arXiv subjects

Ravi Kumar

Publications and source records attributed to Ravi Kumar.

At least 55 records · Page 3Linked to original sources

arXiv2025

Quarter- and half-filled quantum Hall states and their topological orders revealed by daughter states in bilayer graphene

Even-denominator fractional quantum Hall states are promising candidates for fault-tolerant quantum computing due to their underlying non-Abelian topological orders. However, the topological order of these states remains hotly debated. Here, we report transport measurements on ultra-clean bilayer graphene heterostructures, where we observed four quarter-filled states and their corresponding Levin-Halperin daughter states, constraining their topological order. Moreover, we complete the sequence of half-filled plateaus by detecting states at v=-3/2 and v=1/2 whose daughters suggest an alternating sequence of non-Abelian orders. This pattern suggests a universal origin supporting their use in identifying topological order at even-denominator fillings, though further confirmation is needed via direct measurements. The observed quarter- and half-filled states appear in N=0 and N=1 Landau levels, respectively, and thus highlight a competition between interactions favoring paired states of either four- or two-flux composite fermions. Additionally, we observe several 'next-generation' quantum Hall states that require strong interactions between composite fermions.

cond-mat.mes-hall↗

arXiv2025

LAuReL: Learned Augmented Residual Layer

One of the core pillars of efficient deep learning methods is architectural improvements such as the residual/skip connection, which has led to significantly better model convergence and quality. Since then the residual connection has become ubiquitous in not just convolutional neural networks but also transformer-based architectures, the backbone of LLMs. In this paper we introduce Learned Augmented Residual Layer (LAuReL) -- a novel generalization of the canonical residual connection -- with the goal to be an in-situ replacement of the latter while outperforming on both model quality and footprint metrics. Our experiments show that using LAuReL can help boost performance for both vision and language models. For example, on the ResNet-50, ImageNet 1K task, it achieves 60% of the gains from adding an extra layer, while only adding 0.003% more parameters, and matches it while adding 2.6 times fewer parameters. Similarly, when pre-training 1B and 4B parameter LLMs, LAuReL improves performance on a variety of challenging downstream evaluation tasks by 2.54% to 20.05%, while adding only 0.012% and 0.1% additional parameters, respectively.

cs.LG↗

arXiv2025

Phase transitions in $q$-state clock model

The $q-$state clock model, sometimes called the discrete $XY$ model, is known to show a second-order (symmetry breaking) phase transition in two-dimension (2D) for $q\le 4$ ($q=2$ corresponds to the Ising model). On the other hand, the $q\to\infty$ limit of the model corresponds to the $XY$ model, which shows the infinite order (non-symmetry breaking) Berezinskii-Kosterlitz-Thouless (BKT) phase transition in 2D. Interestingly, the 2D clock model with $q\ge 5$ is predicted to show three different phases and two associated phase transitions. There are varying opinions about the actual characters of phases and the associated transitions. In this work, we develop the basic and higher-order mean-field (MF) theories to study the $q$-state clock model systematically. Our MF calculations reaffirm that, for large $q$, there are three phases: (broken) $\mathbb{Z}_q$ symmetric ferromagnetic phase at the low temperature, emergent $U(1)$ symmetric BKT phase at the intermediate temperature, and paramagnetic (disordered) phase at the high temperature. The phase transition at the higher temperature is found to be of the BKT type, and the other transition at the lower temperature is argued to be a large-order spontaneous symmetry-breaking (SSB) type (the largeness of transition order yields the possibility of having some of the numerical characteristics of a BKT transition). The higher-order MF theory developed here better characterizes phases by estimating the spin-spin correlation between two neighbors.

