Ability to specify domains for variables in numeric sampling

[jsharkey13]

This was some sample code I wrote way back that sort of implemented that functionality, not sure how it worked now though:

##### Some test code to implement different limits on the sample space #####
test = numpy.random.random_sample((4, 10))#numpy.ones((4, 10))
lims = {"gamma": (0, 0), "b": (0, 0), "c": (0, 1), "d": (0, 0)}
up_limits = numpy.array([lims[e][1] for e in sorted(lims.keys())])
low_limits = numpy.array([lims[e][0] for e in sorted(lims.keys())])
vals = numpy.dot(numpy.diag(up_limits - low_limits), test)
print vals
vals += numpy.dot(numpy.diag(low_limits), numpy.ones((4, 10)))
tttt = sympy_parser.parse_expr("2*lamda+4*c+5*d+3*b", sympy.abc._clash)
print sorted(tttt.free_symbols, key=lambda x: str(x))
ffff = sympy.lambdify(sorted(tttt.free_symbols, key=lambda x: str(x)), tttt, "numpy")
gggg = ffff(*vals)
print gggg