Reduce Memory Usage Research Articles

Minimum-Cost State-Flipped Control for Reachability of Boolean Control Networks Using Reinforcement Learning.

This article proposes model-free reinforcement learning methods for minimum-cost state-flipped control in Boolean control networks (BCNs). We tackle two questions: 1) finding the flipping kernel, namely, the flip set with the smallest cardinality ensuring reachability and 2) deriving optimal policies to minimize the number of flipping actions for reachability based on the obtained flipping kernel. For Question 1), Q-learning's capability in determining reachability is demonstrated. To expedite convergence, we incorporate two improvements: 1) demonstrating that previously reachable states remain reachable after adding elements to the flip set, followed by employing transfer learning and 2) initiating each episode with special initial states whose reachability to the target state set are currently unknown. For Question 2), it is challenging to encapsulate the objective of simultaneously reducing control costs and satisfying terminal constraints exclusively through the reward function employed in the Q-learning framework. To bridge the gap, we propose a BCN-characteristics-based reward scheme and prove its optimality. Questions 1) and 2) with large-scale BCNs are addressed by employing small memory Q-learning, which reduces memory usage by only recording visited action-values. An upper bound on memory usage is provided to assess the algorithm's feasibility. To expedite convergence for Question 2) in large-scale BCNs, we introduce adaptive variable rewards based on the known maximum steps needed to reach the target state set without cycles. Finally, the effectiveness of the proposed methods is validated on both small-and large-scale BCNs.

An open-source, adaptive solver for particle-resolved simulations with both subcycling and non-subcycling methods

We present the IAMReX (incompressible flow with adaptive mesh refinement for the eXascale), an adaptive and parallel solver for particle-resolved simulations on the multi-level grid. The fluid equations are solved using a finite-volume scheme on the block-structured semi-staggered grids with both subcycling and non-subcycling methods. The particle-fluid interaction is resolved using the multidirect forcing immersed boundary method. The associated Lagrangian markers used to resolve fluid-particle interface only exist on the finest-level grid, which greatly reduces memory usage. The volume integrals are numerically calculated to capture the free motion of particles accurately, and the repulsive potential model is also included to account for the particle–particle collision. We demonstrate the versatility, accuracy, and efficiency of the present multi-level framework by simulating fluid-particle interaction problems with various types of kinematic constraints. The cluster of monodisperse particles case is presented at the end to show the capability of the current solver in handling multiple particles. It is demonstrated that the three-level AMR (Adaptive Mesh Refinement) simulation leads to a 72.46% grid reduction compared with the single-level simulation. The source code and testing cases used in this work can be accessed at https://github.com/ruohai0925/IAMR/tree/development. Input scripts and raw postprocessing data are also available for reproducing all results.

Design and implementation of HDF5-format neutron nuclear database in RMC code

This study developed a new structured nuclear database to improve the readability and extensibility of the ACE (A Compact ENDF) nuclear database, save disk space, reduce memory usage, and enhance the computational efficiency of the Reactor Monte Carlo (RMC) code. A Python package was developed to store nuclear data in HDF5 format. Compared to the ACE database, the HDF5-format database shows significant improvements: an 80% reduction in disk space for continuous energy neutron data and a 60% reduction for neutron thermal scattering data. The HDF5-format database was implemented in RMC’s criticality calculation mode and validated through VERA benchmark problem 2B, demonstrating perfect agreement with the ACE results in keff and neutron flux counts. The Computational results indicate a 7.7% reduction in memory usage and a 20.1% improvement in computational efficiency with the HDF5-format database. Additional tests show that using the database at a single temperature point reduces memory usage by 5.4% and running time by 13.2%. At two temperature points, memory usage decreases by 35.2% and running time by 18.0%. The new data structure reduces temperature-independent redundant data and improves indexing efficiency, leading to greater savings with more temperature points. This development enhances performance of criticality calculation of RMC and addresses ACE database limitations.

