arXiv subjects

Prateek Gupta

Publications and source records attributed to Prateek Gupta.

At least 19 recordsLinked to original sources

arXiv2026

AGNI: A differentiable MHD stability solver & optimizer for magnetic confinement fusion devices

The existence of an ideal MagnetoHydroDynamic (MHD) equilibrium does not guarantee its stability. Finite toroidal mode number (n) instabilities degrade performance in both tokamaks and stellarators and differentiable stability optimization tools to date have operated only in the infinite-n limit. We present AGNI (Analysis of Global Normal modes in Ideal MHD), a GPU-accelerated, automatically differentiable finite-n ideal MHD stability solver and optimizer. AGNI discretizes the ideal MHD energy principle pseudospectrally in real space using differentiation matrices and geometric coefficients from a DESC equilibrium, giving a variational eigenvalue problem for the plasma displacement, and efficiently finds the most unstable modes. Built on JAX, AGNI yields reverse-mode gradients of the growth rate with respect to boundary-shape and profile parameters without re-solving the equilibrium. We benchmark AGNI against the initial-value code NIMSTELL for a modified Landreman-Buller-Drevlak quasi-helically symmetric equilibrium, recovering the dominant m = n = 4 interchange mode with agreement in both growth rate and eigenfunction structure, and verify the automatic differentiation gradients against central finite differences. We quantify CPU and GPU cost for eigenvalue and gradient evaluation, establish the finite-precision limit on resolving near-marginal eigenvalues, and present a robust scheme to impose incompressibility compatible with gradient-based optimization. AGNI will allow us to optimize tokamaks, stellarators, and mirrors against ideal MHD instabilities.

physics.plasm-ph↗

arXiv2026

The Large-Scale Structure of the Universe through the SKA lenses

The large-scale distribution of galaxies in the Universe forms an intricate, interconnected network known as the cosmic web. Cosmological simulations within the standard Lambda-CDM framework successfully reproduce this filamentary structure and predict that the nodes and filaments are filled with tenuous plasma at temperatures ranging from 10^5-10^8 K. The hottest and luminous plasma in the nodes corresponds to the intra-cluster medium, while the cooler, more tenuous, gas extends along filaments and cluster outskirts. Galaxies and galaxy groups form and flow along these filaments before accreting onto galaxy clusters (the nodes), outlining the dynamical evolution of large-scale structures. During this process, an enormous amount of energy is dissipated through complex plasma processes that can be traced by radio emitting electrons. Despite strong theoretical support for this picture, observational validation remains limited. While massive clusters have been widely detected across various wavelengths, cluster outskirts and the diffuse intergalactic medium within filaments has remained elusive due to their extremely faint emission. The advent of highly sensitive radio facilities such as LOFAR, uGMRT, and MeerKAT has recently enabled a few successful detections of emission from comparatively denser regions of the cosmic-web. These include radio megahalos, permeating the entire cluster volume, as well as bridges of radio emission connecting cluster pairs. In this chapter, we summarize current theoretical insights into the cosmic web, discuss observational strategies and recent discoveries, and highlight how the forthcoming Square Kilometre Array (SKA) is expected to transform our understanding of the cosmic web and the distribution of baryons in the Universe.

