Computational Cost Research Articles

Anisotropic multidimensional smoothing using Bayesian tensor product P-splines

Abstract We introduce a highly efficient fully Bayesian approach for anisotropic multidimensional smoothing. The main challenge in this context is the Markov chain Monte Carlo (MCMC) update of the smoothing parameters as their full conditional posterior comprises a pseudo-determinant that appears to be intractable at first sight. As a consequence, existing implementations are computationally feasible only for the estimation of two-dimensional tensor product smooths, which is, however, too restrictive for many applications. In this paper, we break this barrier and derive closed-form expressions for the log-pseudo-determinant and its first and second order partial derivatives. These expressions are valid for arbitrary dimension and very fast to evaluate, which allows us to set up an efficient MCMC sampler with derivative-based Metropolis–Hastings (MH) updates for the smoothing parameters. We derive simple formulas for low-dimensional slices and averages to facilitate visualization and investigate hyperprior sensitivity. We show that our new approach outperforms previous suggestions in the literature in terms of accuracy, scalability and computational cost and demonstrate its applicability through an illustrating temperature data example from spatio-temporal statistics.

Simulation-based optimization of a production system topology - a neural network-assisted genetic algorithm

ABSTRACT There is an abundance of prior research on the optimization of production systems, but there is a research gap when it comes to optimizing which components should be included in a design, and how they should be connected. To overcome this gap, a novel approach is presented for topology optimization of production systems using a genetic algorithm (GA). This GA employs similarity-based mutation and recombination for the creation of offspring, and discrete-event simulation for fitness evaluation. To reduce computational cost, an extension to the GA is presented in which a neural network functions as a surrogate model for simulation. Three types of neural networks are compared, and the type most effective as a surrogate model is chosen based on its optimization performance and computational cost. Both the unassisted GA and neural network-assisted GA are applied to an industrial case study and a scalability case study. These show that both approaches are effective at finding the optimal solution in industrial settings, and both scale well as the number of potential solutions increases, with the neural network-assisted GA having the better scalability of the two.

Tackling the Temporal Stiffness of Kinetic Monte Carlo Simulations of Well-Mixed Chemical Systems via On-the-Fly Scaling and Cost-Error Optimization.

Reaction kinetics in biological systems are often subject to stochastic effects due to the low populations of reacting molecules, necessitating the adoption of kinetic Monte Carlo methods for their study. Such methods, however, can be computationally expensive, especially in the case of stiff systems, where some reactions are executed at much higher frequencies than others. We present an algorithm that reduces the reaction rate constants of the fast processes on-the-fly, thereby saving computational time, while keeping the approximation error within desirable limits. The algorithm couples the Modified Next Reaction Method for simulating stochastic systems with the Common Random Number framework and calculates accurate metrics for both the computational cost and approximation error by generating multiple sets of trajectories that correspond to increasingly reduced (downscaled) reaction rate constants. The optimum downscale factor is chosen via optimization of two conflicting objectives: (a) maximizing the speedup and (b) minimizing the approximation error introduced, and it is straightforward to tune the performance of the method, favoring accuracy versus speed or vice versa. Our approach is demonstrated on a biology-inspired well-mixed stiff system and is shown to accelerate the stochastic simulation thereof from 66 h down to 90 min, achieving a speed-up factor of 44×, without distorting the dynamics of the system studied.

Segmentation of cortical bone, trabecular bone, and medullary pores from micro-CT images using 2D and 3D deep learning models.

Computed tomography (CT) enables rapid imaging of large-scale studies of bone, but those datasets typically require manual segmentation, which is time-consuming and prone to error. Convolutional neural networks (CNNs) offer an automated solution, achieving superior performance on image data. In this methodology-focused paper, we used CNNs to train segmentation models from scratch on 2D and 3D patches from micro-CT scans of otter long bones. These new models, collectively called BONe (Bone One-shot Network), aimed to be fast and accurate, and we expected enhanced results from 3D training due to better spatial context. Contrary to expectations, 2D models performed slightly better than 3D models in labeling details such as thin trabecular bone. Although lacking in some detail, 3D models appeared to generalize better and predict smoother internal surfaces than 2D models. However, the massive computational costs of 3D models limit their scalability and practicality, leading us to recommend 2D models for bone segmentation. BONe models showed potential for broader applications with variation in performance across species and scan quality. Notably, BONe models demonstrated promising results on skull segmentation, suggesting their potential utility beyond long bones with further refinement and fine-tuning.

