Large-scale Calculations Research Articles

Residual-ZZ-coupling suppression and fast two-qubit gate for Kerr-cat qubits based on level-degeneracy engineering

Building large-scale quantum computers requires an interqubit-coupling scheme with a high on–off ratio to avoid unwanted crosstalk coming from residual coupling and to enable fast multi-qubit operations. We propose a ZZ-coupling scheme for two Kerr-cat qubits with a frequency-tunable coupler. By making four relevant states of the two Kerr-cat qubits quadruply degenerate, we can switch off the ZZ coupling. By partially lifting the level degeneracy, we can switch it on. We theoretically show that an experimentally feasible circuit model suppresses the residual ZZ coupling. Moreover, our circuit can realize RZZ(−π/2)-gate fidelity higher than 99.9% within 18 ns when decoherence is ignored. Our model includes the first-order terms in expansion beyond the rotating-wave approximation.

Noncollinear phase of the antiferromagnetic sawtooth chain

The antiferromagnetic sawtooth chain is a prototypical example of a frustrated spin system with vertex-sharing triangles, giving rise to complex quantum states. Depending on the interaction parameters, this system has three phases, of which the gapless noncollinear phase (for strongly coupled basal spins and loosely attached apical spins) has received little theoretical attention so far. In this work, we comprehensively investigate the properties of the noncollinear phase using large-scale tensor network computations which exploit the full SU(2) symmetry of the underlying Heisenberg model. We study the ground state both for finite systems using the density-matrix renormalization group (DMRG) as well as for infinite chains via the variational uniform matrix-product state (VUMPS) formalism. Finite temperatures and correlation functions are tackled via imaginary or real time evolutions, which we implement using the time-dependent variational principle (TDVP). We find that the noncollinear phase is characterized by a low-momentum peak and a diffuse tail for the apex-apex correlations. Deep into the phase, the pattern sharpens into a peak indicating a 90∘ spiral. The apical spins are soft and highly susceptible to external perturbations; they give rise to a large number of gapless magnetic states that are polarized by weak fields and cause a long low-temperature tail in the specific heat. The dynamic spin-structure factor exhibits additive contributions from a two-spinon continuum (excitations of the basal chain) and a gapless peak at k=π/2 (excitations of the apical spins). Small temperatures excite the gapless states and smear the spectral weight of the k=π/2 peak out into a homogeneous flat-band structure. Our results are relevant, e.g., for the material atacamite Cu2Cl(OH)3 in high magnetic fields. Published by the American Physical Society 2025

Quantum machine learning with Adaptive Boson Sampling via post-selection

The implementation of large-scale universal quantum computation represents a challenging and ambitious task on the road to quantum processing of information. In recent years, an intermediate approach has been pursued to demonstrate quantum computational advantage via non-universal computational models. A relevant example for photonic platforms has been provided by the Boson Sampling paradigm and its variants, which are known to be computationally hard while requiring at the same time only the manipulation of the generated photonic resources via linear optics and detection. Beside quantum computational advantage demonstrations, a promising direction towards possibly useful applications can be found in the field of quantum machine learning, considering the currently almost unexplored intermediate scenario between non-adaptive linear optics and universal photonic quantum computation. Here, we report the experimental implementation of quantum machine learning protocols by adding adaptivity via post-selection to a Boson Sampling platform based on universal programmable photonic circuits fabricated via femtosecond laser writing. Our experimental results demonstrate that Adaptive Boson Sampling is a viable route towards dimension-enhanced quantum machine learning with linear optical devices.

Dual-Grid and Mixed-Precision Methods for Accelerating Plane-Wave Hybrid Functional Electronic Structure Calculations.

