Travis Humble

NISQ Device Instability Exceeds Thresholds for Reliable Quantum Algorithm Execution

NISQ devices exhibit significant instability due to non-stationary noise from decoherence, leakage, and crosstalk, undermining error mitigation assumptions. Using similarity metrics on IBM Washington noise data from Jan 2022 to Apr 2023, authors quantify reliability for a 5-qubit Bernstein-Vazirani algorithm, finding fluctuations of 41-92% against a 2.2% stability threshold. This renders the device unreliable for consistently reproducing statistical means in quantum circuits.

Hellinger Distance Bounds Quantum Computation Stability Against Device Noise Drift

Non-stationary noise in quantum devices causes temporal and spatial drift in characterization metrics, destabilizing computation outcomes. The authors quantify distribution differences using Hellinger distance and derive an analytical bound linking this distance to the stability of computed expectation values. Numerical simulations on IBM's transmon-based 'washington' device validate that stability is upper-bounded by the Hellinger distance, enabling tolerance specifications for reliable quantum computing.

Graph Decomposition Enables QAOA to Solve Large MaxCut on Noisy Quantum Hardware

A novel algorithm decomposes large MaxCut problem graphs into subgraphs with ~1/10 the vertices, solvable via QAOA on current noisy quantum devices. Reduced problems yield high approximation ratios (avg. 0.96) using either classical (Gurobi) or quantum (QAOA) subroutines, enabling optimal solutions for 100-vertex graphs via single-layer QAOA on Quantinuum H1-1 with only 500 samples. The method excels on sparse k-regular graphs, reducing to at most nk/(k+1) vertices in polynomial time, and extends to broader combinatorial optimization.

ATARC Quantum Speaker Series - Travis Humble, ORNL/QSC (ATARC Channel)

Quantum Hardware Demonstrates Accurate Singlet Fission Energetics in H4 Molecule Outperforming Tractable Classical Methods

Researchers model singlet fission in the linear H4 molecule on Quantinuum H1-1 hardware using qubit tapering, measurement optimization via shared eigenbases for QWC Pauli strings, and parallel circuit execution across 20 qubits. Results confirm energetic prerequisites for singlet fission with excellent agreement to exact transition energies in the chosen basis. These quantum computations surpass classical methods feasible for such candidates.

QAOA Output Probabilities Follow Approximate Boltzmann Distributions with Effective Temperature Scaling as C_min / (n √p)

Analysis of QAOA circuits on 7,200 random MaxCut instances with 14-23 qubits and p ≤ 12 shows average basis state probabilities scale exponentially with energy (cut value), peaking at the optimal solution, resembling Boltzmann distributions. The effective temperature scales as T ∼ C_min / (n √p), where C_min is the optimal cut value. This scaling enables accurate approximation of output distributions up to 38 qubits, matching exact simulation metrics.

Noise Model for Many-Qubit MS Interactions Matches Trapped-Ion QAOA Experiments

Researchers developed a parameter-free physical model for many-qubit Mølmer-Sørensen (MS) interactions in trapped ions, incorporating noise from vibrational frequency fluctuations, laser power variations, thermal states, and SPAM errors, calibrated from experiments. The model validates against two-ion MS sequences with χ²_red ≈ 2 and predicts MaxCut QAOA approximation ratios of 0.93 and 0.95 for three and six ions, closely matching experimental 91% and 83% of optimal. Projected improvements in measurement and trap frequency control enable 99% of optimal QAOA performance.

Variational Quantum Autoencoder Achieves Subsystem Purification for Hilbert Space Dimensionality Reduction

Proposes a variational autoencoder using parameterized quantum circuits to compress high-dimensional quantum states into tensor product subsystems by minimizing Tr(ρ²). Applies controlled swaps and measurements to halve qubit count while preserving state features. Validates on 8x8 Bars and Stripes dataset, yielding 95% classification accuracy in downstream supervised learning.

