Computational Speed Research Articles

A platinum degradation reaction model for the high-speed computation in the integrated fuel cell system simulator and its validation by the accelerated stress test data of the commercial fuel cells

A model to estimate the Pt degradation rate and the numerical methods for high-speed computation were developed for the year-long and system-scale durability simulation in an acceptable accuracy and computational time. The parameters in the model were determined to reproduce the published accelerated stress test (AST) data with the 1 cm2 MEA of 2nd-generation MIRAI (MIRAI-2), state-of-the-art commercial fuel cell electric vehicle. Accuracy of the simulation results was validated by the published data of the ASTs with the FC stack having 13 cells of MIRAI-2, which consists of the 30000 startup-shutdown cycles in 0–0.88 V and 73000 acceleration-deceleration cycles in 0.6–0.9 V. The experimental and simulation results of the residual fractions of the electrochemical surface area in the cathode catalyst layer after the ASTs were 68 and 69 %, respectively, and they agreed in an acceptable accuracy. It was confirmed from the AST simulation with the different initial particle size distributions that even the small number of the coarse particles significantly accelerated the Pt degradation. The developed model was integrated to the FC system simulator the current authors had presented and a remarkable computational speed for the year-long and system-scale durability simulation was demonstrated with a normal laptop computer.

A platinum degradation reaction model for the high-speed computation in the integrated fuel cell system simulator and its validation by the accelerated stress test data of the commercial fuel cells

A model to estimate the Pt degradation rate and the numerical methods for high-speed computation were developed for the year-long and system-scale durability simulation in an acceptable accuracy and computational time. The parameters in the model were determined to reproduce the published accelerated stress test (AST) data with the 1 cm2 MEA of 2nd-generation MIRAI (MIRAI-2), state-of-the-art commercial fuel cell electric vehicle. Accuracy of the simulation results was validated by the published data of the ASTs with the FC stack having 13 cells of MIRAI-2, which consists of the 30000 startup-shutdown cycles in 0–0.88 V and 73000 acceleration-deceleration cycles in 0.6–0.9 V. The experimental and simulation results of the residual fractions of the electrochemical surface area in the cathode catalyst layer after the ASTs were 68 and 69 %, respectively, and they agreed in an acceptable accuracy. It was confirmed from the AST simulation with the different initial particle size distributions that even the small number of the coarse particles significantly accelerated the Pt degradation. The developed model was integrated to the FC system simulator the current authors had presented and a remarkable computational speed for the year-long and system-scale durability simulation was demonstrated with a normal laptop computer.

Research on the high-performance computing method for the neutron diffusion spatiotemporal kinetics equation based on the convolutional neural network

Due to the uncertainty of computational results and the lack of interpretability of models in solving physical field equations in current deep learning, this paper designs a convolutional neural network that can be used to solve the neutron diffusion spatiotemporal kinetics equation in polar and cylindrical coordinate systems. This algorithm directly utilizes the macroscopic cross-section of the material without using the lattice homogenization method, replaces the finite volume method with the extended matrices, and solves the extended matrices using the convolutional kernels instead of the iterative algorithms. Taking the simplified Tsinghua High Flux Reactor (THFR) as an example, the feasibility of the algorithm is verified on the PyTorch platform and compared with the calculation results of the source iteration method running on the GPU. The calculation results show that when the number of grids in the radial and axial sections of the simplified THFR model is 804,600 and 3,576,000, respectively, and the algorithm is iterated 3000 times, the normalized power of the convolutional neural network and the source iteration method converges to 10−10, and the maximum point by point error of the neutron flux density of the above two algorithms converges to 10−5. The computational time consumed by the convolutional neural network is approximately 880.64 s and 3729.62 s, which reduces the computational time by 4.66% and 5.05% compared to the GPU parallel accelerated source iteration method, and the former consumes 43.75% less memory compared to the latter. The convolutional neural network is mainly used as the virtual physics engine for the THFR digital twin system, in addition to solving the neutron diffusion spatiotemporal kinetics equation and further improving computational speed. The algorithm directly utilizes the neutron macroscopic cross-section of the material to calculate the neutron flux density distribution without using the lattice homogenization, providing theoretical guidance and algorithm support for developing the high-precision multi-physical field coupling model.

