Partitioning Technique Research Articles

Extended stochastic finite element method enhanced by local mesh refinement for random voids analysis

We present in this paper a novel and efficient computational approach in terms of triangular extended stochastic finite element method (T-XSFEM) for simulation of random void problems. The present T-XSFEM is further enhanced by local mesh refinement with the aid of variable-node elements to couple/link different mesh-scales, increasing the efficiency of the developed approach and saving the computational cost. The degrees of freedom are approximated with a truncated generalized polynomial chaos (GPC). The present work depends on the extension of extended finite element method (XFEM) to the stochastic context, containing implicit expression of voids through the random level set functions. A new partition technique is defined to divide the random domain for integration by using a priori knowledge of the void shape function, which can further reduce the computational time. To show the effectiveness and accuracy of the developed approach, numerical experiments are studied and computed results are compared with existing reference solutions.

Energy-efficient Real-time Scheduling on Multicores

With the increasing demand for higher performance, the adoption of multicores has been a major stepping stone in the evolution of hard real-time systems. Though the computational bandwidth is increased due to parallel processing, the indispensable interactivity between the hierarchical memory sub-system and multiple cores has further aggravated the already complex worst case execution time (WCET) analysis of tasks. Furthermore, caches have the biggest influence on task execution time, and the inclusion of shared caches further increases the unpredictability of the system. Cache partitioning techniques have been proposed as a counter-measure to decouple the shared cache latency from the WCET. However, existing energy-efficient scheduling algorithms are oblivious to the unpredictable nature of shared caches or cache partitioning techniques, thus, diminishing their applicability to real-world systems. Without considering inter-task cache contention, directly using existing algorithms or attempting to allocate and schedule a taskset with cache-partition assignments can result in cache violations. To overcome this dilemma, we propose a novel approach to model inter-task cache contention as a dependency graph to be used by well-established algorithms to minimize energy consumption. Extensive simulations demonstrate the effectiveness of our approach to minimize energy consumption while also avoiding cache violations.

Reactive Autonomous Navigation of UAVs for Dynamic Sensing Coverage of Mobile Ground Targets.

This paper addresses a problem of autonomous navigation of unmanned aerial vehicles (UAVs) for the surveillance of multiple moving ground targets. The ground can be flat or uneven. A reactive real-time sliding mode control algorithm is proposed that navigates a team of communicating UAVs, equipped with ground-facing video cameras, towards moving targets to increase some measure of sensing coverage of the targets by the UAVs. Moreover, the Voronoi partitioning technique is adopted to reduce the movement range of the UAVs and decrease the revisit times of the targets. Extensive computer simulations, from the simple case with one UAV and multiple targets to the complex case with multiple UAVs and multiple targets, are conducted to demonstrate the performance of the developed autonomous navigation algorithm. The scenarios where the terrain is uneven are also considered. As shown in the simulation results, although the additional VP technique leads to some extra computation burden, the usage of the VP technique considerably reduces the target revisit time compared to the algorithm without this technique.

Penalty and partitioning techniques to improve performance of QUBO solvers

Quadratic Unconstrained Binary Optimization (QUBO) modeling has become a unifying framework for solving a wide variety of both unconstrained as well as constrained optimization problems. More recently, QUBO (or equivalent −1/+1 Ising Spin) models are a requirement for quantum annealing computers. Noisy Intermediate-Scale Quantum (NISQ) computing refers to classical computing preparing or compiling problem instances for compatibility with quantum hardware architectures. The process of converting a constrained problem to a QUBO compatible quantum annealing problem is an important part of the quantum compiler architecture and specifically when converting constrained models to unconstrained the choice of penalty magnitude is not trivial because using a large penalty to enforce constraints can overwhelm the solution landscape, while having too small a penalty allows infeasible optimal solutions. In this paper we present NISQ approaches to bound the magnitude of the penalty scalar M and demonstrate efficacy on a benchmark set of problems having a single equality constraint and present a QUBO partitioning approach validated by experimentation.