cond-mat.stat-mech↗

arXiv2025

Quantifying Cross-Modality Memorization in Vision-Language Models

Understanding what and how neural networks memorize during training is crucial, both from the perspective of unintentional memorization of potentially sensitive information and from the standpoint of effective knowledge acquisition for real-world, knowledge-intensive tasks. While previous studies primarily investigate memorization within a single modality, such as text memorization in large language models or image memorization in diffusion models, unified multimodal models are becoming increasingly prevalent in practical applications. In this work, we focus on the unique characteristics of cross-modality memorization and conduct a systematic study centered on vision-language models. To facilitate controlled experiments, we first introduce a synthetic persona dataset comprising diverse synthetic person images and textual descriptions. We quantify factual knowledge memorization and cross-modal transferability by training models on a single modality and evaluating their performance in the other. Our results reveal that facts learned in one modality transfer to the other, but a significant gap exists between recalling information in the source and target modalities. Furthermore, we observe that this gap exists across various scenarios, including more capable models, machine unlearning, and the multi-hop case. At the end, we propose a baseline method to mitigate this challenge. We hope our study can inspire future research on developing more robust multimodal learning techniques to enhance cross-modal transferability.

cs.CV↗

arXiv2025

Al$_2$MnCu: A magnetically ordered member of the Heusler alloy family despite having a valence electron count of 24

The magnetic property of the Heusler alloys can be predicted by the famous Slater-Pauling (S-P) rule, which states the total magnetic moment ($m_t$) of such materials can be expressed as $m _t\,=\,(N_V-24)\,μ_B/f.u.$, where $N_V$ is the total valence electron count (VEC). Consequently, no Heusler alloys having VEC = 24 are theoretically expected as well as experimentally reported to have any magnetic ordering. Recently, a special class of Heusler alloys with 50\% concentration of $p$-block elements (anti-Heusler) have been identified, although none of such reported compounds belong to the VEC 24 category. Here, we report a new anti-Heusler alloy, Al$_2$MnCu, that undergoes long-range ferromagnetic (FM) ordering with $T_{\rm C}\sim$315 K and a large magnetic moment of $\sim$1.8 $μ_B$/f.u. despite having VEC 24. A phenomenological model based on molecular orbital hybridization is also proposed to understand the magnetism and unusual deviation from the standard S-P rule.

cond-mat.mtrl-sci↗

arXiv2025

Balls-and-Bins Sampling for DP-SGD

We introduce the Balls-and-Bins sampling for differentially private (DP) optimization methods such as DP-SGD. While it has been common practice to use some form of shuffling in DP-SGD implementations, privacy accounting algorithms have typically assumed that Poisson subsampling is used instead. Recent work by Chua et al. (ICML 2024), however, pointed out that shuffling based DP-SGD can have a much larger privacy cost in practical regimes of parameters. In this work we show that the Balls-and-Bins sampling achieves the "best-of-both" samplers, namely, the implementation of Balls-and-Bins sampling is similar to that of Shuffling and models trained using DP-SGD with Balls-and-Bins sampling achieve utility comparable to those trained using DP-SGD with Shuffling at the same noise multiplier, and yet, Balls-and-Bins sampling enjoys similar-or-better privacy amplification as compared to Poisson subsampling in practical regimes.

cs.LG↗

arXiv2025

On the Differential Privacy and Interactivity of Privacy Sandbox Reports

The Privacy Sandbox initiative from Google includes APIs for enabling privacy-preserving advertising functionalities as part of the effort around limiting third-party cookies. In particular, the Private Aggregation API (PAA) and the Attribution Reporting API (ARA) can be used for ad measurement while providing different guardrails for safeguarding user privacy, including a framework for satisfying differential privacy (DP). In this work, we provide an abstract model for analyzing the privacy of these APIs and show that they satisfy a formal DP guarantee under certain assumptions. Our analysis handles the case where both the queries and database can change interactively based on previous responses from the API.

cs.CR↗

arXiv2025

Crosslingual Capabilities and Knowledge Barriers in Multilingual Large Language Models

