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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980
|
/*
* chern_simons.c
*
* The computation of the Chern-Simons invariant is a little
* delicate because the formula depends on a constant which
* must initially be supplied by the user.
*
* For the UI, this file provides the functions
*
* void set_CS_value( Triangulation *manifold,
* double a_value);
* void get_CS_value( Triangulation *manifold,
* Boolean *value_is_known,
* double *the_value,
* int *the_precision,
* Boolean *requires_initialization);
*
* The UI calls set_CS_value() to pass to the kernel a user-supplied
* value of the Chern-Simons invariant for the current manifold.
*
* The UI calls get_CS_value() to request the current value. If the
* current value is known (or can be computed), get_CS_value() sets
* *value_is_known to TRUE and writes the current value and its precision
* (the number of significant digits to the right of the decimal point)
* to *the_value and *the_precision, respectively. If the current value
* is not known and cannot be computed, it sets *value_is_known to FALSE,
* and then sets *requires_initialization to TRUE if the_value
* is unknown because no fudge factor is available, or
* to FALSE if the_value is unknown because the solution contains
* negatively oriented Tetrahedra. The UI might want to convey
* these situations to the user in different ways.
*
* get_CS_value() normalizes *the_value to the range (-1/4,+1/4].
* This is the ONLY point in code where such an adjustment is made;
* all internal computations are done mod 0.
*
*
* The kernel manages the Chern-Simons computation by keeping track of
* both the current value and the arbitrary constant ("fudge factor")
* which appears in the formula. It uses the following fields of
* the Triangulation data structure:
*
* Boolean CS_value_is_known,
* CS_fudge_is_known;
* double CS_value[2],
* CS_fudge[2];
*
* The Boolean flags indicate whether the corresponding double is
* presently known or unknown. To provide an estimate of precision,
* CS_value[ultimate] and CS_value[penultimate] store the value of the
* Chern-Simons invariant computed relative to the hyperbolic structure
* at the ultimate and penultimate iterations of Newton's method, and
* similarly for the fudge factor CS_fudge[].
*
* For the kernel, this file provides the functions
*
* void compute_CS_value_from_fudge(Triangulation *manifold);
* void compute_CS_fudge_from_value(Triangulation *manifold);
*
* compute_CS_value_from_fudge() computes the CS_value in terms of
* CS_fudge, if CS_fudge_is_known is TRUE. (If CS_fudge_is_known is FALSE,
* it sets CS_value_is_known to FALSE as well.) The kernel calls this
* function when doing Dehn fillings on a fixed Triangulation, where
* the CS_fudge will be known (and constant) but the CS_value will be
* changing.
*
* compute_CS_fudge_from_value() computes the CS_fudge in terms of
* CS_value, if CS_value_is_known is TRUE. (If CS_value_is_known is FALSE,
* it sets CS_fudge_is_known to FALSE as well.) The kernel calls this
* function when it changes a Triangulation without changing the manifold
* it represents.
*
*
* Bob Meyerhoff, Craig Hodgson and Walter Neumann have found at least
* two different algorithms for computing the Chern-Simons invariant.
* The following code allows easy substitution of algorithms, in the
* function compute_CS().
*
* 96/4/16 David Eppstein pointed out that when he does (1,0) Dehn filling
* on m074(1,0), SnapPea quits with the message "The argument in the
* dilogarithm function is too large to guarantee accuracy". I've modified
* the code so that it displays the message "The argument in the dilogarithm
* function is too large to guarantee an accurate value for the Chern-Simons
* invariant" but does not quit. Instead it sets
*
* manifold->CS_value_is_known = FALSE;
* or
* manifold->CS_fudge_is_known = FALSE;
*
* (as appropriate) and continues normally. [By the way, I rejected the
* idea of providing more coefficients for the series. The set of manifolds
* for which the existing coefficients do not suffice is very, very small:
* no problems arise for any of the manifolds in the cusped or closed censuses.
* (Eppstein's example of m074(1,0) is a 3-sphere, but other descriptions
* of the 3-sphere seem to work fine.) So I don't want to slow down the
* computation of the Chern-Simons invariant in the generic case for the
* sake of an almost vanishingly small set of exceptions.]
