Chronological feed of everything captured from Hartmut Neven.
youtube / hneven / Jul 19
Quantum computers leverage quantum physics, specifically the concept of superposition (Multiverse theory), to perform computations with significantly fewer steps than classical computers. This allows them to achieve computational advantages by effectively executing calculations in parallel across many theoretical realities. While practical applications are still emerging, the technology holds promise for signal processing, materials science, and complex optimization problems, with rapid advancements in computational power and error correction.
quantum-computingquantum-physicsquantum-algorithmsquantum-hardwarequantum-applicationsquantum-error-correctiongoogle-quantum-ai
“Quantum computers replace binary logic with quantum physics, utilizing superposition to perform computations in parallel universes.”
paper / hneven / May 27
This paper presents a 69-superconducting qubit analog-digital quantum simulator capable of both universal quantum gates and high-fidelity analog evolution. The simulator was used to study thermalization dynamics and criticality in a 2D XY quantum magnet. Key findings include the observation of the Kosterlitz-Thouless phase transition and deviations from Kibble-Zurek scaling due to the interplay of quantum and classical coarsening.
quantum-simulationanalog-digital-quantum-computingsuperconducting-qubitsthermalizationcriticalityquantum-magnetseigenstate-thermalization-hypothesis
“A 69-superconducting qubit analog-digital quantum simulator has been developed.”
paper / hneven / May 13
This paper extends a known protocol for wormhole teleportation to enable it between two entangled Sachdev-Ye-Kitaev (SYK) models using only a classical communication channel. The research uses finite N simulations to demonstrate holographic features characteristic of wormhole teleportation, relating these features to size winding as a dual description.
quantum-physicswormhole-teleportationholographytheoretical-physicssyk-model
“The existing Gao and Jafferis wormhole teleportation protocol can be extended to SYK models.”
youtube / hneven / Mar 4
A new 3-year, $5 million competition has been launched by XPrize, Google, and Gesda to incentivize the development of quantum algorithms and applications. The initiative aims to accelerate the transition of quantum computing from theoretical concepts to practical solutions for complex global challenges, fostering interdisciplinary collaboration within the quantum community.
quantum-computingquantum-algorithmsinnovation-competitionsscientific-researchtechnology-trends
“Quantum computing began as a theoretical concept, emerging from the limitations of binary transistors.”
youtube / hneven / Jan 1
Hartmut Neven proposes a modification to Penrose and Hameroff's conjecture on consciousness, suggesting that conscious moments arise from the *formation* of quantum superpositions, rather than their collapse. This framework, rooted in the many-worlds interpretation, posits that conscious experience is always of a single classical reality despite a multitude of underlying quantum configurations. His ongoing research with brain organoids and xenon isotopes aims to experimentally validate the quantum nature of consciousness and develop methods for coherent quantum-biological coupling.
quantum-consciousnessquantum-biologybrain-computer-interfaceconsciousness-studiesquantum-computingmindfesthartmut-neven
“Conscious moments are created when quantum superpositions form.”
youtube / hneven / Sep 4
Hartmut Neven from Google Quantum AI proposes a research program to test the conjecture that quantum operations are necessary and sufficient to create sentience. This program is inspired by Roger Penrose and Stuart Hameroff's ideas, particularly the concept that conscious moments arise from the objective reduction of the quantum mechanical wave function. Neven aims to investigate whether quantum computers, with their unique operations like superposition and entanglement, can contribute to or expand conscious experience, in contrast to probabilistic Turing machines.
quantum-computingconsciousness-researchneurosciencequantum-mechanicsexperimental-physics
“Quantum operations are hypothesized to be necessary and sufficient for sentience.”
paper / hneven / Aug 23
This paper introduces a quantum computing protocol for calculating stopping power, a critical property for inertial fusion target design. The approach leverages first-quantized representations and non-Born-Oppenheimer dynamics to address the classical intractability of these calculations in warm-dense conditions. The authors provide estimates for logical qubit requirements and Toffoli costs, demonstrating the potential for quantum computers to perform scientifically significant and classically challenging simulations.
