Searcharxiv⌕ Search

arXiv subjects

Harshit Joshi

Publications and source records attributed to Harshit Joshi.

13 recordsLinked to original sources

Elasto-inertial force exerted on a sphere oscillating in a fluid near a soft interface

Fluid inertia can remain important in confined flows, even at low Reynolds number when convective inertia is negligible; yet its coupling with interface deformability is an open problem. Here, we theoretically investigate the high-frequency oscillations of an immersed sphere near an elastic wall in the lubrication regime. We first reconcile previous predictions for the added mass near a rigid wall by consistently matching the inner lubrication flow with the outer flow. Using this rigid-wall problem as a reference, we then develop a perturbative analysis of wall deformation and derive the leading elasto-inertial correction to the hydrodynamic force. Our results reveal that the coupling between interface softness and fluid inertia produces a dominant non-dissipative force contribution scaling as the power four of the frequency in the high-frequency limit, thus corresponding to the fourth time derivative of the sphere's position. This finding may have implications regarding the short-time dynamics of colloidal particles near deformable interfaces in biological and soft-matter settings.

cond-mat.soft↗

SLIDERS: Evidence Synthesis with Autonomous Database Reconciliation for Systematic Reviews

Systematic reviews requires comprehensive evidence synthesis from large document corpora to answer targeted research questions and are foundational in finance, social sciences, and other technical fields. Manual construction of evidence tables is labor-intensive, and recent LLM-based assistants relying on embedding or keyword search often fail to meet the coverage standards of systematic reviews. We introduce SLIDERS, a novel LLM-based methodology for automatically assembling evidence tables tailored to research questions for systematic reviews. SLIDERS extracts full-text excerpts that serve as direct evidence or as provenance for structured data. Core to SLIDERS is an autonomous evidence reconciliation coding agent that iteratively writes SQL to analyze and reconcile extracted evidence, bringing together information fragmented across documents, resolving inconsistencies and removing duplicates across records. We evaluate SLIDERS on BrowseComp+, achieving 90% accuracy, and show that it outperforms all baselines on additional benchmarks within the LLM's context window. To assess scalability, we introduce two new evidence-synthesis benchmarks, WikiCeleb100 (3.9M tokens) and FinQ100 (36M tokens), which include follow-up questions beyond the original research query. Although SLIDERS synthesizes evidence knowing only the original question, it answers 78.9% and 55.2% of all questions correctly on WikiCeleb100 and FinQ100, respectively

cs.CL↗

SHELDON: Reasoning through Dual Search over Hypothesis and Observation Spaces

Test-time scaling has enabled LLMs to solve increasingly complex reasoning, discovery, and optimization tasks. However, existing approaches often gather observations while seeking evidence for current hypotheses, then generate new hypotheses conditioned on those hypotheses and their associated evidence. This couples observation discovery to existing hypotheses, favoring local refinements over exploration of unknown hypotheses regions. Inspired by the hypothetico-deductive method (Hempel, 1945) and dual-space models of scientific discovery (Klahr & Dunbar, 1988), we introduce SHELDON, a scientific reasoning harness that separately represents and jointly explores hypothesis and observation spaces. By actively seeking new observations along with new hypotheses, SHELDON can discover observations that its current hypotheses do not explain. These unexplained observations expose gaps in its understanding, motivate new investigations, and guide the generation of hypotheses beyond refinements of existing ones. Across diagnosis, abstract reasoning, and optimization benchmarks, SHELDON outperforms evaluated baselines by up to 26.7 points. Additional analysis shows that SHELDON generates up to 4.5x more novel solutions.

cs.AI↗

Quantitative measurement of fluid inertial effects in confined Brownian motion

The hydrodynamic response of Brownian particles in liquids is fundamentally altered by inertial forces arising from unsteady momentum transport in the surrounding fluid. These forces are of two distinct types: the added mass and the history effect. While both are well understood in bulk and weakly-confined geometries, under deterministic driving, their respective behaviours under strong confinement and thermal fluctuations remain scarcely addressed, unclear and often entangled together. The goal of the present study is thus to fill this fundamental gap. The confinement-induced variations of the two distinct inertial contributions are separately investigated in the vicinity of a flat, rigid wall, using a combination of broadrange thermal colloidal-probe atomic-force-microscopy experiments, advanced numerical simulations and theoretical arguments. Specifically, the separation of the added-mass and history-force contributions is achieved through their different frequency-scaling signatures within the assumed form of the generalized dissipation coefficient. Our results pave the way towards a complete picture of Brownian motion at interfaces, in the lubrication regime, with direct relevance to nanofluidics and interfacial biophysics.