Quantum Computing Integrates as Core HEP Discovery Tool with Mutual Ecosystem Benefits

Quantum computing expands HEP capabilities across the Computational Frontier, enabling simulations of quantum field theories, enhanced sensor data analysis for particle searches, and overcoming classical computing bottlenecks. HEP and QIS are interdependent: HEP requires accessible quantum computers and contributes expertise in quantum domain knowledge, superconducting tech, cryogenics, microelectronics, and large-scale management. Co-design of HEP-tailored quantum systems demands robust investment to realize quantum technology promises in HEP over the next decade.

Adaptive Inference of Quantum Channel Parameters Boosts NISQ Circuit Stability

NISQ devices suffer error rates 1e-2 or higher versus classical 1e-17, causing instability and reproducibility issues without user calibration access. The study proposes dynamically inferring critical channel parameters from noisy binary circuit outputs to enable adaptive error mitigation. Efficacy is assessed via Hellinger distance reduction on circuits like uniform superposition, with scalability as an open challenge.

Framing Stability Metrics for Noisy Quantum Programs

Quantum computation demonstrations face noise-induced errors from imperfect hardware. The paper defines computational accuracy, result reproducibility, device reliability, and program stability with intuitive, operationally meaningful bounds on outputs. These metrics underscore the need for statistical analyses to build confidence in quantum information science results.

QAOA Efficiently Prepares Ground States of Frustrated Ising Models on Small Lattices with Few Measurements

QAOA on near-term quantum computers prepares ground states of classical Ising models on 9-spin unit cells of square, Shastry-Sutherland, and triangular lattices, including frustrated regimes. Theoretical success probabilities correlate with ground state structure, requiring ≤100 measurements for ground state identification. Trapped-ion experiments recover Shastry-Sutherland ground states near ideal theoretical values, validating QAOA for materials simulation where classical methods falter.

QIR Alliance Advances Quantum Computing Interoperability and Performance

The QIR Alliance, under the Linux Foundation, is developing a Quantum Intermediate Representation to enhance interoperability and reduce development effort in quantum computing. This initiative aims to integrate classical computations with quantum execution, enabling more expressive programs and optimized hybrid algorithms. By leveraging state-of-the-art compiler tools and existing high-performance computing practices, QIR facilitates advanced hardware-level interactions and supports diverse quantum backends for improved performance and new algorithm design patterns.

Device Noise Characterization Analytically Bounds Quantum Circuit Reproducibility

Noisy quantum circuits exhibit reproducibility limits quantified by Hellinger distance between repeated executions, driven by statistical noise fluctuations. Device characterization metrics provide an analytic upper bound on this Hellinger distance variability. Validation on a superconducting transmon processor with single-qubit circuits confirms the bound via a composite device parameter, enabling efficient reproducibility assessment without exhaustive repetitions.

Optimal Control Enables High-Fidelity Simulation of String Order Melting in Transmon-Based SPT Chains

Optimal control pulses simulate symmetry-protected topological (SPT) states in a tunable transmon architecture by solving one- and two-site optimization problems under leakage constraints. Numerical simulations demonstrate time-dependent melting of perturbed SPT string order in a six-site spin-1 particle chain model. Achieved average state infidelity of 10^{-3} indicates feasibility for current superconducting quantum hardware.

Quantum Computing Blueprint for HEP: Algorithms, Software, and Infrastructure Roadmap

Quantum computing enables novel representations and reasoning for quantum mechanical phenomena in high-energy physics (HEP), supporting modeling, simulation, detection, classification, data analysis, and forecasting. Significant gaps exist in integrating quantum hardware, software, and applications into HEP research programs. The paper outlines challenges and opportunities, prioritizing development of algorithms, applications, software, hardware, and infrastructure for practical and theoretical HEP applications over the next decade.

Diverse Non-Von Neumann Architectures Promise Future HPC Scalability and Efficiency

The escalating demands for computational power in HPC, coupled with the inherent limitations of conventional Von Neumann architectures in terms of energy efficiency and scalability, necessitate a paradigm shift. This panel explores various non-Von Neumann computing approaches, including reversible, in-memory, analog, quantum, neuromorphic, and unary computing, each offering unique strengths to overcome current HPC bottlenecks. These diverse technologies, while in varying stages of maturity, share a common goal of improving energy efficiency and tackling data movement challenges, ultimately aiming to achieve unprecedented performance gains and enable future exascale and zettascale computing capabilities.