quantum-computinginertial-fusionplasma-physicsstopping-powerfirst-principles-calculationquantum-simulationthermodynamic-modeling
“Stopping power calculations are classically challenging due to complex electronic system dynamics and non-equilibrium conditions, especially in warm-dense plasmas.”
paper / hneven / Aug 4
Achieving fault-tolerant quantum computing necessitates scaling quantum gates without exceeding error thresholds. This work introduces a control optimization strategy that effectively manages the complexity of scaling quantum gates in large processors. By choreographing frequency trajectories of 68 superconducting qubits, the method significantly reduces physical error rates, demonstrating a pathway to build large-scale fault-tolerant quantum computers.
quantum-computingquantum-error-correctionquantum-controlsuperconducting-qubitsfault-tolerance
“Scaling quantum gates in large processors traditionally faces challenges due to non-convex, high-constraint, and time-dependent control optimization problems with exponentially expanding configuration spaces.”
paper / hneven / Jun 15
This research investigates the quantum dynamics of magnetization in a 1D Heisenberg spin chain using a 46-superconducting qubit array. While the first two moments of magnetization transfer probability distribution initially suggest Kardar-Parisi-Zhang (KPZ) universality, higher-order moments (third and fourth) empirically contradict this conjecture. The findings emphasize the critical role of analyzing higher moments to accurately classify dynamic universality classes in quantum systems.
quantum-physicsheisenberg-modelspin-chainmagnetizationkpz-universalitysuperconducting-qubitsquantum-systems
“The first two moments of the magnetization transfer probability distribution in a 1D Heisenberg spin chain exhibit superdiffusive behavior.”
paper / hneven / Mar 27
A recent comment reinforces that microscopic mechanisms in experimentally observed quantum teleportation, specifically size winding, thermalization, and scrambling at the time of teleportation, are consistent with a gravitational interpretation of the early-time teleportation dynamics. The objections raised in a prior comment are considered counterfactual scenarios outside the scope of the real experiment.
quantum-physicshigh-energy-physicswormhole-dynamicsquantum-teleportationarxiv
“The microscopic mechanism of experimentally observed teleportation involves size winding.”
paper / hneven / Jan 3
First-quantized quantum algorithms enable exact time evolution of electronic systems with exponential space and polynomial operation advantages over classical real-time time-dependent Hartree-Fock and DFT. Challenges in observable sampling are mitigated by polylogarithmic scaling for k-particle reduced density matrix estimation. This suggests a potential for quantum advantage in finite temperature simulations and specific electron dynamics problems.
quantum-simulationelectron-dynamicsquantum-algorithmsmean-field-methodsquantum-advantagequantum-computing
“Quantum algorithms for simulating electronic ground states are slower than classical mean-field methods but offer higher accuracy.”
blog / hneven / Nov 30 / failed
Google researchers simulated a traversable wormhole's dual CFT on the Sycamore quantum processor using AdS/CFT correspondence, observing a particle traverse via negative energy shockwave. A sparse quantum model was trained with ML techniques like backpropagation and sparsification to capture gravitational dynamics on limited qubits, verifying causal consistency classically. The experiment detected slight asymmetry in information transfer between negative and positive shockwaves, reliant on Sycamore's low error rates; higher noise obscured the signal.
quantum-computingwormholesads-cftholographic-principlequantum-gravitysycamore-processorgoogle-quantum-ai
“The Google Sycamore processor simulated a quantum system dual to a traversable wormhole, allowing a particle to emerge on the other side.”
paper / hneven / Nov 9
Quantum error correction (QEC) is a critical bottleneck for scalable quantum computing due to quantum information leakage from computational states. This work demonstrates a method to effectively remove leakage from all qubits in each cycle of QEC circuits. This technique significantly reduces leakage population and prevents correlated errors, making significant progress towards fault-tolerant quantum computation.
quantum-error-correctionquantum-computingfault-tolerancequantum-informationsycamore-processor
“Leakage of quantum information from computational states poses a major challenge to scalable quantum error correction (QEC).”
paper / hneven / Oct 19
Researchers implemented a generalized stabilizer code on a superconducting processor to create and braid projective non-Abelian Ising anyons. By utilizing unitary gates to manipulate the many-body wavefunction, they experimentally verified anyonic fusion rules and realized braiding statistics to encode an entangled state of three logical qubits. This demonstrates a path toward fault-tolerant quantum computing through the integration of these topological properties with active error correction.
