arXiv subjects

Yulun Wang

Publications and source records attributed to Yulun Wang.

10 recordsLinked to original sources

arXiv2026

Heterogeneous architectures enable a 138x reduction in physical qubit requirements for fault-tolerant quantum computing under detailed accounting

Quantum computer hardware is predicted to scale over hundreds of thousands of qubits coming online in the next decade. Despite significant theoretical and experimental QEC progress, quantum computer architecture has suffered a significant gap, with bottom-up physical-device-driven challenges largely disconnected from top-down QEC-code-driven considerations. In this work, we unify these two views, presenting a complete heterogeneous quantum computing architecture incorporating task-specific hardware selection and QEC encoding, and agnostic to code selection or physical qubit parameters. Our approach further enables special-purpose processing modules, and includes a full microarchitecture for fault-tolerant implementation of interfaces between quantum processing units and quantum memories. Using this architecture and a new fully featured compiler functioning across subsystems at the scale of $1,000$ logical qubits, we schedule and orchestrate a variety of algorithms down to hardware-specific instructions; a detailed accounting of all operations reveals up to 551x reduction in algorithmic logical error and up to 138x reduction in physical-qubit overhead compared to a monolithic baseline architecture. We then consider the factorization of 2048-bit RSA-integers; using an experimentally demonstrated grid-coupling topology, factoring RSA-2048 requires 381k physical qubits and 9.2 days, which can be reduced to 4.9 days via addition of an algorithm-specific accelerator for the Adder subroutine (requiring 439k qubits). Finally, assuming hypothetical long-range coupling, implementing quantum memory using qLDPC codes reduces the resources required for factoring to just 190k qubits and under 10 days. These results and the tooling we have built indicate that heterogeneous quantum-computer architectures can deliver significant, verifiable benefits on realistic hardware.

quant-ph↗

arXiv2025

No need to calibrate: characterization and compilation for high-fidelity circuit execution using imperfect gates

We propose and validate on real quantum computing hardware a new method for extended two-qubit gate set design, replacing iterative, fine calibration with fast characterization of a small number of gate parameters which are then tracked and corrected in circuit compilation. Coherent contributions to the pulse unitary that would traditionally be considered sources of error are treated as part of the gate definition, and compensated in software via single-qubit rotations. This approach enables rapid device-wide generation of high-fidelity two-qubit entangling gates, which are combined with standard calibrated gates to produce an expanded gate set. We show how these gates are directly usable as part of a quantum compiler, synthesizing generic two-qubit circuit blocks into minimal-duration sequences of the characterized gates interleaved with compensating single-qubit rotations. Benchmarking against circuits compiled using the default $CX$ gate alone on 127-qubit IBM hardware shows up to 7X improvement in success probability for Quantum Fourier Transform circuits up to 26 qubits, and up to 9X lower mean-square error in Trotter simulations of the one-dimensional transverse-field Ising model. Our hardware-agnostic characterization and compilation methodology makes it practical to scale up expressive gate sets on quantum computing architectures while minimizing the need for onerous fine-tuning of low-level control waveforms.

quant-ph↗

arXiv2025

Collective Communication for 100k+ GPUs

The increasing scale of large language models (LLMs) necessitates highly efficient collective communication frameworks, particularly as training workloads extend to hundreds of thousands of GPUs. Traditional communication methods face significant throughput and latency limitations at this scale, hindering both the development and deployment of state-of-the-art models. This paper presents the NCCLX collective communication framework, developed at Meta, engineered to optimize performance across the full LLM lifecycle, from the synchronous demands of large-scale training to the low-latency requirements of inference. The framework is designed to support complex workloads on clusters exceeding 100,000 GPUs, ensuring reliable, high-throughput, and low-latency data exchange. Empirical evaluation on the Llama4 model demonstrates substantial improvements in communication efficiency. This research contributes a robust solution for enabling the next generation of LLMs to operate at unprecedented scales.

cs.DC↗

arXiv2025

$\Delta$-Motif: Parallel Subgraph Isomorphism via Tabular Operations for Scalable Layout Selection

