Scalability postulates fault tolerance to be effective. We consider a user-level fault tolerance technique to cope with permanent node failures. It is supported by X10, one of the major Partitioned Global Address Space (PGAS) languages. In Resilient X10, an exception is thrown when a place (node) fails. This paper investigates task pools, which are often used by irregular applications to balance their load. We consider global load balancing with one worker per place. Each worker maintains a private task pool and supports cooperative work stealing. Tasks may generate new tasks dynamically, are free of side-effects, and their results are combined by reduction. Our first contribution is a task pool algorithm that can handle permanent place failures. It is based on snapshots that are regularly written to other workers and are updated in the event of stealing. Second, we implemented the algorithm in the Global Load Balancing framework GLB, which is part of the standard library of X10. We ran experiments with the Unbalanced Tree Search (UTS) and Between ness Centrality (BC) benchmarks. With 64 places on 4 nodes, for instance, we observed an overhead of about 4% for using fault-tolerant GLB instead of GLB. The protocol overhead for a place failure was neglectable.
{"title":"Fault-Tolerant Global Load Balancing in X10","authors":"Marco Bungart, Claudia Fohry, Jonas Posner","doi":"10.1109/SYNASC.2014.69","DOIUrl":"https://doi.org/10.1109/SYNASC.2014.69","url":null,"abstract":"Scalability postulates fault tolerance to be effective. We consider a user-level fault tolerance technique to cope with permanent node failures. It is supported by X10, one of the major Partitioned Global Address Space (PGAS) languages. In Resilient X10, an exception is thrown when a place (node) fails. This paper investigates task pools, which are often used by irregular applications to balance their load. We consider global load balancing with one worker per place. Each worker maintains a private task pool and supports cooperative work stealing. Tasks may generate new tasks dynamically, are free of side-effects, and their results are combined by reduction. Our first contribution is a task pool algorithm that can handle permanent place failures. It is based on snapshots that are regularly written to other workers and are updated in the event of stealing. Second, we implemented the algorithm in the Global Load Balancing framework GLB, which is part of the standard library of X10. We ran experiments with the Unbalanced Tree Search (UTS) and Between ness Centrality (BC) benchmarks. With 64 places on 4 nodes, for instance, we observed an overhead of about 4% for using fault-tolerant GLB instead of GLB. The protocol overhead for a place failure was neglectable.","PeriodicalId":150575,"journal":{"name":"2014 16th International Symposium on Symbolic and Numeric Algorithms for Scientific Computing","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130831303","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Genetic programming can optimise software, including: evolving test benchmarks, generating hyper-heuristics by searching meta-heuristics, generating communication protocols, composing telephony systems and web services, generating improved hashing and C++ heap managers, redundant programming and even automatic bug fixing. Particularly in embedded real-time or mobile systems, there may be many ways to trade off expenses (such as time, memory, energy, power consumption) vs. Functionality. Human programmers cannot try them all. Also the best multi-objective Pareto trade off may change with time, underlying hardware and network connection or user behaviour. It may be GP can automatically suggest different trade offs for each new market. Recent results include substantial speed up by evolving a new version of a program customised for a special case.
{"title":"Genetic Improvement of Programs","authors":"W. Langdon","doi":"10.1109/SYNASC.2014.10","DOIUrl":"https://doi.org/10.1109/SYNASC.2014.10","url":null,"abstract":"Genetic programming can optimise software, including: evolving test benchmarks, generating hyper-heuristics by searching meta-heuristics, generating communication protocols, composing telephony systems and web services, generating improved hashing and C++ heap managers, redundant programming and even automatic bug fixing. Particularly in embedded real-time or mobile systems, there may be many ways to trade off expenses (such as time, memory, energy, power consumption) vs. Functionality. Human programmers cannot try them all. Also the best multi-objective Pareto trade off may change with time, underlying hardware and network connection or user behaviour. It may be GP can automatically suggest different trade offs for each new market. Recent results include substantial speed up by evolving a new version of a program customised for a special case.","PeriodicalId":150575,"journal":{"name":"2014 16th International Symposium on Symbolic and Numeric Algorithms for Scientific Computing","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130620435","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Aster seismology is an emerging branch of astrophysics that studies the interior and global parameters of pulsating stars, based on their natural oscillation frequency. This work proposes different methods for increasing the performance of LNAWENR (Linear Non Adiabatic Non Radial WavEs), a computational intensive pulsational model that studies stars based on their seismic properties. As part of the aster seismological package ROMOSC, LNAWENR is one of the several non-adiabatic models in use now by the international astrophysics and space science community. The model was implemented to study data from NASA's CoRoT mission. The improved version aims to be used for NASA's KEPLER mission. In this context, we employed several serial and parallel code optimizations. The programming frameworks used for the parallel optimizations are Open MP and Open CL.