Large language models (LLMs) are typically multilingual due to pretraining on diverse multilingual corpora. But can these models relate corresponding concepts across languages, i.e., be crosslingual? This study evaluates state-of-the-art LLMs on inherently crosslingual tasks. We observe that while these models show promising surface-level crosslingual abilities on machine translation and embedding space analyses, they struggle with deeper crosslingual knowledge transfer, revealing a crosslingual knowledge barrier in both general (MMLU benchmark) and domain-specific (Harry Potter quiz and TOFU benchmark) contexts. Since simple inference-time mitigation methods offer only limited improvement, we propose fine-tuning of LLMs on mixed-language data, which effectively reduces these gaps, even when using out-of-domain datasets like WikiText. Our findings suggest the need for explicit optimization to unlock the full crosslingual potential of LLMs. Our code is publicly available at https://github.com/google-research/crosslingual-knowledge-barriers.

cs.CL↗

arXiv2025

On Memorization of Large Language Models in Logical Reasoning

Large language models (LLMs) achieve good performance on challenging reasoning benchmarks, yet could also make basic reasoning mistakes. This contrasting behavior is puzzling when it comes to understanding the mechanisms behind LLMs' reasoning capabilities. One hypothesis is that the increasingly high and nearly saturated performance on common reasoning benchmarks could be due to the memorization of similar problems. In this paper, we systematically investigate this hypothesis with a quantitative measurement of memorization in reasoning tasks, using a dynamically generated logical reasoning benchmark based on Knights and Knaves (K&K) puzzles. We find that LLMs could interpolate and memorize the training puzzles (achieving near-perfect accuracy) after fine-tuning, yet they struggle with slight variations of these puzzles. On the other hand, we show that while fine-tuning leads to heavy memorization, it also consistently improves generalization performance. Through in-depth analyses with perturbation tests, cross difficulty-level transferability, probing model internals, and fine-tuning with wrong answers, we establish that LLMs develop reasoning skills on K&K puzzles alongside memorization. Finally, our analysis based on a per-sample memorization score sheds light on how LLMs switch between reasoning and memorization when solving logical puzzles. Our code and data are available at https://memkklogic.github.io.

cs.CL↗

arXiv2025

PREM: Privately Answering Statistical Queries with Relative Error

We introduce $\mathsf{PREM}$ (Private Relative Error Multiplicative weight update), a new framework for generating synthetic data that achieves a relative error guarantee for statistical queries under $(\varepsilon, δ)$ differential privacy (DP). Namely, for a domain ${\cal X}$, a family ${\cal F}$ of queries $f : {\cal X} \to \{0, 1\}$, and $ζ> 0$, our framework yields a mechanism that on input dataset $D \in {\cal X}^n$ outputs a synthetic dataset $\widehat{D} \in {\cal X}^n$ such that all statistical queries in ${\cal F}$ on $D$, namely $\sum_{x \in D} f(x)$ for $f \in {\cal F}$, are within a $1 \pm ζ$ multiplicative factor of the corresponding value on $\widehat{D}$ up to an additive error that is polynomial in $\log |{\cal F}|$, $\log |{\cal X}|$, $\log n$, $\log(1/δ)$, $1/\varepsilon$, and $1/ζ$. In contrast, any $(\varepsilon, δ)$-DP mechanism is known to require worst-case additive error that is polynomial in at least one of $n, |{\cal F}|$, or $|{\cal X}|$. We complement our algorithm with nearly matching lower bounds.

cs.LG↗

arXiv2025

Linear-Time User-Level DP-SCO via Robust Statistics

User-level differentially private stochastic convex optimization (DP-SCO) has garnered significant attention due to the paramount importance of safeguarding user privacy in modern large-scale machine learning applications. Current methods, such as those based on differentially private stochastic gradient descent (DP-SGD), often struggle with high noise accumulation and suboptimal utility due to the need to privatize every intermediate iterate. In this work, we introduce a novel linear-time algorithm that leverages robust statistics, specifically the median and trimmed mean, to overcome these challenges. Our approach uniquely bounds the sensitivity of all intermediate iterates of SGD with gradient estimation based on robust statistics, thereby significantly reducing the gradient estimation noise for privacy purposes and enhancing the privacy-utility trade-off. By sidestepping the repeated privatization required by previous methods, our algorithm not only achieves an improved theoretical privacy-utility trade-off but also maintains computational efficiency. We complement our algorithm with an information-theoretic lower bound, showing that our upper bound is optimal up to logarithmic factors and the dependence on $ε$. This work sets the stage for more robust and efficient privacy-preserving techniques in machine learning, with implications for future research and application in the field.

