scholarly journals RLSchert: An HPC Job Scheduler Using Deep Reinforcement Learning and Remaining Time Prediction

2021 ◽  
Vol 11 (20) ◽  
pp. 9448
Author(s):  
Qiqi Wang ◽  
Hongjie Zhang ◽  
Cheng Qu ◽  
Yu Shen ◽  
Xiaohui Liu ◽  
...  
Keyword(s):  
Reinforcement Learning ◽  
High Performance ◽  
Job Scheduling ◽  
Vital Role ◽  
Runtime Prediction ◽  
Heuristic Scheduling ◽  
The Status ◽  
Computing Platforms ◽  
Job Scheduler ◽  
Policy Optimization

The job scheduler plays a vital role in high-performance computing platforms. It determines the execution order of the jobs and the allocation of resources, which in turn affect the resource utilization of the entire system. As the scale and complexity of HPC continue to grow, job scheduling is becoming increasingly important and difficult. Existing studies relied on user-specified or regression techniques to give fixed runtime prediction values and used the values in static heuristic scheduling algorithms. However, these approaches require very accurate runtime predictions to produce better results, and fixed heuristic scheduling strategies cannot adapt to changes in the workload. In this work, we propose RLSchert, a job scheduler based on deep reinforcement learning and remaining runtime prediction. Firstly, RLSchert estimates the state of the system by using a dynamic job remaining runtime predictor, thereby providing an accurate spatiotemporal view of the cluster status. Secondly, RLSchert learns the optimal policy to select or kill jobs according to the status through imitation learning and the proximal policy optimization algorithm. Extensive experiments on real-world job logs at the USTC Supercomputing Center showed that RLSchert is superior to static heuristic policies and outperforms the learning-based scheduler DeepRM. In addition, the dynamic predictor gives a more accurate remaining runtime prediction result, which is essential for most learning-based schedulers.

scholarly journals Dynamic Integration and Management of Opportunistic Resources for HEP

2019 ◽  
Vol 214 ◽  
pp. 08009 ◽  
Author(s):  
Matthias J. Schnepf ◽  
R. Florian von Cube ◽  
Max Fischer ◽  
Manuel Giffels ◽  
Christoph Heidecker ◽  
...  
Keyword(s):  
High Performance ◽  
High Energy Physics ◽  
Job Scheduling ◽  
High Energy ◽  
Software Environment ◽  
Resource Manager ◽  
Network Bandwidth ◽  
The Status ◽  
Single Entry ◽  
Institute Of Technology

Demand for computing resources in high energy physics (HEP) shows a highly dynamic behavior, while the provided resources by the Worldwide LHC Computing Grid (WLCG) remains static. It has become evident that opportunistic resources such as High Performance Computing (HPC) centers and commercial clouds are well suited to cover peak loads. However, the utilization of these resources gives rise to new levels of complexity, e.g. resources need to be managed highly dynamically and HEP applications require a very specific software environment usually not provided at opportunistic resources. Furthermore, aspects to consider are limitations in network bandwidth causing I/O-intensive workflows to run inefficiently. The key component to dynamically run HEP applications on opportunistic resources is the utilization of modern container and virtualization technologies. Based on these technologies, the Karlsruhe Institute of Technology (KIT) has developed ROCED, a resource manager to dynamically integrate and manage a variety of opportunistic resources. In combination with ROCED, HTCondor batch system acts as a powerful single entry point to all available computing resources, leading to a seamless and transparent integration of opportunistic resources into HEP computing. KIT is currently improving the resource management and job scheduling by focusing on I/O requirements of individual workflows, available network bandwidth as well as scalability. For these reasons, we are currently developing a new resource manager, called TARDIS. In this paper, we give an overview of the utilized technologies, the dynamic management, and integration of resources as well as the status of the I/O-based resource and job scheduling.

scholarly journals The status quo and perspective for improvement of public works procurement performance in Vietnam

2014 ◽  
Vol 6 (1) ◽  
pp. 22 ◽  
Author(s):  
Hai Luong Nguyen ◽  
Tsunemi Watanabe
Keyword(s):  
Construction Industry ◽  
Construction Projects ◽  
High Performance ◽  
Public Procurement ◽  
Poor Performance ◽  
Vital Role ◽  
Public Works ◽  
The Public ◽  
The Status ◽  
Questionnaire Surveys

