Travis Humble

Pre-Fault-Tolerant Quantum Processors Achieve Quantitative Neutron-Scattering Benchmarks for Luttinger Liquids

A 50-qubit superconducting quantum processor simulates dynamical structure factors (DSFs) of KCuF3, a gapless Luttinger liquid, yielding quantitative agreement with inelastic neutron-scattering experiments via a quantum-classical workflow. Simulation accuracy is assessed using multiple metrics, revealing impacts from circuit depth and fidelity. The approach extends to gapped spectra in anisotropic 1D XXZ Heisenberg models relevant to CsCoX3 compounds, establishing a testable framework for strongly entangled quantum materials.

QASMTrans Delivers 100x Faster QASM Compilation with End-to-End Pulse Generation for Noisy Quantum Devices

QASMTrans is a self-contained C++ quantum compiler that achieves over 100x faster transpilation than Qiskit for large, high-depth circuits while maintaining comparable quality, enabling JIT deployment on FPGA/CPU-integrated QPUs. It provides end-to-end compilation from QASM to device pulses with QICK integration, supporting latency-aware Application-tailored Gate Sets (AGS) that optimize critical path pulse schedules and improve fidelity by up to 12% via QuTiP simulation. Noise-adaptive transpilation uses calibration data for qubit placement and critical path focus, plus device partitioning for concurrent circuit execution, facilitating real-time adaptive algorithms like ADAPT-VQE.

Quantum Computing Poised to Overcome Classical Limits in Energy Materials Design

Classical computational methods for energy materials face scaling and time-complexity limits, especially for high-dimensional or strongly correlated systems. Quantum computing leverages superposition and entanglement to tackle intractable problems, enabling hybrid QC-classical approaches for practical material simulation and design. Fault-tolerant QC promises predictive accuracy and quantum advantage for complex systems, though significant challenges persist.

GPU Acceleration with OpenMP Offload Delivers 100x Speedup for Sample-Based Quantum Diagonalization

Sample-based quantum diagonalization (SQD) is a hybrid quantum-HPC algorithm that encodes molecular Hamiltonians into quantum circuits, measures electronic configurations on quantum hardware, and performs iterative diagonalization on filtered subspaces using classical HPC systems. The diagonalization step, handled by the Davidson algorithm on selected electron configurations, is the primary computational bottleneck. By leveraging GPU thread-level parallelism via OpenMP offload on heterogeneous systems like Frontier, the approach achieves ~100x per-node performance gains, reducing classical processing from hours to minutes and enabling efficient ground/excited state energy extraction.

Quantum Computing: Beyond Classical Limits and Towards Integrated Futures

Quantum computing leverages quantum mechanical principles like superposition to revolutionize computation, offering solutions to problems intractable for classical computers. This field is moving from isolated research to integrated systems, seeking to enhance existing high-performance computing infrastructure rather than replace it. Key applications include material science simulations, with a focus on areas like efficient ammonia synthesis, and the potential adoption of quantum cloud platforms for research and education.

Heterogeneous SC-NA Architectures Boost Fault-Tolerant Quantum Efficiency by 752x Speedup and 10x Qubit Reduction

Heterogeneous Quantum Architectures (HQA) integrate superconducting (SC) qubits' speed with neutral-atom (NA) qubits' scalability to overcome limitations of homogeneous systems in fault-tolerant computing. MagicAcc offloads latency-critical Magic State Factories to SC while computing on NA arrays; MCSep uses NA for dense qLDPC memory and SC for fast surface-code processing. End-to-end cost modeling shows 752x average speedup over NA-only baselines and over 10x physical qubit reduction versus SC-only systems via cross-modality interconnects.

IRIS Runtime Evolves into Q-IRIS for Asynchronous Classical-Quantum Workflow Orchestration

Q-IRIS integrates the IRIS task-based runtime with XACC via QIR-EE to enable asynchronous execution of QIR programs across heterogeneous quantum backends and simulators. It supports concurrent classical and quantum tasks, demonstrated by scheduling multiple quantum workloads. Quantum circuit cutting decomposes circuits into subcircuits, reducing simulation load and improving throughput on early quantum hardware.