*/
#include "kernel.h"
#define CS_EPSILON 1e-8
#define LOG_TWO_PI 1.83787706640934548356
static FuncResult compute_CS(Triangulation *manifold, double value[2]);
static FuncResult algorithm_one(Triangulation *manifold, double value[2]);
static Complex alg1_compute_Fu(Triangulation *manifold, int which_approximation, Boolean *Li2_error_flag);
static Complex Li2(Complex w, ShapeInversion *z_history, Boolean *Li2_error_flag);
static Complex log_w_minus_k_with_history(Complex w, int k,
double regular_arg, ShapeInversion *z_history);
static int get_history_length(ShapeInversion *z_history);
static int get_wide_angle(ShapeInversion *z_history, int requested_index);
void set_CS_value(
Triangulation *manifold,
double a_value)
{
manifold->CS_value_is_known = TRUE;
manifold->CS_value[ultimate] = a_value;
manifold->CS_value[penultimate] = a_value;
compute_CS_fudge_from_value(manifold);
}
void get_CS_value(
Triangulation *manifold,
Boolean *value_is_known,
double *the_value,
int *the_precision,
Boolean *requires_initialization)
{
if (manifold->CS_value_is_known)
{
*value_is_known = TRUE;
*the_value = manifold->CS_value[ultimate];
*the_precision = decimal_places_of_accuracy(
manifold->CS_value[ultimate],
manifold->CS_value[penultimate]);
*requires_initialization = FALSE;
/*
* Normalize reported value to the range (-1/4, 1/4].
*/
while (*the_value < -0.25 + CS_EPSILON)
*the_value += 0.5;
while (*the_value > 0.25 + CS_EPSILON)
*the_value -= 0.5;
}
else
{
*value_is_known = FALSE;
*the_value = 0.0;
*the_precision = 0;
*requires_initialization = (manifold->CS_fudge_is_known == FALSE);
}
}
void compute_CS_value_from_fudge(
Triangulation *manifold)
{
double computed_value[2];
if (manifold->CS_fudge_is_known == TRUE
&& compute_CS(manifold, computed_value) == func_OK)
{
manifold->CS_value_is_known = TRUE;
manifold->CS_value[ultimate] = computed_value[ultimate] + manifold->CS_fudge[ultimate];
manifold->CS_value[penultimate] = computed_value[penultimate] + manifold->CS_fudge[penultimate];
}
else
{
manifold->CS_value_is_known = FALSE;
manifold->CS_value[ultimate] = 0.0;
manifold->CS_value[penultimate] = 0.0;
}
}
void compute_CS_fudge_from_value(
Triangulation *manifold)
{
double computed_value[2];
if (manifold->CS_value_is_known == TRUE
&& compute_CS(manifold, computed_value) == func_OK)
{
manifold->CS_fudge_is_known = TRUE;
manifold->CS_fudge[ultimate] = manifold->CS_value[ultimate] - computed_value[ultimate];
manifold->CS_fudge[penultimate] = manifold->CS_value[penultimate] - computed_value[penultimate];
}
else
{
manifold->CS_fudge_is_known = FALSE;
manifold->CS_fudge[ultimate] = 0.0;
manifold->CS_fudge[penultimate] = 0.0;
}
}
static FuncResult compute_CS(
Triangulation *manifold,
double value[2])
{
Cusp *cusp;
/*
* We can handle only orientable manifolds.
*/
if (manifold->orientability != oriented_manifold)
return func_failed;
/*
* Cusps must either be complete, or have Dehn filling
* coefficients which are relatively prime integers.
*/
for (cusp = manifold->cusp_list_begin.next;
cusp != &manifold->cusp_list_end;
cusp = cusp->next)
if (Dehn_coefficients_are_relatively_prime_integers(cusp) == FALSE)
return func_failed;
/*
* Here we plug in the algorithm of our choice.
*/
return algorithm_one(manifold, value);
}
static FuncResult algorithm_one(
Triangulation *manifold,
double value[2])
{
Boolean Li2_error_flag;
int i;
Complex Fu[2],
core_length_sum[2],
complex_volume[2],
length[2];
int singularity_index;
Cusp *cusp;
/*
* This algorithm is taken directly from Craig Hodgson's
* preprint "Computation of the Chern-Simons invariants".
* It extends previous implementations in that it uses
* the shape_histories of the Tetrahedra to compute
* the dilogarithms, which allows solutions with negatively
* oriented Tetrahedra.
