arXiv subjects

Manav Shah

Publications and source records attributed to Manav Shah.

8 recordsLinked to original sources

arXiv2025

Compton amplitude for massive bosons of arbitrary spin

In this work, we write down an analytic expression of electromagnetic tree-level Compton amplitude for a completely symmetric traceless (bosonic) higher spin particle in any dimension. Our analysis is restricted to the three-point function, which is unique in the Infrared and responsible for the Coulomb interactions/soft photon theorem. We propose an analogue of $R_\xi$ gauge in a theory of higher spin particles. We demonstrate that the theory is unitary only for $\xi=1,\infty$.

hep-th↗

arXiv2025

Data Unlearning in Diffusion Models

Recent work has shown that diffusion models memorize and reproduce training data examples. At the same time, large copyright lawsuits and legislation such as GDPR have highlighted the need for erasing datapoints from diffusion models. However, retraining from scratch is often too expensive. This motivates the setting of data unlearning, i.e., the study of efficient techniques for unlearning specific datapoints from the training set. Existing concept unlearning techniques require an anchor prompt/class/distribution to guide unlearning, which is not available in the data unlearning setting. General-purpose machine unlearning techniques were found to be either unstable or failed to unlearn data. We therefore propose a family of new loss functions called Subtracted Importance Sampled Scores (SISS) that utilize importance sampling and are the first method to unlearn data with theoretical guarantees. SISS is constructed as a weighted combination between simpler objectives that are responsible for preserving model quality and unlearning the targeted datapoints. When evaluated on CelebA-HQ and MNIST, SISS achieved Pareto optimality along the quality and unlearning strength dimensions. On Stable Diffusion, SISS successfully mitigated memorization on nearly 90% of the prompts we tested.

cs.LG↗

arXiv2024

Multi-Aspect Reviewed-Item Retrieval via LLM Query Decomposition and Aspect Fusion

While user-generated product reviews often contain large quantities of information, their utility in addressing natural language product queries has been limited, with a key challenge being the need to aggregate information from multiple low-level sources (reviews) to a higher item level during retrieval. Existing methods for reviewed-item retrieval (RIR) typically take a late fusion (LF) approach which computes query-item scores by simply averaging the top-K query-review similarity scores for an item. However, we demonstrate that for multi-aspect queries and multi-aspect items, LF is highly sensitive to the distribution of aspects covered by reviews in terms of aspect frequency and the degree of aspect separation across reviews. To address these LF failures, we propose several novel aspect fusion (AF) strategies which include Large Language Model (LLM) query extraction and generative reranking. Our experiments show that for imbalanced review corpora, AF can improve over LF by a MAP@10 increase from 0.36 to 0.52, while achieving equivalent performance for balanced review corpora.

cs.IR↗

arXiv2023

Adapted poling to break the nonlinear efficiency limit in nanophotonic lithium niobate waveguides

Nonlinear frequency mixing is of critical importance in extending the wavelength range of optical sources. It is also indispensable for emerging applications such as quantum information and photonic signal processing. Conventional lithium niobate with periodic poling is the most widely used device for frequency mixing due to the strong second-order nonlinearity. The recent development of nanophotonic lithium niobate waveguides promises improvements of nonlinear efficiencies by orders of magnitude with sub-wavelength optical conferment. However, the intrinsic nanoscale inhomogeneity in nanophotonic lithium niobate limits the coherent interaction length, leading to low nonlinear efficiencies. Therefore, the performance of nanophotonic lithium niobate waveguides is still far behind conventional counterparts. Here, we overcome this limitation and demonstrate ultra-efficient second order nonlinearity in nanophotonic lithium niobate waveguides significantly outperforming conventional crystals. This is realized by developing the adapted poling approach to eliminate the impact of nanoscale inhomogeneity in nanophotonic lithium niobate waveguides. We realize overall secondharmonic efficiency near 10^4 %/W without cavity enhancement, which saturates the theoretical limit. Phase-matching bandwidths and temperature tunability are improved through dispersion engineering. The ideal square dependence of the nonlinear efficiency on the waveguide length is recovered. We also break the trade-off between the energy conversion ratio and pump power. A conversion ratio over 80% is achieved in the single-pass configuration with pump power as low as 20 mW.

physics.optics↗

arXiv2022

Visible-telecom broadband optical isolator based on dynamic modulation in thin-film lithium niobate

