|
Size: 2588
Comment:
|
Size: 842
Comment:
|
| Deletions are marked like this. | Additions are marked like this. |
| Line 1: | Line 1: |
| ||Previous years: ||[[ScientificSeminar/2015|2015]] ||[[ScientificSeminar/2014|2014]] ||[[ScientificSeminar/2013|2013]] ||[[ScientificSeminar/2012|2012]] ||[[ScientificSeminar/2011|2011]] ||[[ScientificSeminar/2010|2010]] ||[[ScientificSeminar/2009|2009]] ||[[ScientificSeminar/2008|2008]] ||[[ScientificSeminar/2007|2007]] ||[[ScientificSeminar/2006|2006]] ||[[ScientificSeminar/2005|2005]] || | We present a novel parallel algorithm to solve fractional-order systems involving Caputo-types derivative. The numerical method used for finding the solution is Adams-Bashforth-Moulton (ABM) predictor-corrector scheme. Using MPI and OPENMP frameworks the algorithm was implemented to run on a BlueGene/P supercomputer. The same algorithm that ran on the BlueGene/P cluster was adapted to run on GPU and exploit the GPU's capabilities. A comparison between these approaches is presented by running numerical simulations of the same three-dimensional fractional-order systems. |
| Line 3: | Line 3: |
| ---- '''2016''' ---- [[AbstractCrisDoloc|21-th Century Medicine: Digital - Computable - Algorithmic]] [[attachment:CrisDoloc_Slides - 26 July 2016.pdf]] [[BioCrisDoloc|Cris Doloc]], University of Chicago & ALGOMEX & RoGeniX, USA . July 26, 2016, 10:00, room A01 ---- ENTICE Project: dEcentralized repositories for traNsparent and eficienT vIrtual maChine opErations Radu Prodan, University of Innsbruck . July 13, 2016, 12:30, room 045C ---- [[AbstractJosianeZerubia|Marked Point Processes for Object Detection in High Resolution Images: Applications to Earth Observation and Cartography]] [[attachment:JosianeZerubia_Slides - 1 July 2016.pdf]] [[BioJosianeZerubia|Josiane Zerubia]], INRIA Sophia Antipolis Méditerranée, France . July 1, 2016, 10:30, room A01 ''This talk is given with the financial support of [[http://www.signalprocessingsociety.org/newsletter/2015/11/sps-announces-2016-class-of-distinguished-lecturers/|IEEE Signal Processing Society Distinguished Lecturer program for 2016-17]]'' ---- [[AbstractAlexCabuz|Make something people want]] ''Alexandru Cabuz'', [[http://researchforindustry.ro|Research for Industry]], Romania . April 13, 2016, 18:00, room 045C ---- '''Predicting Solar Parks Energy Output using Numerical Weather Prediction Ensemble Systems''' ''Liviu Oana'', West University of Timisoara, Romania . April 6, 2016, 18:00, room 045C ---- '''Big Data Analysis: Problems and Solutions''' ''Adriana Dinis'', West University of Timisoara, Romania . March 23, 2016, 18:00, room 045C ---- '''Projection algorithms for convex feasibility problems''' ''Irina Artinescu'', West University of Timisoara, Romania . March 16, 2016, 18:00, room 045C ---- '''Anderson Localisation for matrix-valued operators''' ''Hakim Boumaza'', University Paris 13, France . March 4, 2016, 11:00, room 045C ---- [[AbstractRochelleTractenberg|Appreciating the role of measurement -and its difficulties- in the validity of scientific research claims]] [[BioRochelleTractenberg|Rochelle Tractenberg]], Georgetown University, USA . January 27, 2016, 12:00, room A01 |
The algorithm implemented in CUDA that approximates the solution using the ABM method was also adapted to use Diethelm's method. Having the same algorithm that runs on the same system, enables an accurate and practical comparison between the numerical methods. |
We present a novel parallel algorithm to solve fractional-order systems involving Caputo-types derivative. The numerical method used for finding the solution is Adams-Bashforth-Moulton (ABM) predictor-corrector scheme. Using MPI and OPENMP frameworks the algorithm was implemented to run on a BlueGene/P supercomputer. The same algorithm that ran on the BlueGene/P cluster was adapted to run on GPU and exploit the GPU's capabilities. A comparison between these approaches is presented by running numerical simulations of the same three-dimensional fractional-order systems.
The algorithm implemented in CUDA that approximates the solution using the ABM method was also adapted to use Diethelm's method. Having the same algorithm that runs on the same system, enables an accurate and practical comparison between the numerical methods.