Efficient Model-Predictive Control for Networked Interval Type-2 T–S Fuzzy System With Stochastic Communication Protocol

In this article, the efficient model-predictive control (EMPC) problem of a class of nonlinear systems in the framework of interval type-2 Takagi-Sugeno (IT2 T-S) fuzzy is investigated. In order to improve the reliability of the data transmission while reducing the network communication burden, a so-called stochastic communication protocol (SCP) governed by a Markov chain is adopted to orchestrate the data transmission order from the controller to the actuator. The purpose of the addressed problem is to design a set of desired EMPC controllers so as to guarantee the mean-square system stability and obtain a good balance among the computation burden, initial feasible region, and the control performance. A novel control model is established for the SCP and IT2 T-S fuzzy nonlinearities in a unified representation by using a fuzzy periodic switching related to the transmission token and the membership function. Then, the system state, the SCP-based control perturbation, and the previous input under the SCP are fully taken into consideration for constructing the objective function. By virtue of the “min-max” strategy, a few optimizations are formulated, and the corresponding EMPC algorithm is provided, where the feedback gain is designed offline, while the control perturbation is obtained online. Furthermore, by means of the matrix partition technique, sufficient conditions are presented to rigidly guarantee the feasibility of the proposed EMPC algorithm and the mean-square stability of the underlying IT2 T-S fuzzy system. Finally, two illustrative examples are utilized to demonstrate the validity of the proposed EMPC strategy.

FedMEC: Improving Efficiency of Differentially Private Federated Learning via Mobile Edge Computing

Federated learning is a recently proposed paradigm that presents significant advantages in privacy-preserving machine learning services. It enables the deep learning applications on mobile devices, where a deep neural network (DNN) is trained in a decentralized manner among thousands of edge clients. However, directly apply the federated learning algorithm to the mobile edge computing environment will incur unacceptable computation costs in mobile edge devices. Moreover, among the training process, frequent model parameters exchanging between participants and the central server will increase the leakage possibility of the users’ sensitive training data. Aiming at reducing the heavy computation cost of DNN training on edge devices while providing strong privacy guarantees, we propose a mobile edge computing enabled federated learning framework, called FedMEC, which integrating model partition technique and differential privacy simultaneously. In FedMEC, the most complex computations can be outsourced to the edge servers by splitting a DNN model into two parts. Furthermore, we apply the differentially private data perturbation method to prevent the privacy leakage from the local model parameters, in which the updates from an edge device to the edge server is perturbed by the Laplace noise. To validate the proposed FedMEC, we conduct a series of experiments on an image classification task under the settings of federated learning. The results demonstrate the effectiveness and practicality of our FedMEC scheme.

Design Analysis of Turbo Decoder Based on One MAP Decoder Using High Level Synthesis Tool

High Level Synthesis (HLS) tool does not only simplify the designing operation and rapid prototyping but also allows the designers to explore large number of design’s techniques such as parallelism, pipeline, memory partitioning and many other techniques. Turbo decoder based on Maximum APosterior Probability (MAP) algorithm is designed in this work using Vivado HLS. The normal turbo decoder with two MAP decoders were implemented with and without parallelism and proposed a new design of turbo decoder with one MAP decoder and it was designed with and without parallelism using different window technique in HLS tool which it is not explored previously. These designs were implemented for different frame size in this work. A comp-arison in latency and resource utilization where done and how a tradeoff done between these two parameters to reach the specific design that we need. The new design produces better results.

Toosendanin relatives, trypanocidal principles from Meliae Cortex.

