May be compiled and run by using hmake (please modify to reflect the location of the user's HPX build directory.):

       ./hmake hpx-net

To run, enter:

   	./hpx-net -tN

Where N refers to the number of processors to be used.

The current build accepts the following numerical inputs from standard input (in order):

    	- Number of neural network input columns (sensors)
	- Number of hidden rows
	- Number of hidden columns
	- Number of output-layer columns
	- Number of iterations
	- Boolean value: Should the execution be purely serial? (1-Y, 0-N)
	- Forward threshold: If parallel, how many rows should be executed before proceeding, purely serial?
	- Backprop threshold: If parallel, how many rows should be executed before proceeding, purely serial?

To avoid the hassle of entering these values, the "input" file has been provided and may be customized to one's liking.

   	 ./hpx-net -tN <input

There is a line at the start of neuron::run redefining int serial as "1" in all cases.

If this line is present, forward pass execution occurs with not-so-fine-grained parallelism (if otherwise specified to be parallel).

In this scenario, only the rows of the ANN are executed in parallel, while the neurons within the rows are executed serially.