Computational Efficiency Research Articles

BRDF-Based Photometric Modeling of LEO Constellation Satellite from Massive Observations

BRDF-Based Photometric Modeling of LEO Constellation Satellite from Massive Observations

TorchMD-Net 2.0: Fast Neural Network Potentials for Molecular Simulations.

Achieving a balance between computational speed, prediction accuracy, and universal applicability in molecular simulations has been a persistent challenge. This paper presents substantial advancements in TorchMD-Net software, a pivotal step forward in the shift from conventional force fields to neural network-based potentials. The evolution of TorchMD-Net into a more comprehensive and versatile framework is highlighted, incorporating cutting-edge architectures such as TensorNet. This transformation is achieved through a modular design approach, encouraging customized applications within the scientific community. The most notable enhancement is a significant improvement in computational efficiency, achieving a very remarkable acceleration in the computation of energy and forces for TensorNet models, with performance gains ranging from 2× to 10× over previous, nonoptimized, iterations. Other enhancements include highly optimized neighbor search algorithms that support periodic boundary conditions and smooth integration with existing molecular dynamics frameworks. Additionally, the updated version introduces the capability to integrate physical priors, further enriching its application spectrum and utility in research. The software is available at https://github.com/torchmd/torchmd-net.

Model-Free Nonstationary Reinforcement Learning: Near-Optimal Regret and Applications in Multiagent Reinforcement Learning and Inventory Control

We consider model-free reinforcement learning (RL) in nonstationary Markov decision processes. Both the reward functions and the state transition functions are allowed to vary arbitrarily over time as long as their cumulative variations do not exceed certain variation budgets. We propose Restarted Q-Learning with Upper Confidence Bounds (RestartQ-UCB), the first model-free algorithm for nonstationary RL, and show that it outperforms existing solutions in terms of dynamic regret. Specifically, RestartQ-UCB with Freedman-type bonus terms achieves a dynamic regret bound of [Formula: see text], where S and A are the numbers of states and actions, respectively, [Formula: see text] is the variation budget, H is the number of time steps per episode, and T is the total number of time steps. We further present a parameter-free algorithm named Double-Restart Q-UCB that does not require prior knowledge of the variation budget. We show that our algorithms are nearly optimal by establishing an information-theoretical lower bound of [Formula: see text], the first lower bound in nonstationary RL. Numerical experiments validate the advantages of RestartQ-UCB in terms of both cumulative rewards and computational efficiency. We demonstrate the power of our results in examples of multiagent RL and inventory control across related products. This paper was accepted by Omar Besbes, revenue management and market analytics. Funding: The research of D. Simchi-Levi and R. Zhu was supported by the MIT Data Science Laboratory. The research of W. Mao, K. Zhang, and T. Başar was supported in part by the U.S. Army Research Laboratory (ARL) Cooperative Agreement W911NF-17-2-0196, in part by the Office of Naval Research (ONR) [MURI Grant N00014-16-1-2710], and in part by the Air Force Office of Scientific Research (AFOSR) [Grant FA9550-19-1-0353]. K. Zhang also acknowledges support from U.S. Army Research Laboratory (ARL) [Grant W911NF-24-1-0085]. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2022.02533 .

Innovative approach to estimate structural damage using linear regression and K-nearest neighbors machine learning algorithms

Conventional structural design methodologies often utilize elastic analysis techniques, such as the equivalent static force method and the response spectrum method. While these methods are known for their simplicity and computational efficiency, they prove inadequate in capturing the extent of structural damage caused by seismic forces. Additionally, employing nonlinear dynamic analysis to estimate structural damage represents a challenging and intricate task, posing difficulties for many structural designers. Consequently, the objective of this paper is to present an innovative methodology for evaluating seismic structural damage of moment-resisting frame structures. This involves the utilization of machine learning algorithms, which have been trained and tested on a large data set generated using a newly developed and numerically efficient simulation procedure. The machine learning algorithms employ both linear regression and K-nearest neighbors approaches to accurately replicate the Park-Ang structural damage index.

