arXiv subjects

Jiyong Yu

Publications and source records attributed to Jiyong Yu.

5 recordsLinked to original sources

arXiv2026

Sympathetic Cooling in Trapped Ions with Spectral Selectivity via the Zeeman Shift

High-fidelity quantum logic operations in trapped ions often require the ions' collective motion to be cooled to near the ground state. Since cooling the ions' motion typically involves dissipative processes such as spontaneous photon scattering, sympathetic cooling is used on select coolant ions between gate sequences to cool the ion chain without affecting the data qubits. Common implementations for coolant ions include different atomic species, different isotopes of the same species or individually addressable ions. Each of these approaches have challenges associated with them, which include increased hardware complexity, reduced efficiency of radial mode cooling and re-ordering events which add additional experimental overhead. We demonstrate a sympathetic cooling scheme leveraging internal metastable atomic levels accessible via a narrow quadrupole transition, utilizing the natural Zeeman shift and individually addressed Raman transitions, to achieve isolation of the non-coolant or ``data ions" from coolant ions. We demonstrate modest decoherence of the data ions due to cooling, while preserving the coherence requirements for high-fidelity gate operations.

quant-ph↗

arXiv2026

Design and Characterization of Compact Acousto-Optic-Deflector Individual Addressing System for Trapped-Ion Quantum Computing

We present a compact design for a beam-steering system based on acousto-optic-deflectors (AODs) used as an individual addressing system for trapped-ion quantum computing. The design targets to minimize the optomechanical degrees of freedom and the optical beam paths to improve optical stability, and we successfully implemented a solution with a compact footprint of less than 1 square foot. The system characterization results show that we achieve clean Gaussian beams at 355nm wavelength with a beam steering range of $\sim$50 times the beam diameter, and an intensity crosstalk of $< 9 \times 10^{-4}$ at all neighboring ions in a five-ion chain. Based on these capabilities, we experimentally demonstrate individual addressing of a 30-ion chain. We estimate the beam switching time of the AOD to be $\sim$240 ns. The compact system design is expected to provide high optical stability, providing the potential for high-fidelity trapped-ion quantum computing with long ion chains.

quant-ph↗

arXiv2025

Tailored Quantum Device Calibration with Statistical Model Checking

Quantum devices require precisely calibrated analog signals, a process that is complex and time-consuming. Many calibration strategies exist, and all require careful analysis and tuning to optimize system availability. To enable rigorous statistical evaluation of quantum calibration procedures, we leverage statistical model checking (SMC), a technique used in fields that require statistical guarantees. SMC allows for probabilistic evaluation of properties of interest, such as a certain parameter's time to failure. We extend the SMC for Processor Analysis (SPA) framework, which uses SMC for evaluation of classical systems, to create SPA for Quantum calibration (SPAQ) enabling simplified tuning and analysis of quantum system calibration. We focus on a directed acyclic graph-based calibration optimization scheme and demonstrate how to craft properties of interest for its analysis. We show how to use SPAQ to find lower bounds of time to failure information, hidden node dependencies, and parameter threshold values and use that information to improve simulated quantum system availability through calibration scheme adjustments.

quant-ph↗

arXiv2020

Speculative Interference Attacks: Breaking Invisible Speculation Schemes

Recent security vulnerabilities that target speculative execution (e.g., Spectre) present a significant challenge for processor design. The highly publicized vulnerability uses speculative execution to learn victim secrets by changing cache state. As a result, recent computer architecture research has focused on invisible speculation mechanisms that attempt to block changes in cache state due to speculative execution. Prior work has shown significant success in preventing Spectre and other vulnerabilities at modest performance costs. In this paper, we introduce speculative interference attacks, which show that prior invisible speculation mechanisms do not fully block these speculation-based attacks. We make two key observations. First, misspeculated younger instructions can change the timing of older, bound-to-retire instructions, including memory operations. Second, changing the timing of a memory operation can change the order of that memory operation relative to other memory operations, resulting in persistent changes to the cache state. Using these observations, we demonstrate (among other attack variants) that secret information accessed by mis-speculated instructions can change the order of bound-to-retire loads. Load timing changes can therefore leave secret-dependent changes in the cache, even in the presence of invisible speculation mechanisms. We show that this problem is not easy to fix: Speculative interference converts timing changes to persistent cache-state changes, and timing is typically ignored by many cache-based defenses. We develop a framework to understand the attack and demonstrate concrete proof-of-concept attacks against invisible speculation mechanisms. We provide security definitions sufficient to block speculative interference attacks; describe a simple defense mechanism with a high performance cost; and discuss how future research can improve its performance.

cs.AR↗

arXiv2018

UCNN: Exploiting Computational Reuse in Deep Neural Networks via Weight Repetition

Convolutional Neural Networks (CNNs) have begun to permeate all corners of electronic society (from voice recognition to scene generation) due to their high accuracy and machine efficiency per operation. At their core, CNN computations are made up of multi-dimensional dot products between weight and input vectors. This paper studies how weight repetition ---when the same weight occurs multiple times in or across weight vectors--- can be exploited to save energy and improve performance during CNN inference. This generalizes a popular line of work to improve efficiency from CNN weight sparsity, as reducing computation due to repeated zero weights is a special case of reducing computation due to repeated weights. To exploit weight repetition, this paper proposes a new CNN accelerator called the Unique Weight CNN Accelerator (UCNN). UCNN uses weight repetition to reuse CNN sub-computations (e.g., dot products) and to reduce CNN model size when stored in off-chip DRAM ---both of which save energy. UCNN further improves performance by exploiting sparsity in weights. We evaluate UCNN with an accelerator-level cycle and energy model and with an RTL implementation of the UCNN processing element. On three contemporary CNNs, UCNN improves throughput-normalized energy consumption by 1.2x - 4x, relative to a similarly provisioned baseline accelerator that uses Eyeriss-style sparsity optimizations. At the same time, the UCNN processing element adds only 17-24% area overhead relative to the same baseline.

cs.NE↗