A clustered federated learning framework for collaborative fault diagnosis of wind turbines

Data-driven approaches demonstrate significant potential in accurately diagnosing faults in wind turbines. To enhance diagnostic performance and reduce communication costs in federated learning with data heterogeneity among different clients, we introduce a clustered federated learning framework to wind turbine fault diagnosis. Initially, a lightweight multiscale separable residual network (LMSRN) model is proposed for each local client. The LMSRN model integrates a multiscale spatial feature derivation unit and a depthwise separable feature extraction unit. Subsequently, to tackle data heterogeneity among clients, canonical correlation coefficients of representations are extracted from the intermediate layers of local LMSRN models, and a representational canonical correlation clustering (RCCC) method is proposed to assess the similarity of local LMSRN models and group them into clusters. Finally, a global model is trained for each cluster. Real-world wind turbine data experiments showcase the superior performance of the proposed clustered federated learning framework over traditional methods in terms of diagnostic accuracy and computational speed. Additionally, the optimal choice of the number of clusters is also discussed.

No Sum (NS) Sequence: A Tool for Quantum-Safe Cryptography

Quantum computing is emerging as a promising technology that can solve the world’s most complex and computationally intensive problems, including cryptography. Conventional cryptography is now facing a severe threat because of the speed of computation offered by quantum computers. There is a need for the standardization of novel quantum-resistant public-key cryptographic algorithms. In this article, the No-Sum (NS) sequence-based public key cryptosystem is proposed, increasing the computational resource requirement against the quantum computer-assisted bruteforce attack. In this paper, we demonstrate an elevation of security of the public key cryptosystem with NS sequence. The proposed methodology works in two phases; in the first phase, the conventional RSA algorithm is used to share the first element and offset parameter between the receiver and sender. In the second phase, prime numbers are generated from the NS sequence and used for encryption/ decryption of messages/ciphers. This paper illustrates the potential of the NS sequence in developing a quantum-resistant cryptosystem, contributing to the advancement of secure communication in the quantum computing era.

Joint contrastive self-supervised learning and weak-orthogonal product quantization for fast image retrieval

Hashing and quantization methods based on deep learning are being widely used in large-scale image retrieval. Most methods use traditional supervised models to train models. Although these methods achieve excellent performance, they rely too much on expensive label information and do not fully utilize data resources. These methods overly encourage binary codes to retain original information, which may lead to models building too much useless information at the expense of discriminative information that is more important for retrieval. To address these issues, we propose a self-supervised method that does not require labeled information, called Contrastive Self-Supervised Weak-Orthogonal Product Quantization (CSWPQ). The method is trained in a label-free self-supervised manner to reduce the need for labels by maximizing the similarity of different views of the same image and minimizing the similarity of different image views, which not only improves the generalization ability of the model but also enhances the robustness. We introduce a quantified contrastive learning method that is different from traditional contrastive learning to learn more discriminative image representations by constructing quantified image features of different views of the image. In addition, in order to reduce the quantization error caused by product quantization, we impose a weak-orthogonal constraint to increase the diversity of representations, reduce the information loss and computational overhead in the quantization process, and improve the computational speed. Extensive experiments on CIFAR-10, NUS-WIDE and FLICKR25K datasets show that our proposed method achieves better performance.

A Multi-Scale Numerical Simulation Method Considering Anisotropic Relative Permeability

Most of the oil reservoirs in China are fluvial deposits with firm reservoir heterogeneity, where differences in fluid flow capacity in individual directions should not be ignored; however, the available commercial reservoir simulation software cannot consider the anisotropy of the relative permeability. To handle this challenge, this paper takes full advantage of the parallelism of the multi-scale finite volume (MsFV) method and establishes a multi-scale numerical simulation approach that incorporates the effects of reservoir anisotropy. The methodology is initiated by constructing an oil–water black-oil model considering the anisotropic relative permeability. Subsequently, the base model undergoes decoupling through a sequential solution, formulating the pressure and transport equations. Following this, a multi-scale grid system is configured, within which the pressure and transport equations are progressively developed in the fine-scale grid domain. Ultimately, the improved multi-scale finite volume (IMsFV) method is applied to mitigate low-frequency error in the coarse-scale grid, thereby enhancing computational efficiency. This paper introduces two primary innovations. The first is the development of a multi-scale solution method for the pressure equation incorporating anisotropic relative permeability. Validated using the Egg model, a comparative analysis with traditional numerical simulations demonstrates a significant improvement in computational speed without sacrificing accuracy. The second innovation involves applying the multi-scale framework to investigate the impact of anisotropy relative permeability on waterflooding performance, uncovering distinct mechanisms by which absolute and relative permeability anisotropy influence waterflooding outcomes. Therefore, the IMsFV method can be used as an effective tool for high-resolution simulation and precise residual oil prediction in anisotropic reservoirs.

