arXiv subjects

Saurav Prakash

Publications and source records attributed to Saurav Prakash.

At least 19 recordsLinked to original sources

arXiv2026

Ni-O hybridization-driven electronic reconstruction across the superconducting dome in an infinite-layer nickelate

Superconductivity in infinite-layer nickelates has drawn wide interest as a cuprate analogue, yet how the electronic structure evolves with hole doping remains unsettled. Here we map the doping- and temperature-dependent unoccupied states of the La-based infinite-layer nickelate La1-xCaxNiO2 using O K-edge and Ni L-edge x-ray absorption spectroscopy. Superconductivity occurs for 0.18<=x<=0.27. Near x~0.20-0.23, low-energy spectral weight redistributes: Ni3d-dominated states decrease while O2p-hybridized states increase, indicating an orbital-selective crossover in Ni-O covalency. This crossover coincides with a sign reversal of the Hall coefficient and precedes the reduction of the superconducting critical temperature at higher doping. By directly linking transport anomalies and the superconducting dome to a measurable Ni-O orbital reorganization, our results provide a key step toward a unified, orbital-resolved phase diagram for infinite-layer nickelates and a practical route to engineer superconductivity via hybridization control.

cond-mat.supr-con↗

arXiv2026

Hole-doped superconductivity above 100 K in infinite-layer cuprate thin films

Since the discovery of superconductivity in (La,Ba)2CuO2 (Ref.~\cite{bednorz1986possible}), a broad family of structurally distinct cuprate superconductors has been proposed or engineered to elucidate the physics of high-temperature superconductivity~\cite{chu2015hole,plakida2010high}. Among them, the infinite-layer cuprate has the simplest structure, consisting only of the essential ingredients for superconductivity: CuO$_2$ square planes separated by spacer ions~\cite{siegrist1988parent}. Despite being proposed nearly 40 years ago, the hole-doped superconductivity via chemical substitution in this compound has not yet been achieved, a fundamental open question in the field. Here, we report the observation of superconductivity in the hole-doped infinite-layer cuprate thin film. Measurements of resistivity and magnetic-field response in Sr1-xRbxCuO2 single-crystal thin films show superconducting transitions with a high onset temperature of 100 K. Hole doping is achieved via the synergistic effect of rubidium substitution and apical oxygen incorporation, as evidenced by structural analysis and transport measurements. As the parent structure of the cuprate family~\cite{chu2015hole}, hole-doped infinite-layer cuprate provides a unique platform for revisiting key puzzles in cuprate superconductors~\cite{keimer2015quantum,tsuei2000pairing,armitage2010progress,dagotto1994correlated}, including strange metal~\cite{proust2019remarkable,taillefer2010scattering} and electron-hole symmetry~\cite{tohyama2004asymmetry,segawa2010zero,lee2014asymmetry}, while bridging to cuprate-nickelate symmetry~\cite{li2019superconductivity,zeng2022superconductivity,chow2025bulk,lechermann2020late}.

cond-mat.supr-con↗

arXiv2026

Federated Learning of Binary Neural Networks: Enabling Low-Cost Inference

Federated Learning (FL) preserves privacy by distributing training across devices. However, using DNNs is computationally intensive at the low-powered edge during inference. Edge deployment demands models that simultaneously optimize memory footprint and computational efficiency, a dilemma where conventional DNNs fail by exceeding resource limits. Traditional post-training binarization reduces model size but suffers from severe accuracy loss due to quantization errors. To address these challenges, we propose FedBNN, a rotation-aware binary neural network framework that learns binary representations directly during local training. By encoding each weight as a single bit $\{+1, -1\}$ instead of a $32$-bit float, FedBNN shrinks the model footprint, significantly reducing runtime (during inference) FLOPs and memory requirements in comparison to federated methods using real models. Evaluations across multiple benchmark datasets demonstrate that FedBNN significantly reduces resource consumption while performing similarly to existing federated methods using real-valued models.

cs.LG↗

arXiv2026

Distributed Perceptron under Bounded Staleness, Partial Participation, and Noisy Communication