cs.LG↗

arXiv2025

Scaling Laws for Differentially Private Language Models

Scaling laws have emerged as important components of large language model (LLM) training as they can predict performance gains through scale, and provide guidance on important hyper-parameter choices that would otherwise be expensive. LLMs also rely on large, high-quality training datasets, like those sourced from (sometimes sensitive) user data. Training models on this sensitive user data requires careful privacy protections like differential privacy (DP). However, the dynamics of DP training are significantly different, and consequently their scaling laws are not yet fully understood. In this work, we establish scaling laws that accurately model the intricacies of DP LLM training, providing a complete picture of the compute-privacy-utility tradeoffs and the optimal training configurations in many settings.

cs.LG↗

arXiv2025

Discovery and Detailed Study of the M31 Classical Nova AT 2023tkw: Evidence for Internal Shocks

We present a detailed analysis of an extragalactic slow classical nova in M31 exhibiting multiple peaks in its light curve. Spectroscopic and photometric observations were used to investigate the underlying physical processes. Shock-induced heating events resulting in the expansion and contraction of the photosphere are likely responsible for the observed multiple peaks. Deviation of the observed spectrum at the peak from the models also suggests the presence of shocks. The successive peaks occurring at increasing intervals could be due to the series of internal shocks generated near or within the photosphere. Spectral modeling suggests a low-mass white dwarf accreting slowly from a companion star. The ejecta mass, estimated from spectral analysis, is $\sim 10^{-4}\mathrm{M_{\odot}}$, which is typical for a slow nova. We estimate the binary, by comparing the archival HST data and eruption properties with stellar and novae models, to comprise a 0.65 $\mathrm{M_{\odot}}$ primary white dwarf and a K III cool evolved secondary star.

astro-ph.HE↗

arXiv2025

Photon-photon coupling induced bound state in the continuum and transparency

This study presents the coherent and dissipative coupling realized in the hybrid photonic resonators that have been achieved via the constructive and destructive interference of the photonic resonator fields with the radiation of a common transmission line fed with microwave photons. In the dissipative coupling regime we have found the coexistence of a peculiar phenomenon bound state in the continuum (BIC) near the crossing of frequency of the uncoupled resonators by satisfying the Friedrich-Wintgen BICs condition. Again just by rotating one of the samples and with the dynamic adjustment of a parameter we have achieved coupling induced transparency between the photonic resonators. This transition from BIC in the absorption regime to transparency opens avenues for different sorts of plain or programmable oscillators, filters, quantum information processors, sensors etc.

quant-ph↗

arXiv2024

Aharonov-Bohm Interference in Even-Denominator Fractional Quantum Hall States

Position exchange of non-Abelian anyons affects the quantum state of their system in a topologically-protected way. Their expected manifestations in even-denominator fractional quantum Hall (FQH) systems offer the opportunity to directly study their unique statistical properties in interference experiments. In this work, we present the observation of coherent Aharonov-Bohm interference at two even-denominator states in high-mobility bilayer graphene-based van der Waals heterostructures by employing the Fabry-Pérot interferometry (FPI) technique. Operating the interferometer at a constant filling factor, we observe an oscillation period corresponding to two flux quanta inside the interference loop, $ΔΦ=2Φ_0$, at which the interference does not carry signatures of non-Abelian statistics. The absence of the expected periodicity of $ΔΦ=4Φ_0$ may indicate that the interfering quasiparticles carry the charge $e^* = \frac{1}{2}e$ or that interference of $e^* = \frac{1}{4}e$ quasiparticles is thermally smeared. Interestingly, at two hole-conjugate states, we also observe oscillation periods of half the expected value, indicating interference of $e^* = \frac{2}{3}e$ quasiparticles instead of $e^* = \frac{1}{3}e$. To probe statistical phase contributions, we operated the FPI with controlled deviations of the filling factor, thereby introducing fractional quasiparticles inside the interference loop. The resulting changes to the interference patterns at both half-filled states indicate that the additional bulk quasiparticles carry the fundamental charge $e^*=\frac{1}{4}e$, as expected for non-Abelian anyons.