Virtualization and energy management optimization of high speed computer network data centers based on optical switching and network technology

With the development of cloud computing and big data technology, the scale and complexity of data centers are increasing, and the thermal energy consumption management of computer processors can effectively improve the energy efficiency of data centers and reduce operating costs. Therefore, this study discusses the application of high-speed computer network architecture based on optical switching and network technology in data center virtualization. By constructing a high-speed computer network model based on optical switching technology, virtualization technology is used to dynamically schedule and manage resources, combined with real-time monitoring system, and the energy consumption data of the processor is collected. Using the data analysis method, the energy consumption performance under different load conditions is evaluated and the optimization strategy is proposed. The experimental results show that the network architecture using optical switching technology significantly reduces the latency of data center and improves the resource utilization. In the thermal management of the processor, dynamic resource scheduling successfully reduces the energy consumption and effectively maintains the stability and performance of the system. As a result, data center virtualization solutions based on optical switching and networking technologies can significantly optimize the power management of computer processors and improve overall energy efficiency.

Applying the Expectation-Maximization Algorithm to Multivariate Signal Extraction

Understanding time series signal extraction is vital for federal agencies, especially as data collection accelerates. One natural extension is moving from univariate to multivariate signal extraction, which offers the promise of reducing extraction error by exploiting cross-sectional relationships. However, such an extension ushers in new computational challenges, viz. larger parameter spaces and more complicated objective functions. The Expectation-Maximization (EM) algorithm provides a methodology to implicitly compute (or approximate) maximum likelihood estimates (MLEs). This paper provides methodology for applying the EM algorithm to a class of latent component multivariate time series models that allow for a nuanced specification of the unobserved signal. We derive an explicit formula for the maximization step, which facilitates computation speed while also improving the stability of the algorithm. Numerical studies demonstrate EM’s ability to efficiently compute MLEs in low-dimensional systems, while also providing feasible estimates in moderate-dimensional systems where MLEs are infeasible to compute. Applications to monthly housing starts and daily immigration are provided.

Recent advances in the integration of protein mechanics and machine learning

Mechanics underlies protein properties and behavior. From a theoretical standpoint, it is possible to derive these based on physical rules. This is appealing because they provide insights into physiology and disease, as well as aid in protein engineering; however, the convoluted nature of the biological system and current computational speeds limit its feasibility. Machine learning (ML) architectures are known for their ability to make inferences on complex data, such as the relationship between protein mechanics, properties, and behavior. Substantial efforts have been made to learn such correlations in tasks such as the prediction of structure, stability, natural frequency, mechanical strength, folding rate, solubility, and function. Each of these properties is interconnected through protein mechanics, and it is not surprising that the methods used in these tasks overlap highly in model input and architecture. In this review, we evaluate ML methods for the seven aforementioned prediction tasks to identify current trends in ML research in the field of protein sciences, focusing on the input and model architecture of each method. A short overview of de novo protein design is also provided. Finally, we highlight trends in the application of ML methods in the field of protein science, as well as directions for future improvements.