Africa Trypanosomiasis remains a serious health problem, but the approved drugs for this disease are so few that novel trypanocidal compounds are demanded. In search for trypanocidal principles from medicinal plants, we found MeOH extracts of Meliae Cortex with potent activity through the screening from about 300 kinds of methanolic extract. By bioassay-guided fractionation from this extract through the liquid-liquid partition and subsequent chromatographic technique using silica gel and ODS, finally we disclosed toosendanin (1) and its relatives as active principles. These active congeners showed not only potent trypanocidal activity but also little cytotoxicity to display the excellent selective index. Taking the isolated amount as well as trypanocidal activity into consideration, 1 was disclosed to be the responsible active principle in Meliae Cortex. Additionally, the derivatives of 1 were chemically prepared from 1 and bioactivity of them were also evaluated. Through the comparison with their trypanocidal activity among the isolated relatives and the synthesized derivatives of 1, the epoxide moiety was revealed to be essential for their potent trypanocidal activity. Furthermore, 3-O-acetyl group and 7-hydroxyl group were presumed to be important functional groups and introduction of methylpropionyl group into hemiacetal hydroxy moiety was clarified to enhance their typanocidal activity.

Robust l2−l∞ dynamic output feedback control design for uncertain discrete‐time switched systems with random time‐varying delay

SummaryThis article deals with the problem of robust output feedback control design for a class of switched systems with uncertainties and random time‐varying delay. Our purpose is focused on designing a full order dynamic output feedback controller and an appropriate switching rule to ensure the exponential mean square stability of the resulting closed‐loop switching system with an l2−l∞ performance level. The appealing aspects of the proposed control scheme include: (a) the development of LMI based delay‐dependent sufficient conditions for the exponential stability analysis of the stochastic hybrid systems using the appropriate choice of the augmented Lyapunov–Krasovskii functional, the partitioning technique and the average dwell time, (b) conservatism reduction of the unified design conditions, and (c) the controller designed exhibiting robustness with l2−l∞ performance against uncertainties and external disturbance. Finally, representative examples are given to demonstrate the validity and the merit of the proposed design technique.

A survey of data partitioning and sampling methods to support big data analysis

Computer clusters with the shared-nothing architecture are the major computing platforms for big data processing and analysis. In cluster computing, data partitioning and sampling are two fundamental strategies to speed up the computation of big data and increase scalability. In this paper, we present a comprehensive survey of the methods and techniques of data partitioning and sampling with respect to big data processing and analysis. We start with an overview of the mainstream big data frameworks on Hadoop clusters. The basic methods of data partitioning are then discussed including three classical horizontal partitioning schemes: range, hash, and random partitioning. Data partitioning on Hadoop clusters is also discussed with a summary of new strategies for big data partitioning, including the new Random Sample Partition (RSP) distributed model. The classical methods of data sampling are then investigated, including simple random sampling, stratified sampling, and reservoir sampling. Two common methods of big data sampling on computing clusters are also discussed: record-level sampling and block-level sampling. Record-level sampling is not as efficient as block-level sampling on big distributed data. On the other hand, block-level sampling on data blocks generated with the classical data partitioning methods does not necessarily produce good representative samples for approximate computing of big data. In this survey, we also summarize the prevailing strategies and related work on sampling-based approximation on Hadoop clusters. We believe that data partitioning and sampling should be considered together to build approximate cluster computing frameworks that are reliable in both the computational and statistical respects.

Contingency set partition-based impact transfer approach for the reliability assessment of composite generation and transmission systems

It is a challenging task to obtain accurate reliability indexes with high computational efficiency for the reliability assessment of composite generation and transmission systems. A contingency set partition-based impact transfer (CSPIT) approach is proposed in this paper. The CSPIT consists of two stages. In the first stage, a contingency set partition technique is developed to combine the merits of the contingency enumeration method and the Monte Carlo simulation method to obtain accurate reliability indexes with higher efficiency compared to the two traditional methods. With this technique, the contingency set is partitioned into the lower-order contingency subset and the higher-order contingency subset. Then, the contingency enumeration method is applied to enumerate all lower-order contingencies, and the Monte Carlo simulation method is utilized to take samples from the higher-order subset. A new reliability index formulation for the technique is also derived. In the second stage, an impact transfer technique is developed to further improve the computational efficiency. With this technique, the derived formulation in the first stage is modified by transferring partial impacts of higher-order contingencies to the corresponding lower-order ones. The impact transfer reduces the impact variance of higher-order contingencies and thus improves the sampling efficiency of the technique proposed in the first stage. Finally, the reliability index computation flow for the entire CSPIT approach is demonstrated. The CSPIT approach is validated using RTS-79 and IEEE-145 test systems. Study results indicate that our approach can obtain accurate reliability indexes with much higher computational efficiency compared with traditional methods.