Quantum Imaginary Time Evolution Achieves 93%+ Success on MaxCut for Graphs up to 50 Vertices

QITE solves MaxCut using a linear unitary Ansatz, unentangled initial state, and imaginary-time-dependent Hamiltonian interpolating from full graph to a two-edge subgraph. Applied to thousands of random graphs up to 50 vertices, it converges to the maximum solution with 93% or higher performance. This outperforms classical greedy and Goemans-Williamson algorithms; ground state overlap serves as a unique quantum performance metric, improvable via higher-order Ansaetze and entanglement.

Parameter Transfer with Rescaling Enables Effective QAOA on Weighted MaxCut

Weighted MaxCut problems introduce poor local optima in QAOA objective landscapes, hindering parameter optimization compared to unweighted cases. A simple rescaling scheme allows transfer of a single typical parameter vector across instances, yielding approximation ratios within 2.0 percentage points of direct optimization on 34,701 graphs up to 20 nodes. Refinements like 10 metadistribution samples or local optimization from transferred parameters match exhaustive optimization in 96.35% of cases.

QAOA Resource Requirements Scale Exponentially with Problem Size, Graph Degree, and Hardware Limitations

QAOA performance on near-term quantum hardware is constrained by exponential scaling in measurement requirements for sampling idealized circuit outputs under noisy gates. These requirements grow with problem size, graph degree, ansatz depth, gate infidelities, and inversely with hardware connectivity. Mitigation strategies include enhancing hardware connectivity or adopting QAOA variants with fewer layers for improved performance.

Empirical Direct Characterization Outperforms Gate Set Tomography and Pauli Reconstruction for Accurate Noisy Quantum Circuit Simulations

Benchmarking on a 27-qubit superconducting transmon device reveals that gate set tomography, Pauli channel noise reconstruction, and empirical direct characterization yield noise models with varying resource costs and information content. Model accuracy in simulating noisy circuits does not correlate with characterization information gained, and the optimal method depends on the specific circuit. Empirical direct characterization scales best and delivers the highest simulation accuracy across benchmarks.

Numerical Simulations Reveal Noise Scaling in UCC Ansatz VQE for NaH Ground State

Numerical simulations assess noisy quantum circuits using unitary coupled cluster (UCC) ansatze in variational quantum eigensolver (VQE) for NaH ground-state energy. Accuracy and fidelity degrade with gate noise levels, inter-molecular configuration, ansatz depth, and optimization methods. Relative energy error and state fidelity provide quantifiable metrics against classical ground truth.

MLIR-Enabled Quantum Compiler for Circuit Transformations and Hardware Diagnostics

Researchers adapt MLIR into a quantum compiler to enable circuit transformations for efficient execution on quantum hardware. They demonstrate mirror circuit insertions during compilation to test hardware performance by measuring quantum circuit accuracy. Validation occurs on superconducting and ion trap platforms, confirming MLIR's utility for hardware-dependent diagnostics via automated transformations. Implementation is open-source at github.com/ORNL-QCI/qcor.

Optimal Control Objective Functions Enable High-Fidelity Quantum Gate Discovery

Traditional optimal control objective functions impose arbitrary high costs on useful quantum gate controls. The proposed framework designs objectives that allow novel gates like echo pulses and locally-equivalent gates. Numerical simulations demonstrate microwave-only entangling gates for transmon qubits with higher fidelity in fewer iterations than standard methods.

Multi-Angle QAOA Boosts Approximation Ratios with Shallower Circuits via Parameter Expansion

The multi-angle QAOA ansatz reduces circuit depth while improving approximation ratios for MaxCut by introducing more classical parameters per layer. It achieves a 33% higher approximation ratio than standard QAOA on an infinite family of MaxCut instances and matches three-layer standard QAOA performance with one layer on 8-vertex graphs. Optimized parameters often zero out, enabling gate removal for even shallower circuits, enhancing viability on NISQ devices. Empirical results confirm superior ratios at equal depths on 50- and 100-vertex graphs.