quantum-computingquantum-physicssuperconducting-processorsnon-abelian-anyonsquantum-braidingtopological-quantum-computation
“Non-Abelian anyons can be created and braided on a superconducting quantum processor using a generalized stabilizer code.”
paper / hneven / Jul 13
This paper presents the first experimental demonstration where quantum error correction improves performance as the number of qubits increases. The research utilizes superconducting qubits to show that a distance-5 surface code logical qubit outperforms an ensemble of distance-3 logical qubits. This advancement is crucial for developing practical quantum computers by mitigating errors effectively.
quantum-error-correctionquantum-computing-hardwaresuperconducting-qubitslogical-qubitssurface-codesquantum-information
“Quantum error correction can improve performance with an increased number of qubits if the error density is sufficiently low.”
paper / hneven / Jun 10
Researchers used a 24-superconducting qubit ring to implement the spin-1/2 XXZ model, demonstrating robust bound states of up to five interacting microwave photons. This experimental setup allowed for the study of photon propagation and the resilience of these bound states even when integrability was broken, challenging previous assumptions about their stability under such conditions. The study introduces a phase-sensitive method to construct the few-body spectrum and extract pseudo-charge via synthetic flux, providing crucial insights into strongly correlated systems.
quantum-physicsmicrowave-photonsbound-statessuperconducting-qubitsquantum-computationcondensed-matter
“Robust bound states of up to five interacting microwave photons can be formed.”
paper / hneven / May 17
Traditional artificial intelligence systems lack the intrinsic regulatory mechanisms found in living organisms. This research introduces a novel homeostatic neural network architecture where the computing substrate is dynamically influenced by the outcomes of its own computations. This inherent vulnerability and self-regulation mechanism paradoxically lead to enhanced adaptability, particularly under conditions of significant concept shift. The framework demonstrates that integrating "skin in the game" for AI agents can lead to superior performance in dynamic and unpredictable environments.
homeostasis-neural-networksconcept-shiftmeta-learningcontinual-learningartificial-intelligence
“Artificial neural networks can be designed with homeostatic features, allowing their internal states and learning rates to be regulated by the outcomes of their computations.”
paper / hneven / Apr 24
Researchers implemented a one-dimensional kicked Ising model using 47 superconducting qubits to demonstrate the existence of non-local Majorana edge modes (MEMs) protected by $\mathbb{Z}_2$ parity symmetry. The system exhibits a uniform late-time decay rate for multi-qubit Pauli operators overlapping with MEMs, regardless of operator size, and demonstrates resilience to symmetry-breaking noise via prethermalization.
quantum-physicssuperconducting-qubitsnoise-resiliencemajorana-edge-modesquantum-computing-research
“Majorana edge modes (MEMs) in the one-dimensional kicked Ising model exhibit a uniform late-time decay rate across multi-qubit Pauli operators.”
paper / hneven / Dec 1
Quantum technology, specifically quantum computation and memory, can significantly improve experimental data processing. This approach minimizes the number of experiments needed compared to classical methods. The advantage is substantial across various tasks, including predicting system properties and quantum principal component analysis, achievable even with current noisy quantum processors.
quantum-computingmachine-learningquantum-advantagequantum-experimentationphysical-systemsnoisy-processorsdata-processing
“Quantum machines can learn from exponentially fewer experiments than classical methods.”
paper / hneven / Nov 26
This paper introduces Nonequilibrium Monte Carlo (NMC), a quantum-inspired algorithm designed to overcome critical slowing down in hard combinatorial optimization problems. NMC employs an adaptive, gradient-free approach to learn cost function features and generate spatially inhomogeneous thermal fluctuations, thereby unfreezing variables and preventing costly exploration-exploitation trade-offs. The method demonstrates significant speedups and robustness in solving complex problems like k-SAT and Quadratic Assignment Problems.
combinatorial-optimizationmonte-carloquantum-inspired-algorithmsk-satisfiability
“NMC algorithms unfreeze variables by employing spatially inhomogeneous thermal fluctuations.”