cond-mat.soft↗

Viscous Settling of Bravais Unit-Cells

We study experimentally and theoretically the Stokesian settling of a well-known class of porous shapes: Bravais lattice unit-cells, whose porosity we vary controllably by changing their lattice spacing. In our experiments, conducted in a square cuboidal container with its long-axis aligned along gravity, we find that the settling speed U and the solid fraction ϕ of these lattice units obey a power-law relationship U $\propto$ ϕ^γ , with an exponent γ = 0.43 independent of their shape. To understand the observed scaling exponent, we analytically and numerically investigate the settling of the simple cubic structure under different approximations. We find that the walls of the container, though far from the sinking object, have a defining effect. Our Stokesian boundary integral simulations show that the Faxen's boundary correction captures the wall-effects accurately and enables us to discount the wall-effect from the experimental data, yielding a power-law exponent γ = 0.30 for settling in an unbounded domain. The power-law relating sinking speed and porosity is a step towards predictively understanding the sedimentation fluxes of complex objects in the clouds and the oceans. However, the applicability of this universal scaling to irregular and biologically richer aggregates found in nature remains an open direction.

cond-mat.soft↗

The significance of two-way coupling in two-dimensional, dusty turbulence

The significance of small-scale forcing of particles on the carrier two-dimensional turbulent flow has been shown to influence the spectral scaling properties of the carrier fluid. We investigate possible consequences of such two-way coupling in a turbulent suspension of inertial particles through one- and two-point Eulerian and Lagrangian statistics. In particular, we find signatures of enhanced intermittency in the vorticity distributions. We characterize the changes in the small-scale geometry of the flow via the Okubo-Weiss parameter. Finally, we examine the scaling properties of the second-order vorticity structure functions and find a non-trivial form of scale-invariance at finite mass loading. Motivated by these observations, we propose an effective multiscale forcing framework in which particle feedback is modeled as a spatially localized small-scale forcing. This dual-scale forcing captures the emergence of modified spectral scaling and provides a minimal Eulerian description of particle-laden turbulence that reproduces key statistical signatures of the system.

physics.flu-dyn↗

Controllable and Reliable Knowledge-Intensive Task-Oriented Conversational Agents with Declarative Genie Worksheets

Large Language Models can carry out human-like conversations in diverse settings, responding to user requests for tasks and knowledge. However, existing conversational agents implemented with LLMs often struggle with hallucination, following instructions with conditional logic, and integrating knowledge from different sources. These shortcomings compromise the agents' effectiveness, rendering them unsuitable for deployment. To address these challenges, we introduce Genie, a programmable framework for creating knowledge-intensive task-oriented conversational agents. Genie can handle involved interactions and answer complex queries. Unlike LLMs, it delivers reliable, grounded responses through advanced dialogue state management and supports controllable agent policies via its declarative specification -- Genie Worksheet. This is achieved through an algorithmic runtime system that implements the developer-supplied policy, limiting LLMs to (1) parse user input using a succinct conversational history, and (2) generate responses according to supplied context. Agents built with Genie outperform SOTA methods on complex logic dialogue datasets. We conducted a user study with 62 participants on three real-life applications: restaurant reservations with Yelp, as well as ticket submission and course enrollment for university students. Genie agents with GPT-4 Turbo outperformed the GPT-4 Turbo agents with function calling, improving goal completion rates from 21.8% to 82.8% across three real-world tasks.

cs.AI↗

Dynamics and clustering of sedimenting disc lattices

Uniform arrays of particles tend to cluster as they sediment in viscous fluids. Shape anisotropy of the particles enriches these dynamics by modifying the mode-structure and the resulting instabilities of the array. A one-dimensional lattice of sedimenting spheroids in the Stokesian regime displays either an exponential or an algebraic rate of clustering depending on the initial lattice spacing (Chajwa et al. 2020). This is caused by an interplay between the Crowley mechanism which promotes clumping, and a shape-induced drift mechanism which subdues it. We theoretically and experimentally investigate the sedimentation dynamics of one-dimensional lattices of oblate spheroids or discs and show a stark difference in clustering behaviour: the Crowley mechanism results in clumps comprised of several spheroids, whereas the drift mechanism results in pairs of spheroids whose asymptotic behavior is determined by pair-hydrodynamic interactions. We find that a Stokeslet, or point-particle, approximation is insufficient to accurately describe the instability and that the corrections provided by the first-reflection are necessary for obtaining some crucial dynamical features. As opposed to a sharp boundary between exponential growth and neutral eigenvalues under the Stokeslet approximation, the first-reflection correction leads to exponential growth for all initial perturbations, but far more rapid algebraic growth than exponential growth at large lattice spacing $d$. For discs with aspect ratio 0.125, corresponding to the experimental value, the instability growth rate is found to decrease with increasing lattice spacing $d$, approximately as $d^{-4.5}$, which is faster than the $d^{-2}$ for spheres (Crowley, 1971). Sedimenting pairs predominantly come together to form '$\perp$', which our theory accounts for through an analysis that builds on Koch & Shaqfeh (1989).