astro-ph.CO↗

arXiv2026

Shaping the Digital Future of ErUM Research: Sustainability & Ethics

This workshop report from "Shaping the Digital Future of ErUM Research: Sustainability & Ethics" (Aachen, 2025) reviews progress on sustainability measures in data-intensive ErUM-Data research since the 2023 call-to-action on resource-aware research. It evaluates short-, medium-, and long-term actions around monitoring and reducing CO2 emissions, improving data and software FAIRness, optimizing workflows and computing infrastructures, and aligning operations with low-carbon energy availability, including concepts such as "breathing" computing centers, long-term data storage strategies, and software efficiency certification. The report stresses the need for systematic teaching, training, mentoring, and new support formats to establish sustainable coding and computing practices, particularly among students and early-career researchers, and highlights the importance of dedicated steering and funding instruments to embed sustainability in project planning. Ethical discussions focus on the transformative use of AI in ErUM-Data, addressing autonomy, bias, transparency, explainability, attribution of responsibility, and the risk of deskilling, while reaffirming that accountability for scientific outcomes remains with human researchers. Finally, the report emphasizes that sustainable transformation requires not only technical measures but also targeted awareness-building, communication strategies, incentives, and community-driven initiatives to move from awareness to action and to integrate sustainability and ethics into everyday scientific practice.

physics.comp-ph↗

arXiv2026

Scalar mixing in non-Markovian homogeneous isotropic synthetic turbulence

We show that non-Markovianity of the velocity field is an essential property of turbulent mixing. We demonstrate this via passive scalar mixing by synthetically generated stochastic velocity fields. Including a separate velocity decorrelation time scale for each spatial scale (random sweeping) yields an essentially non-Markovian velocity field with a finite time memory decaying as -5 (for a decaying spectrum) instead of an exponential decay (Markovian), which is obtained by including a constant time scale for all spatial scales, irrespective of the filtering function. We characterize the Lagrangian mixing statistics of both the Markovian and non-Markovian synthetic fields and compare them against a corresponding incompressible direct numerical simulation (DNS). We also study diffusive passive scalar mixing in the Schmidt number range Sc<1 using the DNS and the synthetic fields. While both the synthetic fields recover the -17/3 scalar spectrum for low Schmidt numbers, the mean gradients in a decaying simulation, as well as the production and dissipation of scalar variance in a statistically stationary simulation, are severely underpredicted by the Markovian fields compared to the non-Markovian fields. Throughout, we compare our results with companion 3D DNS to show the necessity of non-Markovianity in synthetic fields to capture mixing dynamics.

physics.flu-dyn↗

arXiv2025

The Role of Social Learning and Collective Norm Formation in Fostering Cooperation in LLM Multi-Agent Systems

A growing body of multi-agent studies with LLMs explores how norms and cooperation emerge in mixed-motive scenarios, where pursuing individual gain can undermine the collective good. While prior work has explored these dynamics in both richly contextualized simulations and simplified game-theoretic environments, most LLM systems featuring common-pool resource (CPR) games provide agents with explicit reward functions directly tied to their actions. In contrast, human cooperation often emerges without explicit knowledge of the payoff structure or how individual actions translate into long-run outcomes, relying instead on heuristics, communication, and enforcement. We introduce a CPR simulation framework that removes explicit reward signals and embeds cultural-evolutionary mechanisms: social learning (adopting strategies and beliefs from successful peers) and norm-based punishment, grounded in Ostrom's principles of resource governance. Agents also individually learn from the consequences of harvesting, monitoring, and punishing via environmental feedback, enabling norms to emerge endogenously. We establish the validity of our simulation by reproducing key findings from existing studies on human behavior. Building on this, we examine norm evolution across a $2\times2$ grid of environmental and social initialisations (resource-rich vs. resource-scarce; altruistic vs. selfish) and benchmark how agentic societies comprised of different LLMs perform under these conditions. Our results reveal systematic model differences in sustaining cooperation and norm formation, positioning the framework as a rigorous testbed for studying emergent norms in mixed-motive LLM societies. Such analysis can inform the design of AI systems deployed in social and organizational contexts, where alignment with cooperative norms is critical for stability, fairness, and effective governance of AI-mediated environments.

