arXiv subjects

Saurabh Pandey

Publications and source records attributed to Saurabh Pandey.

15 recordsLinked to original sources

arXiv2026

A Tri-Agent Framework for Evaluating and Aligning Question Clarification Capabilities of Large Language Models

Large Language Models (LLMs) are increasingly deployed in interactive systems where understanding user intent precisely is paramount. A key capability for such systems is effective question clarification, especially when user queries are ambiguous or underspecified. This paper introduces a novel tri-agent framework for the robust evaluation of an LLM's ability to engage in clarifying dialogue. Our framework comprises three distinct LLM-based agents: (1) a Question Clarifying Agent (QCA), the system under evaluation, tasked with identifying ambiguities and posing clarifying questions; (2) a Respondent Agent (RA), designed to simulate human user responses, potentially including irrelevant or challenging replies; and (3) an Evaluator Agent (EA), an LLM-as-a-judge, which assesses the quality of the dialogue based on a comprehensive set of metrics. We detail a methodology for synthetic data generation in the supply chain domain as an example. We propose metrics evaluating ambiguity handling, question quality, dialogue efficiency, language appropriateness, and final intent alignment. We also briefly discuss the validation of the EA against human judgments. This work provides a structured approach to benchmark, validate, and improve the clarification capabilities of conversational LLM applications.

cs.CL↗

arXiv2026

TRACE: TRajectory Attribution for Automated Context Engineering

Production AI agents fail when their context sources -- system prompts, knowledge bases, tool descriptions, and procedural skills -- contain errors or gaps. Current maintenance relies on manual log review and ad-hoc debugging, creating a scalability bottleneck as interaction volume grows. We present TRACE (TRajectory Attribution for Automated Context Engineering), an automated feedback loop that mines historical agent trajectories to diagnose and remediate context failures. Our key insight is that trajectories are rich with implicit dissatisfaction signals -- user corrections, rephrasing, abandonment cues -- that reveal precisely where context sources failed, without explicit feedback collection. Unlike model fine-tuning, TRACE operates on the context layer, enabling rapid iteration without retraining. We make four contributions: (1) a trajectory mining framework that systematically extracts diagnostic information from historical agent executions; (2) multi-component causal attribution that extends textual gradients from monolithic prompt optimization to heterogeneous context sources (skills, knowledge bases, tools, prompts); (3) exploratory verification, where agents actively read context sources to distinguish content gaps requiring CREATE from stale content requiring UPDATE, achieving 96% operation accuracy; and (4) a reusable simulation methodology and verifiable benchmark addressing the absence of open datasets for context debugging, with a six-category fault taxonomy, ground truth annotations, and a cross-layer verification protocol. On 60 dissatisfaction traces spanning three complexity tiers (up to 16 execution nodes), TRACE achieves 72.7% root cause attribution and 82% end-to-end fix effectiveness, showing that over 80% of context-layer failures can be automatically diagnosed and remediated by mining historical trajectories, an overlooked resource in production systems.

cs.AI↗

arXiv2026

Cesium Based Laser-Atomic Oscillator

We report the first demonstration of a laser-atomic oscillator with cesium (Cs) atoms. A laser-atomic oscillator (LAO) is analogous to an active mode-locked laser with a self-excited modulator, i.e. atoms, at a ground-state hyperfine transition frequency. Therefore, a LAO can be configured as the simplest active atomic clock or a self-oscillating, earth-field atomic magnetometer that delivers oscillation signals both optically and electrically. With the current experimental Cs-LAO setup, when it is configured as an atomic clock using the 0--0 hyperfine transition, the short-term fractional frequency instability is around 10$^{-10}$ level. When it is configured as a self-oscillating magnetometer using a magnetically-sensitive hyperfine transition, the magnetic field sensitivity is around 100 fT/$\sqrt{\rm{Hz}}$ at 60 Hz. The presented Cs-LAO uses a cavity length from $\sim6.5$ cm to $\sim11.4$ cm. Ultimately, the minimal length of a Cs-LAO device can be $\leq1.63$ cm. Our new efforts unlock the potential of building truly chip-scale atomic clocks and magnetometers.

