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 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286
|
/*
* volume.c
*
* This file contains the function
*
* double volume(Triangulation *manifold, int *precision);
*
* which the kernel provides for the UI. It computes and returns
* the volume of the manifold. If the pointer "precision" is not NULL,
* volume() estimates the number of decimal places of accuracy, and
* places the result in the variable *precision. The error estimate is
* the difference in the computed volumes at the last and next-to-the-last
* iterations of Newton's method (cf. hyperbolic_structures.c).
*
* 94/11/30 JRW This file now contains the function
*
* double birectangular_tetrahedron_volume( O31Vector a,
* O31Vector b,
* O31Vector c,
* O31Vector d);
*
* as well, for use within the kernel.
*/
#include "kernel.h"
static double Lobachevsky(double theta);
double volume(
Triangulation *manifold,
int *precision)
{
int i,
j;
double vol[2]; /* vol[ultimate/penultimate] */
Tetrahedron *tet;
for (i = 0; i < 2; i++) /* i = ultimate, penultimate */
vol[i] = 0.0;
for (tet = manifold->tet_list_begin.next;
tet != &manifold->tet_list_end;
tet = tet->next)
if (tet->shape[filled] != NULL)
for (i = 0; i < 2; i++) /* i = ultimate, penultimate */
for (j = 0; j < 3; j++)
vol[i] += Lobachevsky(tet->shape[filled]->cwl[i][j].log.imag);
if (precision != NULL)
*precision = decimal_places_of_accuracy(vol[ultimate], vol[penultimate]);
return vol[ultimate];
}
/*
* The Lobachevsky() function is based on the formula in
* Milnor's article "Hyperbolic geometry: The first 150 years",
* Bulletin of the American Mathematical Society, volume 6, number 1,
* January 1982, pp. 9-24. The actual formula appears about 2/3 of
* the way down page 18.
*/
static double Lobachevsky(double theta)
{
double term,
sum,
product,
theta_over_pi_squared;
const double *lobcoefptr;
const static double lobcoef[30] = {
5.4831135561607547882413838888201e-1,
1.0823232337111381915160036965412e-1,
4.8444907713545197129262758561472e-2,
2.7891037672165120538296812180796e-2,
1.8199901365960328824311744707278e-2,
1.2823667776324462157674846128817e-2,
9.5243928393815114745643670962394e-3,
7.3530535460250636167039159668208e-3,
5.8479755397266959054962366178048e-3,
4.7619093045811136799814912001842e-3,
3.9525701124525799515137944808229e-3,
3.3333335320272968375315987081340e-3,
2.8490028914574211634331659107080e-3,
2.4630541963678177950454607261557e-3,
2.1505376364114568439132649094016e-3,
1.8939393943803620897114593734851e-3,
1.6806722690053911275277764855335e-3,
1.5015015015233512340706336099638e-3,
1.3495276653220485553945730785293e-3,
1.2195121951230603594927151084491e-3,
1.1074197120711266596728487987281e-3,
1.0101010101010675186059356321779e-3,
9.2506938020352840967144128733280e-4,
8.5034013605442478972252664720549e-4,
7.8431372549019677504189889653457e-4,
7.2568940493468811469129541350086e-4,
6.7340067340067343805464730535677e-4,
6.2656641604010025932192218654462e-4,
5.8445353594389246257711686275038e-4,
5.4644808743169398954500641421420e-4};
/*
* As long as DBL_EPSILON > 5e-22 there will be enough lobcoefs
* for the series. If DBL_EPSILON is smaller than this you'll
* need to add more coefficients to the list.
*/
#if (DBL_DIG > 19)
You need to check DBL_EPSILON.
If it is less than 5e-22 you will need to provide more lobcoefs.
#endif
/*
* Milnor (Lemma 1, p. 17) shows that the Lobachevsky function is
* periodic with period pi. Put theta in the range [-pi/2, +pi/2].
*/
while (theta > PI_OVER_2)
theta -= PI;
while (theta < -PI_OVER_2)
theta += PI;
/*
* Milnor (Lemma 1, p. 17) also shows that the Lobachevsky function
* is an odd function, so we can further restrict theta to the
* range [0, pi/2].
*/
if (theta < 0.0)
return -Lobachevsky(-theta);
/*
* Handle theta == 0.0 specially, to avoid encountering
* log(0.0) later on.