Variable Substitution Cuts QAOA Circuit Depth for 3-SAT

A global variable substitution reduces n-variable monomials in combinatorial optimization to fewer variables, applied to 3-SAT for QAOA implementation. The product formulation with substitution decomposes gates more efficiently than the linear formulation without decomposition. Benchmark 3-SAT instances show strictly lower upper bounds on optimal QAOA circuit depth using the substitution method.

Quantum Variational Imaginary Time Evolution Enables Prime Factorization on Noisy Hardware

Researchers propose prime factorization via variational imaginary time evolution, encoding factors in a Hamiltonian's ground state and iteratively optimizing trial states. Circuit evaluations per iteration scale as O(n^5 d), with n as the number's bit-length and d as circuit depth. The method factorizes numbers using 7-9 qubit Hamiltonians with single-layer entangling gates and runs successfully on IBMQ Lima.

Quantum Annealing Matches Classical Feature Selection for Stress Detection but Excels Under Data Scarcity

Quantum annealing selects optimal feature subsets from physiological signals (foot/hand EDA, ECG, respiration) for stress detection by embedding Pearson correlations into a binary quadratic model solved via D-Wave's clique sampler. It performs equivalently to classical methods under normal conditions. Critically, QA maintains robust performance under data uncertainty like limited training data, where classical techniques degrade significantly.

Quantum Annealing on D-Wave Outperforms Classical ML in Data-Limited, High-Dimensional Classification Tasks

Quantum annealing via D-Wave systems optimizes machine learning pipelines, particularly for classification in real-world applications constrained by limited training data and high-dimensional features. Experimental results demonstrate its use in image recognition, remote sensing, computational biology, and particle physics, where classical methods underperform. The review analyzes advantages over classical computation for such problems while noting implementation challenges.

Quantum Workforce Spans Quantum Specialists, Classical Engineers, Coders, and Communicators to Drive Multidisciplinary Innovation

Panelists define the quantum workforce across three arenas: quantum engineers solving quantum problems (e.g., QKD), quantum engineers applying quantum to classical problems (e.g., inertial sensing), and classical engineers tackling quantum challenges (e.g., precision pulse generators, low-noise power supplies). Essential skills include coding for open-source access and simulations, interdisciplinary communication, teamwork, and passion over deep quantum expertise, enabling translation of theory to deployable systems. Workforce development requires lowering entry barriers via cloud access (IBM, Microsoft), K-12 culture-building, and commoditization for broad adoption, with diversity and specialization evolving as quantum integrates into everyday tools like sensors and computing.

NISQ Devices Show Large Fluctuations in Key Metrics Over 22 Months, Limiting Reliable Scale

Noisy intermediate-scale quantum (NISQ) devices suffer from unstable performance due to fluctuating noise, impacting reproducibility. The study quantifies stability using Hellinger distance to compare gate fidelities, duty cycles, and register addressability across time and space. Data over 22 months reveals significant fluctuations, indicating reliability only at limited scales.

QuaSiMo: Composable Object-Oriented Framework for Hybrid Quantum Simulation Workflows

QuaSiMo introduces an object-oriented design scheme distilling common data structures and methods from existing quantum simulation algorithms, enabling composable hybrid quantum/classical workflows. Implemented in the hardware-agnostic QCOR language as a library, it supports extension, specialization, and dynamic customization for new algorithms. Validation demonstrates utility on IBM and Rigetti quantum processors for prototypical simulations.

Optimal Control Enables Analog Quantum Simulation of Bose-Hubbard Model on Superconducting Hardware

The method combines digital decomposition and optimal control to map arbitrary model Hamiltonians onto superconducting transmon device parameters for analog quantum simulation. It constructs optimal analog controls to emulate extended Bose-Hubbard model dynamics, analyzing impacts of control time, digital error, and pulse complexity. Demonstrated accuracy and robustness suggest applicability to near-term quantum devices.