Benchmarking In-Sensor Machine Learning Computing: An Extension to the MLCommons-Tiny Suite

This paper proposes a new benchmark specifically designed for in-sensor digital machine learning computing to meet an ultra-low embedded memory requirement. With the exponential growth of edge devices, efficient local processing is essential to mitigate economic costs, latency, and privacy concerns associated with the centralized cloud processing. Emerging intelligent sensors equipped with computing assets to run neural network inferences and embedded in the same package, which hosts the sensing elements, present new challenges due to their limited memory resources and computational skills. This benchmark evaluates models trained with Quantization Aware Training (QAT) and compares their performance with Post-Training Quantization (PTQ) across three use cases: Human Activity Recognition (HAR) by means of the SHL dataset, Physical Activity Monitoring (PAM) by means of the PAMAP2 dataset, and superficial electromyography (sEMG) regression with the NINAPRO DB8 dataset. The results demonstrate the effectiveness of QAT over PTQ in most scenarios, highlighting the potential for deploying advanced AI models on highly resource-constrained sensors. The INT8 versions of the models always outperformed their FP32, regarding memory and latency reductions, except for the activations for CNN. The CNN model exhibited reduced memory usage and latency with respect to its Dense counterpart, allowing it to meet the stringent 8KiB data RAM and 32 KiB program RAM limits of the ISPU. The TCN model proved to be too large to fit within the memory constraints of the ISPU, primarily due to its greater capacity in terms of number of parameters, designed for processing more complex signals like EMG. This benchmark aims to guide the development of efficient AI solutions for In-Sensor Machine Learning Computing, fostering innovation in the field of Edge AI benchmarking, such as the one conducted by the MLCommons-Tiny working group.

Neighbor patches merging reduces spatial redundancy to accelerate vision transformer

Vision Transformers (ViTs) deliver outstanding performance but often require substantial computational resources. Various token pruning methods have been developed to enhance throughput by removing redundant tokens; however, these methods do not address the peak memory consumption, which remains equivalent to that of the unpruned networks. In this study, we introduce Neighbor Patches Merging (NEPAM), a method that significantly reduces the maximum memory footprint of ViTs while pruning tokens. NEPAM targets spatial redundancy within images and prunes redundant patches at the onset of the model, thereby achieving the optimal throughput-accuracy trade-off without fine-tuning. Experimental results demonstrate that NEPAM can accelerate the inference speed of the Vit-Base-Patch16-384 model by 25% with a negligible accuracy loss of 0.07% and a notable 18% reduction in memory usage. When applied to VideoMAE, NEPAM doubles the throughput with a 0.29% accuracy loss and a 48% reduction in memory usage. These findings underscore the efficacy of NEPAM in mitigating computational requirements while maintaining model performance.

Assessing progressive collapse regions of reinforced concrete frame structures using Graph Convolutional Networks

In the progressive collapse, the structural topology changed dynamically as failures propagate. Representing reinforced concrete (RC) frames with graph data leverages the topology of beam–column joints and structural components, enabling graph neural networks (GNNs) to predict the failure propagation. Graph convolutional networks (GCNs), a specific type of GNN, offer superior computational efficiency for graph data to traditional GNNs. However, three challenges must be addressed for GCNs to rapidly assess collapse regions: (1) insufficient data, (2) lack of graph representations for structural collapse, and (3) absence of tailored GCN architectures. To address these gaps, this study focused on RC frames with the following initiatives: (1) an efficient and accurate method for generating progressive collapse data was developed; (2) joint-based representation and component-based representation were established, accompanied by a method for converting RC frames into these representations; (3) two GCN architectures were designed to identify failure paths and predicting collapse regions. The performance of the GCN models using various convolutional layer configurations was evaluated on the two graph representations, and the CGR-based model using SAGEConv layers exhibited the best performance with an accuracy and F1 score of 0.9839 and 0.8853, respectively. The proposed method exhibited over a 99 % improvement in calculation efficiency and approximately a 75 % reduction in memory usage compared with traditional FEM. Moreover, it captured the contribution of each component to progressive collapse resistance with better accuracy than previous approaches. Finally, this method was employed to predict the internal and external collapse regions of a real-world structure by incorporating a physics engine-based approach from prior studies.