Modeling multiple wave systems in the eastern equatorial Pacific

While moderate wind and wave conditions prevail in the eastern equatorial Pacific, modeling waves in this area remains challenging due to the presence of multiple wave systems converging from different parts of the ocean. This area is covered by swells originated far away including the storm belts of both hemispheres, coexisting with local generation due to the regular action of both the southern trade winds and the wind jets from Central America. In this context, our ability to predict waves in the area depends on the overall quality (i.e., at Pacific scale) of the meteorological input, and also on the skills of the wave model itself. Clearly any error at the remote generation areas translates into larger errors the further waves go, especially if attention is focused on coastal areas. A relevant aspect is that the traditional integral parameters do not offer the possibility to properly assess the errors associated with the different parts of the spectrum (e.g., wind sea and swell). To gain insight in this direction, we make use of partitioning techniques, which enables us to neatly cross-assign and evaluate three spectral components. Not surprisingly, the performance for the swell part is lower than that of the corresponding wind sea. This is further explored with a couple of tests modifying both the wind input and the wave model physics. We find that although at first sight the initial scheme (i.e., ST4) seems to provide the better estimate, the spectral analysis reveals a substantial underestimation of wind sea, compensated with a substantial overestimation of swell. This suggests a problem with too high winds and wave generation in the storm belts together with a likely lack of dissipation or dispersion of swell. In turn, local waves are generally underestimated due to a corresponding underestimation of the local winds. This insight emphasizes the need and advantages of evaluation methods able to look at the different sectors of the wave spectrum.

Model-Based Recursive Partitioning of Patients' Return Visits to Multispecialty Clinic During the 2009 H1N1 Pandemic Influenza (pH1N1).

Background During the 2009 H1N1 influenza pandemic (pH1N1), the proportion of outpatient visits to emergency departments, clinics and hospitals became elevated especially during the early months of the pandemic due to surges in sick, 'worried well' or returning patients seeking care. We determined the prevalence of return visits to a multispecialty clinic during the 2009 H1N1 influenza pandemic and identify subgroups at risk for return visits using model-based recursive partitioning technique. Methods This study was a retrospective analysis of ILI-related medical care visits to multispecialty clinic in Houston, Texas obtained as part of the Houston Health Department Influenza Sentinel Surveillance Project (ISSP) during the 2009 H1N1 pandemic influenza (April 2009 - March 2010). The data comprised of 2680 individuals who made a total of 2960 clinic visits. Return visit was defined as any visit following the index visit after the wash-out phase prior to the study period. We applied nominal logistic regression and recursive partitioning models to determine the independent predictors and the response probabilities of return visits. The sensitivity and specificity of the outcomes probabilities were determined using receiver operating characteristic (ROC) curve. Results Overall, 4.56% (Prob. 0.0%-17.5%) of the cohort had return visits with significant variations observed attributed to age group (76.0%), type of vaccine received by patients (18.4%) and Influenza A (pH1N1) test result (5.6%). Patients in age group 0-4 years were 9 times (aOR: 8.77, 95%CI: 3.39-29.95, p<0.0001) more likely than those who were 50+ years to have return visits. Similarly, patients who received either seasonal flu (aOR: 1.59, 95% CI 1.01-2.50, p=0.047) or pH1N1 (aOR: 1.74, 95%CI: 1.09-2.75, p=0.022) vaccines were about twice more likely to have return visits compared to those with no vaccination history. Model-based recursive partitioning yielded 19 splits with patients in subgroup I (patients of age group 0-4 years, who tested positive for pH1N1, and received both seasonal flu and pH1N1 vaccines) having the highest risk of return visits (Prob.=17.5%). The area under the curve (AUC) for both return and non-return visits was 72.9%, indicating a fairly accurate classification of the two groups. Conclusions Return visits in our cohort were more prevalent among children and young adults, and those that received either seasonal flu or pH1N1 or both vaccines. Understanding the dynamics in care-seeking behavior during pandemic would assist policymakers with appropriate resource allocation, and in the design of initiatives aimed at mitigating surges and recurrent utilization of the healthcare system.