Beyond CCSD(T) Accuracy at Lower Scaling with Auxiliary Field Quantum Monte Carlo.

We introduce a black-box auxiliary field quantum Monte Carlo (AFQMC) approach to perform highly accurate electronic structure calculations using configuration interaction singles and doubles (CISD) trial states. This method consistently provides more accurate energy estimates than coupled cluster singles and doubles with perturbative triples (CCSD(T)), often regarded as the gold standard in quantum chemistry. This level of precision is achieved at a lower asymptotic computational cost, scaling as O(N6) compared to the O(N7) scaling of CCSD(T). We provide numerical evidence supporting these findings through results for challenging main group and transition metal-containing molecules.

A Systematic Review of CNN Architectures, Databases, Performance Metrics, and Applications in Face Recognition

This study provides a comparative evaluation of face recognition databases and Convolutional Neural Network (CNN) architectures used in training and testing face recognition systems. The databases span from early datasets like Olivetti Research Laboratory (ORL) and Facial Recognition Technology (FERET) to more recent collections such as MegaFace and Ms-Celeb-1M, offering a range of sizes, subject diversity, and image quality. Older databases, such as ORL and FERET, are smaller and cleaner, while newer datasets enable large-scale training with millions of images but pose challenges like inconsistent data quality and high computational costs. The study also examines CNN architectures, including FaceNet and Visual Geometry Group 16 (VGG16), which show strong performance on large datasets like Labeled Faces in the Wild (LFW) and VGGFace, achieving accuracy rates above 98%. In contrast, earlier models like Support Vector Machine (SVM) and Gabor Wavelets perform well on smaller datasets but lack scalability for larger, more complex datasets. The analysis highlights the growing importance of multi-task learning and ensemble methods, as seen in Multi-Task Cascaded Convolutional Networks (MTCNNs). Overall, the findings emphasize the need for advanced algorithms capable of handling large-scale, real-world challenges while optimizing accuracy and computational efficiency in face recognition systems.

Detecting the droplet and bubble in the oil-gas-water three-phase medium via ultrasonic A-scan signal-assisted B-mode imaging method

Abstract Accurately detecting the parameters of bubbles and droplets in the oil-gas-water three-phase medium is essential to critical decision-making about industrial production. An ultrasonic A-scan signal-assisted B-mode imaging (AS-assisted-BI) method that combines a B-mode imaging (BI) method and an A-scan signal (AS) method is developed to simultaneously visualize the oil-gas-water three-phase medium, distinguish the gas bubble and oil droplet, and measure the parameters about their sizes and locations. The applicability of the plane wave imaging (PWI) method for visualizing the oil-gas-water three-phase medium with less computational cost was confirmed. Then, the influence of the bubble and droplet characteristics, such as acoustic impedance difference, size, and location, upon the ultrasonic propagation mechanism are studied to understand the B-mode image and A-scan signal as the design basis of the BI and AS method. The BI method locates the least key points from the B-mode image based on the PWI method compared with the conventional industrial B-mode image method. The AS method automatically locates the key point from the A-scan signal rather than manually marking it in the previous study. Two options for the AS-assisted-BI method, suitable for different detection requirements and applications, have been designed. The numerical simulation and experimental results demonstrated that the proposed method received excellent precision, robustness, and efficiency for detecting the acoustic impedance difference between the bubble and droplet, the longitudinal location of the bubble front surface, and the droplet’s size and center location.