Hybrid functionals that incorporate exact Hartree-Fock exchange (HFX) into density functional theory (DFT) are crucial for accurately predicting the electronic structures of extended systems in condensed-matter physics and materials science. Although the exact exchange contributes only a small fraction of the total energy, HFX calculations in hybrid functionals demand significant computational resources. Here, we introduce dual-grid and mixed-precision techniques, based on two low-rank approximations, adaptively compressed exchange (ACE) and interpolative separable density fitting (ISDF) methods, to significantly improve the computational efficiency of plane-wave hybrid functional calculations in the software package PWDFT (plane wave density functional theory). The dual-grid method introduces a smaller cutoff energy, thereby reducing the number of Fourier and real-space grids involved in the HFX calculations. The mixed-precision method reduces the computational cost by shifting from double precision to single precision during the HFX construction while maintaining double precision for other DFT processes. Numerical results demonstrate that both the dual-grid and mixed-precision techniques can accelerate plane-wave hybrid functional calculations several times, with an acceptable trade-off in accuracy compared to the original low-rank approximations. Furthermore, by utilizing a hybrid MPI and OpenMP parallel implementation, we successfully perform large-scale plane-wave hybrid functional calculations with up to 8,000 silicon atoms using 16,000 CPU cores.

Silicon photonic convolution operator exploiting on-chip nonlinear activation function.

Nonlinear activation functions (NAFs) are essential in artificial neural networks, enhancing learning capabilities by capturing complex input-output relationships. However, most NAF implementations rely on additional optoelectronic devices or digital computers, reducing the benefits of optical computing. To address this, we propose what we believe to be the first implementation of a nonlinear modulation process using an electro-optic IQ modulator on a silicon photonic convolution operator chip as a novel NAF. We validated this operator by constructing a convolutional neural network for radio machine learning classification, achieving 92.5% accuracy-an improvement of 27% over the case without a NAF. Compared with optoelectronic systems that rely on separate components, this fully integrated silicon photonic chip allows the NAF to execute nearly synchronously with the convolution operation, significantly lowering latency and reducing the complexity of the peripheral control system. This work paves the way for a large-scale on-chip optical neural network computation.

Rendering of transparent objects in large-scale full-parallax polygon-based computer holography.

A rendering technique is proposed for creating full-parallax, large-scale computer-generated holograms (CGHs) that can reconstruct a three-dimensional (3D) scene, including transparent objects. The proposed method enables us to simulate optical refraction based on wave optics without relying on Snell's law. Several techniques are also presented to apply the method to creating large-scale CGHs by reducing the computation time. The validity of the proposed techniques is confirmed by fabricating an actual full-parallax, large-scale CGH.

Quick-and-Easy Validation of Protein-Ligand Binding Models Using Fragment-Based Semiempirical Quantum Chemistry.

Electronic structure calculations in enzymes converge very slowly with respect to the size of the model region that is described using quantum mechanics (QM), requiring hundreds of atoms to obtain converged results and exhibiting substantial sensitivity (at least in smaller models) to which amino acids are included in the QM region. As such, there is considerable interest in developing automated procedures to construct a QM model region based on well-defined criteria. However, testing such procedures is burdensome due to the cost of large-scale electronic structure calculations. Here, we show that semiempirical methods can be used as alternatives to density functional theory (DFT) to assess convergence in sequences of models generated by various automated protocols. The cost of these convergence tests is reduced even further by means of a many-body expansion. We use this approach to examine convergence (with respect to model size) of protein-ligand binding energies. Fragment-based semiempirical calculations afford well-converged interaction energies in a tiny fraction of the cost required for DFT calculations. Two-body interactions between the ligand and single-residue amino acid fragments afford a low-cost way to construct a "QM-informed" enzyme model of reduced size, furnishing an automatable active-site model-building procedure. This provides a streamlined, user-friendly approach for constructing ligand binding-site models that requires neither a priori information nor manual adjustments. Extension to model-building for thermochemical calculations should be straightforward.

On Defining Expressions for Entropy and Cross-Entropy: The Entropic Transreals and Their Fracterm Calculus.

Classic formulae for entropy and cross-entropy contain operations x0 and log2x that are not defined on all inputs. This can lead to calculations with problematic subexpressions such as 0log20 and uncertainties in large scale calculations; partiality also introduces complications in logical analysis. Instead of adding conventions or splitting formulae into cases, we create a new algebra of real numbers with two symbols ±∞ for signed infinite values and a symbol named ⟂ for the undefined. In this resulting arithmetic, entropy, cross-entropy, Kullback-Leibler divergence, and Shannon divergence can be expressed without concerning any further conventions. The algebra may form a basis for probability theory more generally.

