Search

The many-body correlation function is a fundamental computation kernel in modern physics computing applications, e.g., Hadron Contractions in Lattice quantum chromodynamics (QCD). This kernel is both computation and memory intensive, involving a series of tensor contractions, and thus usually runs on accelerators like GPUs. Existing optimizations on many-body correlation mainly focus on individual tensor contractions (e.g., cuBLAS libraries and others). In contrast, this work discovers a new optimization dimension for many-body correlation by exploring the optimization opportunities among tensor contractions. More specifically, it targets general GPU architectures (both NVIDIA and AMD) and optimizes many-body correlation’s memory management by exploiting a set of memory allocation and communication redundancy elimination opportunities: first, GPU memory allocation redundancy : the intermediate output frequently occurs as input in the subsequent calculations; second, CPU-GPU communication redundancy : although all tensors are allocated on both CPU and GPU, many of them are used (and reused) on the GPU side only, and thus, many CPU/GPU communications (like that in existing Unified Memory designs) are unnecessary; third, GPU oversubscription: limited GPU memory size causes oversubscription issues, and existing memory management usually results in near-reuse data eviction, thus incurring extra CPU/GPU memory communications. Targeting these memory optimization opportunities, this article proposes MemHC, an optimized systematic GPU memory management framework that aims to accelerate the calculation of many-body correlation functions utilizing a series of new memory reduction designs. These designs involve optimizations for GPU memory allocation, CPU/GPU memory movement, and GPU memory oversubscription, respectively. More specifically, first, MemHC employs duplication-aware management and lazy release of GPU memories to corresponding host managing for better data reusability. Second, it implements data reorganization and on-demand synchronization to eliminate redundant (or unnecessary) data transfer. Third, MemHC exploits an optimized Least Recently Used (LRU) eviction policy called Pre-Protected LRU to reduce evictions and leverage memory hits. Additionally, MemHC is portable for various platforms including NVIDIA GPUs and AMD GPUs. The evaluation demonstrates that MemHC outperforms unified memory management by \( 2.18\times \) to \( 10.73\times \) . The proposed Pre-Protected LRU policy outperforms the original LRU policy by up to \( 1.36\times \) improvement. 1

This paper presents a virtual modelling technique for dynamic safety assessment of practical structures undergoing geometric and material blended nonlinearities. The variational inputs of systematic properties are treated within the 3D dynamic geometric-elastoplastic analyses. To circumvent numerical challenges in solving the coupled nonlinear problems, a freshly developed dynamic virtual modelling (DVM) technique is employed to determine the inherent relationship between the variational input data and the nonlinear structural response by using a new clustering based extended support vector regression (C-XSVR) algorithm with a novel T-spline polynomial kernel function. The virtual modelling models can be constructed at each time step within the Newmark's time integration procedure, which then can be used to predict deflection, force, and stress of the concerned structure at different periods. The DVM is capable of visibly forecasting potential large deformation nonlinear behaviours in an efficient manner, based on the explicit relationship functions. To demonstrate the accuracy and effectiveness of the proposed framework, nonlinear behaviours of two practical applications under future forecasted working conditions are predicted and validated in the numerical investigations.

In this article, some Bohr-type inequalities with one parameter or involving convex combination for bounded analytic functions of Schwarz functions are established. Some previous inequalities are generalized. All the results are sharp.

Electrochemical tests and surface analyses were carried out to study the influence of the external low-voltage spatial electric field (SEF) on the corrosion behaviour of copper in 3.5 wt-% NaCl sol...

We call the solution of a kind of weighted Laplace equation complex-valued kernel $$\alpha $$ -harmonic mappings. In this paper, some geometric properties of the complex-valued kernel $$\alpha $$ -harmonic mappings, such as area, fully starlikeness and fully convexity of order $$\gamma $$ , $$\gamma \in [0,1)$$ , are explored. Furthermore, for a class of given boundary function, the Rado–Kneser–Choquet-type theorem is obtained.

In this paper, some new versions of Bohr-type inequalities for bounded analytic functions of Schwarz functions are established. Most of these inequalities are sharp. Some previous inequalities are generalized.