Enhanced Hybrid Algorithms for Segmentation and Reconstruction of Granular Grains From X‐Ray Micro Computed‐Tomography Images

ABSTRACTAccurate three‐dimensional (3D) reconstruction of granular grains from x‐ray micro‐computed tomography (µCT) images is a long‐standing challenge, particularly for dense soil samples. This study develops a machine learning (ML) enhanced approach to automatically reconstruct granular grains from µCT images. The novel academic contributions of this paper include (a) a hierarchical strategy based on parameter‐independent polygonal approximation, area, and concavity analysis, for the first time, to identify and eliminate both intergranular and intragranular voids; (b) incorporation of a recursive segmentation scheme and ML‐based grain classifier to avoid over‐segmentation; (c) novel modifications on the determination of splitting paths to enhance segmentation accuracy; and (d) an effective approach of assigning initial level set functions for reconstructing granular grains automatically. The hybrid ML algorithm is applied to µCT images of dense Mojave Mars Simulant. The results indicate that the proposed method can accurately segment grain clumps with unclear boundaries. The new automatic reconstruction algorithm eliminates ineffective operations and achieves a three‐fold increase in computational speed than previous methods documented in the literature. Ninety‐one percent of grains with distinct boundaries can be reconstructed and the reconstruction ratio reaches 81% even for grains without distinct boundaries. The overall reconstruction ratio of grains increases by 20% compared with previous methods, achieving a step‐change improvement for one‐to‐one mapping of real soil samples.

Proper orthogonal decomposition reduced-order model of the global oceans

AbstractA reduced-order model (ROM) of the global oceans is developed by projecting the hydrostatic Boussinesq equations of motion onto a proper orthogonal decomposition (POD) basis. Three-dimensional POD modes are calculated from the ocean fields of an ensemble climate reanalysis dataset. The coefficients in the POD ROM are calculated using a regression approach. The performance of various POD ROM configurations are assessed. Each configuration is derived from an alternate sea-water equation of state, linking the density and temperature fields. POD ROM variants incorporating an equation of state in which density is a quadratic function of temperature, are able to reproduce the statistics of the large-scale structures at a fraction of the computational cost required to numerically simulate this flow. Due to the speed and efficiency of calculation, such reduced-order models of the global geophysical system will enable researchers and policy makers to assess the physical risk for a broader range of potential future climate scenarios.

Real-time delivered dose assessment in carbon ion therapy of moving targets

Abstract Purpose:
Real-time adaptive particle therapy is being investigated as a means to maximize the treatment delivery accuracy. To react to dosimetric errors, a system for fast and reliable verification of the agreement between planned and delivered doses is essential. This study presents a clinically feasible, real-time 4D-dose reconstruction system, synchronized with the treatment delivery and motion of the patient, which can provide the necessary feedback on the quality of the delivery.
Methods:
A GPU-based analytical dose engine capable of millisecond dose calculation for carbon ion therapy has been developed and interfaced with the next generation of the Dose Delivery System (DDS) in use at Centro Nazionale di Adroterapia Oncologica (CNAO). The system receives the spot parameters and the motion information of the patient during the treatment and performs the reconstruction of the planned and delivered 4Ddoses. After each iso-energy layer, the results are displayed on a graphical user interface by the end of the spill pause of the synchrotron, permitting verification against the reference dose.
The framework has been verified experimentally at CNAO for a lung cancer case based on a virtual phantom 4DCT. The patient’s motion was mimicked by a moving Ionization Chambers (ICs) 2D-array.
Results:
For the investigated static and 4D-optimized treatment delivery cases, real time dose reconstruction was-achieved with an average pencil beam dose calculation speed up to more than one order of magnitude smaller than the spot delivery. The reconstructed doses have been benchmarked against offline log-file based dose reconstruction with the TRiP98 treatment planning system, as well as QA measurements with the ICs 2D-array, where an average gamma-index passing rate (3%/3mm) of 99.8% and 98.3%, respectively, were achieved.
Conclusion:
This work provides the first real-time 4D-dose reconstruction engine for carbon ion therapy. The framework integration with the CNAO DDS paves the way for a swift transition to the clinics.