Certified Networked Randomness Amplification via Dynamic Remote Probing of 98-Qubit Entangled States

Researchers demonstrate certified randomness amplification over a network using Quantinuum's 98-qubit Helios trapped-ion processor by dynamically streaming quantum gates and delaying measurement basis revelation. The protocol maintains coherence for ~0.9 seconds, limits classical spoofing to 30 ms, and constrains adversaries to a 4,500 km radius. It achieves 0.586 fidelity on 64-qubit random circuits with 276 two-qubit gates, enabling amplification of low-entropy sources into nearly perfect randomness secure against malicious remote devices.

NVQLink: Ethernet-Based Architecture for Microsecond-Latency HPC-QPU Coupling

NVQLink proposes a platform connecting HPC resources to QPU control systems via commercial Ethernet, achieving 3.96 μs max round-trip latency for real-time tasks under 10s of μs tolerance. It supports all QPU modalities and controllers through CUDA-Q extensions enabling real-time callbacks and unified C++ programming of CPU/GPU/FPGA subsystems in the QSC, bypassing slow HTTP interfaces. QSC integration uses MLIR dialects and progressive lowering for kernel-based heterogeneous execution.

Hybrid HPC-QC Architectures Positioned as Future Compute Paradigm for Scalable Scientific Computing

Quantum computing complements rather than replaces HPC, with hybrid systems integrating QC acceleration into classical infrastructures deemed essential for practical scalability. Current QC faces high error rates and limited coherence, necessitating traditional HPC to maximize quantum benefits. The ADAC Institute's Quantum Computing Working Group synthesizes member surveys highlighting ongoing projects and strategic priorities for QC-HPC integration at leading supercomputing centers.

Where Quantum Meets HPC: Challenges, Opportunities, and the Road Ahead ❯ QUANTUM PROGRAM (WISER)

QuBound Predicts Tight Performance Bounds for Noisy Quantum Circuits via Data-Driven Noise Decomposition

QuBound is a data-driven workflow that decomposes historical quantum performance traces to isolate noise sources, embeds circuit and noise data via a novel encoder, and uses LSTM to predict computational performance bounds under fluctuating noise. It outperforms state-of-the-art learning-based predictors, which produce single values falling outside QuBound's bounds, and analytical methods with 10x narrower ranges. QuBound achieves over 10^6 speedup versus noisy simulation while enabling efficient noise characterization for quantum system management like job scheduling.

Quantum Properties Trojans Exploit Unitary Gates and Superposition to Stealthily Attack Pure Quantum Neural Networks

Researchers introduce Quantum Properties Trojans (QuPTs), novel backdoor attacks on fully quantum neural networks (QNNs) that leverage unitary properties of quantum gates for noise insertion and Hadamard gates to induce superposition. These QuPTs achieve superior stealthiness compared to prior methods while severely degrading QNN performance, with the most effective variant reducing binary classifier accuracy by 23% in experiments. This marks the first Trojan attack on purely quantum architectures, independent of hybrid classical-quantum designs.

Quantum Kernel SVM Outperforms Classical ML for Emotion Recognition from Wearable Sensors in Older Adults

Hybrid quantum machine learning with quantum kernel-based SVM exceeds classical ML in classifying emotions from physiological signals of wearable sensors. Achieves F1 scores over 80% across all emotion categories, with up to 36% recall improvement on limited datasets. Offers privacy-preserving, unobtrusive monitoring for ADRD and PTSD patients in clinical settings.

HPCpodcast-79: Travis Humble of ORNL on Quantum Tech (insideHPC Report)

FluxTrap Enables SIMD-Optimized Compilation for Trapped-Ion Quantum Systems

FluxTrap is a SIMD-aware compiler framework for modular trapped-ion quantum machines, unifying segmented intra-trap shift SIMD (S3) and global junction transfer SIMD (JT-SIMD) operations via a SIMD-enriched architectural graph that accounts for transport synchronization, gate-zone locality, and topological constraints. It employs SIMD aggregation and scheduling passes to optimize grouped ion transport and gate execution. On NISQ benchmarks, it achieves up to 3.82x reduction in execution time and orders-of-magnitude fidelity improvements, while scaling to fault-tolerant workloads and informing hardware design.

