The underflow analysis closes with a message about the underflow threshold and two final tests. Paranoia looks for an exponent range that’s too narrow relative to the precision, and tries to compute a value way below the underflow threshold.
print("The Underflow threshold is {:0.17e}, below which"
.format(underflow_threshold))
print("calculation may suffer larger Relative error than ", end="")
print("merely roundoff.")
test_for_narrow_range()
milestone = 130 # ==============================
test_extreme_underflow(underflow_threshold)
The next sections discuss the utility functions that support the underflow exploration. Much of this logic is in-line in the Basic code, often very tersely so, but the program is more malleable and readable when discussed in natural blocks.