Benchmarking Chain Breaks in Quantum Annealing Reveals Hardware-Specific Tuning Opportunities

Quantum annealing on embedded Hamiltonians suffers from chain breaks due to non-adiabatic dynamics causing excitations. Empirical benchmarking identifies optimal parameters minimizing chain break probabilities across problem suites. Localized break rates correlate with embedding methods, enabling targeted post-processing to correct errors and tune Hamiltonians for hardware performance.

ADAPT-VQE and Measurement Caching Boost CMX and PDS Accuracy and Efficiency on NISQ Hardware

ADAPT-VQE constructs shallow quantum circuits that improve ground state energy accuracy in connected moments expansion (CMX) and Peeters-Devreese-Soldatov (PDS) methods without exceeding NISQ constraints. CMX converges to ground state energies given sufficient state overlap, while PDS ensures variational convergence. Measurement caching exploits recurring Hamiltonian terms across moment powers, and coefficient-thresholded measurement further reduces circuit executions for tunable precision.

QAOA Exceeds Classical MaxCut Bounds on Small Graphs with Empirical Performance Scaling

Numerical simulation of QAOA on all non-isomorphic unweighted graphs with ≤9 vertices for MaxCut reveals narrowing approximation ratio distributions and broadening optimal recovery probabilities with increasing layers up to depth 3. QAOA surpasses the Goemans-Williamson classical bound for most graphs. Optimized variational parameters exhibit consistent ensemble patterns enabling efficient MaxCut heuristics; the dataset benchmarks QAOA performance.

Graph Odd Cycles and Symmetry Predict QAOA Performance on MaxCut

QAOA performance on MaxCut is evaluated for all connected non-isomorphic graphs up to 8 vertices at depths ≤3. Strongest predictors of success are presence of odd cycles and graph symmetry levels. Results are shared publicly as a benchmark dataset to identify problem classes likely showing quantum advantage.

NISQ Devices Achieve Routine Chemical Accuracy for Alkali Hydride Molecules via Advanced Algorithms and Error Mitigation

Hybrid quantum-classical methods on IBM NISQ devices, including symmetry-preserving variational eigensolvers, quantum imaginary time evolution with Lanczos, and systematic error cancellation via hidden inverse gates, enable chemical accuracy in electronic structure calculations for alkali hydride molecules. These approaches complement variational techniques and mitigate errors effectively. Results indicate rapid progress from initial quantum computations to routine accuracy for simple molecules, with potential for scaling to larger systems as NISQ hardware improves.

Composable Object-Oriented Framework Enables Modular Hybrid Quantum Simulation Workflows

Presents an object-oriented design scheme for developing hybrid quantum/classical algorithms, using common data structures and methods distilled from existing quantum simulation algorithms. Supports extension, specialization, and dynamic customization to synthesize new workflows. Implemented in hardware-agnostic QCOR language within QuaSiMo library, validated on IBM quantum processors for prototypical simulations.

State-Dependent Noise Model Enables Fidelity-Optimized Routing in NISQ Devices

Researchers validate a composable, state-dependent noise model for CNOT and SWAP gates in NISQ processors, based on correlated binary noise and characterized via pairwise measurements. The model captures non-trivial noise dynamics during quantum state routing, matching tomographic reconstructions from a real device. This enables expected state fidelity estimates to guide optimal routing decisions in near-real-time NISQ operations.

Quantum Annealing RBM Matches Classical Training for Cybersecurity Data Balancing and Classification

Researchers trained restricted Boltzmann machines (RBMs) using D-Wave 2000Q quantum annealing (QA) and classical contrastive divergence (CD) on the imbalanced ISCX cybersecurity dataset. Two balancing schemes—undersampling with majority voting and synthetic data generation—improved classification accuracy, with CD reaching 95.68% and QA 80.04% post-undersampling; synthetic data enabled KNN and NN classifiers to hit 93% accuracy. This 64-bit proof-of-concept demonstrates QA-RBM viability for practical binary classification tasks.