paper / hneven / Oct 20
Typical stochastic solvers for optimization problems often suffer from mode collapse, hindering the discovery of diverse solutions. This work introduces a "time-to-diversity" (TTD) metric to quantify solver performance in sampling approximate solutions for NP-hard problems. The research demonstrates that inhomogeneous quantum annealing schedules, by controlling critical fronts, significantly improve both TTD and traditional "time-to-solution" (TTS) compared to standard schedules, especially for rare solutions. This approach leverages non-equilibrium driving of quantum fluctuations to reduce hard instances and increase solution diversity.
quantum-annealingoptimization-problemsnp-hard-problemsquantum-physicsstatistical-mechanicsmonte-carlo-simulationsalgorithmic-quantum-phase-transitions
“Traditional stochastic solvers suffer from mode collapse, limiting solution diversity for hard optimization problems.”
paper / hneven / Sep 23
A novel classical sampling algorithm has been developed that surpasses the accuracy of recent experimental Gaussian Boson Sampling (GBS) implementations. This method achieves better total variation distance and Kullback-Leibler divergence with a computational cost quadratic in the number of modes. The approach approximates single-mode and two-mode ideal marginals by setting Boltzmann machine parameters via a mean field solution, offering a second-order approximation that improves upon uniform and thermal approximations.
quantum-computinggaussian-boson-samplingclassical-algorithmsquantum-simulationquantum-physics
“A new classical sampling algorithm achieves better total variation distance and Kullback-Leibler divergence than recent experimental Gaussian Boson Sampling (GBS) implementations.”
paper / hneven / Sep 5
This paper introduces a quantum algorithm to infer molecular nuclear spin Hamiltonians from time-resolved Nuclear Magnetic Resonance (NMR) data. The method focuses on learning the anisotropic dipolar term, which is classically challenging to simulate. It demonstrates direct estimation of the Jacobian and Hessian on quantum computers for parameter learning, applicable to both noisy near-term and fault-tolerant quantum devices.
quantum-computingmolecular-structurenuclear-magnetic-resonancequantum-algorithmsspin-hamiltonianmany-body-physics
“A quantum algorithm can infer molecular nuclear spin Hamiltonians from time-resolved NMR measurements of spin-spin correlators.”
paper / hneven / Jul 28
This work presents the experimental observation of a discrete time crystal (DTC) with eigenstate order leveraging a superconducting quantum processor. The study successfully distinguishes stable eigenstate-ordered phases from transient phenomena using a time-reversal protocol and quantum typicality, thereby overcoming the challenges of experimentally identifying stable dynamical phases. They also located the phase transition out of the DTC using finite-size analysis, establishing a scalable method for exploring non-equilibrium phases of matter on current quantum computers.
quantum-computingtime-crystalssuperconducting-qubitsquantum-many-body-systemsnon-equilibrium-physicsquantum-processors
“Discrete Time Crystals (DTCs) are out-of-equilibrium dynamical phases of matter.”
paper / hneven / Jul 19
A novel quantum algorithm is proposed for training wide and deep classical neural networks, potentially offering an exponential speedup over classical gradient descent. This method leverages the efficiency of quantum linear system solvers, achieving O(log n) training time for a dataset of size n, provided efficient data loading and readout. The approach shows promise for accelerating deep learning applications.
quantum-machine-learningneural-networksquantum-algorithmsdeep-learningmnist-datasetsparse-matrix-inversion
“The quantum algorithm trains wide and deep neural networks in O(log n) time.”
paper / hneven / Apr 30
The Entangling Quantum Generative Adversarial Network (EQ-GAN) introduces a novel quantum GAN architecture that leverages entangling operations between generator output and true quantum data. This design ensures convergence to a Nash equilibrium during minimax optimization, addressing limitations of prior quantum GANs. EQ-GAN experimentally demonstrates robustness against coherent errors and effectively learns quantum state representations on a Google Sycamore processor, enabling applications like approximate QRAM and training quantum neural networks.
quantum-machine-learninggenerative-adversarial-networksquantum-computingquantum-circuitsquantum-neural-networksquantum-random-access-memory
“EQ-GANs overcome limitations of previous quantum GANs.”