{"title":"Performance Improvements for the \"Linear Nonadiabatic Nonradial Waves\" Pulsational Model","authors":"Mihai Ovidiu Tirsa, E. Slusanschi, M. Suran","doi":"10.1109/SYNASC.2014.71","DOIUrl":"https://doi.org/10.1109/SYNASC.2014.71","url":null,"abstract":"Aster seismology is an emerging branch of astrophysics that studies the interior and global parameters of pulsating stars, based on their natural oscillation frequency. This work proposes different methods for increasing the performance of LNAWENR (Linear Non Adiabatic Non Radial WavEs), a computational intensive pulsational model that studies stars based on their seismic properties. As part of the aster seismological package ROMOSC, LNAWENR is one of the several non-adiabatic models in use now by the international astrophysics and space science community. The model was implemented to study data from NASA's CoRoT mission. The improved version aims to be used for NASA's KEPLER mission. In this context, we employed several serial and parallel code optimizations. The programming frameworks used for the parallel optimizations are Open MP and Open CL.","PeriodicalId":150575,"journal":{"name":"2014 16th International Symposium on Symbolic and Numeric Algorithms for Scientific Computing","volume":"7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121623378","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
We propose a method for ranking the service compositions according to the good practice of each domain. Knowledge about good practice is modeled in a hierarchical manner inspired from Hierarchical Task Networks. In describing the good practice knowledge we give a model for HTN in N3 notation and we enhanced it with an importance value. Each candidate service composition is checked against good practice in a model checking style. A candidate composition is a sequence of services. The candidate composition is compared to the constraints defined in good practice and is considered good if for each simple task the most important constraints are fulfilled.
{"title":"Using Domain Specific Hierarchical Good Practice for Ranking Service Compositions","authors":"A. Marginean, I. A. Letia, S. Zaporojan","doi":"10.1109/SYNASC.2014.35","DOIUrl":"https://doi.org/10.1109/SYNASC.2014.35","url":null,"abstract":"We propose a method for ranking the service compositions according to the good practice of each domain. Knowledge about good practice is modeled in a hierarchical manner inspired from Hierarchical Task Networks. In describing the good practice knowledge we give a model for HTN in N3 notation and we enhanced it with an importance value. Each candidate service composition is checked against good practice in a model checking style. A candidate composition is a sequence of services. The candidate composition is compared to the constraints defined in good practice and is considered good if for each simple task the most important constraints are fulfilled.","PeriodicalId":150575,"journal":{"name":"2014 16th International Symposium on Symbolic and Numeric Algorithms for Scientific Computing","volume":"35 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116865306","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Marius Barat, Dumitru-Bogdan Prelipcean, Dragos Gavrilut
In this paper we propose a noise detection system based on similarities between instances. Having a data set with instances that belongs to multiple classes, a noise instance denotes a wrongly classified record. The similarity between different labeled instances is determined computing distances between them using several metrics among the standard ones. In order to ensure that this approach is computational feasible for very large data sets, we compute distances between pairs of different labels instances that have a certain degree of similarity. This speed-up is possible through a new clustering method called BDT Clustering presented within this paper, which is based on a supervised learning algorithm.
{"title":"A Practical Approach on Cleaning-Up Large Data Sets","authors":"Marius Barat, Dumitru-Bogdan Prelipcean, Dragos Gavrilut","doi":"10.1109/SYNASC.2014.45","DOIUrl":"https://doi.org/10.1109/SYNASC.2014.45","url":null,"abstract":"In this paper we propose a noise detection system based on similarities between instances. Having a data set with instances that belongs to multiple classes, a noise instance denotes a wrongly classified record. The similarity between different labeled instances is determined computing distances between them using several metrics among the standard ones. In order to ensure that this approach is computational feasible for very large data sets, we compute distances between pairs of different labels instances that have a certain degree of similarity. This speed-up is possible through a new clustering method called BDT Clustering presented within this paper, which is based on a supervised learning algorithm.","PeriodicalId":150575,"journal":{"name":"2014 16th International Symposium on Symbolic and Numeric Algorithms for Scientific Computing","volume":"28 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133205334","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This paper treats the phenomenon of opinion influence in online forum threads. Influence among users' opinions is analyzed by taking into consideration the changes in their opinions. Therefore, a change in a user's opinion is modeled as a change of his/her posts' polarity. The hypothesis that underlies our research is that users' opinions may change over time as a consequence of the interactions between them in online discussions such as online forum threads. Moreover, a user's new post in an online forum thread is considered to have an influence on all the posts sent by other users in reaction to this. Our approach to the opinion influence phenomenon that arises in online forum threads is based on Natural Language Processing techniques, Latent Semantic Analysis and Post-Level Sentiment Analysis. The results obtained by us show that all the previous posts that contain opinions have an influence, more or less significant, on a new post in the same online forum thread.