Quantum Associative Memory Excels in High-Density Particle Track Classification on D-Wave Annealers

Quantum associative memory models based on Ising formulations—QAMM and QCAM—are applied to particle track classification using the D-Wave 2000Q quantum annealer. Energy-based QAMM classification performs well for small pattern densities and low detector inefficiencies. State-based QCAM achieves high recall accuracy for large pattern densities and shows superior robustness to detector noise and inefficiencies. Performance is characterized across detector resolution, pattern library size, and noise levels.

ADAPT-VQE Outperforms Standard VQE in Robustness for Molecular Ground State Simulations

Benchmarking reveals that both VQE and ADAPT-VQE accurately estimate ground-state energies and potential energy curves for H2, NaH, and KH via numerical simulation. ADAPT-VQE demonstrates superior robustness against variations in optimization methods compared to standard VQE. Gradient-based optimizers prove more efficient and effective than gradient-free alternatives, though state fidelity errors increase with molecular size.

Quantum Annealing Solves Small ILP Instances with Optimized Schedules Mitigating Decoherence

Researchers map NP-hard integer linear programming (ILP) problems to quantum annealers, outperforming random guessing on small instances. Optimized anneal schedules reduce decoherence effects, with simulations confirming quantum origins via qualitative reproduction of improvements. Limitations persist due to hardware constraints and decoherence for larger problems.

Correlation-Optimized Virtual Orbitals Enable Compact Plane-Wave Basis for Quantum Many-Body Simulations

Traditional pseudopotential plane-wave Hartree-Fock virtual orbitals capture minimal electron correlation due to weak interactions with occupied orbitals, limiting their use in select CI and coupled cluster methods. COVOs, derived by optimizing orbitals from small pairwise CI Hamiltonians, generate compact virtual spaces that recover substantial correlation energy. For H2, 4 COVOs achieve FCI/cc-pVTZ accuracy in both classical FCI and quantum simulations, extending applicability to quantum computing and other post-HF methods.

Moment-Based Distance Metric Quantifies DiVincenzo Criteria for NISQ Device Stability

Researchers propose a moment-based distance (MBD) metric to assess NISQ device stability by measuring histogram similarity in time (temporal stability) and space (spatial stability). Stability is defined via reproducibility of histograms for identical experiments, grounded in DiVincenzo's quantum computing requirements. The framework is demonstrated using data from IBM's Yorktown device.

QAOA Circuit Depth Lower Bounded by Graph Chromatic Index, Suiting MaxCut but Not Knapsack

QAOA circuit depth for each iteration is lower bounded by the chromatic index of a graph G derived from the combinatorial optimization problem structure. This bound reveals that MaxCut, MaxIndSet, and certain Vertex Cover and SAT instances scale favorably for QAOA on NISQ devices, while Knapsack and TSP exhibit excessive depth requirements. The analysis highlights problem-specific feasibility for quantum advantage given exponential error growth with depth.

Quantum Annealing Controls Benchmarked via Portfolio Optimization on D-Wave 2000Q

Researchers benchmark quantum annealing controls on the D-Wave 2000Q using portfolio optimization instances, comparing empirical results to ground truth solutions. They evaluate forward and reverse annealing methods, identifying optimal control variations that maximize success probability and minimize chain breaks. This approach reveals how controls tune quantum dynamics to improve computational accuracy and understand error mechanisms.

Quantum Annealing Accelerates RBM Training with Comparable Accuracy to Classical CD

Quantum annealing on D-Wave 2000Q computes RBM model expectations for gradient learning faster than MCMC in contrastive divergence (CD). On 64-bit bars and stripes dataset, quantum-trained RBM achieves classification accuracy matching CD-trained models. Quantum sampling eliminates costly MCMC steps while delivering similar image reconstruction and log-likelihood performance.

Quantum Annealing Simulates All Phases of 468-Spin Shastry-Sutherland Ising Model

Researchers simulate a 468-spin frustrated Shastry-Sutherland Ising model on a quantum annealer using mean-field boundary conditions and iterative annealing to overcome finite-size effects and defects. They accurately recover all known phases, including the fractional magnetization plateau, and the static structure factor at phase transitions. This validates quantum annealing for studying frustration-induced exotic phases like spin liquids and stripe phases.