cond-mat.soft↗

Electrostatic interactions between anisotropic particles

We investigate the electrostatic interactions between two charged anisotropic conductors using a combination of asymptotic and numerical methods. For widely separated particles, we employ the method of reflections to analyze the interactions. Although the formulation applies to conductors of arbitrary shapes, it is specifically implemented for spheroid-sphere systems to capture anisotropy effects in a simple configuration. In near-contact cases with axisymmetric configurations, the lubrication approximation is used to extend the analysis. Additionally, we develop a Boundary Integral Method (BIM) to study particle interactions at arbitrary separations, validating the results with asymptotic solutions for both near and far fields. We derive analytical expressions for the electrostatic force and torque on a spheroid due to another spheroid in the far-field regime. When combined with hydrodynamic effects, the electrostatic torque competes with the hydrodynamically favourable alignments of a pair of settling spheroids in certain regions while reinforcing them in others. Consequently, the inclusion of electrostatic effects may influence the instability observed in dilute suspensions of spheroids.

cond-mat.soft↗

Sedimentation Dynamics of Bodies with Two Planes of Symmetry

We show that bodies with two planes of symmetry can display a range of behaviors even without inertia. Any such body supports a conserved quantity in its dynamics, and is either a settler, a drifter or a flutterer, depending only on its shape. At large time, settlers and drifters, respectively, fall vertically and obliquely, while flutterers rotate forever while executing intricate patterns. The dynamics of flutterers decouples into a periodic and a Floquet part with different timescales, giving periodicity or quasiperiodicity. We design a set of bodies and use the boundary integral method to show that settlers, drifters, and flutterers, all lie in this set.

physics.flu-dyn↗

FLAME: A small language model for spreadsheet formulas

Spreadsheets are a vital tool for end-user data management. Using large language models for formula authoring assistance in these environments can be difficult, as these models are expensive to train and challenging to deploy due to their size (up to billions of parameters). We present FLAME, a transformer-based model trained exclusively on Excel formulas that leverages domain insights to achieve competitive performance while being substantially smaller (60M parameters) and training on two orders of magnitude less data. We curate a training dataset using sketch deduplication, introduce an Excel-specific formula tokenizer, and use domain-specific versions of masked span prediction and noisy auto-encoding as pre-training objectives. We evaluate FLAME on formula repair, formula completion, and similarity-based formula retrieval. FLAME can outperform much larger models, such as the Davinci (175B) and Cushman (12B) variants of Codex and CodeT5 (220M), in 10 of 14 evaluation settings for the repair and completion tasks. For formula retrieval, FLAME outperforms CodeT5, CodeBERT, and GraphCodeBERT.

cs.PL↗

Repair Is Nearly Generation: Multilingual Program Repair with LLMs

Most programmers make mistakes when writing code. Some of these mistakes are small and require few edits to the original program -- a class of errors recently termed last mile mistakes. These errors break the flow for experienced developers and can stump novice programmers. Existing automated repair techniques targeting this class of errors are language-specific and do not easily carry over to new languages. Transferring symbolic approaches requires substantial engineering and neural approaches require data and retraining. We introduce RING, a multilingual repair engine powered by a large language model trained on code (LLMC) such as Codex. Such a multilingual engine enables a flipped model for programming assistance, one where the programmer writes code and the AI assistance suggests fixes, compared to traditional code suggestion technology. Taking inspiration from the way programmers manually fix bugs, we show that a prompt-based strategy that conceptualizes repair as localization, transformation, and candidate ranking, can successfully repair programs in multiple languages with minimal effort. We present the first results for such a multilingual repair engine by evaluating on 6 different languages and comparing performance to language-specific repair engines. We show that RING can outperform language-specific repair engines for three of these languages.

cs.SE↗

Neurosymbolic Repair for Low-Code Formula Languages

Most users of low-code platforms, such as Excel and PowerApps, write programs in domain-specific formula languages to carry out nontrivial tasks. Often users can write most of the program they want, but introduce small mistakes that yield broken formulas. These mistakes, which can be both syntactic and semantic, are hard for low-code users to identify and fix, even though they can be resolved with just a few edits. We formalize the problem of producing such edits as the last-mile repair problem. To address this problem, we developed LaMirage, a LAst-MIle RepAir-engine GEnerator that combines symbolic and neural techniques to perform last-mile repair in low-code formula languages. LaMirage takes a grammar and a set of domain-specific constraints/rules, which jointly approximate the target language, and uses these to generate a repair engine that can fix formulas in that language. To tackle the challenges of localizing the errors and ranking the candidate repairs, LaMirage leverages neural techniques, whereas it relies on symbolic methods to generate candidate repairs. This combination allows LaMirage to find repairs that satisfy the provided grammar and constraints, and then pick the most natural repair. We compare LaMirage to state-of-the-art neural and symbolic approaches on 400 real Excel and PowerFx formulas, where LaMirage outperforms all baselines. We release these benchmarks to encourage subsequent work in low-code domains.

cs.SE↗