*/
if (theta == 0.0)
return 0.0;
theta_over_pi_squared = (theta/PI)*(theta/PI);
sum = 0.0;
product = 1.0;
lobcoefptr = lobcoef;
do
{
product *= theta_over_pi_squared;
term = *lobcoefptr++ * product;
sum += term;
}
while (term > DBL_EPSILON);
return theta*(1.0 - log(2*theta) + sum);
}
double birectangular_tetrahedron_volume(
O31Vector a,
O31Vector b,
O31Vector c,
O31Vector d)
{
/*
* Compute the volume of the birectangular tetrahedron with vertices
* a, b, c and d, using the method found in section 4.3 of
*
* E. B. Vinberg, Ob'emy neevklidovykh mnogogrannikov,
* Uspekhi Matematicheskix Nauk, May(?) 1993, 17-46.
*
* Our a, b, c and d correspond to Vinberg's A, B, C and D, as shown
* in his Figure 9. We need to compute the dual basis {aa, bb, cc, dd}
* defined by <aa, a> = 1, <aa, b> = <aa, c> = <aa, d> = 0, etc.
* Let m be the matrix whose rows are the vectors {a, b, c, d}, but
* with the entries in the first column negated to account for the
* indefinite inner product, and let mm be the matrix whose columns
* are the vectors {aa, bb, cc, dd}. Then (m)(mm) = identity, by the
* definition of the dual basis.
*/
GL4RMatrix m,
mm;
O31Vector aa,
bb,
cc,
dd;
double alpha,
beta,
gamma,
delta,
big_delta,
tetrahedron_volume;
int i;
/*
* Set up the matrix m.
*/
for (i = 0; i < 4; i++)
{
m[0][i] = a[i];
m[1][i] = b[i];
m[2][i] = c[i];
m[3][i] = d[i];
}
for (i = 0; i < 4; i++)
m[i][0] = - m[i][0];
/*
* The matrix mm will be the inverse of m, as explained above.
* When m is singular, the birectangular tetrahedron's volume is zero.
*/
if (gl4R_invert(m, mm) != func_OK)
return 0.0;
/*
* Read the dual basis {aa, bb, cc, dd} from mm.
*/
for (i = 0; i < 4; i++)
{
aa[i] = mm[i][0];
bb[i] = mm[i][1];
cc[i] = mm[i][2];
dd[i] = mm[i][3];
}
/*
* Any pair of dual vectors lies in a positive definite 2-plane
* in E^(3,1). Normalize them to have length one, so we can use
* their dot products to compute the dihedral angles.
*/
o31_constant_times_vector(
1.0 / safe_sqrt ( o31_inner_product(aa,aa) ),
aa,
aa);
o31_constant_times_vector(
1.0 / safe_sqrt ( o31_inner_product(bb,bb) ),
bb,
bb);
o31_constant_times_vector(
1.0 / safe_sqrt ( o31_inner_product(cc,cc) ),
cc,
cc);
o31_constant_times_vector(
1.0 / safe_sqrt ( o31_inner_product(dd,dd) ),
dd,
dd);
/*
* Compute the angles alpha, beta and gamma, as shown in
* Vinberg's Figure 9.
*/
alpha = PI - safe_acos(o31_inner_product(aa, bb));
beta = PI - safe_acos(o31_inner_product(bb, cc));
gamma = PI - safe_acos(o31_inner_product(cc, dd));
/*
* Compute big_delta and delta, as in
* Vinberg's Sections 4.2 and 4.3.
*/
big_delta = sin(alpha) * sin(alpha) * sin(gamma) * sin(gamma)
- cos(beta) * cos(beta);
if (big_delta >= 0.0)
uFatalError("birectangular_tetrahedron_volume", "volume");
delta = atan( safe_sqrt( - big_delta) /
(cos(alpha) * cos(gamma)) );
tetrahedron_volume = 0.25 * (
Lobachevsky(alpha + delta)
- Lobachevsky(alpha - delta)
+ Lobachevsky(gamma + delta)
- Lobachevsky(gamma - delta)
- Lobachevsky(PI_OVER_2 - beta + delta)
+ Lobachevsky(PI_OVER_2 - beta - delta)
+ 2.0 * Lobachevsky(PI_OVER_2 - delta)
);
return tetrahedron_volume;
}
|