Enhanced seagull optimization for enhanced accuracy in CUDA-accelerated Levenberg–Marquardt backpropagation neural networks for earthquake forecasting

Hyperparameter tuning is crucial for enhancing the accuracy and reliability of artificial neural networks (ANNs). This study presents an optimization of the Levenberg–Marquardt backpropagation neural network (LM-BPNN) by integrating an improved seagull optimization algorithm (ISOA). The proposed ISOA-LM-BPNN model is designed to forecast earthquakes in the Caribbean region. The study further explores the impact of data and model parallelism, revealing that hybrid parallelism effectively mitigates the limitations of both. This leads to substantial gains in throughput and overall performance. To address computational demands, this model leverages the compute unified device architecture (CUDA) framework, enabling hybrid parallelism on graphics processing units (GPUs). This approach significantly enhances the model’s computational speed. The experimental results demonstrate that the ISOA-LM-BPNN model achieves a 20% improvement in accuracy compared to four baseline algorithms across three diverse datasets. The integration of ISOA with LM-BPNN refines the neural network’s hyperparameters, leading to more precise earthquake predictions. Additionally, the model’s computational efficiency is evidenced by a 56% speed increase when utilizing a single GPU, and an even greater acceleration with dual GPUs connected via NVLink compared to traditional CPU-based computations. The findings underscore the potential of ISOA-LM-BPNN as a robust tool for earthquake forecasting, combining high accuracy with enhanced computational speed, making it suitable for real-time applications in seismic monitoring and early warning systems.

Emergency voltage control strategy for power system transient stability enhancement based on edge graph convolutional network reinforcement learning

Emergency control is essential for maintaining the stability of power systems, serving as a key defense mechanism against the destabilization and cascading failures triggered by faults. Under-voltage load shedding is a popular and effective approach for emergency control. However, with the increasing complexity and scale of power systems and the rise in uncertainty factors, traditional approaches struggle with computation speed, accuracy, and scalability issues. Deep reinforcement learning holds significant potential for the power system decision-making problems. However, existing deep reinforcement learning algorithms have limitations in effectively leveraging diverse operational features, which affects the reliability and efficiency of emergency control strategies. This paper presents an innovative approach for real-time emergency voltage control strategies for transient stability enhancement through the integration of edge-graph convolutional networks with reinforcement learning. This method transforms the traditional emergency control optimization problem into a sequential decision-making process. By utilizing the edge-graph convolutional neural network, it efficiently extracts critical information on the correlation between the power system operation status and node branch information, as well as the uncertainty factors involved. Moreover, the clipped double Q-learning, delayed policy update, and target policy smoothing are introduced to effectively solve the issues of overestimation and abnormal sensitivity to hyperparameters in the deep deterministic policy gradient algorithm. The effectiveness of the proposed method in emergency control decision-making is verified by the IEEE 39-bus system and the IEEE 118-bus system.

A Comparative Study of Artificial Neural Network Training Algorithms to Predict Photovoltaic Module Output Power

Introduction: Solar energy is a crucial component and contributes a large portion of renewable resources, the demand for which has recently increased. During previous years, it was difficult to predict the amount of energy obtained from photovoltaic systems. The angle of inclination of the solar panel, the amount of radiation, the speed of the wind, the amount of humidity, and the temperature of the weather are major factors that effectively affect the production of electrical energy. Scientists have used many strategies to predict the power generated by PV modules accurately, but each method has different pros and cons. Method: This study tested three training algorithms for artificial neural networks (ANN): scaled conjugate gradient (SCG), Levenberg Marquardt (LM), and Bayesian regularization (BR), determining which one performed best in terms of prediction speed and accuracy. Twenty-eight thousand two hundred ninety-six samples of experimental data for primary influencing environmental factors were fed into the artificial neural network, which consists of 15 hidden layers. Before training the network, we preprocessed the data to remove factors that have a secondary effect. Result: The analytical results showed that the artificial neural network trained according to the LM algorithm is the best in terms of accuracy and speed of predicting the resulting photovoltaic energy. Conclusion: The results showed that although the regression evaluation and MSE values for the LM, SCG, and BR algorithms are close (98.129%, 0.0622, 97.849%, 0.0587, 98.151%, and 0.0585, respectively), the LM training algorithm is the best in terms of speed of calculation and display of results.