paper / hneven / Apr 22
Neven, Read, and Rees propose a physicalist framework for consciousness that links subjective experience to agency, arguing that a system with genuine preferences would select outcomes aligned with homeostasis — manifesting as observable forces in third-person physical descriptions. They leverage Knightian uncertainty in large qubit systems — where not even probabilities can be assigned to outcomes — as the mechanism that grants engineered systems the freedom to act on preferences in ways unpredictable to outside observers. This leads to a proposed three-part design for a quantum-processor-powered "animat" that the authors argue could plausibly be considered conscious, agentive, and capable of feelings.
quantum-physicsroboticsconsciousnessagencynon-determinismhuman-robot-interaction
“Agency implementing preferences manifests as a measurable force in a third-person physical theory, with strong preferences producing deterministic evolution and weak preferences producing non-deterministic evolution.”
paper / hneven / Apr 12
Cosmic rays and latent radioactivity induce catastrophic error bursts in large arrays of superconducting qubits. These events generate high-energy phonons causing quasiparticle bursts that destroy qubit coherence across the chip. This phenomenon, which limits the scalability of quantum computing, was directly observed and characterized using a novel rapid space and time-multiplexed measurement method.
quantum-computingsuperconducting-qubitscosmic-rayserror-correctionquantum-hardwarequantum-information
“Cosmic rays and latent radioactivity cause catastrophic error bursts in superconducting qubit arrays.”
paper / hneven / Feb 11
This paper demonstrates exponential suppression of bit or phase-flip errors in superconducting qubits using 1D repetition codes and a 2D surface code. The logical error rate was reduced by over 100 times when scaling from 5 to 21 qubits, and this suppression remained stable over 50 rounds of correction. The findings validate superconducting qubits as a viable platform for fault-tolerant quantum computing development by addressing key challenges in QEC implementation and error characterization.
quantum-computingquantum-error-correctionsuperconducting-qubitsqec-codesfault-tolerant-quantum-computingquantum-physics
“Logical errors in quantum computing can be exponentially suppressed by increasing the number of physical qubits when using Quantum Error Correction (QEC).”
paper / hneven / Jan 23
This paper demonstrates the successful application of a quantum kernel method for high-dimensional data analysis using Google's Sycamore quantum processor. The research showcases the feasibility of quantum machine learning with real, high-dimensional data without classical pre-processing, addressing the challenge of vanishing kernel elements. The study utilizes 17 qubits to classify 67-dimensional cosmological supernova data, achieving competitive accuracy with classical techniques.
quantum-machine-learninghigh-dimensional-datanoisy-quantum-processorsupernova-classificationquantum-kernel-methodssycamore-processorcosmology
“A quantum kernel method can successfully classify high-dimensional data without dimensionality reduction on a noisy quantum processor.”
paper / hneven / Jan 21
This paper experimentally investigates quantum scrambling on a 53-qubit processor, differentiating between operator spreading and operator entanglement. It demonstrates that while operator spreading can be modeled efficiently classically, operator entanglement requires exponentially scaled computational resources, highlighting the utility of near-term quantum processors for complex physical observables.
quantum-scramblingquantum-circuitsquantum-informationquantum-processorsoperator-entanglementout-of-time-order-correlators
“Quantum scrambling involves the spread of initially localized quantum information into the many degrees of freedom of a quantum system.”
youtube / hneven / Jan 1
Google Quantum AI researchers detail diverse career journeys into quantum computing, from astrophysics and startups to theory internships, emphasizing opportunities for non-experts in classical engineering, CS, and support roles. They outline stack-specific debugging—from chip microscopy and automated calibrations to minimal reproducible examples, classical simulation of small instances, and verification of known solutions. Key challenges include achieving sub-threshold error rates for surface code QEC (currently ~1 error per 1000 operations vs. classical's 1 per year), scaling fabrication beyond wafer sizes, control systems for millions of qubits, and developing scalable algorithms amid correlated errors, with TensorFlow Quantum enabling 70+ ML-quantum publications.