Analytical insights into the behavior of finite amplitude waves in plasma fluid dynamics

This study introduces innovative analytical solutions for the [Formula: see text]-dimensional nonlinear Jaulent–Miodek ([Formula: see text]) equation, a governing model elucidating the propagation characteristics of nonlinear shallow water waves with finite amplitude. Employing analytical methodologies such as the Khater II and unified methods, alongside the Adomian decomposition method as a semi-analytical approach, series solutions are derived with the primary aim of elucidating the fundamental physics dictating the evolution of [Formula: see text] waves. Within the realm of nonlinear fluid dynamics, the [Formula: see text] equation encapsulates the behavior of irrotational, inviscid, and incompressible fluid flow, wherein nonlinear effects and dispersion intricately balance to yield stable propagating waves. This equation encompasses terms representing nonlinear convection, dispersion, and nonlinearity effects. The analytical methodologies employed in this investigation yield solutions for various instances of the [Formula: see text] equation, demonstrating convergence, accuracy, and computational efficiency. The outcomes reveal that the Adomian decomposition method yields solutions congruent with those obtained through analytical techniques, thereby affirming the precision of the derived solutions. Furthermore, this study advances the comprehension of the physical implications inherent in the [Formula: see text] equation, serving as a benchmark for evaluating alternative methodologies. The analytical approaches elucidated in this research furnish accessible tools for addressing a diverse array of nonlinear wave equations in mathematical physics and engineering domains. In summary, the introduction of novel exact and approximate solutions significantly contributes to the advancement of knowledge pertaining to the [Formula: see text]-dimensional [Formula: see text] equation. The ramifications of this research extend to the modeling of shallow water waves, offering invaluable insights for researchers and practitioners engaged in the field.

Interatomic Interaction Models for Magnetic Materials: Recent Advances

Abstract Atomistic modeling is a widely employed theoretical method of computational materials science. It has found particular utility in the study of magnetic materials. Initially, magnetic empirical interatomic potentials or spin-polarized density functional theory (DFT) served as the primary models for describing interatomic interactions in atomistic simulations of magnetic systems. Furthermore, in recent years, a new class of interatomic potentials known as magnetic machine-learning interatomic potentials (magnetic MLIPs) has emerged. These MLIPs combine the computational efficiency, in terms of CPU time, of empirical potentials with the accuracy of DFT calculations. In this review, our focus lies on providing a comprehensive summary of the interatomic interaction models developed specifically for investigating magnetic materials. We also delve into the various problem classes to which these models can be applied. Finally, we offer insights into the future prospects of interatomic interaction model development for the exploration of magnetic materials.

TOPAS simulation of photoneutrons in radiotherapy: accuracy and speed with variance reduction

Objective. We provide optimal particle split numbers for speeding up TOPAS Monte Carlo simulations of linear accelerator (linac) treatment heads while maintaining accuracy. In addition, we provide a new TOPAS physics module for simulating photoneutron production and transport. Approach. TOPAS simulation of a Siemens Oncor linac was used to determine the optimal number of splits for directional bremsstrahlung splitting as a function of the field size for 6 MV and 18 MV x-ray beams. The linac simulation was validated against published data of lateral dose profiles and percentage depth-dose curves (PDD) for the largest square field (40 cm side). In separate simulations, neutron particle split and the custom TOPAS physics module was used to generate and transport photoneutrons, called ‘TsPhotoNeutron’. Verification of accuracy was performed by comparing simulations with published measurements of: (1) neutron yields as a function of beam energy for thick targets of Al, Cu, Ta, W, Pb and concrete; and (2) photoneutron energy spectrum at 40 cm laterally from the isocenter of the Oncor linac from an 18 MV beam with closed jaws and MLC. Main results. The optimal number of splits obtained for directional bremsstrahlung splitting enhanced the computational efficiency by two orders of magnitude. The efficiency decreased with increasing beam energy and field size. Calculated lateral profiles in the central region agreed within 1 mm/2% from measured data, PDD curves within 1 mm/1%. For the TOPAS physics module, at a split number of 146, the efficiency of computing photoneutron yields was enhanced by a factor of 27.6, whereas it improved the accuracy over existing Geant4 physics modules. Significance. This work provides simulation parameters and a new TOPAS physics module to improve the efficiency and accuracy of TOPAS simulations that involve photonuclear processes occurring in high-Z materials found in linac components, patient devices, and treatment rooms, as well as to explore new therapeutic modalities such as very-high energy electron therapy.

