The default value is 1e-11
Allows somewhat safer NEURON logical comparisons and integer truncation for floating point numbers. Most NEURON comparisons are treated as true if they are within float_epsilon of being true. eg.
from neuron import h
h("""
proc count_to_1() {for (i = 0; i < 1; i += 0.1) $o1.__call__(i)}
""")
def print_i(i):
print('%3g %g' % (i, int(10*i)))
rv = h.count_to_1(print_i)
h.float_epsilon = 0
# two bugs due to roundoff
rv = h.count_to_1(print_i)
Warning
This has no effect on Python comparisons, which is why the example uses a HOC procedure.
Warning
I certainly haven't gotten every floating comparison in the program to use float_epsilon but I have most of them including all HOC interpreter logical operations, int, array indices, and Vector logic methods.