We study a semi-asynchronous client-server perceptron trained via iterative parameter mixing (IPM-style averaging): clients run local perceptron updates and a server forms a global model by aggregating the updates that arrive in each communication round. The setting captures three system effects in federated and distributed deployments: (i) stale updates due to delayed model delivery and delayed application of client computations (two-sided version lag), (ii) partial participation (intermittent client availability), and (iii) imperfect communication on both downlink and uplink, modeled as effective zero-mean additive noise with bounded second moment. We introduce a server-side aggregation rule called staleness-bucket aggregation with padding that deterministically enforces a prescribed staleness profile over update ages without assuming any stochastic model for delays or participation. Under margin separability and bounded data radius, we prove a finite-horizon expected bound on the cumulative weighted number of perceptron mistakes over a given number of server rounds: the impact of delay appears only through the mean enforced staleness, whereas communication noise contributes an additional term that grows on the order of the square root of the horizon with the total noise energy. In the noiseless case, we show how a finite expected mistake budget yields an explicit finite-round stabilization bound under a mild fresh-participation condition.

cs.LG↗

arXiv2025

Net Magnetization and Inhomogeneous Magnetic Order in a High-Tc Nickelate Superconductor

High-temperature and high-magnetic-field-induced re-entrant superconductivity has been discovered in the infinite-layer nickelate $\mathrm{Sm_{1-x-y} Eu_x Ca_y Ni O_2}$ (SECNO). Infinite-layer nickelates are the closest known analogues of high-$\mathrm{T}_c$ cuprate superconductors, yet they host distinct magnetic ground states. Using low-energy muon spin relaxation and polarized neutron reflectometry, we reveal the magnetic order in SECNO. We find that magnetic freezing occurs at a higher-temperature than in other nickelate compounds, and that a substantial net magnetization of 55 $\,\mathrm{kA}\,\mathrm{m}^{-1}$ $\pm10 \,\mathrm{kA}\,\mathrm{m}^{-1}$ emerges and remains largely unchanged across the superconducting transition. The magnetism in SECNO is disordered and nonuniform.

cond-mat.supr-con↗

arXiv2025

SWAN: Sparse Winnowed Attention for Reduced Inference Memory via Decompression-Free KV-Cache Compression

Large Language Models (LLMs) face a significant bottleneck during autoregressive inference due to the massive memory footprint of the Key-Value (KV) cache. Existing compression techniques like token eviction, quantization, or other low-rank methods often risk information loss, have fixed limits, or introduce significant computational overhead from explicit decompression steps. In this work, we introduce SWAN, a novel, fine-tuning-free framework that eliminates this overhead. Our method uses an offline orthogonal matrix to rotate and prune the KV-cache, which is then used directly in the attention computation without any reconstruction. Our extensive experiments demonstrate that SWAN, augmented with a small dense buffer, offers a robust trade-off, maintaining performance close to the uncompressed baseline even at aggressive 50-60% memory savings per-token on KV-cache. A key advantage is its runtime-tunable compression level, allowing operators to dynamically adjust the memory footprint, a flexibility absent in methods requiring fixed offline configurations. This combination of a decompression-free design, high performance under compression, and adaptability makes SWAN a practical and efficient solution for serving LLMs with long contexts.

cs.LG↗

arXiv2025

AQUA: Attention via QUery mAgnitudes for Memory and Compute Efficient Inference in LLMs

The quadratic complexity of the attention mechanism remains a fundamental barrier to scaling Large Language Models (LLMs) to longer contexts, creating a critical bottleneck in both computation and memory. To address this, we introduce AQUA (Attention via QUery mAgnitudes) a novel and versatile approximation strategy that significantly reduces the cost of attention with a graceful performance trade-off. Our method operates in two phases: an efficient offline step where we compute a universal, language agnostic projection matrix via SVD on a calibration dataset, and an online inference step where we project query and key vectors and dynamically select a sparse subset of dimensions based on the query's magnitude. We provide a formal theoretical analysis of AQUA, establishing the break-even point at which it becomes more computationally efficient than standard attention. Our empirical evaluations on state-of-the-art models like Llama-3.1-8B demonstrate that a 25% reduction in the attention dot-product computation can be achieved with a statistically insignificant impact on performance across a wide range of benchmarks. We further showcase the versatility of AQUA by demonstrating its ability to synergistically accelerate existing token eviction methods like H2O and to directly reduce KV-cache memory size. By offering a controllable knob to balance efficiency and accuracy, AQUA provides a practical and powerful tool for making large-scale LLM inference more accessible and sustainable.

cs.LG↗

arXiv2025

High-field-stabilized reentrant superconductivity in infinite-layer nickelate thin films