Optical isolators are an essential component of photonic systems. Current integrated optical isolators have limited bandwidths due to stringent phase-matching conditions, resonant structures, or material absorption. Here, we demonstrate an ultra-broadband integrated optical isolator in thin-film lithium niobate photonics. We use dynamic standing-wave modulation in a tandem configuration to break Lorentz reciprocity and achieve isolation. We measure an isolation ratio of 15 dB and insertion loss below 0.5 dB for a design wavelength of 1550 nm. In addition, we experimentally show that this isolator can simultaneously operate at visible and telecom wavelengths with comparable performance. Isolation bandwidths ~100 nm can be achieved simultaneously at both visible and telecom wavelengths. Our device's large bandwidth, high flexibility, and real-time tunability can enable novel non-reciprocal functionality on integrated photonic platforms.

physics.optics↗

arXiv2022

Clarification of exceptional point contribution for photonic sensing

Exceptional points, with simultaneous coalescence of eigen-values and eigen-vectors, can be realized with non-Hermitian photonic systems. With the enhanced response, exceptional points have been proposed to improve the performance of photonic sensing. Recently, there are intense debate about the actual sensing advantage of exceptional points. The major concern is that intrinsic noise is also amplified at exceptional points. Here, we aim to clarify the contribution of exceptional points for photonic sensing. This is achieved by analyzing the condition to realize divergent quantum Fisher information in linear non-Hermitian photonic systems. We show that the divergence of quantum Fisher information is the result of lasing threshold, instead of exceptional points. However, exceptional points correspond to the condition that lasing threshold is simultaneously achieved across multiple photonic modes. Therefore, exceptional points can further improve the sensitivity on top of lasing threshold. On the other hand, exceptional points alone cannot provide sensing advantage.

physics.optics↗

arXiv2018

Design, Fabrication and Characterization of nanoplasmonic lattice for trapping of ultracold atoms

Ultracold atom-traps on a chip enhances the practical application of atom traps in quantum information processing, sensing, and metrology. Plasmon mediated near-field optical potentials are promising for trapping atoms. The combination of plasmonic nanostructures and ultracold atoms has the potential to create a two dimensional array of neutral atoms with lattice spacing smaller than that of lattices created from interfering light fields -- the optical lattices. We report the design, fabrication and characterization of a nano-scale array of near-field optical traps for neutral atoms using plasmonic nanostructures. The building block of the array is a metallic nano-disc fabricated on the surface of an ITO-coated glass substrate. We numerically simulate the electromagnetic field-distribution using Finite Difference Time Domain method around the nanodisc, and calculate the intensity, optical potential and the dipole force for $^{87}$Rb atoms. The optical near-field generated from the fabricated nanostructures is experimentally characterized by using Near-field Scanning Optical Microscopy. We find that the optical potential and dipole force has all the desired characteristics to trap cold atoms when a blue-detuned light-field is used to excite the nanostructures. This trap can be used for effective trapping and manipulation of isolated atoms and also for creating a lattice of neutral atoms having sub-optical wavelength lattice spacing. Near-field measurements are affected by the influence of tip on the sub-wavelength structure. We present a deconvolution method to extract the actual near-field profile from the measured data.

physics.optics↗

arXiv2018

A Broadband Superabsorber at Optical Frequencies: Design and Demonstration

Metasurface based super absorbers exhibit near unity absorbance. While the absorption peak can be tuned by the geometry/size of the sub-wavelength resonator, broadband absorption can be obtained by placing multiple resonators of various size or shapes in a unit cell. Metal dispersion hinders high performance broadband absorption at optical frequencies and careful designing is essential to achieve good structures. We propose a novel analytical framework for designing a broadband super absorber which is much faster than the time consuming full wave simulations that are employed so far. Analytical expressions are derived for the wavelength dependency of the design parameters, which are then used in the optimization of broadband absorption. Numerical simulations report an average polarization-independent absorption of ~97 in the 450 to 950 nm spectral region with a near unity absorption (99.36) in the 500 to 850 nm region. Experimentally, we demonstrate an average absorption over 98 in the 450 to 950 nm spectral region at 20 degree incident angle The designed super absorber is polarization insensitive and has a weak launch angle dependency. The proposed framework simplifies the design process and provides a quicker optimal solution for high performance broadband super absorbers.

cond-mat.mtrl-sci↗