Cholesterol gallstone disease is a common global condition. This study investigated the role of plant sterols (PS) in the prevention of gallstone formation and the underlying mechanisms. Adult male mice were fed a lithogenic diet (LD) alone or supplemented with PS (LD-ps), phospholipids (LD-pl) or both PS and phospholipids (LD-ps/pl) for 8 weeks. Incidences of gallstone formation were compared among the groups. Lipids in the bile, liver and serum were analyzed. The expression of genes involved in cholesterol absorption, transport and metabolism in the liver and small intestine was determined. The incidences of gallstone formation were 100% (10/10), 20% (2/10), 100% (10/10) and 40% (4/10) in the LD, LD-ps, LD-pl and LD-ps/pl groups, respectively. Serum cholesterol and intestinal cholesterol absorption were decreased in PS-supplemented mice. The expression of genes related to cholesterol transport and metabolism in the liver was down-regulated by dietary PS. PS supplementation decreased Niemann-Pick C1-like 1 expression in the small intestine and reduced intestinal cholesterol absorption. Our results demonstrated that PS could inhibit intestinal cholesterol absorption and thus prevent cholesterol gallstone formation.

A novel machine learning aided structural reliability analysis for functionally graded frame structures against static loading is proposed. The uncertain system parameters, which include the material properties, dimensions of structural members, applied loads, as well as the degree of gradation of the functionally graded material (FGM), can be incorporated within a unified structural reliability analysis framework. A 3D finite element method (FEM) for static analysis of bar-type engineering structures involving FGM is presented. By extending the traditional support vector regression (SVR) method, a new kernel-based machine learning technique, namely the extended support vector regression (X-SVR), is proposed for modelling the underpinned relationship between the structural behaviours and the uncertain system inputs. The proposed structural reliability analysis inherits the advantages of the traditional sampling method (i.e., Monte-Carlo Simulation) on providing the information regarding the statistical characteristics (i.e., mean, standard deviations, probability density functions and cumulative distribution functions etc.) of any concerned structural outputs, but with significantly reduced computational efforts. Five numerical examples are investigated to illustrate the accuracy, applicability, and computational efficiency of the proposed computational scheme.

Laser impact welding (LIW) is an attractive technique for thin film dissimilar materials solid-state welding. To understand the mechanism of laser impact welding, the wave character of wavy interfaces, the morphology of intermetallics, the diffusion phenomenon and the hardness along the welding direction were investigated with Scanning Electron Microscopy (SEM), Energy Dispersive Spectrometer (EDS), and nanoindentation, respectively. There is a microstructure and mechanical property gradient not only along the welding direction, but also perpendicular to it. The variety of both microstructure and mechanical properties is the combined result of impact pressure and impact angle. The gradient of the nanoindentation demonstrated the variety of the strengthening mechanism nearby the weld interface. The study provides valuable knowledge for welding process understanding and control.

A virtual model aided non-deterministic static analysis (including linear and nonlinear analyses) with polyphase uncertainty is presented in this paper. Within an uncertain system, the polyphase uncertainty integrates both probabilistic and non-probabilistic uncertainties, which is more sophisticated than the conventional uncertainty modelling through a single type. To further improve the computational stableness and robustness of the virtual model, a kernel-based machine learning technique, namely Twin Extended Support Vector Regression (T-X-SVR), is newly developed. The feature of auto-learning is fulfilled through the Bayesian optimization. The proposed approach is capable of providing sufficient statistical information, including the membership functions of mean and standard deviation, fuzzy-valued probabilistic density function (PDF) and cumulative distribution function (CDF) for the upper and lower bounds of the concerned structural response. To demonstrate the effectiveness and computational efficiency of the proposed approach, a verification case, where analytical solutions are available, is tested first. Then, two practically stimulated engineering applications are fully investigated.

A machine learning aided virtual model architecture framework is presented. With great computational stability and preserved convexity features, the Twin Extended Support Vector Regression (T-X-SVR) method is adopted to imitate the underpinned and sophisticated constitutive relationship between the systematic inputs and outcomes in real-world applications. Various numerical simulations can be implemented on the virtual model with greatly reduced computational costs. The multi-type information from heterogeneous sources and multifarious engineering applications are supported by the proposed framework. For the virtual model aided numerical analysis with the multi-type information, fuzzy-valued probabilistic distributional characteristics of the bounds of the concerned structural responses are estimated. The capability of the modularity feature of the virtual model improves the operational flexibility which makes the implementation of such advance technique more user friendly. To demonstrate the applicability and computational efficiency of the proposed framework, three practical engineering stimulated problems (i.e., the mechanical dynamic system , multiphysics with heat transfer and gas flow interaction, and the expansion process of a biomedical stent with both material and geometry nonlinearities), involving multi-type information (i.e., random parameters and fuzzy sets), are fully investigated through the proposed virtual model architecture framework.