physics.atom-ph↗

arXiv2025

Atomtronic Matter-Wave Optics

Matterwaves made up of ultra-cold quantum-degenerate atoms have enabled the creation of tools having unprecedented sensitivity and precision in measuring gravity, rotation or magnetic fields. Applications range from gravitational wave detection and tests of Einstein's equivalence principle to inertial sensing for navigation and gravitational gradient sensing for oil and mineral exploration. In this letter, we introduce atom-optics as a novel tool of manipulating matterwaves in ring-shaped coherent waveguides. We collimate and focus matterwaves derived from Bose-Einstein Condensates (BECs) and ultra-cold thermal atoms in ring-shaped time-averaged adiabatic potentials. We demonstrate `delta-kick cooling' of BECs, reducing their expansion energies by a factor of 34. The atomtronic waveguide ring has a radius of only $485\,\mu m$, compared to other state-of-the-art experiments requiring zero gravity or chambers of ten meter. This level of control with extremely reduced spatial requirements is an important step towards atomtronic quantum sensors.

quant-ph↗

arXiv2025

Multi-loop and Multi-axis Atomtronic Sagnac Interferometry

We report the experimental realization of a large-area and multi-axis atomtronic interferometer in an optical waveguide for rotation sensing. A large enclosed area is achieved through multi-loop operation in a guided atom interferometer using Bose-Einstein condensates. We demonstrate a three-loop interferometer with a total interrogation time of ~ 0.4 s and an enclosed area of 8.7 mm$^2$- the largest reported in a fully guided or one-dimensional setup. High-contrast interference fringes are observed for up to five Sagnac orbits in a smaller loop-area configuration. Our approach enables interleaved rotation measurements about multiple arbitrary axes within the same experimental setup. We present results for area-enclosing interferometers in both horizontal and vertical planes, demonstrating that the interferometer contrast remains comparable across orthogonal orientations of the enclosed area vectors.

physics.atom-ph↗

arXiv2025

Outlier-Robust Multi-Model Fitting on Quantum Annealers

Multi-model fitting (MMF) presents a significant challenge in Computer Vision, particularly due to its combinatorial nature. While recent advancements in quantum computing offer promise for addressing NP-hard problems, existing quantum-based approaches for model fitting are either limited to a single model or consider multi-model scenarios within outlier-free datasets. This paper introduces a novel approach, the robust quantum multi-model fitting (R-QuMF) algorithm, designed to handle outliers effectively. Our method leverages the intrinsic capabilities of quantum hardware to tackle combinatorial challenges inherent in MMF tasks, and it does not require prior knowledge of the exact number of models, thereby enhancing its practical applicability. By formulating the problem as a maximum set coverage task for adiabatic quantum computers (AQC), R-QuMF outperforms existing quantum techniques, demonstrating superior performance across various synthetic and real-world 3D datasets. Our findings underscore the potential of quantum computing in addressing the complexities of MMF, especially in real-world scenarios with noisy and outlier-prone data.

cs.CV↗

arXiv2024

Circuit-based leakage-to-erasure conversion in a neutral atom quantum processor

Leakage out of the computational subspace is a major limitation of current state-of-the-art neutral-atom quantum computers and a significant challenge for scalable systems. In a quantum processor with cesium atoms, we demonstrate proof-of-principle circuit-based conversion of leakage errors to erasure errors via Leakage Detection Units (LDUs), which non-destructively map information about the presence or absence of the qubit onto the state of an ancilla. With a standard LDU circuit, we successfully convert leakage errors to erasure errors for all major leakage pathways while preserving the quantum information in the case that no leakage occurred. We benchmark the performance of the LDU using a three-outcome low-loss state detection method and also explore the advantages of three-outcome measurements for LDUs. We find that the LDU detects atom-loss errors with ~93.4% accuracy, limited by technical imperfections of our apparatus. We further compile and execute a SWAP LDU, wherein the roles of the original data atom and ancilla atom are exchanged under the action of the LDU, providing 'free refilling' of atoms in the case of leakage errors. This circuit-based leakage-to-erasure error conversion is a critical component of a neutral-atom quantum processor where the quantum information may significantly outlive the lifetime of any individual atom in the quantum register.