*/
/*
* To use the Chern-Simons formula, both the complete and filled
* solutions must be geometric, nongeometric or flat.
*/
for (i = 0; i < 2; i++) /* i = complete, filled */
if (manifold->solution_type[i] != geometric_solution
&& manifold->solution_type[i] != nongeometric_solution
&& manifold->solution_type[i] != flat_solution)
return func_failed;
/*
* Initialize the Li2_error_flag to FALSE.
* If the coefficients in Li2() don't suffice to compute the dilogaritm
* to full precision, Li2() will set Li2_error_flag to TRUE.
*/
Li2_error_flag = FALSE;
/*
* Compute F(u) relative to the ultimate and penultimate
* hyperbolic structures, to allow an estimatation of precision.
*/
for (i = 0; i < 2; i++) /* i = ultimate, penultimate */
Fu[i] = alg1_compute_Fu(manifold, i, &Li2_error_flag);
/*
* If Li2() failed, return func_failed;
*/
if (Li2_error_flag == TRUE)
{
uAcknowledge("An argument in the dilogarithm function is too large to guarantee an accurate value for the Chern-Simons invariant.");
return func_failed;
}
/*
* F(u) is
*
* (complex volume) + pi/2 (sum of complex core lengths)
*
* So we subtract off the complex lengths of the core geodesics
* to be obtain the complex volume.
*/
for (i = 0; i < 2; i++) /* i = ultimate, penultimate */
core_length_sum[i] = Zero;
for (cusp = manifold->cusp_list_begin.next;
cusp != &manifold->cusp_list_end;
cusp = cusp->next)
{
compute_core_geodesic(cusp, &singularity_index, length);
switch (singularity_index)
{
case 0:
/*
* The cusp is complete. Do nothing.
*/
break;
case 1:
/*
* Add this core length to the sum.
*/
for (i = 0; i < 2; i++) /* i = ultimate, penultimate */
core_length_sum[i] = complex_plus(
core_length_sum[i],
length[i]);
break;
default:
/*
* We should never arrive here.
*/
uFatalError("algorithm_one", "chern_simons");
}
}
/*
* (complex volume) = F(u) - (pi/2)(sum of core lengths)
*/
for (i = 0; i < 2; i++) /* i = ultimate, penultimate */
{
complex_volume[i] = complex_minus(
Fu[i],
complex_real_mult(
PI_OVER_2,
core_length_sum[i]
)
);
value[i] = complex_volume[i].imag / (2.0 * PI * PI);
}
return func_OK;
}
static Complex alg1_compute_Fu(
Triangulation *manifold,
int which_approximation, /* ultimate or penultimate */
Boolean *Li2_error_flag)
{
Complex Fu;
Tetrahedron *tet;
static const Complex minus_i = {0.0, -1.0};
/*
* We compute the function F(u), which Yoshida has proved holomorphic.
* (See Craig's preprint mentioned above.)
*/
/*
* Initialize F(u) to Zero.
*/
Fu = Zero;
/*
* Add up the log terms.
*/
for (tet = manifold->tet_list_begin.next;
tet != &manifold->tet_list_end;
tet = tet->next)
{
Fu = complex_minus(
Fu,
complex_mult(
tet->shape[ filled ]->cwl[which_approximation][0].log,
tet->shape[ filled ]->cwl[which_approximation][1].log
)
);
Fu = complex_plus(
Fu,
complex_mult(
tet->shape[ filled ]->cwl[which_approximation][0].log,
complex_conjugate(
tet->shape[complete]->cwl[which_approximation][1].log)
)
);
Fu = complex_minus(
Fu,
complex_mult(
tet->shape[ filled ]->cwl[which_approximation][1].log,
complex_conjugate(
tet->shape[complete]->cwl[which_approximation][0].log)
)
);
Fu = complex_minus(
Fu,
complex_mult(
tet->shape[complete]->cwl[which_approximation][0].log,
complex_conjugate(
tet->shape[complete]->cwl[which_approximation][1].log)
)
);
}
/*
* Multiply through by one half.
*/
Fu = complex_real_mult(0.5, Fu);
/*
* Add in the dilogarithms.
*/
for (tet = manifold->tet_list_begin.next;
tet != &manifold->tet_list_end;
tet = tet->next)
{
/*
* To compute the dilogarithm of z, Li2() wants to be
* passed w = log(z) / 2 pi i and the shape_history of z.
