Hi! The *collection execution* runs the program simply as functions over Java collections. It is single threaded, always local, and does not use any Flink memory management, serialization, or so. It is designed to be very lightweight and is tailored towards very small problems.
The *cluster mode* is the regular Flink mode. It spawns a Flink cluster with one worker and multiple slots. It runs programs parallel, uses managed memory, and should behave pretty much like the regular Flink installation (with one worker and little memory). To debug your test, I would first see whether it is parallelism sensitive. The cluster mode uses parallelism 4 by default, the collection execution is single threaded (parallelism 1). You can force the parallelism to be always one by setting it on the execution environment. Stephan On Wed, May 13, 2015 at 12:44 AM, Yi ZHOU <zhouyi0...@hotmail.com> wrote: > Hello, > > Thanks Andra for the gaussian sequence generation. It is a little tricky, > i just leave this part for future work. > > I meet another problem in AffinityPropogation algorithm. I write a few > test code for it. > > https://github.com/joey001/flink/blob/ap_add/flink-staging/flink-gelly/src/test/java/org/apache/flink/graph/test/example/AffinityPropogationITCase.java > < > https://github.com/joey001/flink/blob/ap_add/flink-staging/flink-gelly/src/test/java/org/apache/flink/graph/test/example/AffinityPropogationITCase.java>It > passes the COLLECTION while failed when execution mode = CLUSTER. > I not very clear about the differences and the reason. > > Does anyone give me a clue? > > Thanks, > Best Regards. > > ZHOU Yi > > On 08/05/2015 23:17, Andra Lungu wrote: > >> Hi Yi, >> >> To my knowledge, there is no simple way to generate this kind of >> DataSet(i.e. there is no env.generateGaussianSequence()). >> However, if you look in flink-perf, Till used something like this there: >> >> https://github.com/project-flink/flink-perf/blob/master/flink-jobs/src/main/scala/com/github/projectflink/als/ALSDataGeneration.scala >> Maybe he can give you some tips. >> >> You can also call random.nextGaussian() in Java. >> >> http://docs.oracle.com/javase/7/docs/api/java/util/Random.html#nextGaussian%28%29 >> >> Not sure if this helps, but there is a paper on generating this kind of >> distribution. >> http://ifisc.uib-csic.es/raul/publications/P/P44_tc93.pdf >> >> Best of luck, >> Andra >> >> >> On Fri, May 8, 2015 at 9:45 PM, Yi ZHOU <zhouyi0...@hotmail.com> wrote: >> >> Hello, all >>> >>> when I tested AP algorithm, I had a little question : >>> how to generate a DataSet in gaussian distribution? Is there a >>> implemented funtion? >>> >>> Does any one has a solution? Thank you, >>> >>> ZHOU Yi >>> >>> >