Quantum Assisted Blockchain Security Model Using Artificial Intelligence to Reduce Quantum Attacks

Presently, smart sensors ensure commercial decisions where integrated electronic systems can be securely organized using blockchain and quantum computing because of their unique characteristics and features. In the current scenario, large-scale quantum computers can be built in which most current cryptographic systems can be hacked. Since digital and quantum computers can conduct computations simultaneously, a quantum tool for blockchain framework design is required. Based on these concerns in this research, an enhanced quantum-assisted blockchain security model using the artificial intelligence (EQ-BSM-AI) technique has been proposed. This model validates cryptosystems and blockchain technologies to determine their vulnerability to quantum attacks. Further, in this model, quantum assisted edge computing technique has been used to model the Human-centric Internet of Things (HIoT) system by introducing a quantum key generation process. Based on the post-quantum blockchain (PQB), a secured cryptosystem that is highly resistant to quantum computer attacks has been introduced in this research. This quantum channel with multiple inputs and outputs (MIMO) is designed for a quantum-based communication system to make this model more efficient and withstand errors. In EQ-BSM-AI, an improved quantum encryption algorithm (IQEA) stores the keys for encryption with a generalized probability accumulation model. For the current quantum computers and communications, our proposed system resulted in an improved sampling error reduction of 12.4%, enhanced efficiency of quantum entanglement of 96.3%, information randomness of 93.9%, correlation analysis of 93.2%, and increased resistance to quantum computing attacks of 90.8% when compared with other existing approaches.

Research on the remote sensing large model for large-scale calculation of land spatial parameters

Research on the remote sensing large model for large-scale calculation of land spatial parameters

Correlated Dirac Fermions in Twisted Bilayers of MoS2.

Artificial honeycomb lattices are essential for understanding exotic quantum phenomena arising from the interplay between Dirac physics and electron correlation. This work shows that the top two moiré valence bands in rhombohedral-stacked twisted MoS2 bilayers (tb-MoS2) form a honeycomb lattice with massless Dirac fermions. The hopping and Coulomb interaction parameters are explicitly determined based on large-scale ab initio calculations. The system exhibits significant nonlocal Coulomb repulsion and can be described by the extended Hubbard model. At half-filling, strong Coulomb repulsion in free-standing tb-MoS2 drives the system away from the semimetal phase, resulting in strongly correlated Dirac fermions. By varying the twist angle and dielectric environment, the Hamiltonian parameters can be tuned in a wide range, enabling transitions between distinct quantum phases. The high tunability makes tb-MoS2 a promising simulator for exploring many-body effects of Dirac fermions.

USING NEURAL NETWORKS FOR LANDMINE DETECTION

The issue of large mined areas and their expansion is critical due to their impact on safety and the economy. This article focuses on enhancing the efficiency of optical methods for detecting surface-laid landmines in humanitarian demining tasks. It demonstrates that one of the most promising optical methods for mine detection is hyperspectral imaging, which provides exceptionally precise analysis of the spectral composition of the radiation in the background-target environment. However, practical application of hyperspectral methods is significantly complicated by the vast size of mined territories. Therefore, mine detection must be automated. To address this challenge, the potential of artificial neural networks was investigated. The study proposes the integration of artificial intelligence, specifically RBF (Radial Basis Function) networks, to improve the efficiency and accuracy of object detection. The work compares traditional classification methods with innovative machine learning approaches. Algorithms were tested on both simulated and real data, enabling an evaluation of their ability to identify objects under varying spectral content conditions. The specific characteristics of the study area define the requirements for assessing the performance of technical tools: foremost, the probability of missing a signal must be minimized. At the same time, given the large-scale computations involved, the probability of false alarms should remain low. The study shows that RBF neural networks can detect mines with a low rate of false alarms. During network training with a large spread parameter, the sensitivity of the output to the spectral variability of pixels decreases. This allows the network to detect a target even at low fill coefficients. Thus, the research results indicate that the proposed methods significantly reduce false alarms and ensure high performance under real-world conditions.