An Improved B-hill Climbing Optimization Technique for Solving the Text Documents Clustering Problem.

Considering the increasing volume of text document information on Internet pages, dealing with such a tremendous amount of knowledge becomes totally complex due to its large size. Text clustering is a common optimization problem used to manage a large amount of text information into a subset of comparable and coherent clusters. This paper presents a novel local clustering technique, namely, β-hill climbing, to solve the problem of the text document clustering through modeling the β-hill climbing technique for partitioning the similar documents into the same cluster. The β parameter is the primary innovation in β-hill climbing technique. It has been introduced in order to perform a balance between local and global search. Local search methods are successfully applied to solve the problem of the text document clustering such as; k-medoid and kmean techniques. Experiments were conducted on eight benchmark standard text datasets with different characteristics taken from the Laboratory of Computational Intelligence (LABIC). The results proved that the proposed β-hill climbing achieved better results in comparison with the original hill climbing technique in solving the text clustering problem. The performance of the text clustering is useful by adding the β operator to the hill climbing.

Using unsupervised learning to partition 3D city scenes for distributed building energy microsimulation

Microsimulation is a class of Urban Building Energy Modeling techniques in which energetic interactions between buildings are explicitly resolved. Examples include SUNtool and CitySim+, both of which employ a sophisticated radiosity-based algorithm to solve for radiation exchange. The computational cost of this algorithm increases in proportion to the square of the number of surfaces of which an urban scene is comprised. To simulate large scenes, of the order of 10,000 to 1,000,000 surfaces, it is desirable to divide the scene to distribute the simulation task. However, this partitioning is not trivial as the energy-related interactions create uneven inter-dependencies between computing nodes. To this end, we describe in this paper two approaches ( K-means and Greedy Community Detection algorithms) for partitioning urban scenes, and subsequently performing building energy microsimulation using CitySim+ on a distributed memory High-Performance Computing Cluster. To compare the performance of these partitioning techniques, we propose two measures evaluating the extent to which the obtained clusters exploit data locality. We show that our approach using Greedy Community Detection performs well in terms of exploiting data locality and reducing inter-dependencies among sub-scenes, but at the expense of a higher data preparation cost and algorithm run-time.

Performance Analysis of CUDA by Implementation of Hypergraph Partitioning for Parallelizing Sparse Matrix-Vector Multiplication Using Quadro K4200

This paper addresses the performance analysis of sparse matrix-vector multiplication through hypergraph partitioning techniques using CUDA GPU-based parallel computing. Quadro K4200 is the GPU that is used in this paper. On the implementation of matrix-vector multiplication, various sizes and types of matrices are attempted. Our results show that on the average scenarios with 2 partitions, 4 partitions, 8 partitions, 16 partitions and 32 partitions in 1024 threads, CUDA performs up to 700 × better than sequential programming.

Pengujian Black Box pada Aplikasi Pertokoan Minimarket Menggunakan Metode Equivalence Partitioning

Quality measurements in software are currently required before the software will be used by users. This aims to determine the weakness of the system so that the data generated can match the data entered after the data is executed. This method focuses on data entry, system display, memory usage, and data execution speed so that if the input data does not match what is expected then the system fails.In implementation problems often arise when testers are never sure whether the software being tested has actually passed the test. This happens because there may be several execution paths that have never been tested. The minimarket stock management application will be tested using a black box test where this test is only intended to see if the program matches the function desired by the program without knowing the program code used. To generate test case designs automatically in black box method software testing with the equality.technique partition technique.

