Parallel Scheme Research Articles

Quantum-centric high performance computing for quantum chemistry.

High performance computing (HPC) is renowned for its capacity to tackle complex problems. Meanwhile, quantum computing (QC) provides a potential way to accurately and efficiently solve quantum chemistry problems. The emerging field of quantum-centric high performance computing (QCHPC), which merges these two powerful technologies, is anticipated to enhance computational capabilities for solving challenging problems in quantum chemistry. The implementation of QCHPC for quantum chemistry requires interdisciplinary research and collaboration across multiple fields, including quantum chemistry, quantum physics, computer science and so on. This perspective provides an introduction to the quantum algorithms that are suitable for deployment in QCHPC, focusing on conceptual insights rather than technical details. Parallel strategies to implement these algorithms on quantum-centric supercomputers are discussed. We also summarize high performance quantum emulating simulators, which are considered a viable tool to explore QCHPC. We conclude with challenges and outlooks in this field.

Demonstration-guided deep reinforcement learning for coordinated ramp metering and perimeter control in large scale networks

Effective traffic control methods have great potential in alleviating network congestion. Particularly, in an urban network consisting of heterogeneous roads (e.g., freeways and urban roads), how to integrate and coordinate control policies on different roads is a critical issue in large-scale networks. This study addresses this question from two aspects: modeling and control. From the modeling aspect, we formulate the hybrid traffic modeling in heterogeneous networks with the Asymmetric Cell Transmission Model (ACTM) for freeways and the generalized bathtub model for urban roads. For the control aspect, this study considers two representative control approaches: ramp metering for freeways and perimeter control for urban roads, and we aim to develop a deep reinforcement learning (DRL)-based coordinated control framework for large-scale networks. However, there are two significant challenges in the coordinated control in large-scale networks with DRL methods: non-stationary environment and large search space. To address both issues, we incorporate the demonstration to guide the DRL method for better convergence by introducing the concept of “teacher” and “student” models. The teacher models are traditional controllers that provide control demonstrations. For instance, ALINEA and Gating are two representative feedback controllers for ramp metering and perimeter control which can be “teacher” models. The student models are DRL methods, which learn from teachers and aim to surpass the teachers’ performance. Additionally, we develop a parallel training scheme to accelerate the proposed DRL method. To validate the proposed framework, we conduct two case studies in a small-scale network and a real-world large-scale traffic network in Hong Kong. Numerical results show that the proposed DRL method outperforms demonstrators as well as DRL methods, and the coordinated control is more effective than just controlling ramps or perimeters respectively. The research outcome reveals the great potential of combining traditional controllers with DRL for coordinated control in large-scale networks.

Data migration strategies in mergers and acquisitions: A case study of banking sector

Data migration is a critical component of mergers and acquisitions (M&A), particularly in the banking sector, where seamless integration of information systems is essential for operational continuity and regulatory compliance. This review examines data migration strategies employed during M&A activities, focusing on a case study in the banking sector. The study explores the challenges, methodologies, and best practices that ensure successful data transfer while minimizing risks and disruptions. In the context of banking, M&A transactions often involve integrating vast amounts of sensitive data, including customer records, financial transactions, and regulatory documents. Effective data migration is crucial to preserving data integrity, maintaining customer trust, and achieving the strategic goals of the merger or acquisition. This case study highlights the importance of a well-structured data migration plan, which includes comprehensive data mapping, validation processes, and robust security measures. The review delves into the various strategies adopted for data migration, such as phased migration, parallel run strategies, and the use of automated tools to streamline the process. It also discusses the significance of collaboration between IT and business teams to address data quality issues, ensure compliance with regulatory standards, and align migration efforts with the overall M&A objectives. Furthermore, the study examines the challenges faced during the data migration process, including legacy system integration, data redundancy, and the need for real-time data synchronization. By analyzing the outcomes of the case study, the review provides insights into the lessons learned and the critical success factors that can guide future M&A activities in the banking sector. In conclusion, this study underscores the pivotal role of effective data migration strategies in ensuring a smooth transition during mergers and acquisitions in the banking sector, ultimately contributing to the long-term success of the consolidated entity. Keywords: Data Migration, Mergers and Acquisitions, Banking Sector, Data Integration, Data Mapping, Regulatory Compliance, Phased Migration, Parallel Run Strategy, Legacy Systems, Data Synchronization.