cs.MA↗

arXiv2025

Large Language Models Enable Design of Personalized Nudges across Cultures

Nudge strategies are effective tools for influencing behaviour, but their impact depends on individual preferences. Strategies that work for some individuals may be counterproductive for others. We hypothesize that large language models (LLMs) can facilitate the design of individual-specific nudges without the need for costly and time-intensive behavioural data collection and modelling. To test this, we use LLMs to design personalized decoy-based nudges tailored to individual profiles and cultural contexts, aimed at encouraging air travellers to voluntarily offset CO$_2$ emissions from flights. We evaluate their effectiveness through a large-scale survey experiment ($n=3495$) conducted across five countries. Results show that LLM-informed personalized nudges are more effective than uniform settings, raising offsetting rates by 3-7$\%$ in Germany, Singapore, and the US, though not in China or India. Our study highlights the potential of LLM as a low-cost testbed for piloting nudge strategies. At the same time, cultural heterogeneity constrains their generalizability underscoring the need for combining LLM-based simulations with targeted empirical validation.

cs.CY↗

arXiv2025

Active Scalar Mixing by Homogeneous Isotropic Turbulence

We study the mixing of active scalars by homogeneous isotropic incompressible stochastic velocity fields. We consider both Navier-Stokes generated turbulent fields as well as artificially generated homogeneous isotropic stochastic fields. We use Fourier pseudospectral direct numerical simulations to study the mixing dynamics of two non-reacting species of different density ratios. We use the Atwood number to create a denser mixture and a lighter mixture. We show that in the absence of stirring, a denser mixture homogenizes faster than the lighter mixture. The direction of the density gradient causes the interface across which the molecular diffusion occurs to expand outward for the denser mixture and inward for the lighter mixture. The stirring process, which enhances the diffusion process, increases the rate of homogenization in both mixing methods under study. We define a new mixing metric for studying the mixing evolution of active scalars, which indicates that a denser inhomogeneity in a lighter mixture spreads faster but homogenizes slower. For low Mach number turbulence, there is a negligible coupling between the density gradients and the velocity field responsible for stirring. The post-stirring behavior of active scalars is found to be similar to passive scalars, where the scalar energy spectra decay exponentially and exhibit self-similarity. The turbulence fields generated by solving the Navier-Stokes equation homogenize both the mixtures faster than the synthetic cases. We show that matching the kinetic energy spectra and inertial subrange scaling of a synthetically generated stochastic field with that of a Navier-Stokes generated field is not enough to study mixing dynamics.

physics.flu-dyn↗

arXiv2025

Long-term atomistic finite-temperature substitutional diffusion

Simulating long-term mass diffusion kinetics with atomic precision is important to predict chemical and mechanical properties of alloys over time scales of engineering interest in applications, including (but not limited to) alloy heat treatment, corrosion resistance, and hydrogen embrittlement. We present a new strategy to bridge from the time scale of atomic vibrations to that of vacancy-mediated atomic hops by a combination of statistical mechanics-based Gaussian phase packets (GPP) relaxation and a nudged elastic band (NEB)-facilitated harmonic transition state theory (H-TST) time update. We validate the approach by simulating bulk self-diffusion in copper and the segregation of vacancies and magnesium to a stacking fault and a symmetric tilt grain boundary in aluminum, modeled with an embedded atom method (EAM) potential. The method correctly predicts the kinetics in bulk copper and equilibrium impurity concentrations in aluminum, in agreement with the Langmuir-Mclean solution in the dilute limit. Notably, this technique can reach realistic diffusion time scales of days, weeks, and even years in a computational time of hours, demonstrating its capability to study the long-term chemo-thermo-mechanically coupled behavior of atomic ensembles.

cond-mat.mtrl-sci↗

arXiv2025

Drums of high width

We provide a family of $5$-dimensional prismatoids whose width grows linearly in the number of vertices. This provides a new infinite family of counter-examples to the Hirsch conjecture whose excess width grows linearly in the number of vertices, and answers a question of Matschke, Santos and Weibel.

math.CO↗

arXiv2025

Role of phase distortion in nonlinear saturation of the unstable acoustic modes in hypersonic parallel flow boundary layer