OneAdapt: Resource-Adaptive IR and Compiler for Photonic MBQC Minimizing Hardware and Time under Fusion Constraints

OneAdapt introduces a novel intermediate representation (IR) and optimization passes for measurement-based quantum computing (MBQC) tailored to resource-constrained photonic platforms. The compiler adaptively minimizes required hardware size and execution time while enforcing user-defined limits on fusion devices. It integrates with quantum error correction to enhance efficiency in photonic fault-tolerant quantum computing.

Flexion: Hybrid Bare-Logical Encoding Cuts QEC Overhead for Trapped-Ion MQC

Flexion proposes a hybrid encoding for trapped-ion quantum computers using bare physical qubits for high-fidelity single-qubit gates and QEC-encoded logical qubits only for two-qubit gates, avoiding full encoding overheads like gate synthesis and magic state distillation. It introduces a low-noise bare-to-logical conversion protocol, a tailored hybrid ISA for 2D grid TIQC, and an optimizing compiler that minimizes conversions and improves scheduling. Evaluations on VQA and FTQC benchmarks demonstrate superior performance with reduced resource demands, paving the way for megaquop-scale fault-tolerant computing.

PQI 2025 Public Lecture: Unraveling the Mysteries of Quantum Science & Technology, Dr. Travis Humble (Pittsburgh Quantum Institute)

Trapped-Ion Processor Delivers 71k Certifiably Random Bits via Remote Random Circuit Sampling

Researchers demonstrate certified randomness generation using the 56-qubit Quantinuum H2-1 trapped-ion quantum computer accessed remotely over the internet. The protocol employs client-generated challenge circuits from a small randomness seed, executed on an untrusted quantum server, with classical verification leveraging random circuit sampling hardness. Security analysis against restricted near-term adversaries, backed by supercomputer verification at 1.1e18 FLOPS/s, certifies 71,313 bits of entropy.

ASDF Compiler Enables Efficient Circuit Synthesis for Basis-Oriented Quantum Programming in Qwerty

ASDF is an open-source compiler for Qwerty, a quantum language based on bases and functions rather than circuits, addressing challenges in basis translation and automatic adjoint/predicated function specialization. It leverages a novel high-level quantum IR in MLIR to generate OpenQASM 3 or QIR for simulation or hardware execution. Evaluation shows ASDF circuits match fault-tolerant resource costs of circuit-oriented compilers.

QuSplit Job Splitting Boosts Fidelity and Throughput on Noisy Heterogeneous Quantum Processors

QuSplit employs genetic algorithm-based scheduling to split variational quantum optimization jobs like VQE, executing later stages on high-fidelity processors amid heterogeneous noise in multi-backend platforms. This counters fidelity losses from noise by optimizing resource allocation across backends like IBM Quantum and Amazon Braket. Simulations show sustained high fidelity for all jobs alongside throughput gains and scalability; real experiments on IBM Strasbourg validate fidelity improvements and faster convergence.

Quantum-Centric Krylov Subspace Algorithm Scales Ground State Simulations to 41-Site Impurity Models

A new quantum diagonalization algorithm merges classical diagonalization of quantum samples with Krylov-constructed subspaces, bypassing high-depth quantum phase estimation for noisy quantum hardware. It guarantees polynomial-time convergence under Krylov diagonalization assumptions and ground state sparseness. Demonstrated on Heron quantum processors with Frontier supercomputer, it achieves accurate ground states for single-impurity Anderson model (41 bath sites) and 4-impurity model (7 baths each), matching DMRG benchmarks.

Quantum Hybrid SVMs Outperform Classical Methods in Wearable-Based Stress Detection for Older Adults

Quantum hybrid support vector machines frame stress detection as anomaly detection using smartwatch sensor data, with cortisol as ground truth and TSST protocol validation on 40 older adults. Kernel-based quantum preprocessing explores complex feature spaces efficiently with limited features. Experiments show superior accuracy and recall over classical SVMs, reducing missed stress anomalies critical for timely healthcare interventions.