A Multi-Strategy Improved Sooty Tern Optimization Algorithm for Concrete Dam Parameter Inversion

A original strategy for optimizing the inversion of concrete dam parameters based on the multi-strategy improved Sooty Tern Optimization algorithm (MSSTOA) is proposed to address the issues of low efficiency, low accuracy, and poor optimizing performance. First, computational strategies to improve the traditional Sooty tern algorithm, such as chaos mapping to improve the initial position of the population, a new nonlinear convergence factor, the LIMIT threshold method, and Gaussian perturbation to update the optimal individual position, are adopted to enhance its algorithmic optimization seeking ability. Then, the measured and finite element data are combined to create the optimization inversion fitness function. Based on the MSSTOA, the intelligent optimization inversion model is constructed, the inversion efficiency is improved by parallel strategy, and the optimal parameter inversion is searched. The inversion strategy is validated through test functions, hypothetical arithmetic examples, and concrete dam engineering examples and compared with the inversion results of the traditional STOA and other optimization algorithms. The results show that the MSSTOA is feasible and practical, the test function optimization results and computational time are better than the STOA and other algorithms, the example inversion of the elastic modulus is more accurate than the traditional STOA calculation, and the results of the MSSSTOA inversion are reasonable in the engineering example. Compared with other algorithms, the local extremes are skipped, and the time consumption is reduced by at least 48%. The finite element hydrostatic components calculated from the inversion results are well-fitted to the statistical model with minor errors. The intelligent inversion strategy has good application in concrete dam inverse analysis.

Composer's video for his own work: time parallels, dotted lines, perpendiculars

Relevance of the study is related to the problem of composers' self-illustrations in the YouTube presentations. The novelty of this topic is conditioned by new practice to illustrate music in YouTube. Graphic representation of music leads to the problem of correspondence between music and video. We selected only two Ukrainian composers to compare: Yuri Ishchenko and Sviatoslav Luniov.
 The main objective of the study is the determination of main principles of similarity or dissimilarity of time structures in music and images. The methodology is based on methods of comparison and analogy in the context of phenomenological approach to the Event in the Being Time (by late Heidegger). How the study was done: time structures of selected compositions were compared, through analogies, with the illustrating graphic events as phenomena of existential time. Thereby a number of similarities and dissimilarities were detected.
 Composer Y. Ishchenko preferred to illustrate his music by static images. Such a situation is a paradoxical because of the dynamic essence of his music. It establishes the problem of co-existent two time situations: in the first of them there is an integral and quasi-timeless image of music; but in the other situation there is a running stream of music structures. This approach becomes very and very fruitful because Music is able to use both the continuity, integrity of musical image as well as fluid sound events. The Y. Ishchenko's "Watercolors" for violin and piano are illustrated by several images; this approach makes it possible to mark different parts of total design (i.e. different movements or stages). Another solution one can hear and see in his "Small partita" for violin and harp. I helped him to accompany this design with fragments of animation about baroque époque. There are several GIFs - the periodically repeating images, which creates a cross-rhythm constellation with music. The significant crossfade is present in the video by Ishchenko's 3rd Violin sonata: ruined house through cross-like gap transforms in an abandoned church. The final fade of it lasts a minute. Thus, the minimal graphic processes are extremely expressive. S. Luniov mastered advanced skills in electronic music and its visual presentation. In his video "POST” he combines fragments of his own Panta Rhei for orchestra with sounds of the war. The intensity of the music development is essentially higher than that of graphic illustration, but at the end of the project the situation is opposite. In the video accompanying his Panta Rhei for orchestra S. Luniov synchronized main events and stages of the sound score with more intensive changes of the video stream and its thematic.
 Findings and conclusions. For the composer Y. Ishchenko, the graphic self-illustration on YouTube was mostly the outer necessity. He often limited his projects with static images for the total score or its main parts. Thus, he reinforced the integrity pole of events with the graphic ingredient of YouTube projects. Thereby, the dramaturgic tense of his compositions is concentrated in music. Such an approach was called time perpendicular strategy or time dotted line strategy. For the composer S. Luniov, the graphic self-illustration on YouTube is mostly the parallel stream of events. However, the tempo and intensity of visual and sound streams can be unequal. Such an approach is called time parallel strategy. Significance of the study follows from its conclusions, leading to more skillful and exact solutions within composers' self-illustration, i.e. in clarifying the semantic and structure design in music.