We analyze the role of the relative phasing in the nonlinear saturation of the unstable Mack modes in a hypersonic parallel flow boundary layer in two dimensions (2D). As the linearly unstable Mack modes extract energy from the mean flow, the perturbation energy cascades into higher harmonics as well as the mean flow. The higher harmonics are generated with < 0.5% of total perturbation energy at steady state, indicating a very small role of higher harmonics in 2D. Additionally, the higher harmonics propagate with the same phase speed as the unstable mode, indicating wave steepening and a coherent energy cascade. The mean flow gets decelerated and heated due to the continuous extraction of the perturbation energy into traveling modes and the viscous dissipation of these modes. Unlike unstable modes in classical hydrodynamics, we show that the distortion in relative phasing between the streamwise velocity and wall-normal velocity due to nonlinear distortion of the mean flow is dominant. Using asymptotic reconstruction of the unstable eigenmodes, we compute the perturbation energy budgets in the linear and nonlinear regimes. Through energy budgets, we show that the viscous effects in the wall layer and the viscous effects in the critical layer sufficiently capture the distortion in phase due to the mean-flow distortion. We then combine this in a numerical model for calculating the steady-state perturbation energy and mean-flow distortion through the nonlinear saturation of unstable Mack modes in a hypersonic parallel flow boundary layer in 2D. Throughout, we compare the results of approximate theoretical analysis with 2D direct numerical simulations (DNS).

physics.flu-dyn↗

arXiv2024

OASIS: Open Agent Social Interaction Simulations with One Million Agents

There has been a growing interest in enhancing rule-based agent-based models (ABMs) for social media platforms (i.e., X, Reddit) with more realistic large language model (LLM) agents, thereby allowing for a more nuanced study of complex systems. As a result, several LLM-based ABMs have been proposed in the past year. While they hold promise, each simulator is specifically designed to study a particular scenario, making it time-consuming and resource-intensive to explore other phenomena using the same ABM. Additionally, these models simulate only a limited number of agents, whereas real-world social media platforms involve millions of users. To this end, we propose OASIS, a generalizable and scalable social media simulator. OASIS is designed based on real-world social media platforms, incorporating dynamically updated environments (i.e., dynamic social networks and post information), diverse action spaces (i.e., following, commenting), and recommendation systems (i.e., interest-based and hot-score-based). Additionally, OASIS supports large-scale user simulations, capable of modeling up to one million users. With these features, OASIS can be easily extended to different social media platforms to study large-scale group phenomena and behaviors. We replicate various social phenomena, including information spreading, group polarization, and herd effects across X and Reddit platforms. Moreover, we provide observations of social phenomena at different agent group scales. We observe that the larger agent group scale leads to more enhanced group dynamics and more diverse and helpful agents' opinions. These findings demonstrate OASIS's potential as a powerful tool for studying complex systems in digital environments.

cs.CL↗

arXiv2024

Machine learning and optimization-based approaches to duality in statistical physics

The notion of duality -- that a given physical system can have two different mathematical descriptions -- is a key idea in modern theoretical physics. Establishing a duality in lattice statistical mechanics models requires the construction of a dual Hamiltonian and a map from the original to the dual observables. By using simple neural networks to parameterize these maps and introducing a loss function that penalises the difference between correlation functions in original and dual models, we formulate the process of duality discovery as an optimization problem. We numerically solve this problem and show that our framework can rediscover the celebrated Kramers-Wannier duality for the 2d Ising model, reconstructing the known mapping of temperatures. We also discuss an alternative approach which uses known features of the mapping of topological lines to reduce the problem to optimizing the couplings in a dual Hamiltonian, and explore next-to-nearest neighbour deformations of the 2d Ising duality. We discuss future directions and prospects for discovering new dualities within this framework.

cond-mat.stat-mech↗

arXiv2024

Empirical evidence of Large Language Model's influence on human spoken communication

