A5089711718- Quantum Computing Algorithms and Architecture
- Quantum Information and Cryptography
- Quantum many-body systems
- Gamma-ray bursts and supernovae
- Quantum Mechanics and Applications
- Galaxies: Formation, Evolution, Phenomena
- Spectroscopy and Quantum Chemical Studies
- Generative Adversarial Networks and Image Synthesis
- Cosmology and Gravitation Theories
- Astronomy and Astrophysical Research
- Stellar, planetary, and galactic studies
- Music and Audio Processing
- Quantum and electron transport phenomena
- Stochastic Gradient Optimization Techniques
- AI in cancer detection
- Particle Detector Development and Performance
- Particle accelerators and beam dynamics
- Cold Atom Physics and Bose-Einstein Condensates
- Quantum chaos and dynamical systems
- Cloud Computing and Resource Management
- Radiation Effects in Electronics
- Adaptive optics and wavefront sensing
- Anomaly Detection Techniques and Applications
- Pulsars and Gravitational Waves Research
- Model Reduction and Neural Networks
Harvard University
2022-2025
Harvard University Press
2024-2025
University of California, Berkeley
2021-2024
Los Alamos National Laboratory
2022
We have conducted a search for new strong gravitational lensing systems in the Dark Energy Spectroscopic Instrument Legacy Imaging Surveys' Data Release 8. use deep residual neural networks, building on previous work presented Huang et al. (2020). These surveys together cover approximately one third of sky visible from northern hemisphere, reaching z band AB magnitude ~22.5. compile training sample that consists known as well non-lenses Surveys and Survey. After applying our trained networks...
The newly constructed pseudomagic ensembles possess low nonstabilizerness but are computationally indistinguishable from those with high nonstabilizerness, having implications for quantum scramblers, cryptography, and magic state distillation.
Abstract We study the problem of learning parameters for Hamiltonian a quantum many-body system, given limited access to system. In this work, we build upon recent approaches via derivative estimation. propose protocol that improves scaling dependence prior works, particularly with respect relating structure (e.g., its locality k ). Furthermore, by deriving exact bounds on performance our protocol, are able provide precise numerical prescription theoretically optimal settings hyperparameters...
Extracting information efficiently from quantum systems is crucial for processing. Classical shadows enable predicting many properties of arbitrary states using few measurements. While random single-qubit measurements are experimentally friendly and suitable learning low-weight Pauli observables, they perform poorly nonlocal observables. Introducing a shallow circuit before improves sample efficiency high-weight observables low-rank properties. However, in practice, these circuits can be...
We present GIGA-Lens: a gradient-informed, GPU-accelerated Bayesian framework for modeling strong gravitational lensing systems, implemented in TensorFlow and JAX. The three components, optimization using multi-start gradient descent, posterior covariance estimation with variational inference, sampling via Hamiltonian Monte Carlo, all take advantage of information through automatic differentiation massive parallelization on graphics processing units (GPUs). test our pipeline large set...
Extracting information efficiently from quantum systems is a major component of processing tasks. Randomized measurements, or classical shadows, enable predicting many properties arbitrary states using few measurements. While random single qubit measurements are experimentally friendly and suitable for learning low-weight Pauli observables, they perform poorly nonlocal observables. Prepending shallow circuit before maintains this experimental friendliness, but also has favorable sample...
High-fidelity quantum entanglement enables key networking capabilities such as secure communication and distributed computing, but distributing entangled states through optical fibers is limited by noise loss. Entanglement distillation protocols address this problem extracting high-fidelity Bell pairs from multiple noisy ones. The primary objective minimizing the resource overhead: number of input needed to distill each output pair. While achieving optimal overhead are known in theory, they...
Entanglement serves as a foundational pillar in quantum information theory, delineating the boundary between what is classical and quantum. The common assumption that higher degree of entanglement corresponds to greater “quantumness.” However, this folk belief challenged by fact classically simulable operations, such Clifford circuits, can create highly entangled states. simulability these states raises question: What are differences “low-magic” “high-magic” entanglement? To understand...
We have conducted a search for strong gravitational lensing systems in the Dark Energy Spectroscopic Instrument (DESI) Legacy Imaging Surveys Data Release 9. This is third paper series (following Huang et al. 2020; 2021, Paper I & II, respectively). These surveys together cover $\sim$ 19,000 deg$^2$ visible from northern hemisphere, reaching z-band AB magnitude of 22.5. use deep residual neural network, trained on compilation known and candidates as well non-lenses same footprint. After...
This work uncovers a fundamental connection between doped stabilizer states, concept from quantum information theory, and the structure of eigenstates in perturbed many-body systems. We prove that for Hamiltonians consisting sum commuting Pauli operators (i.e., Hamiltonians) perturbation composed limited number arbitrary terms, can be represented as states with small nullity. result enables application techniques to broad class systems, even highly entangled regimes. Building on this, we...