A full process algebraic representation of Ant Colony Optimization

We present a process algebra capable of specifying parallelized Ant Colony Optimization algorithms in full detail: PA2CO. After explaining the basis of three different ACO algorithms (Ant System, MAX-MIN Ant System, and Ant Colony System), we formally define PA2CO and use it for representing several types of implementations with different parallel schemes. In particular fine-grained and coarse-grained specifications, each one taking advantage of parallel executions at different levels of system granularity, are formalized.

Efficient and Parallel Implementation of Real and Complex Response Functions Employing the Second-Order Algebraic-Diagrammatic Construction Scheme for the Polarization Propagator.

We present the implementation of an efficient matrix-folded formalism for the evaluation of complex response functions and the calculation of transition properties at the level of the second-order algebraic-diagrammatic construction (ADC(2)) scheme. The underlying algorithms, in combination with the adopted hybrid MPI/OpenMP parallelization strategy, enabled calculations of the UV/vis spectra of a guanine oligomer series ranging up to 1032 contracted basis functions, thereby utilizing vast computational resources from up to 32,768 CPU cores. Further analysis of the convergence behavior of the involved iterative subspace algorithms revealed the superiority of a frequency-separated treatment of response equations even for a large spectral window, including 101 frequencies. We demonstrate the applicability to general quantum mechanical operators by the first reported electronic circular dichroism spectrum calculated with a complex polarization propagator approach at the ADC(2) level of theory.

A many-objective evolutionary algorithm assisted by ideal hyperplane

In many-objective optimization problems (MaOPs), balancing convergence and diversity while rapidly converging to the Pareto front is an arduous task for evolutionary algorithms. In addition, with the increase of the number of targets, the number of non-dominant solutions increases exponentially, and the individual selection pressure is insufficient. For this problem, we propose a many-objective evolution algorithm assisted by an ideal hyperplane (MaOEA-IH). To begin, the ideal hyperplane is built from the extremums of each dimension of objective space, guiding the individual to the Pareto front of search. Second, a parallel p-norm mating selection strategy based on the ideal hyperplane is proposed to improve convergence. In addition, two other factors are taken into account: (1) different p-norms are used to measure different spatial scales; and (2) individual selection uncertainty is defined by incorporating a probabilistic perturbation mechanism. Following that, the sum of objectives is applied to shift-based density estimation, which serves as an evaluation criterion in the environmental selection operation. This method increases the chances of solutions with high convergence and diversity entering the next generation, thereby increasing selection pressure. On three benchmark problems of DTLZ, WFG, and MaF, we compare MaOEA-IH with seven excellent algorithms. The results demonstrate that the MaOEA-IH proposed is highly competitive in solving MaOPs.

A new active learning method for reliability analysis based on local optimization and adaptive parallelization strategy

In recent years, an active learning method combining Kriging and Monte Carlo Simulation (AK-MCS), has been developed for calculating the failure probability. However, the original AK-MCS only uses serial computing, which limits its ability to take advantage of distributed computing. Thus, this work introduces a novel adaptive learning approach for reliability analysis by combining local optimization and a parallelization strategy. The new approach identifies the points of greatest uncertainty through local optimization and adds them into the design of experiments. An inner learning loop is implemented, searching for additional best points with a pseudo Kriging model so that the performance function can be evaluated in parallel. Furthermore, an adaptive strategy is proposed to determine the amount of additional points during iteration based on the minimum value of the learning function. We conducted a comparison between the proposed method and the original AK-MCS as well as a few additional methods in order to assess its efficacy and precision. Five examples were considered to assess performance.

P-AK-MCS: Parallel AK-MCS method for structural reliability analysis

In recent years, the active learning reliability method that combines the Kriging model and Monte Carlo simulation (AK-MCS) has emerged as a promising approach due to its computational efficiency and accuracy. However, the commonly used learning functions, such as the expected feasibility function (EFF), U function, H function, and expected risk function (ERF), can only select one training point at each iteration which is time-wasteful when parallel computing is available. Therefore, this paper proposes a parallel active learning Kriging strategy, namely P-AK-MCS, for structural reliability analysis. By introducing an influence function that reflects the impact of the added point on the original learning function, four parallel learning functions are constructed: pseudo-U (PU) function, pseudo-EFF (PEFF), pseudo-H (PH) function, and pseudo-ERF (PERF). These functions aim to identify multiple training points at each iteration without requiring additional functional evaluations. The effectiveness of the proposed method is validated using four examples. The results demonstrate that compared to the standard AK-MCS, the proposed P-AK-MCS significantly reduces the number of computation loops and greatly decreases computational costs. Moreover, the total number of functional evaluations required is similar to that of the standard AK-MCS and remains insensitive to the number of multiple training points.