{"title":"Opinion Influence Analysis in Online Forum Threads","authors":"Dumitru-Clementin Cercel, Stefan Trausan-Matu","doi":"10.1109/SYNASC.2014.38","DOIUrl":"https://doi.org/10.1109/SYNASC.2014.38","url":null,"abstract":"This paper treats the phenomenon of opinion influence in online forum threads. Influence among users' opinions is analyzed by taking into consideration the changes in their opinions. Therefore, a change in a user's opinion is modeled as a change of his/her posts' polarity. The hypothesis that underlies our research is that users' opinions may change over time as a consequence of the interactions between them in online discussions such as online forum threads. Moreover, a user's new post in an online forum thread is considered to have an influence on all the posts sent by other users in reaction to this. Our approach to the opinion influence phenomenon that arises in online forum threads is based on Natural Language Processing techniques, Latent Semantic Analysis and Post-Level Sentiment Analysis. The results obtained by us show that all the previous posts that contain opinions have an influence, more or less significant, on a new post in the same online forum thread.","PeriodicalId":150575,"journal":{"name":"2014 16th International Symposium on Symbolic and Numeric Algorithms for Scientific Computing","volume":"333 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115981599","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
M. Malekimajd, A. M. Rizzi, D. Ardagna, M. Ciavotta, M. Passacantando, A. Movaghar
Nowadays, analyzing large amount of data is of paramount importance for many companies. Big data and business intelligence applications are facilitated by the MapReduce programming model while, at infrastructural layer, cloud computing provides flexible and cost effective solutions for allocating on demand large clusters. Capacity allocation in such systems is a key challenge to providing performance for MapReduce jobs and minimize cloud resource cost. The contribution of this paper is twofold: (i) we formulate a linear programming model able to minimize cloud resources cost and job rejection penalties for the execution of jobs of multiple classes with (soft) deadline guarantees, (ii) we provide new upper and lower bounds for MapReduce job execution time in shared Hadoop clusters. Moreover, our solutions are validated by a large set of experiments. We demonstrate that our method is able to determine the global optimal solution for systems including up to 1000 user classes in less than 0.5 seconds. Moreover, the execution time of MapReduce jobs are within 19% of our upper bounds on average.
{"title":"Optimal Capacity Allocation for Executing MapReduce Jobs in Cloud Systems","authors":"M. Malekimajd, A. M. Rizzi, D. Ardagna, M. Ciavotta, M. Passacantando, A. Movaghar","doi":"10.1109/SYNASC.2014.58","DOIUrl":"https://doi.org/10.1109/SYNASC.2014.58","url":null,"abstract":"Nowadays, analyzing large amount of data is of paramount importance for many companies. Big data and business intelligence applications are facilitated by the MapReduce programming model while, at infrastructural layer, cloud computing provides flexible and cost effective solutions for allocating on demand large clusters. Capacity allocation in such systems is a key challenge to providing performance for MapReduce jobs and minimize cloud resource cost. The contribution of this paper is twofold: (i) we formulate a linear programming model able to minimize cloud resources cost and job rejection penalties for the execution of jobs of multiple classes with (soft) deadline guarantees, (ii) we provide new upper and lower bounds for MapReduce job execution time in shared Hadoop clusters. Moreover, our solutions are validated by a large set of experiments. We demonstrate that our method is able to determine the global optimal solution for systems including up to 1000 user classes in less than 0.5 seconds. Moreover, the execution time of MapReduce jobs are within 19% of our upper bounds on average.","PeriodicalId":150575,"journal":{"name":"2014 16th International Symposium on Symbolic and Numeric Algorithms for Scientific Computing","volume":"98 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126088969","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
We detail the integration in SPIKE, an implicit induction theorem prover, of two reasoning modules operating over naturals combined with interpreted symbols. The first integration schema is à la Boyer-Moore, based on the combination of a congruence closure procedure with a decision procedure for linear arithmetic over rationals/reals. The second follows a 'black-box' approach and is based on external SMT solvers. It is shown that the two extensions significantly increase the power of SPIKE, their performances are compared when proving a non-trivial application.