Vertical Slowness-Constrained Joint Anisotropic Parameters and Event-Location Inversion for Downhole Microseismic Monitoring

The construction of accurate anisotropic velocity models is essential for effective microseismic monitoring in hydraulic fracturing. Ignoring anisotropy can result in significant distortions in microseismic event locations and their interpretation. Although methods exist to simultaneously invert anisotropic parameters and event locations using microseismic arrival times, the results heavily depend on accurate initial models and sufficient ray coverage due to strong trade-offs among multiple parameters. Microseismic waveform inversion for anisotropic parameters remains challenging due to the low signal-to-noise ratio of the data and the high computational cost. To address these challenges, we propose a method for jointly inverting event locations and velocity updates based on arrival times and vertical slowness estimates, under the assumption of small horizontal velocity variations. Vertical slowness estimates, which are independent of source information and easily obtainable, provide an additional constraint that enhances inversion stability. We test the proposed method in four synthetic examples under various conditions. The results demonstrate that incorporating vertical slowness effectively constrains and stabilizes conventional travel-time inversion, especially in scenarios with poor raypath coverage. Additionally, we apply this method to a field case and find that it produces more reasonable event locations compared to inversions using arrival times alone. This joint inversion method can enhance the accuracy of anisotropic structures and event locations, which thus help with fracture characterization in tight and low-permeability reservoirs. It may serve as an effective downhole monitoring approach for hydrocarbon and geothermal energy production.

Development of interatomic potential suitable for molecular dynamics simulation of Ni oxidation and Ni-NiO interface.

Large-scale molecular dynamics (MD) simulations enabled by computationally efficient semiempirical potentials are an invaluable tool for materials modeling. In the case of metallic alloys, embedded atom method (EAM) and Finnis-Sinclair (FS) potentials are a reasonable choice based on their good balance of quality and computational cost. However, these semiempirical potentials are not suitable for simulating ionic systems, which prevents their use in studying many technologically relevant metal-oxide systems. The charge transfer ionic potential (CTIP), which can utilize EAM/FS potentials available in the literature together with a variable charge representation of electrostatic interactions, should be a reasonable choice for performing reliable and computationally efficient MD simulations of such systems. However, only a few such potentials are available in the literature, and their computational cost is much higher compared to EAM/FS potentials. In the present work, we have attempted to remedy these deficiencies by combining several modifications to the CTIP model proposed in the literature and efficiently implementing them into the widely used Large-scale Atomic/Molecular Massively Parallel Simulator MD code. Using these modifications, we have developed a new Ni-O CTIP parameterization, which has been tested in several different scenarios of interest. First, the early stages of Ni surface oxidation were simulated, demonstrating the nucleation and growth of a crystalline NiO film across the surface. Second, solidification and vitrification in the Ni-O system were investigated, demonstrating that the new CTIP parameterization provides reasonable agreement with the experimentally determined equilibrium phase diagram. Finally, we studied the interaction of dislocations in a Ni matrix with a NiO inclusion using a simulation cell with an unprecedented number of atoms for a variable charge MD simulation. Thus, the approach utilized in the present study is an efficient method to simulate large scale atomic mechanisms in metal-oxide systems.

Data Augmentation Techniques Using Generative Adversarial Networks: Employing GANs to Create Synthetic Data for Enhancing Machine Learning Model Training

The capability of GANs to increase data size has been a revelation to people in the augmentation field. It has saved machine learning algorithms with scarcity, imbalance, or poor data variation. Unlike the previous approaches based on transformation techniques, GANs can learn complex patterns and generate realistic synthetic data, which have virtually never been possible for fields and applications. In this paper, 12 new GAN-augmentation strategies are proposed and implemented. These are image-to-image translation, text-to-image synthesis, style transfer, and domain adaptation. These techniques prove that using GANs can enhance the quality of the data and models, balance the data, and maintain the confidentiality of the data while applying any of these techniques. The use of GANs in generating new data to enrich data sets has proved promising, especially in specific fields like health, where medical images complement diagnostic models while respecting patients' rights to privacy. It has also been used in image recognition when it creates diverse outputs to solve problems such as class imbalance, besides solving regular problems faced in natural language processing, including text-to-image synthesis. Compared to previous approaches, GANs provide better and more complex solutions and fill the data with newcomers instead of transforming the given ones. This capability leads to the creation of much more diverse datasets that can be closer to the real environment. Nevertheless, there are some key issues of concern, such as computational cost, mode collapse, and, most notably, social impact issues, where GANs may be misused using deepfake technologies. The proposed techniques show great potential and are based on a new approach to using data augmentation to solve modern tasks. However, the lack of detailed experiments and decreased accuracy compared to machine learning algorithms emphasize the need for future research to confirm the efficiency of the proposed approach in various applications. Overcoming these limitations via solid architectures, appropriate measures of performance, and policies is crucial. While GANs are still a relatively young type of AI technology, their enhancement and scaling can create significant opportunities for using these models in combination with other complex models, such as diffusion and reinforcement learning.