*/
Fu = complex_plus
(
Fu,
Li2
(
complex_div
(
tet->shape[filled]->cwl[which_approximation][0].log,
TwoPiI
),
tet->shape_history[filled],
Li2_error_flag
)
);
}
/*
* Multiply by -i.
*/
Fu = complex_mult(minus_i, Fu);
return Fu;
}
static Complex Li2(
Complex w,
ShapeInversion *z_history,
Boolean *Li2_error_flag)
{
/*
* Compute the dilogarithm of z = exp(2 pi i w) as explained
* in Craig's preprint mentioned above. Note that we use
* the variable w instead of the z which appears in Craig's
* preprint, to avoid confusion with the z which appears
* in the formula for F(u).
*
* The term Craig calls "S" we compute in two parts
*
* s0 = sum from i = 1 to infinity . . .
* s1 = sum from k = 1 to N . . . + Nw
*
* The remaining part of the formula we call
*
* t = pi^2/6 + 2 pi i w - 2 pi i w log(-2 pi i w) + (pi w)^2
*/
Complex s0,
s1,
s,
t,
w_squared,
two_pi_i_w,
kk,
k_plus_w,
k_minus_w,
result;
int i,
k;
static const Complex pi_squared_over_6 = {PI*PI/6.0, 0.0},
four_pi_i = {0.0, 4.0*PI},
minus_pi_i = {0.0, -PI},
log_minus_two_pi_i = {LOG_TWO_PI, -PI_OVER_2};
/*
* The array a[] contains the coefficients for the infinite series
* in s0. The constant num_terms tells how many we need to use to
* insure accuracy (see Analysis of Convergence below).
*
* The following Mathematica code computed these coefficients
* for N = 2. (I use "n" where Craig used "N" to conform both
* to proramming conventions regarding capital letters, and also
* to Mathematica's conventions.)
*
* a[i_, n_] :=
* N[(Zeta[2i] - Sum[k^(-2i), {k,1,n}]) / (2i(2i + 1)), 60]
* a2[i_] := a[i, 2]
* Array[a2, 30]
*
* Note that to get 20 significant digits in a2[30] = 6.4e-33, we
* must request at least 53 decimal places of accuracy from
* Mathematica, and probably a little more since the accuracy we
* request is the accuracy to which the intermediate calculations
* are truncated -- the final accuracy could be a little worse.
* By the way, we really do need a lot of that accuracy even in
* the tiny coefficients, because they will be multiplied by high
* powers of w, and |w| may be greater than one.
*/
static const int num_terms = 30;
static const int n = 2;
static const double a[] ={
0.0,
6.58223444747044060787e-2,
9.91161685556909575800e-4,
4.09062377249795170123e-5,
2.37647497144915803729e-6,
1.63751161982593974054e-7,
1.24738994105660169102e-8,
1.01418480335632980259e-9,
8.62880373230578403363e-11,
7.59064144690016509252e-12,
6.85041587014555123901e-13,
6.30901702974110744035e-14,
5.90712644809102073367e-15,
5.60732930747841393884e-16,
5.38501558411235458177e-17,
5.22344536523359867175e-18,
5.11092595568460128406e-19,
5.03912265560217431595e-20,
5.00200835767964640183e-21,
4.99518851712940000071e-22,
5.01545492014257760830e-23,
5.06048349504093155712e-24,
5.12862546072263579933e-25,
5.21876054821516289501e-26,
5.33019249317297967524e-27,
5.46257395282628942810e-28,
5.61585233364316675625e-29,
5.79023077981676178469e-30,
5.98614037451033538648e-31,
6.20422080422041301360e-32,
6.44530754870034591211e-33};
/*
* Analysis of convergence.
*
* The i-th coefficient in the (partial) zeta function is
*
* (N+1)^(-2i) + (N+2)^(-2i) + (N+3)^(-2i) + ...
*
* Lemma. For large i, this series may be approximated by its first
* term (N+1)^(-2i).
*
* Proof. [Probably not worth reading, but I figured I ought to
* include it.] Get an upper bound on the sum of the neglected
* terms by comparing them to an integral:
*
* (N+2)^(-2i) + (N+3)^(-2i) + ...