SWEMniCS: a software toolbox for modeling coastal ocean circulation, storm surges, inland, and compound flooding

Flooding from storm surges, rainfall-runoff, and their interaction into compounding events are major natural hazards in coastal regions. To assess risks of damages to life and properties alike, numerical models are needed to guide emergency responses and future assessments. Numerical models, such as ADCIRC have over many decades shown their usefulness in such assessments. However, these models have a high threshold in terms of new user engagement as development and compilation is not trivial for users trained in compiled programming languages. Here, we develop a new open-source finite element solver for the numerical simulation of flooding. The numerical solution of the underlying PDEs is developed using the finite element framework FEniCSx. The goal is a framework where new methods can be rapidly tested before time-consuming development into codes like ADCIRC. We validate the framework on several test cases, including large-scale computations in the Gulf of Mexico for Hurricane Ike (2008).

Environmental impact study of the sightseeing electric vehicle supply chain based on the B2C e-commerce model and LCA framework

Studying the impact of the electric vehicle supply chain on the environment is crucial for determining the future development direction of the industry. We have developed a method for evaluating the impact of supply chains on the environment based on a lifecycle framework. This method innovatively seeks the connection between the lifecycle process of physical products and the supply chain, and organizes the environmental impact assessment factors of the electric vehicle supply chain from three aspects: physical resources, power energy, and waste emissions, in order to construct an LCA fuzzy comprehensive evaluation model for the electric vehicle supply chain. For the first time, the research method of transforming qualitative analysis into quantitative data was introduced into the life cycle environmental impact assessment, and empirical research was conducted using the supply chain of sightseeing electric vehicles as an example. The results indicate that the scrapping stage of electric vehicles has the most severe impact on the environment. Strengthening research on strategies or technologies for handling waste batteries and automobiles is key to improving the environmental performance of the supply chain. This method breaks through the requirements and limitations of traditional life cycle assessment methods on data sources and parameters, avoids large-scale calculations that cannot be separated from subjective factors in traditional methods, simplifies the process of supply chain environmental impact assessment, shortens the evaluation time, and improves the efficiency of environmental impact assessment. It is more practical and has good application prospects.

COMBO: compressed block-wise out-of-core diffraction computation for tera-scale holography.

Generating large-scale holograms using computer-generated holography (CGH) requires vast memory resources, often exceeding available system memory. While out-of-core processing offers a solution, it introduces significant I/O bottlenecks during diffraction, a core operation in CGH. To address this challenge, we present the COMBO system, a novel out-of-core processing framework designed to accelerate large-scale diffraction computation. COMBO integrates block-wise data handling with GPU-accelerated compression to significantly enhance I/O efficiency, further optimized through the use of multiple SSDs. Experimental results show that COMBO achieves up to 4.16 times faster performance compared to prior out-of-core methods while maintaining high-quality holographic reconstructions. Additionally, we successfully generated a 256K hologram, requiring tera-scale computational space (e.g., 4TB), on a system with only 64GB of system memory.

Finding dense sublattices as low energy states of a Hamiltonian

Lattice-based cryptography has emerged as one of the most prominent candidates for postquantum cryptography, projected to be secure against the imminent threat of large-scale fault-tolerant quantum computers. The Shortest Vector Problem (SVP) is to find the shortest nonzero vector in a given lattice. It is fundamental to lattice-based cryptography and believed to be hard even for quantum computers. We study a natural generalization of the SVP known as the K-Densest Sublattice Problem (K-DSP): to find the densest K-dimensional sublattice of a given lattice. We formulate K-DSP as finding the first excited state of a Z-basis Hamiltonian, making K-DSP amenable to investigation via an array of quantum algorithms, including Grover search, quantum Gibbs sampling, adiabatic, and variational quantum algorithms. The complexity of the algorithms depends on the basis through which the input lattice is presented. We present a classical polynomial-time algorithm that takes an arbitrary input basis and preprocesses it into inputs suited to quantum algorithms. With preprocessing, we prove that O(KN2) qubits suffice for solving K-DSP for N-dimensional input lattices. We empirically demonstrate the performance of a quantum approximate optimization algorithm K-DSP solver for low dimensions, highlighting the influence of a good preprocessed input basis. We then discuss the hardness of K-DSP in relation to the SVP, to see if there is a reason to build postquantum cryptography on K-DSP. We devise a quantum algorithm that solves K-DSP with runtime exponent (5KNlog2N)/2. Therefore, for fixed K, K-DSP is no more than polynomially harder than the SVP. The central insight we use is similar in spirit to those underlying the best-known classical K-DSP algorithm due to Dadush and Micciancio. Whether the exponential dependence on K can be lowered remains an open question. Published by the American Physical Society 2024

