Parallel and distributed algorithms in Numerical analysis

Algoritmi numerične analize so široko uporabni v znanosti. Predvsem reševanje velikih sistemov enačb je stalnica znanstvenih simulacij na mnogih področjih, predvsem na področju fizike, strojništva, meteorologije in astronomije. Taki sistemi enačb so pogosto tako veliki, da jih zgolj z enim računalnikom ne moremo rešiti. V diplomskem delu smo preučili in opisali numerične algoritme za računanje določenih integralov, množenje redkih in gostih matrik ter reševanje linearnih sistemov na distribuiranih arhitekturah. Še več, izvedli smo tudi prototip oblačne storitve za reševanje teh problemov s pomočjo aktorskega modela izračunavanja in ogrodja Akka.The algorithms studied in numerical analysis are widely used in science and engineering. They are especially common with large scientific simulations in many fields, ranging from astronomy and meteorology to physics and mechanical engineering. Systems of equations are commonly so big we are unable to solve them using just a single computer. In this thesis we have examined the algorithms for numerical integration, both sparse and dense matrix multiplication and solving of systems of linear equations on a distributed memory machine. Moreover, we have also engineered a prototype of a cloud service for solving these kinds of problems using the actor model of concurrency