From the printing press to social media, innovations in communication technology have repeatedly reshaped how ideas spread through human culture. Chatbots powered by generative artificial intelligence constitute a new medium, encoding cultural patterns in their neural representations and disseminating them in conversations with hundreds of millions of people. Whether these patterns transmit into human language, and ultimately shape human culture, is a fundamental question. While fully quantifying the causal impact of a chatbot like ChatGPT on human culture is challenging, lexical shifts in human spoken communication may offer an early indicator. Here we show that words preferentially generated by ChatGPT, such as delve, showcase, boast, intricacies and meticulous, increased abruptly in spontaneous human speech. A synthetic-control analysis of 737,083 hours of conversation from 824,634 podcast episodes, screened for unscripted speech, causally links this shift to ChatGPT's release. The measurable influence on spontaneous speech suggests that humans internalize the lexical choices of large language models (LLMs). A preregistered experiment (N = 496) confirms they do, as a brief chatbot interaction led participants to adopt its words as their own, persisting past a distractor task and confirmed in forced lexical choice, indicating entrenchment in the active vocabulary. Together these results show that machines trained on human data now feed their own traits back into human language, integrating LLMs into the ongoing processes of cultural evolution.. This coupling raises concerns about linguistic homogenization and the capacity of a few major AI providers for latent cultural influence at scale.

cs.CY↗

arXiv2024

Mixing of active scalars due to random shock waves in two dimensions

In this work, we investigate the mixing of active scalars in two dimensions by the stirring action of stochastically generated shock waves. We use direct numerical simulations (DNS) of the interaction of shock waves with two non-reacting species to analyse the mixing dynamics for different Atwood numbers (At). Unlike passive scalars, the presence of density gradients in active scalars makes the species diffusion nonlinear, introducing a concentration gradient-driven term and a density gradient-driven nonlinear dissipation term in the concentration evolution equation. We show that the direction of the concentration gradient causes the interface across which molecular diffusion occurs to expand outward or inward, even without any stirring action. Shock waves enhance the mixing process by increasing the perimeter of the interface and by sustaining concentration gradients. Negative Atwood number mixtures sustain concentration gradients for longer time than positive Atwood number mixtures due to the so-called nonlinear dissipation terms. We estimate the time till when the action of stirring is dominant over molecular mixing. We also highlight the role of baroclinicity in increasing the interface perimeter in the stirring dominant regime. We compare the stirring effect of shock waves on mixing of passive scalars with active scalars and show that the vorticity generated by baroclinicity is responsible for the folding and stretching of the interface in the case of active scalars. We conclude by showing that lighter mixtures with denser inhomogeneities (At < 0) take longer time to homogenise than the denser mixtures with lighter inhomogeneities (At > 0).

physics.flu-dyn↗

arXiv2024

Generative Active Learning for the Search of Small-molecule Protein Binders

Despite substantial progress in machine learning for scientific discovery in recent years, truly de novo design of small molecules which exhibit a property of interest remains a significant challenge. We introduce LambdaZero, a generative active learning approach to search for synthesizable molecules. Powered by deep reinforcement learning, LambdaZero learns to search over the vast space of molecules to discover candidates with a desired property. We apply LambdaZero with molecular docking to design novel small molecules that inhibit the enzyme soluble Epoxide Hydrolase 2 (sEH), while enforcing constraints on synthesizability and drug-likeliness. LambdaZero provides an exponential speedup in terms of the number of calls to the expensive molecular docking oracle, and LambdaZero de novo designed molecules reach docking scores that would otherwise require the virtual screening of a hundred billion molecules. Importantly, LambdaZero discovers novel scaffolds of synthesizable, drug-like inhibitors for sEH. In in vitro experimental validation, a series of ligands from a generated quinazoline-based scaffold were synthesized, and the lead inhibitor N-(4,6-di(pyrrolidin-1-yl)quinazolin-2-yl)-N-methylbenzamide (UM0152893) displayed sub-micromolar enzyme inhibition of sEH.