The public procurement sector plays a vital role in the economic development in developing countries such as Vietnam. However, public procurement activities usually perform poorly. This situation can be attributed to ineffective procedures and system (“hardware”) and human resource management (“software”), which occurs at every stage in project purchasing. The poor performance has reduced the effectiveness and efficiency of project delivery in the construction industry, causing delays, cost over-runs, and defects in construction projects. This paper, through working experience and observation by the first author, problems of public procurement were obtained as hypotheses and then validated based on questionnaire surveys and CIS (Construction Industry Structure) model analysis. The survey results indicated a relative correlation with CIS model in description of current construction industry. The study aims to identify issues of public procurement at all stages: pre bid, bid information, evaluation, and award. Based on identified major problems and determined risks, the results are expected to provide a valuable perspective, and thus, to propose necessary strategies to deliver high performance, competition and transparency for the public procurement. In further studies, it is relevant to propose a new model for sustainable public procurement based on the best value approach.

scholarly journals Deep Reinforcement Learning for the Computation Offloading in MIMO-based Edge Computing

2021 ◽  
Author(s):  
Abdeladim Sadiki ◽  
Jamal Bentahar ◽  
Rachida Dssouli ◽  
Abdeslam En-Nouaary
Keyword(s):  
Reinforcement Learning ◽  
High Performance ◽  
Multiple Input Multiple Output ◽  
Edge Computing ◽  
Base Stations ◽  
Computation Offloading ◽  
Discrete Action ◽  
Computationally Intensive ◽  
Multi Access ◽  
Policy Optimization

Multi-access Edge Computing (MEC) has recently emerged as a potential technology to serve the needs of mobile devices (MDs) in 5G and 6G cellular networks. By offloading tasks to high-performance servers installed at the edge of the wireless networks, resource-limited MDs can cope with the proliferation of the recent computationally-intensive applications. In this paper, we study the computation offloading problem in a massive multiple-input multiple-output (MIMO)-based MEC system where the base stations are equipped with a large number of antennas. Our objective is to minimize the power consumption and offloading delay at the MDs under the stochastic system environment. To this end, we formulate the problem as a Markov Decision Process (MDP) and propose two Deep Reinforcement Learning (DRL) strategies to learn the optimal offloading policy without any prior knowledge of the environment dynamics. First, a Deep Q-Network (DQN) strategy to solve the curse of the state space explosion is analyzed. Then, a more general Proximal Policy Optimization (PPO) strategy to solve the problem of discrete action space is introduced. Simulation results show that the proposed DRL-based strategies outperform the baseline and state-of-the-art algorithms. Moreover, our PPO algorithm exhibits stable performance and efficient offloading results compared to the benchmark DQN strategy.

scholarly journals Deep Reinforcement Learning for the Computation Offloading in MIMO-based Edge Computing

2021 ◽  
Author(s):  
Abdeladim Sadiki ◽  
Jamal Bentahar ◽  
Rachida Dssouli ◽  
Abdeslam En-Nouaary
Keyword(s):  
Reinforcement Learning ◽  
High Performance ◽  
Multiple Input Multiple Output ◽  
Edge Computing ◽  
Base Stations ◽  
Computation Offloading ◽  
Discrete Action ◽  
Computationally Intensive ◽  
Multi Access ◽  
Policy Optimization

Multi-access Edge Computing (MEC) has recently emerged as a potential technology to serve the needs of mobile devices (MDs) in 5G and 6G cellular networks. By offloading tasks to high-performance servers installed at the edge of the wireless networks, resource-limited MDs can cope with the proliferation of the recent computationally-intensive applications. In this paper, we study the computation offloading problem in a massive multiple-input multiple-output (MIMO)-based MEC system where the base stations are equipped with a large number of antennas. Our objective is to minimize the power consumption and offloading delay at the MDs under the stochastic system environment. To this end, we formulate the problem as a Markov Decision Process (MDP) and propose two Deep Reinforcement Learning (DRL) strategies to learn the optimal offloading policy without any prior knowledge of the environment dynamics. First, a Deep Q-Network (DQN) strategy to solve the curse of the state space explosion is analyzed. Then, a more general Proximal Policy Optimization (PPO) strategy to solve the problem of discrete action space is introduced. Simulation results show that the proposed DRL-based strategies outperform the baseline and state-of-the-art algorithms. Moreover, our PPO algorithm exhibits stable performance and efficient offloading results compared to the benchmark DQN strategy.

