These are tools to experiment with representations and approaches to handling
rulesets.

Rules have support and then a bit vector (or GMP bignum) representing the
samples for which the rule applies.

A ruleset is a collection of rules and a captures vector/bignum associated
with each rule indicating which samples get captured by which rule.

makedata.py: Transform data sets into something easily read into a C program
	Assumes input files in the *.TAB and *.Y formats and produces a .out
	format of <rule, truthtable> tuples where entry i in the truthtable
	contains either an ascii 1 or ascii 0 indicating if the rule applies
	to the ith sample in the training data.

	Usage: (from within python)
	from makedata import *
	get_freqitemsets("basename")

	where basename would be something like adult2_train and the files
	adult2_train.TAB and adult2_train.Y exist. The function will output
	adult2_train.out


analyze.c:	Driver program that:
	1. Calls rules_init to read in the data produced by makedata
	2. Executes [-i iterations] (default 10) of:
		3. Create random ruleset of [-s size] (default 3)
		4. Performs size^2 adjacent swaps
		5. Performs size delete/add pairs

rulelib.c:	Library of routines for manipulating rules and rulesets.
	See rule.h for function prototypes exported.


Compile options:

This package compiles both with and without the GMP library.  Without it,
bit vector operations are coded manually as arrays of long longs. With -D GMP,
we store the vectors as bignums.