BackgroundHepatic lipid metabolism regulates biliary composition and influences the formation of cholesterol gallstones. The genes Hmgcr and Cyp7a1, which encode key liver enzymes, are regulated by circadian rhythm-related transcription factors. We aimed to investigate the effect of circadian rhythm disruption on hepatic cholesterol and bile acid metabolism and the incidence of cholesterol stone formation.MethodsAdult male C57BL/6J mice were fed either a lithogenic diet (LD) only during the sleep phase (time-restricted lithogenic diet feeding, TRF) or an LD ad libitum (non-time-restricted lithogenic diet feeding, nTRF) for 4 weeks. Food consumption, body mass gain, and the incidence of gallstones were assessed. Circulating metabolic parameters, lipid accumulation in the liver, the circadian expression of hepatic clock and metabolic genes, and the gut microbiota were analyzed.ResultsTRF caused a dysregulation of the circadian rhythm in the mice, characterized by significant differences in the circadian expression patterns of clock-related genes. In TRF mice, the circadian rhythms in the expression of genes involved in bile acid and cholesterol metabolism were disrupted, as was the circadian rhythm of the gut microbiota. These changes were associated with high biliary cholesterol content, which promoted gallstone formation in the TRF mice.ConclusionDisordered circadian rhythm is associated with abnormal hepatic bile acid and cholesterol metabolism in mice, which promotes gallstone formation.

The double descent curve is one of the most intriguing properties of deep neural networks. It contrasts the classical bias-variance curve with the behavior of modern neural networks, occurring where the number of samples nears the number of parameters. In this work, we explore the connection between the double descent phenomena and the number of samples in the deep neural network setting. In particular, we propose a construction which augments the existing dataset by artificially increasing the number of samples. This construction empirically mitigates the double descent curve in this setting. We reproduce existing work on deep double descent, and observe a smooth descent into the overparameterized region for our construction. This occurs both with respect to the model size, and with respect to the number epochs.

Trimethylamine-N-oxide (TMAO) is a metabolite derived from trimethylamine (TMA), which is first produced by gut microbiota and then oxidized by flavin-containing monooxygenase 3 (FMO3) in the liver. TMAO may contribute to the development of diseases such as atherosclerosis because of its role in regulating lipid metabolism. In this study, we found that high plasma TMAO levels were positively associated with the presence of gallstone disease in humans. We further found increased hepatic FMO3 expression and elevated plasma TMAO level in a gallstone-susceptible strain of mice C57BL/6J fed a lithogenic diet (LD), but not in a gallstone-resistant strain of mice AKR/J. Dietary supplementation of TMAO or its precursor choline increased hepatic FMO3 expression and plasma TMAO levels and induced hepatic canalicular cholesterol transporters ATP binding cassette (Abc) g5 and g8 expression in mice. Up-regulation of ABCG5 and ABCG8 expression was observed in hepatocytes incubated with TMAO in vitro. Additionally, in AKR/J mice fed a LD supplemented with 0.3% TMAO, the incidence of gallstones rose up to 70% compared with 0% in AKR/J mice fed only a LD. This was associated with increased hepatic Abcg5 and g8 expression induced by TMAO. Our study demonstrated TMAO could be associated with increased hepatic Abcg5/g8 expression, biliary cholesterol hypersecretion and gallstone formation.

The nonlinear dynamic responses of a nanocomposite organic solar cell (NCOSC) are developed through the classical plate theory. The investigated NCOSC consists of five layers which are including Al, P3HT: PCBM, PEDOT: PSS, IOT and glass. A uniformly distributed external excitation is exerted on the simply supported NCOSC. The impacts of the Winkler-Pasternak elastic foundation, thermal environment and damping on the nonlinear dynamic responses of the NCOSC are investigated. The equations of motion and geometric compatibility of the NCOSC with the consideration of the von Karman nonlinearity are derived. The governing equation of the dynamic system is formulated by employing the Galerkin and the fourth-order Runge-Kutta methods. Several numerical experiments are thoroughly presented to report the effects of damping ratio, temperature variations, and elastic foundation parameters on the frequency–amplitude curves and nonlinear dynamic response of the NCOSC. The numerical studies indicate that the existence of the Winkler-Pasternak elastic foundation effectively reduces the dynamic response of the NCOSC. In addition, the damping and thermal variation depress the vibration of the NCOSC but with relatively less efficiency compared with the Winkler- Pasternak elastic foundation.