LLMEffiChecker:Understanding and Testing Efficiency Degradation of Large Language Models

Large Language Models (LLMs) have received much recent attention due to their human-level accuracy. While existing works mostly focus on either improving accuracy or testing accuracy robustness, the computation efficiency of LLMs, which is of paramount importance due to often vast generation demands and real-time requirements, has surprisingly received little attention. In this paper, we make the first attempt to understand and test potential computation efficiency robustness in state-of-the-art LLMs. By analyzing the working mechanism and implementation of 20,543 public-accessible LLMs, we observe a fundamental property in LLMs that could be manipulated in an adversarial manner to reduce computation efficiency significantly. Our interesting observation is that the output length determines the computation efficiency of LLMs instead of the input, where the output length depends on two factors: an often sufficiently large yet pessimistic pre-configured threshold controlling the max number of iterations and a runtime generated end of sentence (EOS) token. Our key motivation is to generate test inputs that could sufficiently delay the generation of EOS such that LLMs would have to go through enough iterations to satisfy the pre-configured threshold. We present LLMEffiChecker , which can work under both white-box setting and black-box setting. In the white-box scenario, LLMEffiChecker develops a gradient-guided technique that searches for a minimal and unnoticeable perturbation at character-level, token-level, and structure-level. In the black-box scenario, LLMEffiChecker employs a causal inference-based approach to find critical tokens and similarly applies three levels of imperceptible perturbation to them. Both the white-box and black-box settings effectively delay the appearance of EOS, compelling these inputs to reach the naturally-unreachable threshold. To demonstrate the effectiveness of LLMEffiChecker , we conduct a systematic evaluation on nine public-available LLMs: Google T5, AllenAI WMT14, Helsinki-NLP translator, Facebook FairSeq, UNICAMP-DL translator, MarianMT, Google FLAN-T5, MBZUAI LaMini-GPT and Salesforce CodeGen. Experimental results show that LLMEffiChecker can increase on average LLMs’ response latency and energy consumption by 325% to 3244% and 344% to 3616%, respectively, by perturbing just one character or token in the input sentence. Our case study shows that inputs generated by LLMEffiChecker significantly affect the battery power in real-world mobile devices ( i.e. , drain more than 30 times battery power than normal inputs).

Semi-TSGAN: Semi-Supervised Learning for Highlight Removal Based on Teacher-Student Generative Adversarial Network.

Despite recent notable advancements in highlight image restoration techniques, the dearth of annotated data and the lightweight deployment of highlight removal networks pose significant impediments to further advancements in the field. In this paper, to the best of our knowledge, we first propose a semi-supervised learning paradigm for highlight removal, merging the fusion version of a teacher-student model and a generative adversarial network, featuring a lightweight network architecture. Initially, we establish a dependable repository to house optimal predictions as pseudo ground truth through empirical analyses guided by the most reliable No-Reference Image Quality Assessment (NR-IQA) method. This method serves to assess rigorously the quality of model predictions. Subsequently, addressing concerns regarding confirmation bias, we integrate contrastive regularization into the framework to curtail the risk of overfitting on inaccurate labels. Finally, we introduce a comprehensive feature aggregation module and an extensive attention mechanism within the generative network, considering a balance between network performance and computational efficiency. Our experimental evaluations encompass comprehensive assessments on both full-reference and non-reference highlight benchmarks. The results demonstrate conclusively the substantive quantitative and qualitative enhancements achieved by our proposed algorithm in comparison to state-of-the-art methodologies.

