{"id":784,"date":"2012-05-31T17:02:20","date_gmt":"2012-05-31T15:02:20","guid":{"rendered":"https:\/\/webs.uab.cat\/hpc4eas\/news\/thesis-rsolar\/"},"modified":"2012-05-31T17:02:20","modified_gmt":"2012-05-31T15:02:20","slug":"thesis-rsolar","status":"publish","type":"news","link":"https:\/\/webs.uab.cat\/hpc4eas\/news\/thesis-rsolar\/","title":{"rendered":"Thesis R.Solar"},"content":{"rendered":"<p><strong>Abstract:<\/strong><\/p>\n<p><span id=\"result_box\" class=\"long_text\" lang=\"en\"><span class=\"hps\">Partitioning<\/span> <span class=\"hps\">and load balancing<\/span> <span class=\"hps\">are issues<\/span> <span class=\"hps\">of great interest in <\/span><span class=\"hps\">distributed simulations<\/span> <span class=\"hps\">based on <\/span><span class=\"hps\">spatially explicit<\/span> <span class=\"hps\">individual-oriented<\/span><span> models.<\/span> <span class=\"hps\">The decomposition<\/span> <span class=\"hps\">of the problem domain<\/span> <span class=\"hps\">and<\/span> <span class=\"hps\">the efficient data distribution <\/span><span class=\"hps\">on<\/span> <span class=\"hps\">the computing nodes<\/span> <span class=\"hps\">of<\/span> <span class=\"hps\">the parallel architecture<\/span> <span class=\"hps\">\/<\/span> <span class=\"hps\">distributed<\/span> <span class=\"hps\">are crucial factors in<\/span> <span class=\"hps\">performance figures<\/span> <span class=\"hps\">for<\/span> <span class=\"hps\">distributed simulation.<\/span><\/span><\/p>\n<p><span class=\"long_text\" lang=\"en\"><span class=\"hps\">In<\/span> <span class=\"hps\">this work we<\/span> <span class=\"hps\">have developed a<\/span> <span class=\"hps\">new methodology for<\/span> <span class=\"hps\">partitioning<\/span> <span class=\"hps\">and load balancing<\/span> <span class=\"hps\">for<\/span> <span class=\"hps\">large-scale<\/span> <span class=\"hps\">distributed simulations<\/span> <span class=\"hps\">of<\/span> <span class=\"hps\">individual-oriented<\/span> <span class=\"hps\">models<\/span> <span class=\"hps\">that show <\/span><span class=\"hps\">spatially explicit<\/span> <span class=\"hps\">movement patterns. <span id=\"result_box\" class=\"long_text\" lang=\"en\"><span class=\"hps\">In order to validate our strategies, <\/span><span class=\"hps\">the model of<\/span> <span class=\"hps\">Huth<\/span> <span class=\"hps\">&amp;<\/span> <span class=\"hps\">Wissel<\/span><span>, which <\/span><span class=\"hps\">represents the<\/span> <span class=\"hps\">coordinated and<\/span> <span class=\"hps\">polarized<\/span> <span class=\"hps\">movement<\/span> <span class=\"hps\">of fish<\/span>, has been used. <\/span><\/span><\/span><\/p>\n<p><span class=\"long_text\" lang=\"en\"><span class=\"hps\"><span id=\"result_box\" class=\"long_text\" lang=\"en\"><span class=\"hps\">The<\/span> <span class=\"hps\">partitioning method<\/span> <span class=\"hps\">is to<\/span> <span class=\"hps\">decompose the<\/span> <span class=\"hps\">problem domain into<\/span> <span class=\"hps\">compact<\/span> <span class=\"hps\">partitions<\/span> <span class=\"hps\">generated from the<\/span> <span class=\"hps\">radial<\/span> <span class=\"hps\">blanket<\/span> <span class=\"hps\">approach<\/span> <span class=\"hps\">and<\/span> <span class=\"hps\">Voronoi<\/span> <span class=\"hps\">diagrams<\/span><span>.<\/span> <span class=\"hps\">The distribution<\/span> <span class=\"hps\">of partitions<\/span> <span class=\"hps\">is performed<\/span> <span class=\"hps\">by means of<\/span> <span class=\"hps\">the<\/span> <span class=\"hps\">proximity<\/span> <span class=\"hps\">cluster<\/span> <span class=\"hps\">partitions<\/span> using <span id=\"result_box\" class=\"long_text\" lang=\"en\">a<\/span> <span class=\"hps atn\">new definition meta-partitions <\/span><span class=\"hps atn\">equals to the number<\/span> <span class=\"hps\">of<\/span> <span class=\"hps\">computing <\/span><span class=\"hps\">cores. <\/span><span class=\"hps\">The strategy for<\/span> <span class=\"hps\">dynamic<\/span> <span class=\"hps\">load balancing<\/span> <span class=\"hps\">is to detect the<\/span> <span class=\"hps\">imbalance<\/span> <span class=\"hps\">through<\/span> <span class=\"hps\">an algorithm based on<\/span> <span class=\"hps\">thresholds and<\/span> <span class=\"hps\">re<\/span><span class=\"atn\">&#8211;<\/span><span>confi<\/span><span class=\"hps\">gure<\/span> <span class=\"hps atn\">meta-<\/span><span>partitions<\/span> <span class=\"hps\">to achieve<\/span> <span class=\"hps\">rebalancing<\/span><span>.<\/span> <span class=\"hps\">Finally<\/span><span>, there has been<\/span> <span class=\"hps\">extensive<\/span> <span class=\"hps\">experimentation<\/span> <span id=\"result_box\" class=\"long_text\" lang=\"en\">to validate<\/span> <span class=\"hps\">and verify<\/span> <span class=\"hps\">the viability of<\/span> <span id=\"result_box\" class=\"long_text\" lang=\"en\">the distributed<\/span> <span class=\"hps\">simulation<\/span> <span class=\"hps\">in different scenarios.<\/span><\/span> <br \/><\/span><\/span><\/p>\n","protected":false},"excerpt":{"rendered":"<p>Abstract: Partitioning and load balancing are issues of great interest in distributed simulations based on spatially explicit individual-oriented models. The decomposition of the problem domain and the efficient data distribution on the computing nodes of the parallel architecture \/ distributed are crucial factors in performance figures for distributed simulation. In this work we have developed [&hellip;]<\/p>\n","protected":false},"author":20,"featured_media":0,"template":"","class_list":["post-784","news","type-news","status-publish","hentry"],"_links":{"self":[{"href":"https:\/\/webs.uab.cat\/hpc4eas\/wp-json\/wp\/v2\/news\/784","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/webs.uab.cat\/hpc4eas\/wp-json\/wp\/v2\/news"}],"about":[{"href":"https:\/\/webs.uab.cat\/hpc4eas\/wp-json\/wp\/v2\/types\/news"}],"author":[{"embeddable":true,"href":"https:\/\/webs.uab.cat\/hpc4eas\/wp-json\/wp\/v2\/users\/20"}],"wp:attachment":[{"href":"https:\/\/webs.uab.cat\/hpc4eas\/wp-json\/wp\/v2\/media?parent=784"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}