Magnetic fields typically suppress superconductivity through Pauli and orbital limiting effects. However, there are rare instances of magnetic-field-induced superconductivity, as seen in Chevrel phase compounds [1], organic conductors [2], uranium-based heavy-fermion systems [3, 4], and moire graphene [5], though these materials possess inherently low superconducting transition temperatures (Tc). Here, we demonstrate high field-stabilized superconductivity in a class of materials with a significantly higher Tc (up to 40 K): the infinite-layer nickelates [6]. Both low-field and high-field superconducting states can be plausibly explained by a compensation mechanism akin to the Jaccarino-Peter effect. These findings demonstrate the possibility of achieving substantially enhanced upper critical fields in high-temperature superconductors.

cond-mat.supr-con↗

arXiv2025

Federated Nonlinear System Identification

We consider federated learning of linearly-parameterized nonlinear systems. We establish theoretical guarantees on the effectiveness of federated nonlinear system identification compared to centralized approaches, demonstrating that the convergence rate improves as the number of clients increases. Although the convergence rates in the linear and nonlinear cases differ only by a constant, this constant depends on the feature map $\phi$, which can be carefully chosen in the nonlinear setting to increase excitation and improve performance. We experimentally validate our theory in physical settings where client devices are driven by i.i.d. control inputs and control policies exhibiting i.i.d. random perturbations, ensuring non-active exploration. Experiments use trajectories from nonlinear dynamical systems characterized by real-analytic feature functions, including polynomial and trigonometric components, representative of physical systems including pendulum and quadrotor dynamics. We analyze the convergence behavior of the proposed method under varying noise levels and data distributions. Results show that federated learning consistently improves convergence of any individual client as the number of participating clients increases.

cs.LG↗

arXiv2025

Persistent paramagnons in high-temperature infinite-layer nickelate superconductors

The recent discovery of high-temperature superconductivity in hole-doped SmNiO$_2$, exhibiting the record-high transition temperature $T_c$ among infinite-layer (IL) nickelates, has opened a new avenue for exploring design principles of superconductivity. Experimentally determining the electronic structure and magnetic interactions in this new system is crucial to elucidating the mechanism behind the enhanced superconductivity. Here, we report a Ni $L$-edge resonant inelastic x-ray scattering (RIXS) study of superconducting Sm-based IL nickelate thin films Sm$_{1-x-y-z}$Eu$_x$Ca$_y$Sr$_z$NiO$_2$ (SECS). Dispersive paramagnonic excitations are observed in both optimally and overdoped SECS samples, supporting a spin-fluctuation-mediated pairing scenario. However, despite the two-fold enhancement of $T_c$ in the Sm-based nickelates compared to their Pr-based counterparts, the effective exchange coupling strength is reduced by approximately $20\%$. This behavior contrasts with hole-doped cuprates, where magnetic interactions correlate positively with $T_c$, highlighting essential differences in their superconducting mechanisms.

cond-mat.supr-con↗

arXiv2024

Room Temperature Strong Orbital Moments in Perpendicularly Magnetized Magnetic Insulator

The balance between the orbital and spin magnetic moments in a magnetic system is the heart of many intriguing phenomena. Here, we show experimental evidence of a large orbital moment, which competes with its spin counterpart in a ferrimagnetic insulator thulium iron garnet, Tm3Fe5O12. Leveraging element-specific X-ray magnetic circular dichroism (XMCD), we establish that the dominant contribution to the orbital moment originates from 4f orbitals of Tm. Besides the large Tm orbital moment, intriguingly, our results also reveal a smaller but evident non-zero XMCD signal in the O K edge, suggesting additional spin-orbit coupling and exchange interactions with the nearest neighbour Fe atoms. The unquenched orbital moment is primarily responsible for a significant reduction in g-factor, typically 2 in transition metals, as determined independently using ferromagnetic resonance spectroscopy. Our findings reveal a non-linear reduction in the g-factor from 1.7 at 300 K to 1.56 at 200 K in Tm3Fe5O12 thin films. These results provide critical insights into the role of the f orbitals in long-range magnetic order and stimulate further exploration in orbitronics.

cond-mat.mtrl-sci↗

arXiv2024

Embracing Federated Learning: Enabling Weak Client Participation via Partial Model Training