quantum-computinggoogle-quantum-aiquantum-error-correctionquantum-hardwarediversity-outreachquantum-educationtechnical-challenges
“Google Quantum AI superconducting qubits exhibit roughly one error per thousand operations.”
youtube / hneven / Jan 1
Google Quantum AI is expanding its Santa Barbara campus, doubling team size to over 100 hires, and targeting milestone 2 on its roadmap to a million physical qubit machine by demonstrating error suppression (lambda >1) in surface codes via improved CNOT fidelity (0.2%), repetitive readout, and reduced crosstalk. In NISQ, they near beyond-classical out-of-time-order correlator (OTOC) computations at 420 iSWAPs with 2x gate fidelity gains, validate random circuit sampling's exponential classical hardness, and offer cloud access to 50+ qubit processors. Speculative applications span fusion reactor design via BBGKY hierarchies, isotopolog pharmacophore tuning, wide-layer neural network training speedups, and quantum GANs for enhanced graphics.
quantum-computinggoogle-quantum-aiquantum-error-correctionsurface-codenisi-eraquantum-algorithmshiring-expansion
“Google plans to double team size in two years by hiring over 100 employees across diverse quantum and engineering backgrounds”
paper / hneven / Nov 9
Achieving quantum advantage with early-generation fault-tolerant quantum computers (FTQCs) necessitates algorithms with at least quartic polynomial speedups. Quadratic speedups are insufficient due to the substantial constant factor overheads associated with current error-correction techniques, specifically the surface code. This challenge persists even with significant improvements in logical gate rates.
quantum-computingerror-correctionquantum-advantagequantum-algorithmsfault-tolerancesurface-codesquantum-physics
“For modest fault-tolerant quantum computers to achieve a runtime advantage, algorithms need more than small polynomial speedups.”
paper / hneven / Nov 3
This Nature Communications paper establishes a rigorous theoretical framework for evaluating quantum advantage in machine learning, showing that classical models equipped with sufficient training data can match quantum models even on problems tailored to quantum computation. The authors derive tight prediction error bounds that are both asymptotically rigorous and empirically predictive. To recover a clear quantum edge, they propose a "projected quantum model" that guarantees a provable speed-up in the fault-tolerant regime, and demonstrate a meaningful near-term advantage on engineered datasets at up to 30 qubits.
quantum-machine-learningquantum-computingclassical-machine-learningquantum-advantageprediction-error-boundsfault-tolerant-quantum-computingnear-term-quantum
“Classical machine learning models, when given access to data, can be competitive with quantum models even on problems specifically designed for quantum computation.”
paper / hneven / Oct 22
Quantum computers can sample output distributions exponentially faster than classical computers. This paper investigates whether similar exponential separation exists in quantum sample learning. It evaluates various generative models for learning quantum data, demonstrating the superior performance of LSTM and identifying exponential complexity growth for other models.
quantum-computingmachine-learninggenerative-modelscomputational-complexitydeep-learninglstm
“Quantum computers offer an exponential speedup in sampling output distributions compared to classical computers.”
paper / hneven / Oct 15
Researchers successfully simulated the one-dimensional Fermi-Hubbard model on a 16-qubit superconducting quantum processor. This experiment observed the separation of charge and spin dynamics in a highly excited regime, phenomena beyond conventional quasiparticle descriptions. The work highlights a viable approach to quantum simulating strongly correlated systems, mitigating errors through advanced gate calibration and error-mitigation techniques.
quantum-simulationfermi-hubbard-modelsuperconducting-qubitserror-mitigationquantum-computingquantum-physicsstrongly-correlated-systems
“The one-dimensional Fermi-Hubbard model can be simulated on a 16-qubit superconducting quantum processor.”
paper / hneven / Aug 19
This paper introduces a framework for understanding quantum optimization algorithms by connecting them to quantum system dynamics, quantum walks, and classical continuous relaxations. It uses 'kinetic energy' on a graph to analyze the mechanisms behind algorithmic success and failure. The authors identify specific pitfalls stemming from kinetic energy opposing optimization goals, such as wavefunction confinement and phase randomization, and propose new methods for algorithmic improvement.