A Frugal Approach Toward Modeling of Defects in Metal 3D Printing Through Statistical Methods in Finite Element Analysis

Metal additive manufacturing has emerged as a revolutionary technology for the fabrication of high-complexity components. However, this technique presents unique challenges related to the structural integrity and final strength of the parts produced due to inherent defects, such as porosity, cracks, and geometric deviations. These defects significantly impact the fatigue life of the material by acting as stress concentrators that accelerate failure under cyclic loading. On the one hand, this type of model is very complicated in its approach, since, even with encouraging results, the complexity of the calculation with these variables makes it difficult to obtain a simple result that allows for a generalized interpretation. On the other hand, using more familiar methods, it is possible to qualitatively guess the behavior that helps obtain results with better applicability, even at limited levels of precision. This paper presents a simplified finite element method combined with a statistical approach to model the presence of porosity in metal components produced by additive manufacturing. The proposed model considers a two-dimensional square plate subjected to tensile stress, with randomly introduced defects characterized by size, shape, and orientation. The percentage of porosity that affects each aspect determines the adjustment of the mechanical properties of finite elements. A series of simulations were performed to generate multiple models with random defect distributions to estimate maximum stress values. This approach demonstrates that complex models are not always necessary for a preliminary practical estimate of the effects of new manufacturing techniques. Furthermore, it demonstrates the potential for the extension of frugal computational techniques, which aim to minimize computational and experimental costs in the engineering field. The article discusses future research directions, particularly those related to potential business applications, including commercial uses. This follows a discussion of the existing limitations of this study.

An explainable deep learning model for diabetic foot ulcer classification using swin transformer and efficient multi-scale attention-driven network

Diabetic Foot Ulcer (DFU) is a severe complication of diabetes mellitus, resulting in significant health and socio-economic challenges for the diagnosed individual. Severe cases of DFU can lead to lower limb amputation in diabetic patients, making their diagnosis a complex and costly process that poses challenges for medical professionals. Manual identification of DFU is particularly difficult due to their diverse visual characteristics, leading to multiple cases going undiagnosed. To address this challenge, Deep Learning (DL) methods offer an efficient and automated approach to facilitate timely treatment and improve patient outcomes. This research proposes a novel feature fusion-based model that incorporates two parallel tracks for efficient feature extraction. The first track utilizes the Swin transformer, which captures long-range dependencies by employing shifted windows and self-attention mechanisms. The second track involves the Efficient Multi-Scale Attention-Driven Network (EMADN), which leverages Light-weight Multi-scale Deformable Shuffle (LMDS) and Global Dilated Attention (GDA) blocks to extract local features efficiently. These blocks dynamically adjust kernel sizes and leverage attention modules, enabling effective feature extraction. To the best of our knowledge, this is the first work reporting the findings of a dual track architecture for DFU classification, leveraging Swin transformer and EMADN networks. The obtained feature maps from both the networks are concatenated and subjected to shuffle attention for feature refinement at a reduced computational cost. The proposed work also incorporates Grad-CAM-based Explainable Artificial Intelligence (XAI) to visualize and interpret the decision making of the network. The proposed model demonstrated better performance on the DFUC-2021 dataset, surpassing existing works and pre-trained CNN architectures with an accuracy of 78.79% and a macro F1-score of 80%.

RADE: A reduced approach to density-functional expansion.

Density-functional theory (DFT) has become an extensively and successfully used tool in the studies of molecules and materials. However, DFT remains computationally expensive, especially for exploring the conformational space of molecular systems comprising a few hundred atoms. Here, we present a Reduced Approach to Density-functional Expansion (RADE), devised to substantially reduce the computational cost of standard DFT methods. RADE can be implemented fully non-empirically as an efficient first-principles electronic structure method. Preliminary results for molecules containing elements H, C, N, and O indicate that this method can, in general, reproduce well the results from standard DFT calculations.