Pengujian Black Box pada Aplikasi Sistem Informasi Pengelolaan Masjid Menggunakan Teknik Equivalence Partitions

Black-box testing is very important because this technique is able to identify errors in functions, interfaces, data models, and access to external data sources. In implementation problems often arise when testers are never sure whether the software being tested has actually passed the test. This happens because there may be several execution paths that have never been tested. Testers must make every possible input data combination for testing. The selection of input data to find errors is a problem for testers because it has a high probability, so the test case design can automatically become a solution. In this implementation the application to be tested using black-box testing is the Mosque Management Information System. The Mosque Management application will be tested using black box testing where the test is only intended to see if the program matches the function desired by the program without knowing the program code used. To produce test case designs automatically in black box method software testing with equality partition technique. Equivalence Partitions technique is a test based on entering data on each form in the mosque management information system, each menu will be tested and grouped according to function whether it is valid or not.

UNAT: UNstructured Acceleration Toolkit on SW26010 many-core processor

PurposeThe purpose of this paper is to provide an automatic parallelization toolkit for unstructured mesh-based computation. Among all kinds of mesh types, unstructured meshes are dominant in engineering simulation scenarios and play an essential role in scientific computations for their geometrical flexibility. However, the high-fidelity applications based on unstructured grids are still time-consuming, no matter for programming or running.Design/methodology/approachThis study develops an efficient UNstructured Acceleration Toolkit (UNAT), which provides friendly high-level programming interfaces and elaborates lower level implementation on the target hardware to get nearly hand-optimized performance. At the present state, two efficient strategies, a multi-level blocks method and a row-subsections method, are designed and implemented on Sunway architecture. Random memory access and write–write conflict issues of unstructured meshes have been handled by partitioning, coloring and other hardware-specific techniques. Moreover, a data-reuse mechanism is developed to increase the computational intensity and alleviate the memory bandwidth bottleneck.FindingsThe authors select sparse matrix-vector multiplication as a performance benchmark of UNAT across different data layouts and different matrix formats. Experimental results show that the speed-ups reach up to 26× compared to single management processing element, and the utilization ratio tests indicate the capability of achieving nearly hand-optimized performance. Finally, the authors adopt UNAT to accelerate a well-tuned unstructured solver and obtain speed-ups of 19× and 10× on average for main kernels and overall solver, respectively.Originality/valueThe authors design an unstructured mesh toolkit, UNAT, to link the hardware and numerical algorithm, and then, engineers can focus on the algorithms and solvers rather than the parallel implementation. For the many-core processor SW26010 of the fastest supercomputer in China, UNAT yields up to 26× speed-ups and achieves nearly hand-optimized performance.

Pengujian Black Box pada Aplikasi Perpustakaan Berbasis Web Menggunakan Teknik Equivalence Partitioning

In determining the level of quality of a software it is necessary to test the software. Tests tested can include specifications, design, and coding of the software. This test is conducted to test web-based library software. This test uses the Black Box Testing method using the Equivalence Partitioning technique. Black Box Testing is applied in order to find errors in software related to erroneous functions, errors in interface display, data structure errors or access rights in external databases, sub-optimal performance in a process, and errors in initialization or termination. The Equivalence Partitioning technique defines the cases to be tested by finding some errors and minimizing the number of cases that must be made. The Equivalence Partitioning test case design is based on the premise of the input and output of a component which is partitioned into classes according to the specifications of that component, which are treated equally (equivalence). It can also be assumed that the results of an input produce the same response. The expected result of testing this software is to ensure that this software is free from errors so that in its use there will be no obstacles that prevent the user from doing his work. And inform that the software is in accordance with predetermined specifications.