In Federated Learning (FL), clients may have weak devices that cannot train the full model or even hold it in their memory space. To implement large-scale FL applications, thus, it is crucial to develop a distributed learning method that enables the participation of such weak clients. We propose EmbracingFL, a general FL framework that allows all available clients to join the distributed training regardless of their system resource capacity. The framework is built upon a novel form of partial model training method in which each client trains as many consecutive output-side layers as its system resources allow. Our study demonstrates that EmbracingFL encourages each layer to have similar data representations across clients, improving FL efficiency. The proposed partial model training method guarantees convergence to a neighbor of stationary points for non-convex and smooth problems. We evaluate the efficacy of EmbracingFL under a variety of settings with a mixed number of strong, moderate (~40% memory), and weak (~15% memory) clients, datasets (CIFAR-10, FEMNIST, and IMDB), and models (ResNet20, CNN, and LSTM). Our empirical study shows that EmbracingFL consistently achieves high accuracy as like all clients are strong, outperforming the state-of-the-art width reduction methods (i.e. HeteroFL and FjORD).

cs.LG↗

arXiv2024

All Rivers Run to the Sea: Private Learning with Asymmetric Flows

Data privacy is of great concern in cloud machine-learning service platforms, when sensitive data are exposed to service providers. While private computing environments (e.g., secure enclaves), and cryptographic approaches (e.g., homomorphic encryption) provide strong privacy protection, their computing performance still falls short compared to cloud GPUs. To achieve privacy protection with high computing performance, we propose Delta, a new private training and inference framework, with comparable model performance as non-private centralized training. Delta features two asymmetric data flows: the main information-sensitive flow and the residual flow. The main part flows into a small model while the residuals are offloaded to a large model. Specifically, Delta embeds the information-sensitive representations into a low-dimensional space while pushing the information-insensitive part into high-dimension residuals. To ensure privacy protection, the low-dimensional information-sensitive part is secured and fed to a small model in a private environment. On the other hand, the residual part is sent to fast cloud GPUs, and processed by a large model. To further enhance privacy and reduce the communication cost, Delta applies a random binary quantization technique along with a DP-based technique to the residuals before sharing them with the public platform. We theoretically show that Delta guarantees differential privacy in the public environment and greatly reduces the complexity in the private environment. We conduct empirical analyses on CIFAR-10, CIFAR-100 and ImageNet datasets and ResNet-18 and ResNet-34, showing that Delta achieves strong privacy protection, fast training, and inference without significantly compromising the model utility.

cs.CR↗

arXiv2024

ATP: Enabling Fast LLM Serving via Attention on Top Principal Keys

We propose a new attention mechanism with linear complexity, ATP, that fixates \textbf{A}ttention on \textbf{T}op \textbf{P}rincipal keys, rather than on each individual token. Particularly, ATP is driven by an important observation that input sequences are typically low-rank, i.e., input sequences can be represented by a few principal bases. Therefore, instead of directly iterating over all the input tokens, ATP transforms inputs into an orthogonal space and computes attention only on the top principal bases (keys). Owing to the observed low-rank structure in input sequences, ATP is able to capture semantic relationships in input sequences with a few principal keys. Furthermore, the attention complexity is reduced from \emph{quadratic} to \emph{linear} without incurring a noticeable performance drop. ATP further reduces complexity for other linear layers with low-rank inputs, leading to more speedup compared to prior works that solely target the attention module. Our evaluations on various models (e.g., BERT and Llama) demonstrate that ATP achieves comparable accuracy with much lower computation and memory complexity than the standard attention mechanism. In particular, ATP barely loses accuracy with only $1/2$ principal keys, and only incurs around $2\%$ accuracy drops with $1/4$ principal keys.

cs.LG↗

arXiv2024

On the origin of topotactic reduction effect for superconductivity in infinite-layer nickelates

Topotactic reduction utilizing metal hydrides as reagents emerges as an effective approach to achieve exceptionally low oxidization states of metal ions and unconventional coordination networks. This method opens avenues to the development of entirely new functional materials, with one notable example being the infinite-layer nickelate superconductors. However, the reduction effect on the atomic reconstruction and electronic structures -- crucial for superconductivity -- remains largely unresolved. We design two sets of control Nd$_{0.8}$Sr$_{0.2}$NiO$_2$ thin films and implement secondary ion mass spectroscopy to highlight the absence of reduction-induced hydrogen intercalation. X-ray absorption spectroscopy shows a significant linear dichroism with dominant Ni 3d$_{x2{-}y2}$ orbitals on superconducting samples, indicating a Ni single-band nature of infinite-layer nickelates. Consistent with the superconducting $T_c$, the Ni 3d orbitals asymmetry manifests a dome-like reduction duration dependence. Our results unveil the critical role of reduction in modulating the Ni-3d orbital polarization and its impact on the superconducting properties.

