arXiv subjects

Apoorva Singh

Publications and source records attributed to Apoorva Singh.

6 recordsLinked to original sources

arXiv2026

Deal Me Maybe: The Role of Emotions in Multi-Agent Negotiation

Negotiation is a demanding social task for LLM agents, requiring strategic reasoning, persuasion, and interpersonal adaptation. Yet existing benchmarks often treat agents as emotionally neutral, overlooking a key driver of human bargaining behavior. We study how prompt-conditioned emotions affect LLM-based price negotiation. In a controlled framework, buyer and seller agents are independently assigned one of six emotional states and negotiate over 350 real consumer products under two budget conditions. Across 36 emotion-pair settings and five widely used LLMs, we find that emotions strongly shape outcomes. Angry buyers almost never reach agreement (0.39% deal rate), while happy buyers agree most often (28.91%), but obtain worse prices than fearful buyers. Emotion effects are role-dependent: buyer emotion mainly drives acceptance and rejection, whereas seller emotion shapes concession dynamics. These effects influence not only language, but also termination behavior and price trajectories, raising concerns for emotion-conditioned agents in commerce.

cs.AI↗

arXiv2025

Large-scale patterns of small-scale vorticity interactions foster moist convection during cyclogenesis

The formation and intensification of a tropical cyclone is a complex phenomenon involving several feedback interactions between momentum and energetics of the storm, and across multiple spatio-temporal scales. Background vorticity interactions in the turbulent atmosphere play a crucial role in the formation of cyclones. How these vorticity interactions lead to convective organization and sustain a disastrous cyclonic vortex amidst a turbulent atmosphere remains elusive. Moreover, what processes distinguish depressions that develop into a cyclone from those that do not? Here, we investigate the role of small-scale vorticity interactions in the background flow in sustaining large-scale organization during the emergence of a cyclone. We construct time-varying complex networks where geographical locations are nodes and connections between nodes represent short-time vorticity correlations. Only those nodes are connected that are in spatial proximity corresponding to sub-meso length scales. Each network is constructed for 29 hours of data; consecutive networks are separated by three hours, thus revealing the evolution of local coherence in vorticity dynamics. We discover that small-scale vorticity interactions manifest as large-scale emergent patterns. Further, we establish that organized moist convection is significantly correlated to regions of locally coherent vorticity dynamics during the intensification of a depression that forms a cyclone; however, such correlations are not sustained during non-developing cases. Using modal analysis of time-evolving network connectivity, we show that these large-scale patterns are essentially large-scale modes of propagation of coherence in small-scale vorticity dynamics. We explain that such propagation is facilitated by moisture feedback at small-scales and self-organized patterns at large-scales.

physics.ao-ph↗

arXiv2025

Talk, Snap, Complain: Validation-Aware Multimodal Expert Framework for Fine-Grained Customer Grievances

Existing approaches to complaint analysis largely rely on unimodal, short-form content such as tweets or product reviews. This work advances the field by leveraging multimodal, multi-turn customer support dialogues, where users often share both textual complaints and visual evidence (e.g., screenshots, product photos) to enable fine-grained classification of complaint aspects and severity. We introduce VALOR, a Validation-Aware Learner with Expert Routing, tailored for this multimodal setting. It employs a multi-expert reasoning setup using large-scale generative models with Chain-of-Thought (CoT) prompting for nuanced decision-making. To ensure coherence between modalities, a semantic alignment score is computed and integrated into the final classification through a meta-fusion strategy. In alignment with the United Nations Sustainable Development Goals (UN SDGs), the proposed framework supports SDG 9 (Industry, Innovation and Infrastructure) by advancing AI-driven tools for robust, scalable, and context-aware service infrastructure. Further, by enabling structured analysis of complaint narratives and visual context, it contributes to SDG 12 (Responsible Consumption and Production) by promoting more responsive product design and improved accountability in consumer services. We evaluate VALOR on a curated multimodal complaint dataset annotated with fine-grained aspect and severity labels, showing that it consistently outperforms baseline models, especially in complex complaint scenarios where information is distributed across text and images. This study underscores the value of multimodal interaction and expert validation in practical complaint understanding systems. Resources related to data and codes are available here: https://github.com/sarmistha-D/VALOR

cs.CL↗

arXiv2023

Helson-Lowdenslager and de Branges type theorems in the setting of continuous rotationally symmetric norms

A Helson-Lowdenslager type result has been proved by Chen in the context of Lebesgue spaces of the unit circle equipped with a continuous rotationally symmetric norm by studying the simply invariant subspaces of the operator of multiplication by the coordinate function $z$. In this paper, we generalize Chen's result by obtaining a description of simply invariant subspaces for multiplication by $z^n$. A de Branges type result is also proved for Hardy spaces equipped with continuous rotationally symmetric norms.

math.FA↗

arXiv2022

Multiplication by finite Blaschke factors on a general class of Hardy spaces

A broader class of Hardy spaces and Lebesgue spaces have been introduced recently on the unit circle by considering continuous $\|.\|_1$-dominating normalized gauge norms instead of the classical norms on measurable functions and a Beurling type result has been proved for the operator of multiplication by the coordinate function. In this paper, we generalize the above Beurling type result to the context of multiplication by a finite Blaschke factor $B(z)$ and also derive the common invariant subspaces of $B^2(z)$ and $B^3(z)$. These results lead to a factorization result for all functions in the Hardy space equipped with a continuous rotationally symmetric norm.

math.FA↗

arXiv2019

Performance Bounds of Nanoparticles Laden Volumetric Absorption Solar Thermal Platforms in Laminar Flow Regime

Recent success in synthesizing thermally stable nanofluids at low costs is a significant breakthrough in the evolution of volumetric absorption based solar thermal systems. However, we have yet not been able to clearly identify the range of operating and design parameters in which volumetric absorption could prove to be beneficial. One of the key reasons being that we have not been able to fully understand the heat transfer mechanisms involved in these novel systems. The present work takes a few steps further in this direction wherein we have developed a comprehensive and mechanistic theoretical framework which is robust enough to account for coupled transport phenomena and orders of magnitudes of operating parameters for host of receiver design configurations. Moreover, we have also modeled equivalent surface absorption based systems to provide a comparison between volumetric and surface absorption processes under similar operating conditions. Performance characteristics reveal that particularly at high solar concentration ratios, volumetric absorption-based receivers could have 35% - 49% higher thermal efficiencies compared to their surface absorption-based counterparts. Finally, the present work serves to define optimal performance domains of these solar thermal systems, particularly in the laminar flow regime (200 < Re < 1600) and over a wide range of solar concentration ratios (5-100) and inlet fluid temperatures (293-593K).

physics.app-ph↗