Prediction and Analysis of Transient Turbine Tip Clearance Using Long Short-Term Memory Neural Network

Abstract The transient turbine tip clearance (δ) throughout the engine process is crucial to modern high-performance aero engines. However, there is still a lack of efficient and accurate transient prediction models of tip clearances with active thermal control (ATC) system, especially for the tip clearances of the complex turbine structures with various parameters. This study develops a transient prediction model for the tradeoff between computational efficiency and accuracy, which includes an offline dataset generation process and an online δ prediction process. The offline dataset is first generated using an in-house finite element analysis code, which is validated against a transient tip clearance experiment, and data splicing and sensitivity analysis are applied to enrich the sample features and reduce the input parameters' dimensionality. Then, the long short-term memory neural network (LSTM) is employed to learn the transient tip clearances' timing information. The time consumption for the transient prediction model is significantly shorter than that for the tip clearance calculation method by three orders, and the maximum relative error is as low as 3.59%. In addition, the transient characteristics, including the overshoot value (σ) and the response time (ts), are investigated with different jet Reynolds numbers (Rec) and temperatures (Tfc) of ATC cooling flow. The ts decreases with larger Rec and smaller Tfc due to a more significant cooling effect. However, the σ increases with the increase of Rec and Tfc due to the different sensitivity of cooling parameters. This study provides a reference for the transient tip clearance prediction and the adjustments in the cooling strategies.

Lightweight Hash-Based Authentication Protocol for Smart Grids.

Smart grids integrate information and communications technology into the processes of electricity production, transportation, and consumption, thereby enabling interactions between power suppliers and consumers to increase the efficiency of the power grid. To achieve this, smart meters (SMs) are installed in households or buildings to measure electricity usage and allow power suppliers or consumers to monitor and manage it in real time. However, SMs require a secure service to address malicious attacks during memory protection and communication processes and a lightweight communication protocol suitable for devices with computational and communication constraints. This paper proposes an authentication protocol based on a one-way hash function to address these issues. This protocol includes message authentication functions to address message tampering and uses a changing encryption key for secure communication during each transmission. The security and performance analysis of this protocol shows that it can address existing attacks and provides 105,281.67% better computational efficiency than previous methods.

SIAlex: Species identification and monitoring based on bird sound features

The combination of deep learning and bird sound recognition is widely employed in bird species conservation monitoring. A complex network structure is not conducive for deploying bird sound recognition devices, resulting in problems such as long inference time and low efficiency. Using AlexNet as the backbone model, we explore the potential of shallow and straightforward models without complex connection techniques or attention mechanisms, named SIAlex, to recognise and classify 20 bird sound datasets, which are simultaneously validated on a 10 class UrbanSound8k dataset. Using the structural re-parameterization method, the number of model layers is reduced, computational efficiency is improved, and the inference time is significantly reduced, achieving a decoupling of training and inference time in the structure. To increase the nonlinearity of the model, a cascaded approach is utilised to increase the number of activation functions, thereby significantly improving the generalisation performance of the model. Simultaneously, in the classifier section, convolutional layer replaces the original fully connected layer, thereby reducing the inference time and increasing the feature extraction ability of the model, improving accuracy, and effectively recognising bird speech. The experimental data show that the SIAlex network on the Birdsdata dataset improves the accuracy to 93.66%, and the inference time for a piece of data is only 2.466 ms. The accuracy of the UrbanSound8k dataset reaches 96.04%, and the inference time for a piece of data is 3.031 ms. A large number of experimental comparisons have shown that the method proposed in this paper achieves good results in reducing the inference time of the model, bringing breakthroughs in the application of shallow, simple models.

Dynamic phase-differencing profilometry with number-theoretical phase unwrapping and interleaved projection

High-speed 3D measurement is receiving increasing attention. However, simultaneously achieving high computational efficiency, algorithmic robustness, and reconstructing ratio is challenging. Therefore, a dynamic phase-differencing profilometry (DPDP) is proposed. By capturing the minimum three phase-shifting sinusoidal deformed patterns and establishing a brand-new model, the phase difference between the object on the reference plane and the reference plane is directly resolved to effectively improve computational efficiency. Although it is wrapped, by using only two auxiliary complementary gratings with a purposely designed lower frequency, a DPDP-based number-theoretical temporal phase unwrapping (NT-TPU) algorithm is also proposed to unwrap the wrapped phase difference rather than the phase itself with high robustness. Furthermore, compared to existing PSP-based NT-TPU, the proposed NT-TPU can normally work under more relaxed restrictions. In order to accomplish a high reconstructing ratio, a pentabasic interleaved projection (PIP) strategy based on time division multiplexing is proposed. It can improve the reconstructing ratio from one reconstruction per every five patterns to an equivalent of one reconstruction per every 1.67 patterns. Experimental results demonstrate that the proposed method achieves high computational efficiency, high algorithmic robustness, and high reconstructing ratio simultaneously and has prospective application in high-speed 3D measurement.