Ultra-wide viewing angle holographic display system based on spherical diffraction

In the field of holographic displays, achieving a large viewing angle (LVA) has long been an essential and formidable challenge due to the limited diffraction angle of planar spatial light modulators (SLMs). Spherical holography, which allows observation from all perspectives, represents a practical approach to achieve the LVA. However, the physical limitations of commercial SLMs constrain direct implementation, making the realization of spherical holographic displays almost impractical and thus hindering technological advancement. This paper introduces an optical solution for spherical holographic display system, enabling the accurate reproduction of ultra-wide viewing angles and large objects. Our system utilizes a novel technique involving a convex parabolic mirror to convert an incident plane wave into a spherical wave. This innovative strategy permits the use of a planar SLM to create a 360°spherical holographic display. Additionally, we address the sampling issue for generating spherical holograms by constraining the point spread function (PSF), reducing the number of samples, and adapting it for visible light. Numerical simulations and optical experiments validate our success in achieving an ultra-wide viewing angle with a 360°horizontal angle and an 80°zenith angle range. Our system outperforms the state-of-the-art holographic-optical-elements approach both in computation speed and object size. We believe that our work significantly advances spherical holographic displays and offers a practical solution with vast potential applications in medicine, geology, and astronomy.

The Pico Balloon Archive: Establishing the First Super Pressure Balloon Satellite Network for Atmospheric Research

Abstract We present the Pico Balloon Archive (PBA), a specialized platform for archiving, visualizing, and verifying data from pico balloons— lightweight, super pressure balloons capable of operating in the upper troposphere and lower stratosphere for durations ranging from weeks to years. With a latitudinal range from 88.77°S to 89.60°N and data collection ongoing since 2021, the PBA stands as the most spatially extensive repository for super pressure balloons to date. It offers a centralized, user-friendly online portal (http://picoballoonarchive.org) for data access, featuring a summary of flight details, downloadable raw data, and individual flight visualizations. In this study, we validate the PBA’s wind speed and direction calculations against the Integrated Radiosonde Archive (IGRA), showing strong correlations (r2 = 0.66 and 0.78 for wind speed and direction, respectively). We also highlight the PBA’s effectiveness in charting global atmospheric circulation and the capacity of certain balloons to traverse hemispheres, a feat made possible by unique stratospheric dynamics. Furthermore, we demonstrate the PBA’s effectiveness in validating Numerical Weather Prediction (NWP) Lagrangian trajectories, quantifying errors based on model initialization latitude. This continuously updated super pressure balloon network is poised to significantly aid the atmospheric science community, facilitating a deeper understanding of global atmospheric processes.

RASPA3: A Monte Carlo code for computing adsorption and diffusion in nanoporous materials and thermodynamics properties of fluids.

We present RASPA3, a molecular simulation code for computing adsorption and diffusion in nanoporous materials and thermodynamic and transport properties of fluids. It implements force field based classical Monte Carlo/molecular dynamics in various ensembles. In this article, we introduce the new additions and changes compared to RASPA2. RASPA3 is rewritten from the ground up in C++23 with speed and code readability in mind. Transition-matrix Monte Carlo is added to compute the density of states and free energies. The Monte Carlo code for rigid molecules is based on quaternions, and the atomic positions needed in the energy evaluation are recreated from the center of mass position and quaternion orientation. The expanded ensemble methodology for fractional molecules, with a scaling parameter λ between 0 and 1, now also keeps track of analytic expressions of dU/dλ, allowing independent verification of the chemical potential using thermodynamic integration. The source code is freely available under the MIT license on GitHub. Using this code, we compare four Monte Carlo (MC) insertion/deletion techniques: unbiased Metropolis MC, Configurational-Bias Monte Carlo (CBMC), Continuous Fractional Component MC (CFCMC), and CB/CFCMC. We compare particle distribution shapes, acceptance ratios, accuracy and speed of isotherm computation, enthalpies of adsorption, and chemical potentials, over a wide range of loadings and systems, for the grand canonical ensemble and for the Gibbs ensemble.

On the expressiveness and efficiency of guarded lists in Bach

Concurrency theory has received considerable attention, but mostly in the scope of synchronous process algebras such as CCS, CSP, and ACP. As another way of handling concurrency, data-based coordination languages aim to provide a clear separation between interaction and computation by synchronizing processes asynchronously by means of information being available or not on a shared space. Although these languages enjoy interesting properties, verifying program correctness remains challenging. In particular, model checking logic formulae is known to raise performance issues due to the state space explosion problem. In this paper, we propose a guarded list construct as a solution to address this problem. Beyond increasing performance, it also enriches the expressiveness of data-based coordination languages and allows for program transformations that further increase the speed of computations.