in many areas of science and engineering, where a central open challenge is developing experimental complexity measurements to distinguish random structures from evolved or engineered materials. Assembly Theory (AT) was developed to measure complexity produced by selection, evolution and technology. Here, we extend the fundamentals of AT to quantify complexity in inorganic molecules and solid-state periodic objects such as crystals, minerals and microprocessors, showing how the framework of AT can be used to distinguish naturally formed materials from evolved and engineered ones by quantifying the amount of assembly using the assembly equation defined by AT. We show how tracking the Assembly of repeated structures within a material allows us formalizing the complexity of materials in a manner accessible to measurement. We confirm the physical relevance of our formal approach, by applying it to phase transformations in crystals using the HCP to FCC transformation as a model system. To explore this approach, we introduce random stacking faults in closed-packed systems simplified to one-dimensional strings and demonstrate how Assembly can track the phase transformation. We then compare the Assembly of closed-packed structures with random or engineered faults, demonstrating its utility in distinguishing engineered materials from randomly structured ones. Our results have implications for the study of pre-genetic minerals at the origin of life, optimization of material design in the trade-off between complexity and function, and new approaches to explore material technosignatures which can be unambiguously identified as products of engineered design.

cond-mat.mtrl-sci↗

arXiv2025

Accelerated parameter estimation of supermassive black hole binaries in LISA using a meshfree approximation

The Laser Interferometer Space Antenna (LISA) will be capable of detecting gravitational waves (GWs) in the milli-Hertz band. Among various sources, LISA will detect the coalescence of supermassive black hole binaries (SMBHBs). Accurate and rapid inference of parameters for such sources will be important for potential electromagnetic follow-up efforts. Rapid Bayesian inference with LISA includes additional complexities as compared to current generation terrestrial detectors in terms of time and frequency dependent antenna response functions. In this work, we extend a recently developed, computationally efficient technique that uses meshfree interpolation methods to accelerate Bayesian reconstruction of compact binaries. Originally developed for second-generation terrestrial detectors, this technique is now adapted for LISA parameter estimation. Using the full inspiral, merger, and ringdown waveform (PhenomD) and assuming rigid adiabatic antenna response function, we show faithful inference of SMBHB parameters from GW signals embedded in stationary, Gaussian instrumental noise. We discuss the computational cost and performance of the meshfree approximation method in estimating the GW source parameters.

gr-qc↗