quant-ph↗

arXiv2022

A Three-Stage Self-Training Framework for Semi-Supervised Semantic Segmentation

Semantic segmentation has been widely investigated in the community, in which the state of the art techniques are based on supervised models. Those models have reported unprecedented performance at the cost of requiring a large set of high quality segmentation masks. To obtain such annotations is highly expensive and time consuming, in particular, in semantic segmentation where pixel-level annotations are required. In this work, we address this problem by proposing a holistic solution framed as a three-stage self-training framework for semi-supervised semantic segmentation. The key idea of our technique is the extraction of the pseudo-masks statistical information to decrease uncertainty in the predicted probability whilst enforcing segmentation consistency in a multi-task fashion. We achieve this through a three-stage solution. Firstly, we train a segmentation network to produce rough pseudo-masks which predicted probability is highly uncertain. Secondly, we then decrease the uncertainty of the pseudo-masks using a multi-task model that enforces consistency whilst exploiting the rich statistical information of the data. We compare our approach with existing methods for semi-supervised semantic segmentation and demonstrate its state-of-the-art performance with extensive experiments.

cs.CV↗

arXiv2019

AEDGE: Atomic Experiment for Dark Matter and Gravity Exploration in Space

We propose in this White Paper a concept for a space experiment using cold atoms to search for ultra-light dark matter, and to detect gravitational waves in the frequency range between the most sensitive ranges of LISA and the terrestrial LIGO/Virgo/KAGRA/INDIGO experiments. This interdisciplinary experiment, called Atomic Experiment for Dark Matter and Gravity Exploration (AEDGE), will also complement other planned searches for dark matter, and exploit synergies with other gravitational wave detectors. We give examples of the extended range of sensitivity to ultra-light dark matter offered by AEDGE, and how its gravitational-wave measurements could explore the assembly of super-massive black holes, first-order phase transitions in the early universe and cosmic strings. AEDGE will be based upon technologies now being developed for terrestrial experiments using cold atoms, and will benefit from the space experience obtained with, e.g., LISA and cold atom experiments in microgravity. This paper is based on a submission (v1) in response to the Call for White Papers for the Voyage 2050 long-term plan in the ESA Science Programme. ESA limited the number of White Paper authors to 30. However, in this version (v2) we have welcomed as supporting authors participants in the Workshop on Atomic Experiments for Dark Matter and Gravity Exploration held at CERN: ({\tt https://indico.cern.ch/event/830432/}), as well as other interested scientists, and have incorporated additional material.

gr-qc↗

arXiv2019

Hypersonic Bose-Einstein Condensates in Accelerator Rings

Some of the most sensitive and precise measurements to date are based on matterwave interferometry with freely falling atomic clouds. Examples include high-precision measurements of inertia, gravity and rotation. In order to achieve these very high sensitivities, the interrogation time has to be very long and consequently the experimental apparatus has to be very tall, in some cases reaching ten or even one hundred meters. Cancelling gravitational acceleration, e.g. in atomtronic circuits6,7 and matterwave guides, will result in compact devices having much extended interrogation times and thus much increased sensitivity both for fundamental and practical measurements. In this letter, we demonstrate extremely smooth and controllable matterwave guides by transporting Bose-Einstein condensates (BEC) over macroscopic distances: We use a novel neutral-atom accelerator ring to bring BECs to very high speeds (16x their velocity of sound) and transport them in a magnetic matterwave guide for 15 cm whilst fully preserving their internal coherence. The high angular momentum of more than 40000h per atom gives access to the higher Landau levels of quantum Hall states. The hypersonic velocities combined with our ability to control the potentials with pico-Kelvin precision open new perspectives in the study of superfluidity and give rise to new regimes of tunnelling and transport. Coherent matterwave guides are expected to enable interaction times of several seconds in highly compact devices. These developments will result in portable guided-atom interferometers for applications such as inertial navigation and gravity mapping.

