Supercomputing: 9th Russian Supercomputing Days, RuSCDays 2023, Moscow, Russia, September 25–26, 2023, Revised Selected Papers, Part II (Lecture Notes in Computer Science)

✍ Scribed by Vladimir Voevodin (editor), Sergey Sobolev (editor), Mikhail Yakobovskiy (editor), Rashit Shagaliev (editor)

• Publisher: Springer
• Year: 2024
• Tongue: English
• Leaves: 346
• Category: Library

✦ Synopsis

The two-volume set LNCS 14388 and 14389 constitutes the refereed proceedings of the 9th Russian Supercomputing Days International Conference (RuSCDays 2023) held in Moscow, Russia, during September 25-26, 2023.

The 44 full papers and 1 short paper presented in these proceedings were carefully reviewed and selected from 104 submissions. The papers have been organized in the following topical sections: supercomputer simulation; distributed computing; and HPC, BigData, AI: algorithms, technologies, evaluation.

✦ Table of Contents

Preface
Organization
Contents – Part II
Contents – Part I
Distributed Computing
Benchmarking DAG Scheduling Algorithms on Scientific Workflow Instances
1 Introduction
2 DAG Scheduling
2.1 Problem Statement
2.2 Algorithms
3 Benchmark
3.1 Workflow Instances
3.2 System Configurations
3.3 Algorithm Comparison Metrics
4 Simulation Model
5 Benchmark Results
6 Conclusion
References
Classification of Cells Mapping Schemes Related to Orthogonal Diagonal Latin Squares of Small Order
1 Introduction
2 Preliminaries
2.1 Mutually Orthogonal Diagonal Latin Squares
2.2 Cells Mapping Schemes
3 List of Distinct Multisets of Cycle Lengths
4 Searching for MODLS of Order 10 via ESODLS CMS and SAT
5 Conclusions
References
Comparative Analysis of Digitalization Efficiency Estimation Methods Using Desktop Grid
1 Introduction
2 Digitalization Efficiency Estimation Methods
2.1 Modified Method of Digitalization Efficiency Estimation
3 Model
4 Numerical Experiments Based on Desktop Grid
4.1 Results and Discussion
5 Conclusion
References
Diagonalization and Canonization of Latin Squares
1 Introduction
2 Basic Concepts and Definitions
3 Embedding of LS and DLS Isomorphism Classes
4 Diagonalization and Canonization of Latin Squares
5 Practical Application of Canonization and Diagonalization
6 Conclusion
References
Probabilistic Modeling of the Behavior of a Computing Node in the Absence of Tasks on the Project Server
1 Introduction
1.1 Related Work
1.2 Using Random Exponential Backoff in BOINC
2 Probabilistic Characteristics of Node Behavior
2.1 Basix Quantities
2.2 Method for Calculating the Probability of Query Execution
3 Distribution Function Calculation
3.1 Distribution Function
3.2 Recursive Way to Calculate Multiple Integrals
4 Mathematical Expectation
5 Conclusions
References
Using Virtualization Approaches to Solve Deep Learning Problems in Voluntary Distributed Computing Projects
1 Introduction
2 Distributed Deep Learning
3 Voluntary Distributed Computing
3.1 Distributed Computing Platforms
3.2 Desktop Grid System
4 Virtualization
4.1 Virtualization in the Desktop Grid on BOINC
4.2 Application Virtualization
5 Results and Discussion
5.1 Configure the BOINC-Server
5.2 Experiment with Virtual Machine Approach
6 Conclusion
References
Workflows of the High-Throughput Virtual Screening as a Service
1 Introduction
2 HiTViSc Concept and Related Work
3 Workflows of the HiTViSc
3.1 Virtual Screening
3.2 Results Analysis
3.3 Management of the Computational Resources
4 Conclusion
References
HPC, BigData, AI: Algorithms, Technologies, Evaluation
3D Seismic Inversion for Fracture Model Reconstruction Based on Machine Learning
1 Introduction
2 3D Seismic Models and Images
2.1 Discrete Fracture Model
2.2 3D Seismic Migration and Ideal Images
3 Fracture Model Reconstruction Using Machine Learning
3.1 The Dataset for Machine Learning
3.2 The UNet Neural Network for the Machine Learning Algorithm
4 Numerical Examples
4.1 Simplified Training Dataset
4.2 Sophisticated Training Dataset
5 Conclusions
References
A Computational Model for Interactive Visualization of High-Performance Computations
1 Introduction
2 Problem Statement
2.1 Formalization of Online Visualization
3 Computation Model. Basic Level
4 Computation Model. Service Level
4.1 Service for Managing Reactions via Messages
4.2 Query Service
4.3 Task Service
4.4 Payload Service
5 Computation Model. Promise Level
6 Prototype Implementation
7 Example Application
8 Related Works
9 Conclusion and Future Work
References
An Algorithm for Mapping of Global Adjacency Lists to Local Numeration in a Distributed Graph in the GridSpiderPar Tool
1 Introduction
2 The Parallel Incremental Algorithm
3 Graph Storage
4 The Algorithm for Mapping of Global Adjacency Lists to Local Numeration in a Distributed Graph
5 Results
6 Conclusion
References
Construction of Locality-Aware Algorithms to Optimize Performance of Stencil Codes on Heterogeneous Hardware
