File: misc.rst

package info (click to toggle)
neuron 8.2.6-2
  • links: PTS, VCS
  • area: main
  • in suites: forky
  • size: 34,760 kB
  • sloc: cpp: 149,571; python: 58,465; ansic: 50,329; sh: 3,510; xml: 213; pascal: 51; makefile: 35; sed: 5
file content (47 lines) | stat: -rwxr-xr-x 1,181 bytes parent folder | download | duplicates (4)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
Miscellaneous
-------------

.. data:: float_epsilon


    Syntax:
        ``h.float_epsilon = 1e-11``


    Description:
        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. 
         

        .. code::

            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.