Simulation-Driven Universal Surrogates of Coupled Mechanical Systems: Real-Time Simulation of a Forestry Crane

Abstract Preparing simulation-driven surrogates for a coupled mechanical system can be challenging because the associated mechanical and actuator dynamics demand high-fidelity numerical solutions. Proposed here is a universal hydraulic surrogate (UHS), which can provide solutions to high-fidelity mechanical systems with a universal actuator in a surrogate-assisted monolithic approach. The UHS acts as an alternative to the standard lumped fluid theory by eliminating the hydraulic pressures differential equations. A surrogate-assisted universal actuator uses an approximated model to define hydraulic force in high-fidelity mechanical systems. The approximated force model was developed through training against the dynamics of a one-dimensional (1D) hydraulic cylinder and spring-damper. A covariance matrix adaption evolutionary strategy (CMA-ES) was used as an optimization algorithm to minimize differences between the standard dynamics and UHS approaches at the position and velocity levels. The robustness of resulting UHS was validated to predict the behaviors of the simple four-bar mechanism and the forestry crane. The focus was on numerical accuracy and computational efficiency. The maximum percent normalized root mean square error (PN-RMSE) between the states of the approximated force model and lumped fluid theory were approximately 2.04% and 6.95%, respectively. The proposed method was approximately 52 times faster than the standard lumped fluid theory method. By providing accurate predictions outside the training data, the simulation-driven UHS promises better computational performance leading to real-time simulation solutions for the coupled mechanical systems. The UHS can be applied in simulation, optimization, control, state and parameter estimation, and Artificial Intelligence (AI) implementations for coupled mechanical systems.

Discrete Boltzmann model with split collision for nonequilibrium reactive flows

Abstract A multi-relaxation-time discrete Boltzmann model (DBM) with split collision is proposed for both subsonic and supersonic compressible reacting flows, where chemical reactions take place among various components. The physical model is based on a unified set of discrete Boltzmann equations that describes the evolution of each chemical species with adjustable acceleration, specific heat ratio, and Prandtl number. On the righ-hand side of discrete Boltzmann equations, the collision, force, and reaction terms denote the change rates of distribution functions due to self- and cross-collisions, external forces, and chemical reactions, respectively. The source terms can be calculated in three ways, among which the matrix inversion method possesses the highest physical accuracy and computational efficiency. Through Chapman-Enskog analysis, it is proved that the DBM is consistent with the reactive Navier-Stokes equations, Fick's law and Stefan-Maxwell diffusion equation in the hydrodynamic limit. Compared with the one-step-relaxation model, the split collision model offers a detailed and precise description of hydrodynamic, thermodynamic, and chemical nonequilibrium effects. Finally, the model is validated by six benchmarks, including multicomponent diffusion, mixture in the force field, Kelvin-Helmholtz instability, flame at constant pressure, opposing chemical reaction, and steady detonation.

Admission Prediction Using Time Series Analysis

Efficient resource allocation and strategic planning in educational institutions heavily rely on accurate predictions of student admissions. This paper presents a detailed investigation into the application of time series analysis techniques for admission prediction. We explore the utilization of historical admission data, spanning multiple academic cycles, to train and evaluate several time series models.The study encompasses a comparative analysis of traditional statistical models such as autoregressive integrated moving average (ARIMA) and seasonal autoregressive integrated moving average (SARIMA) alongside more advanced deep learning techniques like long short-term memory (LSTM) networks. By evaluating the performance metrics, including accuracy, robustness, and computational efficiency, we aim to identify the most suitable model for admission prediction tasks. Moreover, the research delves into the impact of various external factors such as changes in the academic calendar, socio-economic indicators, and demographic shifts on admission patterns. Understanding these factors is crucial for enhancing the predictive capabilities of the models and enabling institutions to adapt their strategies accordingly. The experimental results and comparative analysis provide valuable insights for educational institutions, enabling them to make data-driven decisions regarding enrollment management strategies and resource allocation. Ultimately, this research contributes to the advancement of admission prediction methodologies, facilitating more efficient and informed decision-making processes in the educational domain.