Subgraph isomorphism is a fundamental problem in graph analysis that seeks to find all instances of a pattern graph within a larger data graph while preserving structural relationships. This NP-complete problem is central to domains such as biological network analysis, social network mining, and quantum circuit optimization. Traditional approaches rely on backtracking algorithms like VF2, which suffer from sequential bottlenecks that limit their ability to exploit modern parallel hardware. In this work, we introduce $\Delta$-Motif, a GPU-accelerated subgraph isomorphism algorithm that reformulates the task through the lens of database operations. Our key insight is to represent both data and pattern graphs in tabular form, turning subgraph isomorphism into database primitives including joins, sorts, merges, and filters. $\Delta$-Motif decomposes graphs into small building blocks called motifs and systematically combines them using scalable relational operations. By leveraging mature, optimized libraries from the NVIDIA RAPIDS ecosystem and Pandas framework, our solution achieves massive parallelism while remaining portable across systems supporting standard relational primitives. Benchmarks show that $\Delta$-Motif outperforms established algorithms like VF2, achieving speedups of up to $595\times$ on GPUs. We further demonstrate its impact by applying it to quantum circuit compilation, addressing a critical bottleneck in quantum computing and enabling scaling to near- and medium-term devices. Our approach democratizes high-performance graph processing by exposing it through familiar database abstractions, eliminating the need for low-level programming while delivering exceptional computational efficiency.

cs.DS↗

arXiv2024

Integrated error-suppressed pipeline for quantum optimization of nontrivial binary combinatorial optimization problems on gate-model hardware at the 156-qubit scale

We introduce a novel hybrid quantum-classical variational optimization method for unconstrained binary combinatorial optimization problems on gate-model quantum computers, integrating a custom variational ansatz, staged feedback-based dual variational parameter update strategies, efficient parametric compilation, automated error suppression during hardware execution, and scalable O($n$) classical post-processing to correct for bitflip errors. Without this integrated approach, we show that standard circuit execution at scale produces output indistinguishable from random sampling, establishing the necessity of each pipeline component. We benchmark the method on IBM superconducting quantum computers for classically nontrivial optimization problems, where the optimization is conducted on hardware with no use of classical simulation or prior knowledge of the solution. For Max-Cut on random regular graphs with topologies not matched to device connectivity, the method achieves approximation ratios of 100% for unweighted 3-regular graphs up to 156 nodes, weighted regular graphs up to 80 nodes, and weighted 7-regular graphs up 50 nodes. Applied to higher-order binary optimization, the method finds the ground state energy of 127- and 156-qubit spin-glass models matched to device topology with linear, quadratic, and cubic interaction terms, achieving approximation ratios of at least 99.5% across all instances tested. The method consistently outperforms a classical local solver across all problems. Where published results on identical problem instances are available, our method demonstrates competitive or superior performance. These results demonstrate that an appropriately engineered approach enables gate-model quantum computers to produce high-quality solutions for nontrivial binary optimization problems at the 156 qubit scale, where naive implementations are insufficient for good performance.

quant-ph↗

arXiv2023

Experimental benchmarking of an automated deterministic error suppression workflow for quantum algorithms

Excitement about the promise of quantum computers is tempered by the reality that the hardware remains exceptionally fragile and error-prone, forming a bottleneck in the development of novel applications. In this manuscript, we describe and experimentally test a fully autonomous workflow designed to deterministically suppress errors in quantum algorithms from the gate level through to circuit execution and measurement. We introduce the key elements of this workflow, delivered as a software package called Fire Opal, and survey the underlying physical concepts: error-aware compilation, automated system-wide gate optimization, automated dynamical decoupling embedding for circuit-level error cancellation, and calibration-efficient measurement-error mitigation. We then present a comprehensive suite of performance benchmarks executed on IBM hardware, demonstrating up to > 1000X improvement over the best alternative expert-configured techniques available in the open literature. Benchmarking includes experiments using up to 16 qubit systems executing: Bernstein Vazirani, Quantum Fourier Transform, Grover's Search, QAOA, VQE, Syndrome extraction on a five-qubit Quantum Error Correction code, and Quantum Volume. Experiments reveal a strong contribution of Non-Markovian errors to baseline algorithmic performance; in all cases the deterministic error-suppression workflow delivers the highest performance and approaches incoherent error bounds without the need for any additional sampling or randomization overhead, while maintaining compatibility with all additional probabilistic error suppression techniques.