State estimation with quantum extreme learning machines beyond the scrambling time

Quantum extreme learning machines (QELMs) leverage untrained quantum dynamics to efficiently process information encoded in input quantum states, avoiding the high computational cost of training more complicated nonlinear models. On the other hand, quantum information scrambling (QIS) quantifies how the spread of quantum information into correlations makes it irretrievable from local measurements. Here, we explore the tight relation between QIS and the predictive power of QELMs. In particular, we show efficient state estimation is possible even beyond the scrambling time, for many different types of dynamics — in fact, we show that in all the cases we studied, the reconstruction efficiency at long interaction times matches the optimal one offered by random global unitary dynamics. These results offer promising venues for robust experimental QELM-based state estimation protocols, as well as providing novel insights into the nature of QIS from a state estimation perspective.

Atom-wise formulation of the many-body dispersion problem for linear-scaling van der Waals corrections

A common approach to modeling dispersion interactions and overcoming the inaccurate description of long-range correlation effects in electronic structure calculations is the use of pairwise-additive potentials, as in the Tkatchenko-Scheffler [] method. In previous work [H. Muhli , ], we have shown how these are amenable to highly efficient atomistic simulation by machine learning their local parametrization. However, the atomic polarizability and the electron correlation energy have a complex and nonlocal many-body character and some of the dispersion effects in complex systems are not sufficiently described by these types of pairwise-additive potentials. Currently, one of the most widely used rigorous descriptions of the many-body effects is based on the many-body dispersion (MBD) model [A. Tkatchenko , ]. In this work, we show that the MBD model can also be locally parametrized to derive a local approximation for the highly nonlocal many-body effects. With this local parametrization, we develop an atomwise formulation of MBD that we refer to as linear MBD (lMBD), as this decomposition enables linear scaling with system size. This model provides a transparent and controllable approximation to the full MBD model with tunable convergence parameters for a fraction of the computational cost observed in electronic structure calculations with popular density-functional theory codes. We show that our model scales linearly with the number of atoms in the system and is easily parallelizable. Furthermore, we show how using the same machinery already established in previous work for predicting Hirshfeld volumes with machine learning enables access to large-scale simulations with MBD-level corrections. Published by the American Physical Society 2025

Ensemble noise properties of the European Pulsar Timing Array

Abstract The null hypothesis in Pulsar Timing Array (PTA) analyses includes assumptions about ensemble properties of pulsar time-correlated noise. These properties are encoded in prior probabilities for the amplitude and the spectral index of the power-law power spectral density of temporal correlations of the noise. Because multiple realizations of time-correlated noise processes are found in pulsars, these ensemble noise properties could and should be modelled in the full-PTA observations by parameterising the respective prior distributions using the so-called hyperparameters. This approach is known as the hierarchical Bayesian inference. In this work, we introduce a new procedure for numerical marginalisation over hyperparameters. The procedure may be used in searches for nanohertz gravitational waves and other PTA analyses to resolve prior misspecification at negligible computational cost. Furthermore, we infer the distribution of amplitudes and spectral indices of the power spectral density of spin noise and dispersion measure variation noise based on the observation of 25 millisecond pulsars by the European Pulsar Timing Array (EPTA). Our results may be used for the simulation of realistic noise in PTAs.

Autoencoder Reconstruction of Cosmological Microlensing Magnification Maps

Abstract Enhanced modeling of microlensing variations in light curves of strongly lensed quasars improves measurements of cosmological time delays, the Hubble Constant, and quasar structure. Traditional methods for modeling extragalactic microlensing rely on computationally expensive magnification map generation. With large data sets expected from wide-field surveys like the Vera C. Rubin Legacy Survey of Space and Time, including thousands of lensed quasars and hundreds of multiply imaged supernovae, faster approaches become essential. We introduce a deep-learning model that is trained on pre-computed magnification maps covering the parameter space on a grid of κ, γ, and s. Our autoencoder creates a low-dimensional latent space representation of these maps, enabling efficient map generation. Quantifying the performance of magnification map generation from a low dimensional space is an essential step in the roadmap to develop neural network-based models that can replace traditional feed-forward simulation at much lower computational costs. We develop metrics to study various aspects of the autoencoder generated maps and show that the reconstruction is reliable. Even though we observe a mild loss of resolution in the generated maps, we find this effect to be smaller than the smoothing effect of convolving the original map with a source of a plausible size for its accretion disk in the red end of the optical spectrum and larger wavelengths and particularly one suitable for studying the broad-line region of quasars. Used to generate large samples of on-demand magnification maps, our model can enable fast modeling of microlensing variability in lensed quasars and supernovae.