The Role of Entrepreneurship in Local Economic Growth: Example of Gaziantep

2012 ◽  
Author(s):  
Mehmet Vahit Eren ◽  
Erdinç Tutar ◽  
Filiz Tutar ◽  
Çisil Erkan
Keyword(s):  
Swot Analysis ◽  
Vital Role ◽  
Local Economy ◽  
Government Policies ◽  
Optimum Level ◽  
Economic Sectors ◽  
Local Economies ◽  
The Status ◽  
Financial Growth

In order to avoid social inequality of opportunity and improvement of local economies have become government policies in Turkey, as it is in other countries around the world. Incentives, regional development agencies, techno parks and also local entrepreneurs play crucial role in the improvement process of local economies. The increasing rivalry and globalization concept necessitate entrepreneurs to take more risks, to reach innovations to seize opportunities in optimum level. Entrepreneurship is a motor vessel in financial growth and in development, and entrepreneurship is also the source of innovation and creativity. In this regard, the more entrepreneurship develops in a country, the higher level of welfare possesses the chance to advance. The purpose of this report, in which it has been aimed to reveal vital role of entrepreneurship in the progress of local economies, is emphasizing the status of entrepreneurship that transformed Gaziantep’s socio-economic level of development into its present position. Thus with this aim a SWOT analysis, in terms of Gaziantep’s economic entrepreneurship has been carried out. Positive contributions of Gaziantep’s immensely developed industry, facilitation of local innovative entrepreneurs’ involvement in various local economic sectors and in accordance channeling immigration into deployment in local economy have been observed in this study. Significantly it has been observed that plenitude of “opportunist entrepreneurship” or in other words “the entrepreneurs with strategic growth plans” in this region contributed local economy positively.

Multi-level Parallelization of Genotype Imputation on Supercomputers

2020 ◽  
Vol 15 ◽  
Author(s):  
Weiwen Zhang ◽  
Long Wang ◽  
Theint Theint Aye ◽  
Juniarto Samsudin ◽  
Yongqing Zhu
Keyword(s):  
Association Study ◽  
Message Passing ◽  
High Performance ◽  
Message Passing Interface ◽  
Genome Wide Association Study ◽  
Job Scheduling ◽  
Genotype Imputation ◽  
Job Level ◽  
Multi Level ◽  
High Performance Requirement

Background: Genotype imputation as a service is developed to enable researchers to estimate genotypes on haplotyped data without performing whole genome sequencing. However, genotype imputation is computation intensive and thus it remains a challenge to satisfy the high performance requirement of genome wide association study (GWAS). Objective: In this paper, we propose a high performance computing solution for genotype imputation on supercomputers to enhance its execution performance. Method: We design and implement a multi-level parallelization that includes job level, process level and thread level parallelization, enabled by job scheduling management, message passing interface (MPI) and OpenMP, respectively. It involves job distribution, chunk partition and execution, parallelized iteration for imputation and data concatenation. Due to the design of multi-level parallelization, we can exploit the multi-machine/multi-core architecture to improve the performance of genotype imputation. Results: Experiment results show that our proposed method can outperform the Hadoop-based implementation of genotype imputation. Moreover, we conduct the experiments on supercomputers to evaluate the performance of the proposed method. The evaluation shows that it can significantly shorten the execution time, thus improving the performance for genotype imputation. Conclusion: The proposed multi-level parallelization, when deployed as an imputation as a service, will facilitate bioinformatics researchers in Singapore to conduct genotype imputation and enhance the association study.

scholarly journals A novel PET tracer 18F-deoxy-thiamine: synthesis, metabolic kinetics, and evaluation on cerebral thiamine metabolism status

2020 ◽  
Vol 10 (1) ◽  
Author(s):  
Changpeng Wang ◽  
Siwei Zhang ◽  
Yuefei Zou ◽  
Hongzhao Ma ◽  
Donglang Jiang ◽  
...  
Keyword(s):  
High Performance ◽  
Specific Activity ◽  
High Accumulation ◽  
Metabolic Kinetics ◽  
Thiamine Metabolism ◽  
Pet Tracer ◽  
The Status ◽  
The Brain