Focus-Attention Approach in Optimizing DETR for Object Detection from High-Resolution Images

Transformer-based detectors have recently achieved remarkable success in object detection, revolutionizing the field with their efficiency and accuracy. However, applying these models to high-resolution images presents significant challenges due to the increased computational demands and complexity of processing dense, high-resolution data. In this paper, we introduce a novel model specifically designed for object detection from high-resolution imagery. This model incorporates a multi-layer object-focus network along with a transformer encoder-decoder structure. Specifically, the model employs a dual-head strategy in the object-focus network to balance the detailed analysis of small objects with computational efficiency. This is achieved by leveraging data sparsity to reduce unnecessary computations in massive background regions. Additionally, to improve detection performance for small objects, we propose a method to effectively apply the transformer encoder-decoder structure combined with the object-focus network on the multi-layer feature maps of the feature pyramid. Our extensive evaluations of the VisDrone, MS-COCO, and UAVid datasets demonstrate that our model outperforms other DETR-based detectors in both detection accuracy and computational speed, highlighting its superior performance. These results indicate a significant advancement in the field of high-resolution object detection utilizing transformer architecture.

The Alternating Direction Method of Multipliers for Sufficient Dimension Reduction

The minimum average variance estimation (MAVE) method has proven to be an effective approach to sufficient dimension reduction. In this study, we apply the computationally efficient optimization algorithm named alternating direction method of multipliers (ADMM) to a particular approach (MAVE or minimum average variance estimation) to the problem of sufficient dimension reduction (SDR). Under some assumptions, we prove that the iterative sequence generated by ADMM converges to some point of the associated augmented Lagrangian function. Moreover, that point is stationary. It also presents some numerical simulations on synthetic data to demonstrate the computational efficiency of the algorithm.

A fast bootstrap algorithm for causal inference with large data.

Estimating causal effects from large experimental and observational data has become increasingly prevalent in both industry and research. The bootstrap is an intuitive and powerful technique used to construct standard errors and confidence intervals of estimators. Its application however can be prohibitively demanding in settings involving large data. In addition, modern causal inference estimators based on machine learning and optimization techniques exacerbate the computational burden of the bootstrap. The bag of little bootstraps has been proposed in non-causal settings for large data but has not yet been applied to evaluate the properties of estimators of causal effects. In this article, we introduce a new bootstrap algorithm called causal bag of little bootstraps for causal inference with large data. The new algorithm significantly improves the computational efficiency of the traditional bootstrap while providing consistent estimates and desirable confidence interval coverage. We describe its properties, provide practical considerations, and evaluate the performance of the proposed algorithm in terms of bias, coverage of the true 95% confidence intervals, and computational time in a simulation study. We apply it in the evaluation of the effect of hormone therapy on the average time to coronary heart disease using a large observational data set from the Women's Health Initiative.

Quantum Vision Transformers for Quark–Gluon Classification

We introduce a hybrid quantum-classical vision transformer architecture, notable for its integration of variational quantum circuits within both the attention mechanism and the multi-layer perceptrons. The research addresses the critical challenge of computational efficiency and resource constraints in analyzing data from the upcoming High Luminosity Large Hadron Collider, presenting the architecture as a potential solution. In particular, we evaluate our method by applying the model to multi-detector jet images from CMS Open Data. The goal is to distinguish quark-initiated from gluon-initiated jets. We successfully train the quantum model and evaluate it via numerical simulations. Using this approach, we achieve classification performance almost on par with the one obtained with the completely classical architecture, considering a similar number of parameters.