SymBreak Breaks Quantum Degeneracy Symmetry to Boost qLDPC Decoder Performance

SymBreak enhances belief propagation (BP) decoders for quantum LDPC (qLDPC) codes by adaptively modifying the decoding graph to mitigate quantum degeneracy, a key cause of BP convergence failures. This graph-level symmetry breaking enables fast, accurate decoding with 16.17× lower logical error rates than vanilla BP and 3.23× better than BP+OSD across qLDPC families. SymBreak incurs only 18.97% runtime overhead over BP while outperforming more complex BP+OSD in speed.

CaliScalpel Enables Concurrent In-Situ Qubit Calibration in Surface Codes via Code Deformation

CaliScalpel integrates fine-grained qubit calibration directly into surface code quantum error correction by deforming the code to isolate calibrating qubits from logical patches. This permits calibration to run concurrently with computation, using code enlargement to preserve error correction with minimal overhead. Optimized schedules from device characterization reduce physical error rates, yielding negligible execution time impact.

Day 1: Integrating Quantum Computing with High-performance Computing for Scientific Discovery (NERSC)

PowerMove Compiler Revolutionizes Neutral Atom QC by Integrating Zoned Architecture and Qubit Movement

PowerMove is a compiler for neutral atom quantum computers (NAQCs) that optimizes gate scheduling, qubit allocation, movement, and inter-zone communication by exploiting interdependencies in the zoned architecture (ZA). It enhances qubit movement frameworks to unlock scalability and fidelity improvements. Evaluations show fidelity gains of several orders of magnitude over state-of-the-art, with up to 3.46x faster execution and 213.5x reduced compilation time.

Quantum Computing: From Utility-Scale Quantum Computing to Quantum-Centric Supercomputing (Hudsonforum)

Greedy Topology-Aware Partitioning Accelerates Quantum Circuit Division with Comparable Quality

GTQCP employs a greedy heuristic on the qubit dependency graph to partition quantum circuits for execution on distributed hardware. It outperforms QuickPartitioner by 18% in runtime and ScanPartitioner by 96%, while matching ScanPartitioner's partition quality and exceeding QuickPartitioner's by 38%. This enables faster synthesis without sacrificing partition optimality in Berkeley Quantum Synthesis Toolkit benchmarks.

Visual Analytics Enables Spatial-Temporal Analysis and Optimization of Quantum Device Performance

Presents a visual analytics framework for exploring spatial and temporal patterns in quantum device metrics like qubit coherence and gate fidelity. Computes similarities and variances in performance data, with detailed qubit error analysis. Includes visualization for quantum circuit optimization to enhance algorithm efficiency and interpretability.

Hardware-Agnostic Framework for Integrating Quantum Accelerators into Classical HPC Workflows

Quantum computing enhances scientific HPC in quantum chemistry, optimization, and AI despite NISQ noise challenges. The proposed framework integrates QC as a hardware-agnostic accelerator using diverse simulators and hardware within classical HPC systems. Leveraging ORNL and DOE expertise, it incorporates QC into existing workflows via hardware, software, and user interfaces to enable synergistic quantum-classical computing.

Ancillary Entangling Kicks Double Quantum Annealing Speed with Enhanced Accuracy

Ancillary entangling Floquet kicks accelerate adiabatic quantum annealing by coupling primary system qubits to ancillary qubits via digital multi-qubit gates. Tuning ancillary gauge degrees enhances functionality in the system subspace, overcoming excitation gap shrinkage at scale. Numerical simulations on Ising models and encoded H2 yield 100% time-to-solution improvement with higher fidelity, validated by time-averaged Hamiltonian theory.

Engineering Dependable Hybrid HPC-Quantum Computing Systems

Quantum computing integration with HPC forms the HPC-QC paradigm, offering enhanced workloads but introducing novel dependability challenges. The paper identifies integration issues across QC, cloud, HPC, and network security, proposing an interdisciplinary framework for prescriptive co-design to prevent pitfalls and accelerate QC maturation. Key efforts focus on resilient systems, adapting classical techniques to quantum domains, and developing paradigm-specific performance-reliability metrics for scalable hybrids.