cond-mat.mes-hall↗

arXiv2024

Fock-space perturbed relativistic coupled-cluster calculations of electric dipole polarizability and nuclear spin-dependent parity non-conservation in Cs

We implement the Fock-space perturbed relativistic coupled-cluster theory to compute the electric dipole polarizability of ground and low lying excited states, and nuclear spin-dependent parity violating (NSD-PNC) transition amplitudes in Cs. Moreover, to check the accuracy of the wavefunctions used in the calculations, we compute the excitation energies, E1 transition amplitudes and magnetic dipole hyperfine constants for ground and low lying excited states. To improve the accuracy of the computed properties, we have incorporated the corrections from the relativistic and QED effects in our calculations. The contributions from triple excitations are accounted perturbatively. Our results on excitation energies, E1 transition amplitudes and hyperfine constants are in good agreement with the available experimental results. Our polarizability results using FS-PRCC theory match well with the experimental values. The values of parity-violating transition amplitudes from our calculations are, in general, on the lower side of the previous values. From the detail analysis of electron correlations, we find that the corrections from the Breit interaction and QED effects are important to get accurate results of NSD-PNC amplitudes in Cs. The largest cumulative contribution from the Breit and QED corrections is found to be $\approx$ 3.2\% of the total value. The upper bound on the theoretical uncertainty in our calculated NSD-PNC amplitudes is estimated to be about 1\%.

physics.atom-ph↗

arXiv2024

How Unique is Whose Web Browser? The role of demographics in browser fingerprinting among US users

Browser fingerprinting can be used to identify and track users across the Web, even without cookies, by collecting attributes from users' devices to create unique "fingerprints". This technique and resulting privacy risks have been studied for over a decade. Yet further research is limited because prior studies used data not publicly available. Additionally, data in prior studies lacked user demographics. Here we provide a first-of-its-kind dataset to enable further research. It includes browser attributes with users' demographics and survey responses, collected with informed consent from 8,400 US study participants. We use this dataset to demonstrate how fingerprinting risks differ across demographic groups. For example, we find lower income users are more at risk, and find that as users' age increases, they are both more likely to be concerned about fingerprinting and at real risk of fingerprinting. Furthermore, we demonstrate an overlooked risk: user demographics, such as gender, age, income level and race, can be inferred from browser attributes commonly used for fingerprinting, and we identify which browser attributes most contribute to this risk. Our data collection process also conducted an experiment to study what impacts users' likelihood to share browser data for open research, in order to inform future data collection efforts, with responses from 12,461 total participants. Female participants were significantly less likely to share their browser data, as were participants who were shown the browser data we asked to collect. Overall, we show the important role of user demographics in the ongoing work that intends to assess fingerprinting risks and improve user privacy, with findings to inform future privacy enhancing browser developments. The dataset and data collection tool we provide can be used to further study research questions not addressed in this work.

cs.CY↗

arXiv2024

Scalable DP-SGD: Shuffling vs. Poisson Subsampling

We provide new lower bounds on the privacy guarantee of the multi-epoch Adaptive Batch Linear Queries (ABLQ) mechanism with shuffled batch sampling, demonstrating substantial gaps when compared to Poisson subsampling; prior analysis was limited to a single epoch. Since the privacy analysis of Differentially Private Stochastic Gradient Descent (DP-SGD) is obtained by analyzing the ABLQ mechanism, this brings into serious question the common practice of implementing shuffling-based DP-SGD, but reporting privacy parameters as if Poisson subsampling was used. To understand the impact of this gap on the utility of trained machine learning models, we introduce a practical approach to implement Poisson subsampling at scale using massively parallel computation, and efficiently train models with the same. We compare the utility of models trained with Poisson-subsampling-based DP-SGD, and the optimistic estimates of utility when using shuffling, via our new lower bounds on the privacy guarantee of ABLQ with shuffling.

cs.LG↗