Abstract Background Some neuropsychological diseases are associated with abnormal thiamine metabolism, including Korsakoff–Wernicke syndrome and Alzheimer’s disease. However, in vivo detection of the status of brain thiamine metabolism is still unavailable and needs to be developed. Methods A novel PET tracer of 18F-deoxy-thiamine was synthesized using an automated module via a two-step route. The main quality control parameters, such as specific activity and radiochemical purity, were evaluated by high-performance liquid chromatography (HPLC). Radiochemical concentration was determined by radioactivity calibrator. Metabolic kinetics and the level of 18F-deoxy-thiamine in brains of mice and marmosets were studied by micro-positron emission tomography/computed tomography (PET/CT). In vivo stability, renal excretion rate, and biodistribution of 18F-deoxy-thiamine in the mice were assayed using HPLC and γ-counter, respectively. Also, the correlation between the retention of cerebral 18F-deoxy-thiamine in 60 min after injection as represented by the area under the curve (AUC) and blood thiamine levels was investigated. Results The 18F-deoxy-thiamine was stable both in vitro and in vivo. The uptake and clearance of 18F-deoxy-thiamine were quick in the mice. It reached the max standard uptake value (SUVmax) of 4.61 ± 0.53 in the liver within 1 min, 18.67 ± 7.04 in the kidney within half a minute. The SUV dropped to 0.72 ± 0.05 and 0.77 ± 0.35 after 60 min of injection in the liver and kidney, respectively. After injection, kidney, liver, and pancreas exhibited high accumulation level of 18F-deoxy-thiamine, while brain, muscle, fat, and gonad showed low accumulation concentration, consistent with previous reports on thiamine distribution in mice. Within 90 min after injection, the level of 18F-deoxy-thiamine in the brain of C57BL/6 mice with thiamine deficiency (TD) was 1.9 times higher than that in control mice, and was 3.1 times higher in ICR mice with TD than that in control mice. The AUC of the tracer in the brain of marmosets within 60 min was 29.33 ± 5.15 and negatively correlated with blood thiamine diphosphate levels (r = − 0.985, p = 0.015). Conclusion The 18F-deoxy-thiamine meets the requirements for ideal PET tracer for in vivo detecting the status of cerebral thiamine metabolism.

scholarly journals Air Learning: a deep reinforcement learning gym for autonomous aerial robot visual navigation

2021 ◽  
Author(s):  
Srivatsan Krishnan ◽  
Behzad Boroujerdian ◽  
William Fu ◽  
Aleksandra Faust ◽  
Vijay Janapa Reddi
Keyword(s):  
Reinforcement Learning ◽  
Embedded System ◽  
Broad Class ◽  
Visual Navigation ◽  
Raspberry Pi ◽  
Latency Distribution ◽  
Hardware In The Loop ◽  
Resource Constrained ◽  
Aerial Robot ◽  
Policy Optimization

AbstractWe introduce Air Learning, an open-source simulator, and a gym environment for deep reinforcement learning research on resource-constrained aerial robots. Equipped with domain randomization, Air Learning exposes a UAV agent to a diverse set of challenging scenarios. We seed the toolset with point-to-point obstacle avoidance tasks in three different environments and Deep Q Networks (DQN) and Proximal Policy Optimization (PPO) trainers. Air Learning assesses the policies’ performance under various quality-of-flight (QoF) metrics, such as the energy consumed, endurance, and the average trajectory length, on resource-constrained embedded platforms like a Raspberry Pi. We find that the trajectories on an embedded Ras-Pi are vastly different from those predicted on a high-end desktop system, resulting in up to $$40\%$$ 40 % longer trajectories in one of the environments. To understand the source of such discrepancies, we use Air Learning to artificially degrade high-end desktop performance to mimic what happens on a low-end embedded system. We then propose a mitigation technique that uses the hardware-in-the-loop to determine the latency distribution of running the policy on the target platform (onboard compute on aerial robot). A randomly sampled latency from the latency distribution is then added as an artificial delay within the training loop. Training the policy with artificial delays allows us to minimize the hardware gap (discrepancy in the flight time metric reduced from 37.73% to 0.5%). Thus, Air Learning with hardware-in-the-loop characterizes those differences and exposes how the onboard compute’s choice affects the aerial robot’s performance. We also conduct reliability studies to assess the effect of sensor failures on the learned policies. All put together, Air Learning enables a broad class of deep RL research on UAVs. The source code is available at: https://github.com/harvard-edge/AirLearning.

