Hmmm… sounds like you are currently running a Hadoop map operation and running out of memory for the map. If you are running out of memory in Xcalar, please let me know. There are some cluster configuration tweaks we can make to impact this.
There is some possibility that the limitation is Python’s processing using this algorithm. My instinct is that you would achieve a higher level of parallelization and performance if you were to replicate the geographic proximity analysis algorithm using native Xcalar operations.
Did that answer your question?