quantum-optimizationquantum-algorithmsquantum-mechanismsquantum-walksqaoakinetic-energygraph-laplacian
“Quantum optimization algorithms can be intuitively understood through connections to quantum system dynamics, quantum walks, and classical continuous relaxations.”
paper / hneven / Jul 14
This paper evaluates the practicality of various quantum heuristic algorithms for combinatorial optimization on near-term fault-tolerant quantum computers. The authors compile and optimize circuits for several algorithms, including quantum accelerated simulated annealing and the quantum approximate optimization algorithm. Their findings suggest that achieving a practical quantum advantage over classical algorithms, even for problems with quadratic speedup, will require substantial advancements in surface code implementations and resource budgets.
quantum-computingfault-tolerant-quantum-computingcombinatorial-optimizationquantum-algorithmsquantum-simulated-annealingquantum-approximate-optimization
“Quantum optimization heuristics providing only a quadratic speedup are unlikely to outperform classical algorithms on current modest superconducting qubit surface code processors.”
paper / hneven / Jun 8
The Snake Optimizer is a new algorithm designed to overcome the limitations of traditional global optimizers in calibrating quantum processors. It leverages AI, dynamic programming, and graph optimization to efficiently solve complex, non-convex, high-dimensional, and highly constrained optimization problems inherent in quantum computing. This optimizer has demonstrated state-of-the-art performance, notably contributing to quantum supremacy on a 53-qubit processor, and exhibits favorable scaling for larger systems.
quantum-computingoptimization-algorithmsquantum-processorsmachine-learningsuperconducting-qubitsquantum-supremacy
“Traditional global optimizers are inefficient for quantum processor calibration due to the complexity of the optimization problems.”
paper / hneven / Apr 8
This paper evaluates the Quantum Approximate Optimization Algorithm (QAOA) on the Google Sycamore processor for both planar (hardware-native) and non-planar graph problems. It demonstrates that while QAOA shows promising results for problems mapped directly to the hardware's connectivity, its performance significantly degrades when problems require extensive compilation to fit the hardware architecture. This highlights a key challenge for current near-term quantum computers in addressing real-world optimization problems that often lack planar graph structures.
quantum-computingqaoasuperconducting-qubitscombinatorial-optimizationquantum-algorithmshardware-software-co-design
“The Google Sycamore superconducting qubit quantum processor can apply the Quantum Approximate Optimization Algorithm (QAOA) to combinatorial optimization problems.”
paper / hneven / Apr 8
This paper details the execution of Hartree-Fock (HF) calculations for fermionic systems on a superconducting quantum computer. The authors successfully simulated the binding energy of hydrogen chains (H6-H12) and diazene isomerization using a 12-qubit system with a variational Givens rotation ansatz. Crucially, they employed N-representability-based error mitigation techniques to significantly enhance experimental fidelity, demonstrating a foundation for scaling complex correlated quantum simulations.
quantum-computingquantum-chemistrysuperconducting-qubitshartree-fockerror-mitigationfermionic-systemsquantum-simulation
“Variational simulations of fermionic systems are a promising application for noisy intermediate-scale quantum (NISQ) devices.”
paper / hneven / Mar 6
TensorFlow Quantum (TFQ) is an open-source library designed for rapid prototyping of hybrid quantum-classical machine learning models. It provides high-level abstractions for designing and training both discriminative and generative quantum models within the TensorFlow ecosystem, including support for high-performance quantum circuit simulators. The framework enables exploration of diverse quantum learning tasks, aiming to facilitate research in quantum computing and machine learning.
tensorflow-quantumquantum-machine-learninghybrid-quantum-classical-modelsquantum-computingmachine-learning-frameworksquantum-algorithms
“TensorFlow Quantum (TFQ) is an open-source library for hybrid quantum-classical machine learning.”
paper / hneven / Dec 9
This work addresses the challenge of characterizing non-Markovian quantum noise originating from two-level system (TLS) defects in amorphous dielectrics, which are a major source of decoherence in solid-state qubits. It introduces a novel gate-dependent error model for SWAP-like two-qubit gates in frequency-tunable Transmon qubits. The research combines Moiré-enhanced swap spectroscopy with deep evolutionary algorithms and neural networks to accurately learn TLS parameters from noisy experimental data, improving upon existing methods for quantum control optimization.