In this paper, a robust interval free vibration analysis for 3D functionally graded frame type engineering structure is presented through the finite element method (FEM). Uncertain material properties, which are including the Young's modulus and material density, of the functionally graded material are considered. Unlike the conventional uncertainty quantification through stochastic approach, the uncertain system inputs are modelled by the interval approach. Instead of straining on the precise statistical information of the uncertain parameters, only upper and lower bounds of the uncertain system inputs are required for valid structural safety assessment. By implementing the mathematical programming approach combined with the intrinsic characteristics of the non-deficient engineering structures, the upper and lower bounds of the natural frequencies of 3D functionally graded frame structure can be explicitly formulated by two independent eigen-problems. The sharpness and physical feasibility of the interval natural frequencies of the functionally graded structure can be well preserved. To demonstrate the competence of the proposed method, two numerical examples have been thoroughly investigated. In addition, diverse numerical investigations have been conducted to explore the impacts of uncertain material properties and the power-law index of the functionally graded materials on the overall structural performance.

This paper proposes a hyperplane modelling technique aided polymorphic uncertainty quantification strategy for various engineering applications. Accumulative experiences from industrial applications have continuously revealed that probability distribution characteristics for system properties cannot always be precisely determined, due to the scarcity of available information. Thus, the polymorphic uncertainty is introduced to consider fuzzy and random uncertainty simultaneously. To tackle such uncertainty quantification challenges, Multiple Target Strategy (MTS) and Single Target Strategy (STS) are proposed. MTS is a generalized polymorphic uncertainty quantification strategy, capable of providing sufficient amounts of the statistical information of the concerned structural response. On the contrary, STS considers the polymorphic uncertainty quantification in a more precise and efficient manner. Both strategies are implemented based on their respective hyperplane models. For hyperplane model construction, a newly developed ensemble meta-regression technique, namely AdaBoost Extended Support Vector Regression (Ada-X-SVR), is embedded with a novel kernel function. The proposed hyperplane modelling aided polymorphic uncertainty quantification framework provides the feasibility of both generalized and targeted estimations, auto-learning and information update features. Furthermore, to demonstrate the applicability and computational efficiency of the proposed strategies, a benchmark analysis with an available analytical solution and three engineering applications with linear and nonlinear performance are fully investigated.

The system containing ordinary Portland cement and fly ash (FA) tends to be vulnerable against carbonation, where hydration estimation is crucial for assessing the carbonation resistance of the system. To date, few available models targeting FA-cement hydration are free from empirical or epistemic parameters, leading to unresolved issues of model sensitivity in estimating carbonation. In this study, a well-acclaimed hydration model is coupled with chemophysical modelling to numerically interpret the time-dependent mechanisms of system decay under carbonation. Furthermore, leveraging the power of extended support vector regression method, a robust and efficient framework is proposed for the investigation of hydration model sensitivity in carbonation assessment. The developed method is extensively validated against experimental studies and a carefully crafted benchmark test. Through sensitivity analyses, key findings on the propagation of hydration model sensitivity as well as the detailed levels of impact from respective model parameters are unravelled.

We present a Minimal Event Distance Aneuploidy Lineage Tree (MEDALT) algorithm that infers the evolution history of a cell population based on single-cell copy number (SCCN) profiles, and a statistical routine named lineage speciation analysis (LSA), whichty facilitates discovery of fitness-associated alterations and genes from SCCN lineage trees. MEDALT appears more accurate than phylogenetics approaches in reconstructing copy number lineage. From data from 20 triple-negative breast cancer patients, our approaches effectively prioritize genes that are essential for breast cancer cell fitness and predict patient survival, including those implicating convergent evolution.The source code of our study is available at https://github.com/KChen-lab/MEDALT.