LiveLattice: Real-time visualisation of tilted light-sheet microscopy data using a memory-efficient transformation algorithm.

Light-sheet fluorescence microscopy (LSFM), a prominent fluorescence microscopy technique, offers enhanced temporal resolution for imaging biological samples in four dimensions (4D; x, y, z, time). Some of the most recent implementations, including inverted selective plane illumination microscopy (iSPIM) and lattice light-sheet microscopy (LLSM), move the sample substrate at an oblique angle relative to the detection objective's optical axis. Data from such tilted-sample-scan LSFMs require subsequent deskewing and rotation for proper visualisation and analysis. Such data preprocessing operations currently demand substantial memory allocation and pose significant computational challenges for large 4D dataset. The consequence is prolonged data preprocessing time compared to data acquisition time, which limits the ability for live-viewing the data as it is being captured by the microscope. To enable the fast preprocessing of large light-sheet microscopy datasets without significant hardware demand, we have developed WH-Transform, a memory-efficient transformation algorithm for deskewing and rotating the raw dataset, significantly reducing memory usage and the run time by more than 10-fold for large image stacks. Benchmarked against the conventional method and existing software, our approach demonstrates linear runtime compared to the cubic and quadratic runtime of the other approaches. Preprocessing a raw 3D volume of 2 GB (512 × 1536 × 600 pixels) can be accomplished in 3 s using a GPU with 24 GB of memory on a single workstation. Applied to 4D LLSM datasets of human hepatocytes, lung organoid tissue and brain organoid tissue, our method provided rapid and accurate preprocessing within seconds. Importantly, such preprocessing speeds now allow visualisation of the raw microscope data stream in real time, significantly improving the usability of LLSM in biology. In summary, this advancement holds transformative potential for light-sheet microscopy, enabling real-time, on-the-fly data preprocessing, visualisation, and analysis on standard workstations, thereby revolutionising biological imaging applications for LLSM and similar microscopes.

A graphics processing unit accelerated sparse direct solver and preconditioner with block low rank compression

We present the GPU implementation efforts and challenges of the sparse solver package STRUMPACK. The code is made publicly available on github with a permissive BSD license. STRUMPACK implements an approximate multifrontal solver, a sparse LU factorization which makes use of compression methods to accelerate time to solution and reduce memory usage. Multiple compression schemes based on rank-structured and hierarchical matrix approximations are supported, including hierarchically semi-separable, hierarchically off-diagonal butterfly, and block low rank. In this paper, we present the GPU implementation of the block low rank (BLR) compression method within a multifrontal solver. Our GPU implementation relies on highly optimized vendor libraries such as cuBLAS and cuSOLVER for NVIDIA GPUs, rocBLAS and rocSOLVER for AMD GPUs and the Intel oneAPI Math Kernel Library (oneMKL) for Intel GPUs. Additionally, we rely on external open source libraries such as SLATE (Software for Linear Algebra Targeting Exascale), MAGMA (Matrix Algebra on GPU and Multi-core Architectures), and KBLAS (KAUST BLAS). SLATE is used as a GPU-capable ScaLAPACK replacement. From MAGMA we use variable sized batched dense linear algebra operations such as GEMM, TRSM and LU with partial pivoting. KBLAS provides efficient (batched) low rank matrix compression for NVIDIA GPUs using an adaptive randomized sampling scheme. The resulting sparse solver and preconditioner runs on NVIDIA, AMD and Intel GPUs. Interfaces are available from PETSc, Trilinos and MFEM, or the solver can be used directly in user code. We report results for a range of benchmark applications, using the Perlmutter system from NERSC, Frontier from ORNL, and Aurora from ALCF. For a high frequency wave equation on a regular mesh, using 32 Perlmutter compute nodes, the factorization phase of the exact GPU solver is about 6.5× faster compared to the CPU-only solver. The BLR-enabled GPU solver is about 13.8× faster than the CPU exact solver. For a collection of SuiteSparse matrices, the STRUMPACK exact factorization on a single GPU is on average 1.9× faster than NVIDIA’s cuDSS solver.