quant-ph↗

arXiv2023

Collisional S-Matrix for the Vibrational Dynamics of H+H2 by Quantum Computing

An algorithm and a system of quantum circuits is developed and applied to compute accurately the S matrix for the transitions between vibrational states of H2 for collisions with H. The algorithm was applied to 100 eV laboratory collision energy at a quantum circuit simulator. The effects of the discretized dissociative continuum to the transition cross sections are carefully studied and accuracy and convergence of the results with the chosen parameters of the algorithm and the collision system are verified by comparison with a solution of the time-dependent Schrodinger equation using the classical algorithm as well as comparison with a few results from the literature.

quant-ph↗

arXiv2022

Multistate Transition Dynamics by Strong Time-Dependent Perturbation in NISQ era

We develop a quantum computing scheme utilizing McLachlan variational principle in a hybrid quantum-classical algorithm to accurately calculate the transition dynamics of a closed quantum system with many excited states subject to a strong time-dependent perturbation. A systematic approach for optimal construction of a general N-state ansatz with unary N-qubit encoding is refined. We also utilize qubit efficient encoding in McLachlan variational quantum algorithm to reduce the number of qubits to log2 N, simultaneously diminishing depths of the quantum circuits. The significant reduction of the number of time steps is achieved by use of the second order marching method. Instrumental in obtaining high accuracy are adaptations of the circuits to include time-dependent global phase correction. We illustrated, tested and optimized our quantum computing algorithm on a set of 16 bound hydrogenic eigenstates exposed to a strong laser attosecond pulse. Results for transition probabilities are obtained with accuracy better than 1%, as established by comparison to the benchmark data. Use of interaction representation of the Hamiltonian reduces the effect of both NISQ noise and sampling errors accumulation while the quantum system evolves in time.

quant-ph↗

arXiv2021

Variational Quantum Linear Solver with Dynamic Ansatz

Variational quantum algorithms have found success in the NISQ era owing to their hybrid quantum-classical approach which mitigate the problems of noise in quantum computers. In our study we introduce the dynamic ansatz in the Variational Quantum Linear Solver for a system of linear algebraic equations. In this improved algorithm, the number of layers in the hardware efficient ansatz circuit is evolved, starting from a small and gradually increasing until convergence of the solution is reached. We demonstrate the algorithm advantage in comparison to the standard, static ansatz by utilizing fewer quantum resources and with a smaller quantum depth on average, in presence and absence of quantum noise, and in cases when the number of qubits or condition number of the system matrix are increased. The numbers of iterations and layers can be altered by a switching parameter. The performance of the algorithm in using quantum resources is quantified by a newly defined metric.

quant-ph↗

arXiv2020

Prospect of using Grover's search in the noisy-intermediate-scale quantum-computer era

In order to understand the bounds of utilization of the Grover's search algorithm for the large unstructured data in presence of the quantum computer noise, we undertake a series of simulations by inflicting various types of noise, modelled by the IBM QISKit. We apply three forms of Grover's algorithms: (1) the standard one, with 4-10 qubits, (2) recently published modified Grover's algorithm, set to reduce the circuit depth, and (3) the algorithms in (1) and (2) with multi-control Toffoli's modified by addition of an ancilla qubit. Based on these simulations, we find the upper bound of noise for these cases, establish its dependence on the quantum depth of the circuit and provide comparison among them. By extrapolation of the fitted thresholds, we predict what would be the typical gate error bounds when apply the Grover's algorithms for the search of a data in a data set as large as thirty two thousands.

quant-ph↗