Additional file 1: Fig. S1. Methodology of the framework. a. Illustration of minimal event distance (MED) calculation. b. Average lineage partitioning accuracy (LPA) on 100 simulation datasets without noise. c. Estimating lineage specific cumulative fold level (CFL). d. Estimating significance of CFL in an individual sample. e. AUC of non-random fitness-associated alterations (FAAs) detection based on LSA, permutated SCCN matrix rather than reconstructing tree, GISTIC test and one-side Wilcoxon signed-rank test on 100 simulation datasets without noise. f. Identification of non-random fitness-associated CNAs in a cohort of samples. g. Identification of parallel evolution CNAs in an individual sample. Fig. S2. The efficiency of MEDALT based on 9 × 3 × 20 simulation datasets with the population size from 400 to 2000, genome size from 100 to 1000. Fig. S3. Simulation and evaluation of CNA evolution model. a. Illustration of simulated genomic structural rearrangements in the evolution of a tumor. K represents the number of CNAs during ∆t period. r represents the number of adjacent regions which are affected by a CNA. TD: tandem duplication. TER: terminal deletion. DEL: interstitial deletion. BFB: breakage fusion bridge. b. Simulated and inferred copy number evolution distance between two genomes. Compared with MED are commonly used distance metrics Hamming, Euclidean and Manhattan. c. The AUC for identifying FAAs based on different combinations of models. Wilcox represents one-side Wilcoxon signed-rank test. d. The effects of noise on FAAs detection. Fig. S4. SCCN profile of TNBC patient KTN102. Each row represents a cell from pre-, mid-, or post-treatment. Fig. S5. Average distance between root node and cells from pre-, mid- or post-treatment based on MEDALT, maximal parsimony (MP), neighbor-joining (NJ) and maximum likelihood tree. FC refers to the fold changes between the average distance to root of the mid−/post- cells and that of the pre-treatment cells. Fig. S6. Stratified average CNA rates and fractions of DDR genes loss among lineages (distinguished by colors) in 6 primary TNBC samples. Fig. S7. Gene set enrichment analysis (GSEA) for genes identified by LSA in patient t1. Colors correspond to branches. Fig. S8. Significant genes identified through cohort LSA from the TNBC scDNA-seq data. a. Venn diagram of the genes identified by the MEDALT, MP and GISTIC but not reported in oncoKB, COSMIC and intOGen. b. Overall survival (OS) analysis of breast cancer patients in TCGA. c. Progression free survival (PFS) analysis of breast cancer patients in TCGA. d. Overall survival analysis of breast cancer patients in the METABRIC. e. The fraction of cancer genes overlapping with events which were significant in single lineage (#Lineage = 1), multiple lineages (#Lineage > 1), parallel evolution test ((#Lineage > 1& PLSA

Additional file 6.

Additional file 7. Review history.

Sparse matrix vector multiplication is an important and commonly used computing kernel in scientific computing. Irregular arrangement of non zeros in sparse matrix leads to irregular memory access pattern, which in turn affects the running speed. In the past ten years, there have been many optimization methods of sparse matrix vector multiplication based on different ideas and techniques. In this paper, the commonly used performance optimization techniques of sparse matrix vector multiplication in multi/many-core architecture are comprehensively investigated. We classify technical methods according to their common characteristics, and discuss the problems encountered by researchers of various methods. In addition, We provide a typical large sparse matrix set for testing. This paper also provides a theoretical basis for our subsequent sparse matrix calculation in Sunway TaihuLight archiecture.

© 2020 Elsevier B.V. A machine learning aided stochastic nonlocal damage analysis framework is proposed for quasi-brittle materials. The uncertain system parameters, including the material properties and loading actions, have been incorporated and analysed within a unified safety assessment framework against various working conditions. A three-dimensional integral-type nonlocal damage model through finite element method (FEM) has been adopted. For the purpose of investigating the probabilistic damage analysis problems, a freshly established machine learning approach, namely the capped-extended-support vector regression method (C-X-SVR), is proposed to eliminate the influences of random outliers in the first step, then establish the relationship between the uncertain systemic inputs and structural responses. Such that the training robustness and computational adaptability of the proposed regression model can be reinforced. Moreover, the proposed approach is competent of efficiently predicting the statistical information (i.e., means, standard deviations, probability density functions and cumulative density functions) of structural behaviours under continuous information update of the uncertain working condition from mercurial environment. One real-life experimental validation and two numerical investigations are implemented to further verify the effectiveness and efficiency of the uncertainty quantification framework against probabilistic damage analysis.