Practical Applications of Grand-canonical Electronic Structure Calculations in Electrochemical Simulation.

Modeling electrified interfaces has long been a great challenge in electrochemistry. In recent years, the grand-canonical treatment for electrons has gradually been developed, and its combination with density functional theory has been widely used to simulate electrochemical processes on an atomistic scale. In this Perspective, we aim to discuss several practical applications of this powerful technique after a short review of necessary fundamentals. We will begin with capacitor-based parametrization method of grand-canonical calculated results. If considering the electrodes under different applied potentials as different materials, the parametrization can be viewed as a kind of "quadratic scaling relation", which might reduce the overall computational costs by data postanalysis rather than algorithm development. Following an example of the abnormal potential-independent energetic curve within the bandgap area, we turn the topic to the semiconducting electrodes. Meanwhile, the specific behaviors of the bandgap also indicate that besides the reaction thermodynamics and kinetics, the detailed electronic structure of the system can also be well described by the grand-canonical treatment on electrons. Several possibilities for further applications are proposed correspondingly and summarized at the end of paper. We believe that the grand-canonical treatment for electronic structure calculations can greatly enrich our understanding of the fundamental mechanisms under electrochemical environments.

A Fog Computing and Blockchain-based Anonymous Authentication Scheme to Enhance Security in VANET Environments

Authentication of vehicles and users, integrity of exchanged messages, and privacy preservation are essential features in VANETs. VANETs are used to collect information on road conditions, vehicle location and speed, and traffic congestion data. The open exchange of information within VANETs poses serious security threats. Furthermore, existing schemes have higher communication and computational costs, making them incompatible with resource-constrained VANET applications. This study proposes a multifactor authentication and privacy-preserving security scheme for VANETs based on blockchain and fog computing to meet all these requirements. The proposed scheme uses fingerprints and Quick Response (QR) codes as a multifactor to authenticate vehicle users and fog-cloud computing techniques to reduce the computational burden on RSUs and improve service quality and resilience. Additionally, the scheme synchronizes a consistent ledger across all RSUs using blockchain technology to store and distribute vehicle authentication statuses. Through a thorough comparison with relevant current protocols, the scheme shows a much-reduced computing expense and communication burden in situations with high vehicle density within a timeframe of 6.3846 ms and 544 bytes for communication costs. In addition, the proposed scheme demonstrates a successful balance between efficacy and complexity, protecting confidentiality, anonymous authentication, and ensuring integrity and conditional tracking. Formal and informal security analysis showed that the proposed scheme is more reliable, practical, and secure against many hostile attacks, such as modification attacks, 51% attacks, Sybil attacks, and MITM attacks.

Efficient Elliptic-Curve-Cryptography-Based Anonymous Authentication for Internet of Things: Tailored Protocols for Periodic and Remote Control Traffic Patterns

IoT-based applications require effective anonymous authentication and key agreement (AKA) protocols to secure data and protect user privacy due to open communication channels and sensitive data. While AKA protocols for these applications have been extensively studied, achieving anonymity remains a challenge. AKA schemes using one-time pseudonyms face resynchronization issues after desynchronization attacks, and the high computational overhead of bilinear pairing and public key encryption limits its applicability. Existing schemes also lack essential security features, causing issues such as vulnerability to ephemeral secret leakage attacks and key compromise impersonation. To address these issues, we propose two novel AKA schemes, PUAKA and RCAKA, designed for different IoT traffic patterns. PUAKA improves end device anonymity in the periodic update pattern by updating one-time pseudonyms with authenticated session keys. RCAKA, for the remote control pattern, ensures anonymity while reducing communication and computation costs using shared signatures and temporary random numbers. A key contribution of RCAKA is its ability to resynchronize end devices with incomplete data in the periodic update pattern, supporting continued authentication. Both protocols’ security is proven under the Real-or-Random model. The performance comparison results show that the proposed protocols exceed existing solutions in security features and communication costs while reducing computational overhead by 32% to 50%.