Approximate Quantum Adders Boost NISQ Fidelity by up to 371% with Reduced Circuit Depth

Approximate computing is applied to quantum adder circuits to enhance noise resilience and minimize depth for NISQ devices. Five designs are proposed: three with carryout and two without, using pass-through approximations (zero depth) and single-CNOT constant-depth approaches. Experiments on IBM Qiskit with noise models (thermal, depolarizing, etc.) show fidelity improvements of 8.34%-219.22% over exact ripple carry adders without carryout, and 8.23%-371% with carryout.

RNS-DQC Enables Low-Depth, Noise-Resilient Distributed Quantum Addition

RNS-based distributed quantum addition replaces high-depth quantum adders with parallel modulo adders across multiple quantum computers, reducing circuit depth and enhancing noise resilience in NISQ devices. The QSMART tool generates optimized RNS adder sets based on depth, range, and efficiency, incorporating a novel low-depth Quantum Diminished-1 Modulo (2n+1) Adder (QDMA). Simulations on Quantinuum H1 demonstrate 11.36%-133.15% higher output fidelity for 6-10 bit additions versus non-distributed full adders, with scalability beyond 20 qubits.

Refining QPU Task Management to Bridge Quantum Computing and HPC

Quantum computing programming advances toward HPC-compatible workflows, yet faces core integration hurdles with disparate technologies. QC languages, runtimes, and algorithms share commonalities but require QPU refinements for many-task and asynchronous management. The paper explores these enhancements via examples to unlock QC's scientific potential within HPC ecosystems.

Surf-Deformer Achieves 35x-70x Lower Failure Rates in Surface Code QEC with Half the Qubits

Surf-Deformer integrates adaptive defect mitigation into surface code workflows using basic deformation instructions from gauge transformations, enabling optimized repairs for specific defects with minimal qubits. It features an adaptive code layout supporting efficient logical operations. Evaluations demonstrate 35x-70x end-to-end failure rate reductions over prior methods at 50% qubit usage, with ablation studies confirming superior QEC preservation and near-optimal communication throughput.

[[4,2,2]] Error Detection Code Enables VQE to Achieve Chemical Accuracy for H2 on NISQ Devices

The [[4,2,2]] quantum error detection code, combined with readout error detection and post-selection, mitigates noise in NISQ VQE simulations of H2 ground state energy. Encoded simulations yield energy estimates over 1 mHa (0.027 eV) more accurate than unencoded ones, with comparable precision and higher state fidelity. Unlike unencoded results, encoded outputs fall within the 1.6 mHa chemical accuracy threshold relative to the exact energy.

QIR-EE Enables Cross-Platform Execution of Hybrid Quantum-Classical Programs via LLVM and XACC

QIR-EE is a cross-platform execution engine that parses, interprets, and executes QIR programs using LLVM for hybrid quantum-classical instructions. It provides extension points for custom runtimes and hardware, integrating with the XACC library to dispatch programs across IonQ, Quantinuum, IBM hardware, and numerical simulators. Results demonstrate efficient handling of mixed instructions and data in quantum frameworks.

Bayesian Adaptation Boosts PEC Stability by 60% Against Non-Stationary Quantum Noise

Non-stationary noise undermines probabilistic error cancellation (PEC) in quantum computing by destabilizing observable estimates. A Bayesian method adaptively enhances PEC, improving both accuracy and stability. On IBM Kolkata with a 5-qubit Bernstein-Vazirani algorithm, it achieves 42% higher accuracy and 60% greater stability versus non-adaptive PEC.

TANQ-Sim Pioneers Double-Precision Tensor Cores for Scalable Noisy Quantum Circuit Simulation on GPUs

TANQ-Sim is a density matrix simulator for noisy quantum circuits that leverages double-precision tensor cores on NVIDIA Ampere/Hopper GPUs, marking the first non-AI/ML use of this hardware. It employs gate fusion optimizations and NVSHMEM for efficient scaling on HPC systems like NERSC Perlmutter. Evaluations confirm its ability to handle deep circuits with coherent/non-coherent noise, demonstrated via teleportation, entanglement distillation, and Ising model case studies.