* < (N+2)^(-2i) + integral from x = N+2 to infinity of x^(-2i) dx
* = (N+2)^(-2i) + (N+2)^(-2i+1)/(2i-1)
* < (N+2)^(-2i) + (N+2)^(-2i)
* = 2 (N+2)^(-2i)
*
* Therefore the ratio (error)/(first term) is less than
*
* [2 (N+2)^(-2i)] / [(N+1)^(-2i)]
* = 2 ((N+1)/(N+2))^(2i)
*
* Thus, for example, if N = 2 and i >= 10, the ratio
* (error)/(first term) will be less than 2 (3/4)^10 = 1%.
* When N = 4 we need i >= 15 to obtain 1% accuracy.
* Q.E.D.
*
*
* The preceding lemma implies that the infinite series
* for S has the same convergence behavior as the series
*
* S' = (w/(N+1))^2i / i^2
*
* so we analyze S' instead of S. The error introduced
* by truncating the series after some i = i0 is bounded
* by the corresponding error in the geometric series
*
* S" = |w/(N+1)|^2i / i0^2
*
* The latter error is
*
* (first neglected term) / (1 - ratio)
*
* = (|w/(N+1)|^2i0 / i0^2) / (1 - |w/(N+1)|^2)
*
* = |w/(N+1)|^2i0 / (i0^2 (1 - |w/(N+1)|^2))
*
* A quick calculcation in Mathematica shows that if
* we are willing to calculate 30 terms in the series,
* then |w/(N+1)| < 0.5 implies the error will be
* less than 1e-20. In other words, the series can
* be used successfully for |w| < (N+1)/2. What values
* of z (i.e. what actual simplex shapes) does this
* correspond to?
*
* Letting w = x + iy, we get
*
* z = exp(2 pi i (x + i y))
* = exp(-2 pi y + 2 pi i x)
* = exp(-2 pi y) * (cos(2 pi x) + i sin(2 pi x))
*
* In other words, at an argument of 2 pi x, the acceptable
* parameters z are those with moduli between
* exp(-2 pi sqrt(((N+1)/2)^2 - x^2)) and
* exp(+2 pi sqrt(((N+1)/2)^2 - x^2)).
*
* For N = 2:
* When x = 0 we get values of z along the positive real axis
* from 0.00008 to 12000.
* When x = 1/2 we get values of z along the negative real axis
* from -0.0001 to -7000.
* When x = 1 we get values of z along the positive real axis
* from 0.0008 to 1000.
*
* This is good news: it means that the 30-term series for S will
* be accurate to 1e-20 for all (reasonable) nondegenerate values
* of z. I don't foresee the need for a greater radius of
* convergence, but if one is ever needed, just switch to N = 4.
*/
/*
* According to the preceding Analysis of Convergence, our
* computations will be accurate to 1e-20 whenever
* |w| < (N+1)/2 = 3/2.
*/
if (complex_modulus(w) > 1.5)
{
*Li2_error_flag = TRUE;
return Zero;
}
/*
* Note the values of w^2 and 2 pi i w.
*/
w_squared = complex_mult(w, w);
two_pi_i_w = complex_mult(TwoPiI, w);
/*
* Compute t.
*
* In the third term, - 2 pi i w will lie in the strip
* 0 < Im(- 2 pi i w) < - pi i, so we choose the argument
* in its log to be in the range (0, - pi).
*/
t = pi_squared_over_6;
t = complex_plus(
t,
two_pi_i_w
);
t = complex_minus(
t,
complex_mult(
two_pi_i_w,
complex_plus(
log_minus_two_pi_i,
log_w_minus_k_with_history(w, 0, 0.0, z_history)
)
)
);
t = complex_plus(
t,
complex_real_mult(PI * PI, w_squared)
);
/*
* Compute s0.
*
* Start with the high order terms and work backwards.
* It's a little faster, because fewer multiplications
* are required, and might also be a little more accurate.
*/
s0 = Zero;
for (i = num_terms; i > 0; --i)
{
s0.real += a[i];
s0 = complex_mult(s0, w_squared);
}
s0 = complex_mult(s0, w);
/*
* Compute s1.