quantum-computingmachine-learningquantum-noisequbit-characterizationevolutionary-algorithmsquantum-error-correction
“Two-level system (TLS) defects in amorphous dielectrics are a significant source of noise and decoherence in solid-state qubits.”
paper / hneven / Nov 8
Quantum annealing (QA) exhibits a significant scaling advantage over path-integral Monte Carlo (PIMC) for simulating geometrically frustrated magnets. This advantage, observed on a 1440-qubit system, is especially pronounced with increasing system size and decreasing temperature. The findings suggest that near-term quantum devices can provide practical acceleration for complex computational tasks in condensed matter physics.
quantum-computingquantum-simulationgeometrically-frustrated-magnetsquantum-annealingcomputational-advantagequantum-materials
“Quantum annealing (QA) demonstrates a dynamical advantage over Path-Integral Monte Carlo (PIMC) in simulating geometrically frustrated magnets.”
paper / hneven / Oct 23
This paper introduces an optimal fermion-to-qubit mapping utilizing ternary trees, which significantly reduces the number of qubits required for representing fermionic systems. This method enables more efficient learning of k-fermion reduced density matrices (RDMs) in quantum simulations by reducing the quantum circuit repetitions needed to determine RDM elements. The technique also improves upon existing qubit RDM determination schemes by making the process independent of system size.
fermion-to-qubit-mappingternary-treesquantum-states-learningreduced-density-matrixquantum-simulationquantum-computingquantum-physics
“A novel fermion-to-qubit mapping is introduced, based on ternary trees.”
paper / hneven / Oct 23
This document provides updated supplementary information for the "Quantum supremacy using a programmable superconducting processor" article published in Nature. Key updates include the addition of the qFlex source code URL, an Erratum section, a new figure (S41) comparing statistical and total uncertainty for log and linear XEB, new references, and various minor clarifications and formatting corrections. The original article details a significant milestone in quantum computing where a programmable superconducting processor demonstrated the ability to perform a computational task beyond the capabilities of classical supercomputers, as detailed in the referenced Nature publication.
quantum-computingquantum-supremacysuperconducting-processorsquantum-physicsscientific-articlegoogle-ai
“The document is an updated version of supplementary information for a Nature article on quantum supremacy.”
blog / hneven / Oct 23 / failed
Google's Sycamore quantum processor performed a specific computation in 200 seconds that would require 10,000 years on the world's fastest supercomputer, demonstrating quantum supremacy as published in Nature. This milestone leverages quantum mechanics to tackle problems intractable for classical computers, such as molecular simulation for drug discovery and battery design. The achievement advances research through open-sourced tools like Cirq, enabling broader exploration of quantum applications.
quantum-computingquantum-supremacysycamore-chipgoogle-aiquantum-researchnature-paper
“Sycamore performed a computation in 200 seconds that would take the world’s fastest supercomputer 10,000 years.”
paper / hneven / Jul 11
Quantum Neural Networks (QNNs) face challenges with parameter initialization for rapid convergence. This work explores using classical neural networks (meta-learning) to optimize the parameter landscape of quantum variational algorithms. This approach significantly reduces optimization iterations and generalizes across problem sizes, enabling training on smaller instances and applying to larger, intractable problems on quantum devices.
quantum-neural-networksmeta-learningquantum-algorithmshybrid-quantum-classicaloptimization-strategies
“Classical neural networks can assist in optimizing quantum neural networks.”
paper / hneven / Jul 2
This paper introduces a novel structural property in quantum spin glasses, termed asymptotic orthogonality. This property results in an eigen-spectrum divided into alternating bands of two state types (x and z), each exhibiting distinct non-ergodic extended eigenstates (NEEs). The discovery suggests new quantum search algorithms by leveraging these NEE bands for population transfer in the spin-configuration space.
quantum-spin-glassn P-hard-problemsquantum-algorithmseigen-spectrumnon-ergodic-extended-eigenstatesmany-body-localizationrandom-energy-model
“Quantum spin glasses exhibit asymptotic orthogonality, leading to a split eigen-spectrum.”