Incorporating edge convolution and correlative self-attention into graph neural network for material properties prediction

Abstract Predicting material properties is a crucial problem for new material designs. Traditional methods either based on trail-and-error experiments or large-scale density function theory calculations have their limitations. The recent machine learning methods shed light on resolving this problem efficiently. However, most machine learning models only consider the local atomic environment, while ignoring the non-local correlations between atoms. Even the periodic patterns of crystal structure are not taken into serious consideration. These issues lead to insufficient grasping of the feature information of atoms and bonds.In this paper, we propose a crystal graph convolutional neural network based on edge convolution and correlative self-attention, namely EdgeConv-GANN. The network is able to better extract atomic and bonding feature information and also effectively learning the importance weights of all neighboring nodes. Numerical experiments on predicting electronic structural properties of metal-organic frameworks show that our model achieves state-of-the-art performance. In addition, we applied the proposed model to predict the heat capacity and thermal decomposition temperature of materials, both results show that our method generalize well in multi-scale prediction tasks.

Realizing multi-level phase-change storage by monatomic antimony

With the growing need for extensive data storage, enhancing the storage density of nonvolatile memory technologies presents a significant challenge for commercial applications. This study explores the use of monatomic antimony (Sb) in multi-level phase-change storage, leveraging its thickness-dependent crystallization behavior. We optimized nanoscale Sb films capped with a 4-nm SiO2 layer, which exhibit excellent amorphous thermal stability. The crystallization temperature ranges from 165 to 245 °C as the film thickness decreases from 5 to 3 nm. These optimized films were then assembled into a multilayer structure to achieve multi-level phase-change storage. A typical multilayer film consisting of three Sb layers was fabricated as phase-change random access memory (PCRAM), demonstrating four distinct resistance states with a large on/off ratio (∼102) and significant variation in operation voltage (∼0.5 V). This rapid, reversible, and low-energy multi-level storage was achieved using an electrical pulse as short as 20 ns at low voltages of 1.0, 2.1, 3.0, and 3.6 V for the first, second, and third SET operation, and RESET operation, respectively. The multi-level storage capability, enabled by segregation-free Sb with enhanced thermal stability through nano-confinement effects, offers a promising pathway toward high-density PCRAM suitable for large-scale neuromorphic computing.

Introducing the Second-Order Features Adjoint Sensitivity Analysis Methodology for Neural Ordinary Differential Equations—II: Illustrative Application to Heat and Energy Transfer in the Nordheim–Fuchs Phenomenological Model for Reactor Safety

This work presents an illustrative application of the newly developed “Second-Order Features Adjoint Sensitivity Analysis Methodology for Neural Ordinary Differential Equations (2nd-FASAM-NODE)” methodology to determine most efficiently the exact expressions of the first- and second-order sensitivities of NODE decoder responses to the neural net’s underlying parameters (weights and initial conditions). The application of the 2nd-FASAM-NODE methodology will be illustrated using the Nordheim–Fuchs phenomenological model for reactor safety, which describes a short-time self-limiting power transient in a nuclear reactor system having a negative temperature coefficient in which a large amount of reactivity is suddenly inserted. The representative model responses that will be analyzed in this work include the model’s time-dependent total energy released, neutron flux, temperature and thermal conductivity. The 2nd-FASAM-NODE methodology yields the exact expressions of the first-order sensitivities of these decoder responses with respect to the underlying uncertain model parameters and initial conditions, requiring just a single large-scale computation per response. Furthermore, the 2nd-FASAM-NODE methodology yields the exact expressions of the second-order sensitivities of a model response requiring as few large-scale computations as there are features/functions of model parameters, thereby demonstrating its unsurpassed efficiency for performing sensitivity analysis of NODE nets.

Joint Association, Deployment and Flight Trajectory Optimization for Multi-UAV-Enabled Large-Scale Mobile Edge Computing

Joint Association, Deployment and Flight Trajectory Optimization for Multi-UAV-Enabled Large-Scale Mobile Edge Computing