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
|
/*
* canonize.h
*
* In canonize_part_1.c, canonize_part_2.c, and canonize_result.c, the sum
* of the tilts of two Tetrahedra relative to their common face is
* considered to be zero iff it lies between -CONCAVITY_EPSILON and
* +CONCAVITY_EPSILON.
*
* 95/10/20 cusp_neighborhoods.c now includes this file as well,
* so it can supress the drawing of faces which serve only to subdivide
* the canonical cell decomposition into tetrahedra.
*
* 97/3/1 The processing of the data for 16-crossing knots led
* to several examples which sometimes were reported to have
* canonical decompositions with cells other than tetrahedra,
* and sometimes not. This led me to suspect that CONCAVITY_EPSILON
* was too big, so I changed it from 1e-6 to 1e-7. (This problem
* hadn't arisen with snappea 1.3.x because the latter normalized
* cusp volumes to 1.0, whereas SnapPea 2.x normalizes them to
* (3/16)sqrt(3).) Eventually I may need to adopt a more sophisticated
* approach to using CONCAVITY_EPSILON, somehow taking into account
* the volume of the cusp.
*/
#define CONCAVITY_EPSILON 1e-7
|
