Dear Spark Devs, I have written an experimental 1d laplace parallel Spark solver <http://myunsoo-dataworks.blogspot.kr/2015/06/solve-differential-equations-with-spark.html> , out of curiousity regarding this <http://apache-spark-user-list.1001560.n3.nabble.com/Solving-Systems-of-Linear-Equations-Using-Spark-td13674.html#a23354> thread. It still may have some unknown bugs and only works for very special cases but still, i think it shows it can be done with Spark.
When solving large scale differential equations, one may consider using MPI/PVM based parallel machines because of their raw power and efficiency. But then again I think Spark (in-memory) map-reduce can also be reasonably fast and easy to use, with some tweaks. - For example a tweak such as transferring data only between specific clusters will significantly improve both memory efficiency and speed. (partial broadcasting?) But still I do not know how many people may also be interested in this and how much contribution to Spark scene it can make. So if you are interested building Differential Equation Spark Solver, or can suggest where I can discuss this with potentially interested parties, or have any other suggestions / information, please let me know! Thank you. Best regards, Myunsoo -- View this message in context: http://apache-spark-developers-list.1001551.n3.nabble.com/Differential-Equation-Spark-Solver-tp13005.html Sent from the Apache Spark Developers List mailing list archive at Nabble.com. --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@spark.apache.org For additional commands, e-mail: dev-h...@spark.apache.org