High-efficiency computation for electromagnetic forming process: An explicit-implicit GPU approach

Low computational efficiency is one of the crucial challenges in electromagnetic forming process simulation. However, the huge gap in switching from the electromagnetic field to mechanics hinders the development of high-efficiency parallel coupling. A graphics processing unit (GPU)-based explicit-implicit coupling method is proposed to address the high-computational cost in simulations of the electromagnetic forming process, in which the large deformation (explicit) and electromagnetic field (implicit) are intimately coupled. The electromagnetic-mechanical coupling is achieved by constructing five parallel algorithms, which include the parallel operations of vector-vector, matrix-matrix, matrix-vector, vector assembly, and matrix-free assembly parallel preconditioned conjugate gradient algorithm. The high-efficiency GPU parallel computing is realized through the special data storage structure and parallel strategies. The explicit-implicit GPU method is validated against the experimental results, and several examples are presented for simulations of the electromagnetic forming process to illustrate both the accuracy and high-efficiency parallel acceleration performance of the present program.

Parallel implement of real-time delta-sigma modulation for digital mobile fronthaul.

We propose and experimentally demonstrate a novel FPGA-based parallel architecture for delta-sigma modulation (DSM) for digital mobile fronthaul employing the DSM interface. This architecture breaks the limitations of the feedback loop in DSM, is not constrained by critical paths, and supports fully parallel processing, so it can deal with high sampling rates at low hardware operating speeds. In contrast to other parallel schemes, the proposed bit-by-bit quantization DSM avoid significant storage resources requirements for buffering. A real-time experimental system using Xilinx Kintex Ultrascale FPGA was implemented to validate the feasibility of the proposed architecture. 14 carrier aggregated orthogonal frequency division multiplexing (OFDM) signals are digitized by DSM into a 5Gb/s PAM4 signal and transmitted over a 20 km single-mode fiber (SMF). As a waveform-agnostic digitization interface, we also experimentally demonstrated the DSM with 14 carrier aggregated filter-bank-multicarrier (FBMC) signals, which can achieve better EVM performance.

Modoru: Clos nanosecond optical switching for distributed deep training [Invited

Distributed deep training has become a significant consumer of bandwidth across datacenter-scale networks. The diverse parallel strategies employed in deep training require different communication patterns, necessitating the periodic adaptation of dynamic topologies. Since electrical switching approaches its capacity limit due to high bandwidths and has difficulties in regard to topology adaptation (i.e., logical and physical topologies are isomorphic), optical switching has become an attractive option to address these bottlenecks. In this paper, we propose Modoru, a wavelength- and datarate-agnostic Clos architecture with a switching speed of O(10ns). Modoru is a drop-in replacement solution that has no constraints on achieving a high radix. To verify its topological flexibility, we also develop topology-as-a-service, which provisions sequentially dynamic topologies for training jobs and guarantees high topology availability over the entire network. Large-scale simulations show a basic 7.9× acceleration in deep training jobs using Modoru. Additionally, experiments on the Modoru prototype demonstrate acceleration of deep training jobs through the provisioning of adaptive topologies.

An autonomic parallel strategy for exhaustive search tree algorithms on shared or heterogeneous systems

SummaryBacktracking branch‐and‐prune (BP) algorithms and their variants are exhaustive search tree techniques widely employed to solve optimization problems in many scientific areas. However, they characteristically often demand significant amounts of computing power for problem sizes representative of real‐world scenarios. Given that their search domains can often be partitioned, these algorithms are frequently designed to execute in parallel by harnessing distributed computing systems. However, to achieve efficient parallel execution times, an effective strategy is required to balance the nonuniform partition workloads across the available resources. Furthermore, with the increasing integration of servers with heterogeneous resources and the adoption of resource sharing, balancing workloads is becoming complex. This paper proposes a strategy to execute parallel BP algorithms more efficiently on even shared or heterogeneous distributed systems. The approach integrates a self‐adjusting dynamic partitioning method in the BP algorithm with a dynamic scheduler, provided by an application middleware, which manages the parallel execution while addressing any issues of imbalance. Empirical results indicate better scalability with efficiencies above 90% for instances of an application case study for the discretizable molecular distance geometry problem (DMDGP). Improvements of up to 38% were obtained in execution speed‐ups compared to a more traditional parallel BP implementation for DMDGP.