cond-mat.supr-con↗

arXiv2024

Federated Classification in Hyperbolic Spaces via Secure Aggregation of Convex Hulls

Hierarchical and tree-like data sets arise in many applications, including language processing, graph data mining, phylogeny and genomics. It is known that tree-like data cannot be embedded into Euclidean spaces of finite dimension with small distortion. This problem can be mitigated through the use of hyperbolic spaces. When such data also has to be processed in a distributed and privatized setting, it becomes necessary to work with new federated learning methods tailored to hyperbolic spaces. As an initial step towards the development of the field of federated learning in hyperbolic spaces, we propose the first known approach to federated classification in hyperbolic spaces. Our contributions are as follows. First, we develop distributed versions of convex SVM classifiers for Poincaré discs. In this setting, the information conveyed from clients to the global classifier are convex hulls of clusters present in individual client data. Second, to avoid label switching issues, we introduce a number-theoretic approach for label recovery based on the so-called integer $B_h$ sequences. Third, we compute the complexity of the convex hulls in hyperbolic spaces to assess the extent of data leakage; at the same time, in order to limit communication cost for the hulls, we propose a new quantization method for the Poincaré disc coupled with Reed-Solomon-like encoding. Fourth, at the server level, we introduce a new approach for aggregating convex hulls of the clients based on balanced graph partitioning. We test our method on a collection of diverse data sets, including hierarchical single-cell RNA-seq data from different patients distributed across different repositories that have stringent privacy constraints. The classification accuracy of our method is up to $\sim 11\%$ better than its Euclidean counterpart, demonstrating the importance of privacy-preserving learning in hyperbolic spaces.

cs.LG↗

arXiv2023

Lottery Aware Sparsity Hunting: Enabling Federated Learning on Resource-Limited Edge

Edge devices can benefit remarkably from federated learning due to their distributed nature; however, their limited resource and computing power poses limitations in deployment. A possible solution to this problem is to utilize off-the-shelf sparse learning algorithms at the clients to meet their resource budget. However, such naive deployment in the clients causes significant accuracy degradation, especially for highly resource-constrained clients. In particular, our investigations reveal that the lack of consensus in the sparsity masks among the clients may potentially slow down the convergence of the global model and cause a substantial accuracy drop. With these observations, we present \textit{federated lottery aware sparsity hunting} (FLASH), a unified sparse learning framework for training a sparse sub-model that maintains the performance under ultra-low parameter density while yielding proportional communication benefits. Moreover, given that different clients may have different resource budgets, we present \textit{hetero-FLASH} where clients can take different density budgets based on their device resource limitations instead of supporting only one target parameter density. Experimental analysis on diverse models and datasets shows the superiority of FLASH in closing the gap with an unpruned baseline while yielding up to $\mathord{\sim}10.1\%$ improved accuracy with $\mathord{\sim}10.26\times$ fewer communication, compared to existing alternatives, at similar hyperparameter settings. Code is available at \url{https://github.com/SaraBabakN/flash_fl}.

cs.LG↗

arXiv2023

Federated Learning of Large Models at the Edge via Principal Sub-Model Training

Federated Learning (FL) is emerging as a popular, promising decentralized learning framework that enables collaborative training among clients, with no need to share private data between them or to a centralized server. However, considering many edge clients do not have sufficient computing, memory, or communication capabilities, federated learning of large models still faces significant bottlenecks. To keep such weak but crucial clients in the loop, prior works either consider a heterogeneous-client setting where clients train models with different sizes; or offload training to the server. However, the heterogeneous-client setting requires some clients to train full model, which is not aligned with the resource-constrained setting; while the latter ones break privacy promises in FL when sharing intermediate representations or labels with the server. To overcome these limitations, in this work, we formulate a realistic, but much less explored, cross-device FL setting in which no client can train a full large model nor is willing to share any intermediate information with the remote server. Under such a formulation, we develop a principal sub-model (PriSM) training methodology to collaboratively train a full large model, while assigning each client a small sub-model that is a probabilistic low-rank approximation to the full server model. When creating sub-models, PriSM first performs a principal kernel analysis in the orthogonal kernel space to obtain importance of each kernel. Then, PriSM adopts a novel importance-aware sampling process to select a subset of kernels (i.e., a kernel with high importance is assigned with a higher sampling probability). This sampling process ensures each sub-model is still a low-rank approximation to the full model, while all sub-models together achieve nearly full coverage on the principal kernels.

cs.LG↗