Learning-Augmented Sketching Offers Improved Performance for Privacy Preserving and Secure GWAS.

The introduction of trusted execution environments (TEEs), such as secure enclaves provided by the Intel SGX technology has enabled secure and privacy-preserving computation on the cloud. The stringent resource limitations, such as memory constraints, required by some TEEs necessitates the development of computational approaches with reduced memory usage, such as sketching. One example is the SkSES method for GWAS on a cohort of case and control samples from multiple institutions, which identifies the most significant SNPs in a privacy-preserving manner without disclosing sensitive genotype information to other institutions or the cloud service provider. Here we show how to improve the performance of SkSES on large datasets by augmenting it with a learning-augmented approach. Specifically, we show how individual institutions can perform smaller scale GWAS on their own datasets and identify two sets of variants according to certain criteria, which are then used to guide the sketching process to more accurately identify significant variants over the collective dataset. The new method achieves up to 40% accuracy gain compared to the original SkSES method under the same memory constraints on datasets we tested on. The code is available at https://github.com/alreadydone/sgx-genome-variants-search.

FedCPG: A class prototype guided personalized lightweight federated learning framework for cross-factory fault detection

Industrial equipment condition monitoring and fault detection are crucial to ensure the reliability of industrial production. Recently, data-driven fault detection methods have achieved significant success, but they all face challenges due to data fragmentation and limited fault detection capabilities. Although centralized data collection can improve detection accuracy, the conflicting interests brought by data privacy issues make data sharing between different devices impractical, thus forming the problem of industrial data silos. To address these challenges, this paper proposes a class prototype guided personalized lightweight federated learning framework(FedCPG). This framework decouples the local network, only uploading the backbone model to the server for model aggregation, while employing the head model for local personalized updates, thereby achieving efficient model aggregation. Furthermore, the framework incorporates prototype constraints to steer the local personalized update process, mitigating the effects of data heterogeneity. Finally, a lightweight feature extraction network is designed to reduce communication overhead. Multiple complex industrial data distribution scenarios were simulated on two benchmark industrial datasets. Extensive experiments have demonstrated that FedCPG can achieve an average detection accuracy of 95% in complex industrial scenarios, while simultaneously reducing memory usage and the number of parameters by 82%, surpassing existing methods in most average metrics. These findings offer novel perspectives on the application of personalized federated learning in industrial fault detection.

ECDSA-secured blockchain architecture for interoperable healthcare in the Internet of Robotic Things (IoRT)

ABSTRACT The ever-growing Internet of Robotic Things (IoRT) in healthcare settings introduces new challenges in securing sensitive patient data. Existing communication networks within IoRT environments are vulnerable to security breaches. This paper proposes an ECDSA-secured blockchain architecture specifically designed to address these challenges. A key innovation is the implementation of a private blockchain network on ZedBoard development boards for access control and data integrity verification. IoT devices collect patient health data, which is then securely transmitted to a Mobius server for storage, minimizing resource impact on ZedBoards. To further strengthen the security framework, the architecture integrates the Elliptic Curve Digital Signature Algorithm (ECDSA) with Radix- 2 w optimizations. These optimizations result in a 28% reduction in CPU usage and an 8.96% reduction in memory usage, significantly enhancing cryptographic performance. Practical implementation validates the effectiveness of the proposed approach, providing a robust solution for securing IoRT networks. This research offers valuable insights for industries reliant on the seamless integration of IoT and robotics and concludes with key findings and future research directions, including exploring energy-efficient cryptographic protocols, enhancing scalability through advanced consensus mechanisms, and developing user-friendly interfaces for managing IoRT networks.