scholarly journals An Efficiency Enhancing Methodology for Multiple Autonomous Vehicles in an Urban Network Adopting Deep Reinforcement Learning

2021 ◽  
Vol 11 (4) ◽  
pp. 1514 ◽  
Author(s):  
Quang-Duy Tran ◽  
Sang-Hoon Bae
Keyword(s):  
Reinforcement Learning ◽  
Traffic Congestion ◽  
Autonomous Vehicles ◽  
Penetration Rate ◽  
Autonomous Vehicle ◽  
Effective Means ◽  
Urban Network ◽  
Learning Agents ◽  
Policy Optimization ◽  
The Impact

To reduce the impact of congestion, it is necessary to improve our overall understanding of the influence of the autonomous vehicle. Recently, deep reinforcement learning has become an effective means of solving complex control tasks. Accordingly, we show an advanced deep reinforcement learning that investigates how the leading autonomous vehicles affect the urban network under a mixed-traffic environment. We also suggest a set of hyperparameters for achieving better performance. Firstly, we feed a set of hyperparameters into our deep reinforcement learning agents. Secondly, we investigate the leading autonomous vehicle experiment in the urban network with different autonomous vehicle penetration rates. Thirdly, the advantage of leading autonomous vehicles is evaluated using entire manual vehicle and leading manual vehicle experiments. Finally, the proximal policy optimization with a clipped objective is compared to the proximal policy optimization with an adaptive Kullback–Leibler penalty to verify the superiority of the proposed hyperparameter. We demonstrate that full automation traffic increased the average speed 1.27 times greater compared with the entire manual vehicle experiment. Our proposed method becomes significantly more effective at a higher autonomous vehicle penetration rate. Furthermore, the leading autonomous vehicles could help to mitigate traffic congestion.

scholarly journals Effects of Molecular Weight of Functionalized Liquid Butadiene Rubber as a Processing Aid on the Properties of SSBR/Silica Compounds

Polymers ◽  
2021 ◽  
Vol 13 (6) ◽  
pp. 850
Author(s):  
Donghyuk Kim ◽  
Byungkyu Ahn ◽  
Kihyun Kim ◽  
JongYeop Lee ◽  
Il Jin Kim ◽  
...  
Keyword(s):  
Molecular Weight ◽  
Anionic Polymerization ◽  
High Performance ◽  
Crosslink Density ◽  
Functional Group ◽  
Vital Role ◽  
Butadiene Rubber ◽  
Molecular Weights ◽  
Mooney Viscosity ◽  
Silica Dispersion

Liquid butadiene rubber (LqBR) which used as a processing aid play a vital role in the manufacturing of high-performance tire tread compounds. However, the studies on the effect of molecular weight, microstructure, and functionalization of LqBR on the properties of compounds are still insufficient. In this study, non-functionalized and center-functionalized liquid butadiene rubbers (N-LqBR and C-LqBR modified with ethoxysilyl group, respectively) were synthesized with low vinyl content and different molecular weights using anionic polymerization. In addition, LqBR was added to the silica-filled SSBR compounds as an alternative to treated distillate aromatic extract (TDAE) oil, and the effect of molecular weight and functionalization on the properties of the silica-filled SSBR compound was examined. C-LqBR showed a low Payne effect and Mooney viscosity because of improved silica dispersion due to the ethoxysilyl functional group. Furthermore, C-LqBR showed an increased crosslink density, improved mechanical properties, and reduced organic matter extraction compared to the N-LqBR compound. LqBR reduced the glass transition temperature (Tg) of the compound significantly, thereby improving snow traction and abrasion resistance compared to TDAE oil. Furthermore, the energy loss characteristics revealed that the hysteresis loss attributable to the free chain ends of LqBR was dominant.

Sign in / Sign up

Export Citation Format

Share Document