q-bio.BM↗

arXiv2024

Finite-temperature grain boundary properties from quasistatic atomistics

Grain boundary (GB) properties greatly influence the mechanical, electrical, and thermal response of polycrystalline materials. Most computational studies of GB properties at finite temperatures use molecular dynamics (MD), which is computationally expensive, limited in the range of accessible timescales, and requires cumbersome techniques like thermodynamic integration to estimate free energies. This restricts the reasonable computation (without incurring excessive computational expense) of GB properties to regimes that are often unrealistic, such as zero temperature or extremely high strain rates. Consequently, there is a need for simulation methodology that avoids the timescale limitations of MD, while providing reliable estimates of GB properties. The Gaussian Phase-Packet (GPP) method is a temporal coarse-graining technique that can predict relaxed atomic structures at finite temperature in the quasistatic limit. This work applies GPP, combined with the quasiharmonic approximation for computing the free energy, to the problem of determining the free energy and shear coupling factor of grain boundaries in metals over a range of realistic temperatures. Validation is achieved by comparison to thermodynamic integration, which confirms that the presented approach captures relaxed-energy GB structures and shear coupling factors at finite temperature with a high degree of accuracy.

cond-mat.mtrl-sci↗

arXiv2024

Dissipation of nonlinear acoustic waves in thermoviscous pores

We derive a nonlinear acoustic wave propagation model for analysing the thermoviscous dissipation in narrow pores with wavy walls. As the nonlinear waves propagate in the thermoviscous pores, the wave-steepening effect competes with the bulk dissipation, as well as the thermoviscous heat transfer and shear from the pore walls. Consequently, the length scale of the wave is modified. We use the characteristic nonlinear wave thickness scale to obtain linear and nonlinear wave equations governing the unsteady shock-wall interaction. We also perform two-dimensional shock-resolved DNS of the wave propagation inside the pores and compare the results with model equations. We show that for flat-walls and shock strength parameter $\epsilon$, the dimensional wall heat-flux and shear scale as $\epsilon$. For wavy walls, the scaling becomes $\epsilon^{3/2 - n(k)}$ where $k$ is the wall-waviness wavenumber and the exponent $n$ increases from $0.5$ for $k=0$ to $n(k)\approx0.65$ for $k=10$, $n(k)\approx 0.75$ for $k=20$, and $n(k)\approx0.85$ for $k=40$. Hence, increasing the wall waviness reduces the dependence of the wall heat-flux and shear on nonlinear acoustic wave strength.

physics.flu-dyn↗

arXiv2023

Adaptive friends-of-friends algorithm for identifying gravitationally bound cosmological structures

The Universe at the present epoch is found to be a network of matter over-dense and under-dense regions. To date, this picture of the Universe is best revealed through cosmological large-volume simulations and large-scale galaxy redshift surveys, in which, the most important step is the appropriate identification of structures. So far, these structures are identified using various group finding codes, mostly based on the friends-of-friends (FoF) or spherical over-density (SO) algorithms. Although, the main purpose is to identify gravitationally bound structures, surprisingly, the mass information has hardly been used effectively by these codes. Moreover, the methods used so far either constrain the over-density or use the real unstructured geometry only. Even though these are key factors in the accurate determination of structures-mass information, hardly any attempt has been made as yet to consider these important parameters together while formulating the grouping algorithms. In this paper, we present our proposed algorithm which takes care of all the above-mentioned relevant features and ensures the bound structures by means of physical quantities, mainly mass and the total energy information. We introduced a novel concept of physically relevant arm-length for each element depending on their individual gravity leading to a distinct linking length for each unique pair of elements. This proposed algorithm is thus fundamentally new that, not only able to catch the gravitationally bound, real unstructured geometry, it does identify it roughly within a predefined physically motivated density threshold. Such a thing could not be simultaneously achieved before by any of the usual FoF or SO-based methods. We also demonstrate the unique ability of the code in the appropriate identification of structures, both from large volume cosmological simulations as well as from galaxy redshift surveys.

astro-ph.CO↗