Open-AI model Efficient Memory Reduce Management for the Large Language Models (LLMs) Serving with Paged Attention of sharing the KV Cashes

Abstract: High throughput serving of large language models (LLMs) requires batching sufficiently many requests at a time. However, existing systems struggle because the key-value cache (KV cache) memory for each request is huge and grows and shrinks dynamically. When managed inefficiently, this memory can be significantly wasted by fragmentation and redundant duplication, limiting the batch size. To address this problem we proposed a Paged Attention. An alternative algorithm inspired by the classical virtual memory and paging techniques in operating systems. An LLM serving system that achieves (1) near-zero waste in KV cache memory and (2) flexible sharing of KV cache within and across requests to further reduce memory usage. Our evaluations show that LLM improves the throughput of popular LLMs by 2-4× with the same level of latency compared to the state-of-the-art systems, such as Faster Transformer and Orca. The improvement is more pronounced with longer sequences, larger models, and more complex decoding algorithms

Sparsity-Independent Lyapunov Exponent in the Sachdev-Ye-Kitaev Model.

The saturation of a recently proposed universal bound on the Lyapunov exponent has been conjectured to signal the existence of a gravity dual. This saturation occurs in the low-temperature limit of the dense Sachdev-Ye-Kitaev (SYK) model, N Majorana fermions with q body (q>2) infinite-range interactions. We calculate certain out-of-time-order correlators (OTOCs) for N≤64 fermions for a highly sparse SYK model and find no significant dependence of the Lyapunov exponent on sparsity up to near the percolation limit where the Hamiltonian breaks up into blocks. This provides strong support to the saturation of the Lyapunov exponent in the low-temperature limit of the sparse SYK. A key ingredient to reaching N=64 is the development of a novel quantum spin model simulation library that implements highly optimized matrix-free Krylov subspace methods on graphical processing units. This leads to a significantly lower simulation time as well as vastly reduced memory usage over previous approaches, while using modest computational resources. Strong sparsity-driven statistical fluctuations require both the use of a much larger number of disorder realizations with respect to the dense limit and a careful finite size scaling analysis. The saturation of the bound in the sparse SYK points to the existence of a gravity analog that would enlarge substantially the number of field theories with this feature.

DELTA: Memory-Efficient Training via Dynamic Fine-Grained Recomputation and Swapping

To accommodate the increasingly large-scale models within limited-capacity GPU memory, various coarse-grained techniques, such as recomputation and swapping, have been proposed to optimize memory usage. However, these methods have encountered limitations, either in terms of inefficient memory reduction or diminished training performance. In response to this, our paper introduces DELTA, an innovative approach for memory-efficient large-scale model training that combines fine-grained memory optimization and prefetching technology to reduce memory usage while maintaining high training throughput concurrently. Initially, we formulate the problem of memory-throughput joint optimization as an easy-solving 0/1 Knapsack problem. Leveraging this formalization, we use an improving polynomial complexity heuristic algorithm to address the problem effectively. Furthermore, we introduce a novel bidirectional prefetching technology into dynamic memory management, which significantly accelerates the model training when compared to relying solely on recomputation or swapping. Finally, DELTA offers users an automated training execution library, eliminating the need for manual configuration or specialized expertise. Experimental results demonstrate the effectiveness of DELTA in reducing GPU memory consumption. Compared to state-of-the-art methods, DELTA achieves substantial memory savings ranging from 40% to 72%, while maintaining comparable convergence performance for various models, including ResNet-50, ResNet-101, and BERT-Large. Notably, DELTA enables the training of GPT2-Large and GPT2-XL with batch sizes increased by 5.5 × and 6 ×, respectively, showcasing its versatility and practicality in enabling large-scale model training on GPU hardware.