1 Introduction
2 Problem Statement
2.1 Lattice Boltzmann Method
2.2 Manycore CPU
3 LRnLA Methods
3.1 ConeTorre LRnLA Algorithm
3.2 Special Features
3.3 Localization Properties
4 Construction of an Algorithm for Locality-Aware Parallelism
4.1 Data Structure
4.2 Data Exchange
4.3 Time-Domain Decomposition
4.4 FArShFold Algorithm
4.5 Heterogeneous Cores
4.6 L3 Localization
4.7 Roofline Analysis for Zen3
5 Benchmarks
5.1 Code Implementation
5.2 Performance Tests
5.3 Related Works
6 Conclusion
References
Development of Components for Monitoring and Control Intelligent Information System
1 Introduction
2 Materials and Methods
3 Results
3.1 General Educational and Methodological Support
3.2 Development of the Sugar Yield Assessment Module
3.3 Development of a Sugar Yield Optimization Module
4 Conclusion
References
Image Segmentation Algorithms Composition for Obtaining Accurate Masks of Tomato Leaf Instances
1 Introduction
2 Related Work
2.1 Instance Segmentation
2.2 High-Resolution Segmentation
2.3 Agricultural Industry
3 Dataset
3.1 Agricultural Industry
3.2 Data Markup Methodology
3.3 Data Sampling
3.4 Semi-automatic Annotation
3.5 Preprocessing
4 Method
4.1 Mask R-CNN Settings
4.2 Masks Refinement
4.3 Suppression of False Positive Hypotheses
4.4 Algorithms Composition Inference
4.5 Training Parameters
5 Experiments
6 Semi-automatic Data Markup
7 Conclusion
References
Implementation of Dusty Gas Model Based on Fast and Implicit Particle-Mesh Approach SPH-IDIC in Open-Source Astrophysical Code GADGET-2
1 Introduction
2 Numerical Methods
2.1 Monaghan and Kocharyan Scheme (MK)
2.2 SPH-IDIC Algorithm
2.3 Energy Equation
3 Results
3.1 One-Dimensional Tests
3.2 Three-Dimensional Tests
3.3 Parallelization Efficiency
4 Conclusions
References
MDProcessing.jl: Julia Programming Language Application for Molecular Dynamics Trajectory Processing
1 Introduction
2 Julia Programming Language for Simulation Processing
2.1 Optional Typing and Multiple Dispatch
2.2 Code Compilation
2.3 Library Tools
3 MDProcessing.jl Package
3.1 Data Types
3.2 Customization
3.3 Performance
4 Conclusions
References
Methods and Algorithms for Intelligent Video Analytics in the Context of Solving Problems of Precision Pig Farming
1 Introduction
2 Related Works
3 Dataset
4 Pipeline
4.1 Data Pre-preprocessing
4.2 Instance Segmentation
4.3 Tracking
4.4 Weight Estimation
4.5 Density Maps
5 Tools of Analysis
6 Conclusion
References
Nucleic Acid-Protein Interaction Prediction Using Geometric Deep Learning
1 Introduction
2 Method
2.1 Architecture
2.2 Dataset
2.3 Training
2.4 Implementation
3 Results
3.1 Identification of NA Binding Sites
3.2 Identification of Nucleotide Specificity
3.3 Interaction Prediction
3.4 Parallel Training
4 Conclusion
5 Code Availability
References
Parallel Algorithm for Incompressible Flow Simulation Based on the LS-STAG and Domain Decomposition Methods
1 Introduction
2 The Test Problem Statement
3 Main Ideas of the LS-STAG Method
4 Domain Decomposition Method
4.1 Main Ideas
4.2 Computing LS-STAG Matrices for Subdomains and Interface Problems
5 Numerical Experiments
6 Conclusions
References
Parallel Algorithm for Source Type Recovering by the Time Reversal Mirror
1 Introduction
2 Mathematical Statement
3 Numerical Simulation
4 Description of Algorithm
5 Numerical Test
6 Paralleling Technique
7 Conclusions
References
Recognition of Medical Masks on People's Faces in Difficult Decision-Making Conditions
1 Introduction
2 Related Works
3 Complex Approach to Solving the Problem of Recognizing the Presence and Wearing Correctness of PPE
4 Recognizing the Presence and Correct Wearing of Medical Masks
4.1 Data and Dataset Characteristics
4.2 Automatic Data Markup System for the Detection Task
4.3 Human Head Neural Network Detector
4.4 Classifier
4.5 Examples
5 Conclusion
References
Use of Different Metrics to Generate Training Datasets for a Numerical Dispersion Mitigation Neural Network
1 Introduction
2 Neural Network Architecture
3 Input Data
4 Methods for Creating Training Dataset
4.1 Distance Between the Sources
4.2 Dataset Based on the Velocity Model
4.3 Dataset Based on Distances Between Seismograms
4.4 Numerical Results
5 Global Sensitivity Analysis of Operator G
6 Conclusion
References
Validity and Limitations of Supervised Learning for Phase Transition Research
1 Introduction
2 Previous Work
3 Models
3.1 Ising Models on Square and Triangular Lattices
3.2 Baxter-Wu Model
3.3 Phase Transitions and Universality
4 Data Generation and Deep Learning
4.1 Data Generation
4.2 Neural Network Architectures and Output Data
5 Learning and Testing
6 Influence of Anisotropy
6.1 Ising Model on Square Lattice
6.2 Ising Model on Triangular Lattice
7 Influence of Number of Epochs for Training
7.1 Ordered Phase Prediction in Spin Systems
8 Discussion
References
Author Index