Adiabatic Quantum SVMs Achieve Order-of-Magnitude Training Speedup with Comparable Accuracy

Adiabatic quantum computers train support vector machines by leveraging their optimization capabilities for quadratic unconstrained binary optimization problems. The quantum approach demonstrates time complexity an order of magnitude better than classical methods and matches Scikit-learn accuracies on Iris, WBC, Wine, Digits, and Lambeq datasets. Scalability analysis reveals 3.5-4.5x training time speedups over classical baselines on high-dimensional datasets with millions of features.

Quantum-Centric Supercomputing Emerges as Key Enabler for Materials Science Challenges

Quantum computing promises to accelerate hard computational tasks in materials science that strain current high-performance supercomputers, such as simulation, analysis, and data processing for novel material design. Quantum-centric supercomputing integrates quantum processors with classical HPC via result validation, hard problem identification, and synergistic workflows. The paper outlines representative use cases, implementation challenges, and future research directions to realize these benefits.

High-Performance CPU Simulations Enable Verification of Deep Quantum Circuits for Nuclear Physics

Researchers developed optimized numerical simulation techniques, including 1- and 2-qubit gate fusion and mid-circuit measurement management, to verify deep quantum circuits for low-energy nuclear state preparation. These methods scale to 21-qubit circuits with over 115 million gates across high-performance computing systems. The advances address memory and processing bottlenecks in classical simulation of quantum nuclear physics applications.

QuPAD Achieves 270x Faster Calibration for High-Fidelity Quantum Learning on Noisy Devices

QuPAD addresses runtime fidelity loss in NISQ devices for variational quantum algorithms by replacing CNOT gates with robust parameterized Rzx gates and using pulse-level calibration via fitting functions and evolutionary optimization. This in-situ calibration predicts and corrects gate deviations efficiently without costly parameter-shift optimization. On 8-10 qubit devices, QuPAD runs in under 15 minutes, yielding 59% accuracy gains in classification and 66% better energy estimates in molecular simulation versus vanilla VQC.

Bayesian Adaptive PEC Outperforms Static Methods 4.5x on Time-Varying Quantum Noise

Quantum computers face non-stationary noise channels with high error rates that degrade reliability. A Bayesian inference algorithm using Dirichlet distributions models Pauli channel stochasticity and dynamically infers parameters to adapt probabilistic error cancellation (PEC). This approach achieves 4.5x better performance than non-adaptive methods, measured by Hellinger distance to ideal distributions, highlighting the need for temporal noise characterization.

QASMTrans Delivers 369X Transpilation Speedup Over Qiskit for Large NISQ Circuits

QASMTrans is a C++ framework for quantum circuit transpilation targeting NISQ devices, converting high-level QASM circuits to machine-specific implementations. It achieves up to 369X speedup over Qiskit, transpiling large circuits like uccsd_n24 in 69s and qft_n320 in 31s versus Qiskit's >1 hour. This enables faster design space exploration for dense circuits with O(10^6) gates.

QECC-Synth Automates Optimal QEC Layouts on Sparse Quantum Hardware via Ancilla Bridges and MaxSAT

QECC-Synth is an automated compiler that synthesizes quantum error correction (QEC) code layouts for sparse hardware architectures by leveraging ancilla bridge techniques and optimizing their design flexibilities with MaxSAT. It addresses the connectivity mismatch between dense QEC requirements and sparse hardware, avoiding underutilization of circuit features and manual designs limited to specific codes. Evaluations demonstrate superior performance and broader applicability across diverse QEC codes and architectures compared to prior methods.

Photons Drive Entanglement-Based Quantum Information Technologies

Entanglement, a uniquely quantum phenomenon originating from 1935 debates, powers revolutionary quantum information technologies (QIT) with photons as ideal carriers due to room-temperature operation, telecom compatibility, and interfacing with solid-state systems. The review details entanglement generation theory, historical experiments, and photonic implementations for sensing, imaging, spectroscopy, data processing, and communication. It forecasts architectures for next-generation entanglement-based QIT applications.