Sparse FIR Filter Design using Double Generalized Orthogonal Matching Pursuit (DGOMP)

In this paper, sparse FIR filter was designed using Double Generalized Orthogonal Matching Pursuit (DGOMP) to reduce memory usage and increasing the speed thereby decreasing computational complexity of the algorithm. Mathematical models were formulated and simulations were conducted to validate the performance of the proposed method. The performance was compared with BOMP and Conventional FIR filter. The results showed that the DGOMP method achieved higher sparsity and a better approximation of an ideal filter. Additionally, the designed sparse FIR filters using DGOMP showed better performance in terms of time of execution when the signal lengths keep increasing, giving a 10% faster execution time when compared to BOMP. The passband and stopband attenuation, as well as ripple values were better, offering the flexibility of parameter adjustment. The results showed that DGOMP is a promising approach for designing sparse FIR filters.

Sensorless control using single current sensor with PFC assisted SRM drive for cost-effective domestic appliances

This paper presents a sensorless control designed for low to medium speed operation of an 8/6 pole switched reluctance motor (SRM) drive, utilizing only a single current and voltage sensor while ensuring grid-side power quality. It emphasizes a cost-effective solution suitable for low-power applications in both domestic and industrial appliances, employing SRM drive technology. In this, AC input is fed to Cuk converter that suppresses grid current total harmonic distortion (THD). Using symmetrical structural properties of SRM allows for a cost-friendly sensorless control approach. Position estimation for four-phase SRM is achieved by comparing two-position reference fluxes with running flux. In case of four-phase SRM, phases A and C are conducted sequentially, enabling a single voltage and current sensor to estimate flux for both phases. This method contrasts with lookup table-based rotor position estimation, requiring only two sets of position flux data, thus reducing memory usage, controller complexity, and speeding up response. A laboratory prototype of a four-phase 8/6 SRM is used to estimate rotor position in sensorless control. Finally, test results are used to validate effectiveness of presented PFC assisted sensorless control approach.

Enhancing computational efficiency in 3-D seismic modelling with half-precision floating-point numbers based on the curvilinear grid finite-difference method

SUMMARY Large-scale and high-resolution seismic modelling are very significant to simulating seismic waves, evaluating earthquake hazards and advancing exploration seismology. However, achieving high-resolution seismic modelling requires substantial computing and storage resources, resulting in a considerable computational cost. To enhance computational efficiency and performance, recent heterogeneous computing platforms, such as Nvidia Graphics Processing Units (GPUs), natively support half-precision floating-point numbers (FP16). FP16 operations can provide faster calculation speed, lower storage requirements and greater performance enhancement over single-precision floating-point numbers (FP32), thus providing significant benefits for seismic modelling. Nevertheless, the inherent limitation of fewer 16-bit representations in FP16 may lead to severe numerical overflow, underflow and floating-point errors during computation. In this study, to ensure stable wave equation solutions and minimize the floating-point errors, we use a scaling strategy to adjust the computation of FP16 arithmetic operations. For optimal GPU floating-point performance, we implement a 2-way single instruction multiple data (SIMD) within the floating-point units (FPUs) of CUDA cores. Moreover, we implement an earthquake simulation solver for FP16 operations based on curvilinear grid finite-difference method (CGFDM) and perform several earthquake simulations. Comparing the results of wavefield data with the standard CGFDM using FP32, the errors introduced by FP16 are minimal, demonstrating excellent consistency with the FP32 results. Performance analysis indicates that FP16 seismic modelling exhibits a remarkable improvement in computational efficiency, achieving a speedup of approximately 1.75 and reducing memory usage by half compared to the FP32 version.