We introduce a quantum information theory-inspired method to improve the characterization of many-body Hamiltonians on near-term devices. design new class similarity transformations that, when applied as preprocessing step, can substantially simplify Hamiltonian for subsequent analysis hardware. By design, these be identified and efficiently using purely classical resources. In practice, allow us shorten requisite physical circuit-depths, overcoming constraints imposed by imperfect...
Abstract We have conducted a search for strong gravitational lensing systems in the Dark Energy Spectroscopic Instrument (DESI) Legacy Imaging Surveys Data Release 9. This is third paper series. These surveys together cover ∼19,000 deg 2 visible from Northern Hemisphere, reaching z -band AB magnitude of ∼22.5. use deep residual neural network, trained on compilation known and high-grade candidates as well nonlenses same footprint. After applying our network to survey data, we visually...
We study the problem of learning parameters for Hamiltonian a quantum many-body system, given limited access to system. In this work, we build upon recent approaches via derivative estimation. propose protocol that improves scaling dependence prior works, particularly with respect relating structure (e.g., its locality $k$). Furthermore, by deriving exact bounds on performance our protocol, are able provide precise numerical prescription theoretically optimal settings hyperparameters in such...
Variational Quantum Algorithms (VQAs) are a promising approach for practical applications like chemistry and materials science on near-term quantum computers as they typically reduce resource requirements. However, in order to implement VQAs, an efficient classical optimization strategy is required. Here we present new stochastic gradient descent method using adaptive number of shots at each step, called the global Coupled Adaptive Number Shots (gCANS) method, which improves prior art both...
Notions of nonstabilizerness, or "magic", quantify how non-classical quantum states are in a precise sense: exhibiting low nonstabilizerness preclude advantage. We introduce 'pseudomagic' ensembles that, despite computationally indistinguishable from those with high nonstabilizerness. Previously, such computational indistinguishability has been studied respect to entanglement, introducing the concept pseudoentanglement. However, we demonstrate that pseudomagic neither follows...
Entanglement serves as a foundational pillar in quantum information theory, delineating the boundary between what is classical and quantum. The common assumption that higher entanglement corresponds to greater degree of 'quantumness'. However, this folk belief challenged by fact classically simulable operations, such Clifford circuits, can create highly entangled states. simulability these states raises question: are differences 'low-magic' entanglement, 'high-magic' entanglement? We answer...
Motzkin spin-chains, which include 'colorless' (integer spin $s=1$) and 'colorful' ($s \geq 2$) variants, are one-dimensional (1D) local integer models notable for their lack of a conformal field theory (CFT) description low-energy physics, despite being gapless. The colorful variants particularly unusual, as they exhibit power-law violation the area-law entanglement entropy (as $\sqrt{n}$ in system size $n$), rather than logarithmic seen CFT. In this work, we analytically discover several...
Quantum chaos is a quantum many-body phenomenon that associated with number of intricate properties, such as level repulsion in energy spectra or distinct scalings out-of-time ordered correlation functions. In this work, we introduce novel class "pseudochaotic" Hamiltonians fundamentally challenges the conventional understanding and its relationship to computational complexity. Our ensemble computationally indistinguishable from Gaussian unitary (GUE) strongly-interacting Hamiltonians,...
Abstract We study the problem of learning parameters for Hamiltonian a quantum many-body system, given limited access to system. In this work, we build upon recent approaches via derivative estimation. propose protocol that improves scaling dependence prior works, particularly with respect relating structure (e.g., its locality k). Furthermore, by deriving exact bounds on performance our protocol, are able provide precise numerical prescription theoretically optimal settings hyperparameters...
In this work, we introduce a novel deep learning architecture, Variable Length Embeddings (VLEs), an autoregressive model that can produce latent representation composed of arbitrary number tokens. As proof concept, demonstrate the capabilities VLEs on tasks involve reconstruction and image decomposition. We evaluate our experiments mix iNaturalist ImageNet datasets find achieve comparable results to state art VAE, using less than tenth parameters.
We introduce a quantum information theory-inspired method to improve the characterization of many-body Hamiltonians on near-term devices. design new class similarity transformations that, when applied as preprocessing step, can substantially simplify Hamiltonian for subsequent analysis hardware. By design, these be identified and efficiently using purely classical resources. In practice, allow us shorten requisite physical circuit-depths, overcoming constraints imposed by imperfect...
We introduce a novel approach to video modeling that leverages controlled differential equations (CDEs) address key challenges in tasks, notably interpolation and mask propagation. apply CDEs at varying resolutions leading continuous-time U-Net architecture. Unlike traditional methods, our does not require explicit optical flow learning, instead makes use of the inherent features produce highly expressive model. demonstrate competitive performance against state-of-the-art models for...