PHyL v1.0: A parallel, flexible, and advanced software for hydrological and slope stability modeling at a regional scale

Physically-based hydrological-geotechnical modeling at large scales is difficult, especially due to the time-consuming nature of flow routing and 3D soil stability models. Although parallelization techniques are commonly used for each model individually, there is currently no concurrent parallelization strategy for both. This study proposed an open-source, Parallelized, and modular modeling software for regional Hydrologic processes and Landslides simulation and prediction (PHyL v1.0). It offers parallel computation in both hydrological and 3D slope stability modules, cross-scale modeling ability via a soil moisture downscaling method, and advanced input/output (I/O) and post-processing visualization. Additionally, PHyL v1.0 is flexible and extensible, making it compatible with all mainstream operating systems. We applied PHyL v1.0 in the Yuehe River Basin, where the computational efficiencies, parallel performance, parameter sensitivity analysis, and predictive capabilities were evaluated. The PHyL v1.0 is therefore appropriately used as an advanced software for high-resolution and complex simulations of regional floods and landslides.

Architectures of 4x1 and 4x2 OFDM With Two-path Parallel ICI Cancellation Full-Rate Schemes in Mobile Channels

Architectures of 4x1 and 4x2 OFDM With Two-path Parallel ICI Cancellation Full-Rate Schemes in Mobile Channels

ParBFT: An Optimized Byzantine Consensus Parallelism Scheme

ParBFT: An Optimized Byzantine Consensus Parallelism Scheme

Saturn: An Optimized Data System for Multi-Large-Model Deep Learning Workloads

Large models such as GPT-3 and ChatGPT have transformed deep learning (DL), powering applications that have captured the public's imagination. Such models must be trained on multiple GPUs due to their size and computational load, driving the development of a bevy of "model parallelism" techniques and tools. Navigating such parallelism choices, however, is a new burden for DL users such as data scientists, domain scientists, etc., who may lack the necessary systems knowhow. The need for model selection , which leads to many models to train due to hyper-parameter tuning or layer-wise finetuning, compounds the situation with two more burdens: resource apportioning and scheduling. In this work, we unify these three burdens by formalizing them as a joint problem that we call SPASE: Select a Parallelism, Allocate resources, and Schedule. We propose a new information system architecture to tackle the SPASE problem holistically, exploiting the performance opportunities presented by joint optimization. We devise an extensible template for existing parallelism schemes and combine it with an automated empirical profiler for runtime estimation. We then formulate SPASE as an MILP. We find that direct use of an MILP-solver is significantly more effective than several baseline heuristics. We optimize the system runtime further with an introspective scheduling approach. We implement all these techniques into a new data system we call Saturn. Experiments with benchmark DL workloads show that Saturn achieves 39-49% lower model selection runtimes than current DL practice.

Possibility of decryption speed-up by parallel processing in CCA secure hashed ElGamal.

In order to prove the ElGamal CCA(Chosen Ciphertext Attack) security in the random oracle model, it is necessary to use the group where ICDH(Interactive Computational Diffie Hellman) assumption holds. Until now, only bilinear group with complex algebraic structure has been known as the ICDH group. In this paper, we introduce the ICDH group with simple algebraic structure. In other words, we prove that ICDH assumption holds in the integer group with composite modulus. On the basis of this, we propose the CCA secure hashed ElGamal and its fast variant to speed up decryption by parallel processing. Our parallel scheme has the fastest decryption among all CCA secure PKE(Public Key Encryption) schemes implemented in integer group and gives the possibility that ElGamal protocol could be practical when the big modulus numbers are used to resist the quantum attack.

Efficient Inverse Fractional Neural Network-Based Simultaneous Schemes for Nonlinear Engineering Applications

Finding all the roots of a nonlinear equation is an important and difficult task that arises naturally in numerous scientific and engineering applications. Sequential iterative algorithms frequently use a deflating strategy to compute all the roots of the nonlinear equation, as rounding errors have the potential to produce inaccurate results. On the other hand, simultaneous iterative parallel techniques require an accurate initial estimation of the roots to converge effectively. In this paper, we propose a new class of global neural network-based root-finding algorithms for locating real and complex polynomial roots, which exploits the ability of machine learning techniques to learn from data and make accurate predictions. The approximations computed by the neural network are used to initialize two efficient fractional Caputo-inverse simultaneous algorithms of convergence orders ς+2 and 2ς+4, respectively. The results of our numerical experiments on selected engineering applications show that the new inverse parallel fractional schemes have the potential to outperform other state-of-the-art nonlinear root-finding methods in terms of both accuracy and elapsed solution time.