Improving long-term electricity time series forecasting in smart grid with a three-stage channel-temporal approach

—Transformer-based models have shown progress in addressing electricity time series forecasting challenges. However, as the forecasting horizon extends, the computational complexity required to capture long-term global correlations may limit their ability to utilize extensive historical data. This paper proposes a non-Transformer model named Three-Stage Channel-Temporal (TSCT), designed to be lightweight and capable of handling longer look-back windows for long-term electricity time series forecasting (LTESF) in smart grid contexts. TSCT sequentially derives feature maps along two dimensions, channel and temporal, focusing on ‘which' and ‘when', respectively. Moreover, its dynamic capacity to decompose and fuse information enables the disentanglement of intricate temporal patterns, highlighting the fundamental characteristics inherent in the time series. Extensive experiments demonstrate that our proposed TSCT outperforms state-of-the-art methods in smart grid scenarios using a commonly used Electricity dataset. Notably, the TSCT approach exhibits significantly higher efficiency compared to Transformer-based methods: an impressive 85% reduction in trainable parameters, a substantial 99% reduction in GPU memory usage, a 94% reduction in running time, and a 49% reduction in inference time. Code is available at: https://github.com/Zhao-Sun/TSCT.

Enhancing image data security with chain and non-chain Galois ring structures

The local ring-based cryptosystem is built upon the core mathematical operations of the algebraic structure of local rings, which provides a significant advantage in ensuring security against advanced cryptanalysis. However, using the entire set of units within this structure would be computationally impractical in practical applications. Thus, we present a novel approach for designing 16×16 and 32×32 substitution boxes over chain and non-chain Galois ring respectively, by utilizing subgroups of local rings in a computationally feasible manner. Our proposed scheme significantly reduces memory usage by integrating both chain and non-chain rings. Specifically, the 16×16 and 32×32 S-boxes require only 16×28and 32×28 memory cells, respectively, whereas S-boxes of these dimensions over fields have been shown to be highly inefficient due to their extensive memory requirements (16×216and 32×232,respectively). The proposed method offers a more efficient solution for constructing S-boxes over large Galois fields and integrates chain and non-chain local Galois rings, as well as finite fields, for efficient transmission designs in smart devices. The algebraic structures used in this approach are used to establish the most vital aspect of a block cipher, the substitution box. We use each of these algebraic structures, each with a different bit size, to construct three distinct S-boxes. We discuss the performance and sensitivity of the proposed S-boxes to demonstrate their effectiveness in data protection. In this technique, the substitution boxes established by the non-chain ring, maximal cyclic subgroup of the Galois ring, and Galois field are applied for the processes of substitution, exclusive-or function, and diffusion in image encryption, respectively. Furthermore, we have performed numerous standard analyses on the encrypted image, including statistical analysis, differential analysis, and NIST tests. The results of these tests demonstrate the effectiveness of the proposed approach for various cryptographic purposes. Overall, this work provides a valuable contribution to the field of cryptography, particularly in constructing efficient S-boxes over large Galois fields.

Enhanced Multi-Beam Echo Sounder Simulation through Distance-Aided and Height-Aided Sound Ray Marching Algorithms

The study proposes two innovative algorithms in the field of multi-beam echo sounder (MBES) simulation: distance-aided sound ray marching (DASRM) and height-aided sound ray marching (HASRM). These algorithms aim to enhance the efficiency and accuracy of MBES simulations, particularly when dealing with long-distance propagation and real-time processing limitations. DASRM addresses issues related to simulation accuracy by efficiently utilizing the KD-tree for spatial indexing and intersection detection instead of the signed distance field (SDF). Building upon the further analysis of DASRM, HASRM is proposed, which improves the search strategy for ray intersections and utilizes a height field pyramid for sampling and retrieval, thereby reducing memory usage while enhancing indexing efficiency. The experimental results demonstrate that both algorithms significantly outperform traditional methods in terms of simulation time, with HASRM exhibiting particular advantages in parallel computing due to its data structure and improved strategies. Additionally, DASRM is well suited for applications requiring complex scene construction, while HASRM proves especially effective in simulating MBES with a focus on underwater terrain due to its effectiveness in handling large incident angles and long-distance propagation.