我们详细介绍了在SPIKE(一个隐式归纳法定理证明器)中，两个推理模块在结合了解释符号的自然上运行的集成。第一种积分模式是基于同余闭包过程与有理数/实数上线性算法的决策过程的结合的 la Boyer-Moore。第二种方法采用“黑盒”方法，并基于外部SMT求解器。结果表明，这两种扩展都显著提高了SPIKE的性能，并在一个非平凡应用中对其性能进行了比较。
{"title":"Implementing Reasoning Modules in Implicit Induction Theorem Provers","authors":"Sorin Stratulat","doi":"10.1109/SYNASC.2014.26","DOIUrl":"https://doi.org/10.1109/SYNASC.2014.26","url":null,"abstract":"We detail the integration in SPIKE, an implicit induction theorem prover, of two reasoning modules operating over naturals combined with interpreted symbols. The first integration schema is à la Boyer-Moore, based on the combination of a congruence closure procedure with a decision procedure for linear arithmetic over rationals/reals. The second follows a 'black-box' approach and is based on external SMT solvers. It is shown that the two extensions significantly increase the power of SPIKE, their performances are compared when proving a non-trivial application.","PeriodicalId":150575,"journal":{"name":"2014 16th International Symposium on Symbolic and Numeric Algorithms for Scientific Computing","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121366408","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
D. Onchis, D. Frunzaverde, Mihail Gaianu, Relu Ciubotariu
This paper presents an effective algorithm for the identification of multiple phases in microstructures images. The procedure is based on an efficient image segmentation using the fuzzy c-means algorithm. Furthermore, the algorithm is accelerated on a GPU cluster in order to obtain optimal computing times for large size images. The results are compared on the same experimental images with the ones obtained from a commercial software and the accuracy of the proposed algorithm is demonstrated.
{"title":"Multi-phase Identification in Microstructures Images Using a GPU Accelerated Fuzzy C-Means Segmentation","authors":"D. Onchis, D. Frunzaverde, Mihail Gaianu, Relu Ciubotariu","doi":"10.1109/SYNASC.2014.86","DOIUrl":"https://doi.org/10.1109/SYNASC.2014.86","url":null,"abstract":"This paper presents an effective algorithm for the identification of multiple phases in microstructures images. The procedure is based on an efficient image segmentation using the fuzzy c-means algorithm. Furthermore, the algorithm is accelerated on a GPU cluster in order to obtain optimal computing times for large size images. The results are compared on the same experimental images with the ones obtained from a commercial software and the accuracy of the proposed algorithm is demonstrated.","PeriodicalId":150575,"journal":{"name":"2014 16th International Symposium on Symbolic and Numeric Algorithms for Scientific Computing","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116151843","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
We investigate the semantics of a biologically-inspired formalism. This formalism was initially introduced by Cardelli as a "strand algebra" for DNA computing. For such a language we study and relate new formal semantic models. The mathematical framework is given by complete metric spaces in which the Banach fixed point theorem is used, various semantic functions are defined as fixed points of appropriate higher-order mappings. We define a new denotational semantics and compare it with the operational semantics introduced by Cardelli. We establish the formal relation between the operational and the denotational semantics by using an abstraction operator and a fixed point argument. In this way we establish the correctness of the denotational semantics with respect to the operational semantics.
{"title":"Correct Metric Semantics for a Biologically-Inspired Formalism","authors":"Gabriel Ciobanu, E. Todoran","doi":"10.1109/SYNASC.2014.50","DOIUrl":"https://doi.org/10.1109/SYNASC.2014.50","url":null,"abstract":"We investigate the semantics of a biologically-inspired formalism. This formalism was initially introduced by Cardelli as a \"strand algebra\" for DNA computing. For such a language we study and relate new formal semantic models. The mathematical framework is given by complete metric spaces in which the Banach fixed point theorem is used, various semantic functions are defined as fixed points of appropriate higher-order mappings. We define a new denotational semantics and compare it with the operational semantics introduced by Cardelli. We establish the formal relation between the operational and the denotational semantics by using an abstraction operator and a fixed point argument. In this way we establish the correctness of the denotational semantics with respect to the operational semantics.","PeriodicalId":150575,"journal":{"name":"2014 16th International Symposium on Symbolic and Numeric Algorithms for Scientific Computing","volume":"9 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126707959","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：