Note that this is a work-in-progress. When this is finished, this will allow researchers to perform the fastLSU program in parallel on individual machines. See the paper in this repository for more information on the algorithm itself.
The main program begins with FastBHConcurrent.java - from there a ResourceMonitor monitors the number of p-values in each instance of the WorkerThread. The threads then return PValues, which hold the final result.
To run the program, simply double click the executable JAR file and follow the instructions. There are windows binaries as well which should allow for more memory usage (OS X wrapper to come).
--Simplified WorkerThread to prevent parallel disk reads on the same file (this caused too many seeks on a given hard-drive) --Added small performance tweaks --Added case when results do not fit into memory --Fixed ResourceMonitor bug that prevented it from being cleared after each iteration