Please ignore all small commits stemming from inefficient workflow. The most recent commit (12/20/2017) is the important one.
Added numba just in time (JIT) compilation to slow four lines of numpy indexing. Re-wrote these lines in loop form so the JIT compiler could optimize them. This section is tested to be 2.8-2.9x faster on Haswell and KNL.
Please ignore all small commits stemming from inefficient workflow. The most recent commit (12/20/2017) is the important one.
Added numba just in time (JIT) compilation to slow four lines of numpy indexing. Re-wrote these lines in loop form so the JIT compiler could optimize them. This section is tested to be 2.8-2.9x faster on Haswell and KNL.