quant-ph↗

arXiv2019

Bi-chromatic adiabatic shells for atom interferometry

We demonstrate bi-chromatic adiabatic magnetic shell traps as a novel tool for matterwave interferometry. Using two strong RF fields, we dress the $|1,-1\rangle $ and $ |2,1\rangle$ states of Rubidium Bose-Einstein Condensates thus creating two independently controllable shell traps. This allows us to match the two traps and, using microwave pulses, create a state-dependent clock-type interferometer. Given the low horizontal confinement of the interferometer, the atoms can be made to spread out thus yielding a 2D sheet, which could be used in a direct imaging interferometer. This interferometer can be sensitive to spatially varying electric or magnetic fields, which could be DC, AC, RF fields or microwaves, or even local variations in gravity. We demonstrate that the trap-matching afforded by the independent control of the shell traps allows long coherence times which will result in highly sensitive imaging matterwave interferometers.

quant-ph↗

arXiv2019

Blocking Mechanism of Porn Website in India: Claim and Truth

In last few years, the addiction of internet is apparently recognized as the serious threat to the health of society. This internet addiction gives an impetus to pornographic addiction because most of the pornographic content is accessible through internet. There have been ethical concerns on blocking the contents over internet. In India Uttarakhand High court has taken initiative for the blocking of pornographic content over internet. Technocrats are coming up with various innovative mechanisms to block the content over internet with various techniques, although long ago in 2015. The Supreme Court of India has already asked to block some of the websites but it could not be materialized. The focus of this research paper is to review the effectiveness of blocking existing web content blocking mechanism of pornographic websites in Indian context.

cs.CY↗

arXiv2019

Simple precision measurements of optical beam sizes

We present a simple precision-method to quickly and accurately measure the diameters of Gaussian beams, Airy spots, and central peak of Bessel beams ranging from the sub-millimeter to many centimeters without specialized equipment. Simply moving a wire through the beam and recording the relative losses using an optical power meter, one can easily measure the beam diameters with a precision of 1%. The accuracy of this method has been experimentally verified for Gaussian beams down to the limit of a commercial slit-based beam profiler (3%).

physics.ins-det↗

arXiv2016

Matter-wave interferometers using TAAP rings

We present two novel matter-wave Sagnac interferometers based on ring- shaped time-averaged adiabatic potentials (TAAP). For both the atoms are put into a superposition of two different spin states and manipulated independently using elliptically polarized rf-fields. In the first interferometer the atoms are accelerated by spin-state-dependent forces and then travel around the ring in a matter-wave guide. In the second one the atoms are fully trapped during the entire interferometric sequence and are moved around the ring in two spin-state-dependent "buckets". Corrections to the ideal Sagnac phase are investigated for both cases. We experimentally demonstrate the key atom-optical elements of the interferometer such as the independent manipulation of two different spin states in the ring-shaped potentials under identical experimental conditions.

quant-ph↗

arXiv2011

Detection of a single cobalt microparticle with a microfabricated atomic magnetometer

We present magnetic detection of a single, 2 μm diameter cobalt microparticle using an atomic magnetometer based on a microfabricated vapor cell. These results represent an improvement by a factor of 105 in terms of the detected magnetic moment over previous work using atomic magnetometers to detect magnetic microparticles. The improved sensitivity is due largely to the use of small vapor cells. In an optimized setup, we predict detection limits of 0.17 μm^3.

physics.atom-ph↗