*/
s1 = Zero;
for (k = 1; k <= n; k++)
{
kk = complex_real_mult(k, One);
k_plus_w = complex_plus (kk, w);
k_minus_w = complex_minus(kk, w);
s1 = complex_plus(
s1,
complex_real_mult(log(k), w)
);
s1 = complex_minus(
s1,
complex_real_mult(
0.5,
complex_mult(
k_plus_w,
log_w_minus_k_with_history(w, -k, 0.0, z_history)
)
)
);
s1 = complex_plus(
s1,
complex_real_mult(
0.5,
complex_mult(
k_minus_w,
/*
* We write Craig's log(k - w), which had an
* argument of 0 for the regular case, as
* log(-1) + log(w - k), and choose arg(log(-1)) = -pi
* and arg(log(w - k)) = +pi for the regular case.
*/
complex_plus(
minus_pi_i,
log_w_minus_k_with_history(w, k, PI, z_history)
)
)
)
);
}
s1 = complex_plus(
s1,
complex_real_mult(n, w)
);
/*
* Add t + (4 pi i)(s0 + s1) to get the final answer.
*/
s = complex_plus(s0, s1);
result = complex_plus(
t,
complex_mult(four_pi_i, s)
);
return result;
}
static Complex log_w_minus_k_with_history(
Complex w,
int k,
double regular_arg,
ShapeInversion *z_history)
{
int which_strip;
double estimated_argument;
int i;
/*
* This function computes log(w - k), taking into account the "history"
* of the shape z from which w is derived (z = exp(2 pi i w), as
* explained above). That is, it takes into account z's precise
* path through the parameter space, up to isotopy.
*
* regular_arg supplies the correct argument for the case of a
* regular ideal tetrahedron, with z = (1/2) + (sqrt(3)/2)i,
* w = 1/6, and a trivial "history". Typically regular_arg
* will be 0 for k <= 0, and +pi for k > 0.
*
* To understand what's going on here, it will be helpful to make
* yourself pictures of the z- and w-planes, as follows:
*
* z-plane. Draw axes for the complex plane representing z.
* Draw small circles at 0 and 1 to show where
* z is singular.
* Color the real axis blue from -infinity to 0, and label
* it '0' to indicate that z crosses this segment when
* there is a ShapeInversion with wide_angle == 0.
* Color the real axis red from 1 to +infinity, and label
* it '1' to indicate that z crosses this segment when
* there is a ShapeInversion with wide_angle == 1.
* Color the real axis green from 0 to 1, and label
* it '2' to indicate that z crosses this segment when
* there is a ShapeInversion with wide_angle == 2.
*
* w-plane. Draw the preimage of the z-plane picture under the
* map z = exp(2 pi i w).
* The singularities occur at the integer points on
* the real axis.
* Red half-lines labeled 1 extend from each singularity
* downward to infinity.
* Green half-lines labeled 2 extend from each singularity
* upward to infinity.
* Blue lines labelled 0 pass vertically through each
* half-integer point on the real axis.
*
* We will use the z_history to trace the path of w through the
* w-plane picture, keeping track of the argument of log(w - k)
* as we go. We begin with the shape of a regular ideal tetrahedron,
* namely z = (1/2) + (sqrt(3)/2) i, w = 1/6 + 0 i.
*
* It suffices to keep track of the approximate argument to the
* nearest multiple of pi, since the true argument will be within
* pi/2 of that estimate.
*
* The vertical strips in the w-plane (which are preimages of
* the halfplane z.imag > 0 and z.imag < 0 in the z-plane)
* are indexed by integers. Strip n is the strip extending
* from w.real = n/2 to w.real = (n+1)/2.
*/
/*
* We begin at w = 1/6, and set the estimated_argument to
* regular_arg (this will typically be 0 if k <= 0, or pi if k > 0,
* corresponding to Walter and Craig's choices for the case of
* positively oriented Tetrahedra).
*/
which_strip = 0;
estimated_argument = regular_arg;
/*
* Now we read off the z_history, adjusting which_strip
* and estimated_argument accordingly.
*
* Typically the z_history will be NULL, so nothing happens here.
*
* Technical note: this isn't the most efficient way to read
* a linked list backwards, but clarity is more important than
* efficiency here, but the z_histories are likely to be so short.
*/
for (i = 0; i < get_history_length(z_history); i++)
switch (get_wide_angle(z_history, i))
{
case 0:
/*
* If we're in an even numbered strip, move to the right.
* If we're in an odd numbered strip, move to the left.
* The estimated_argument does not change.
*/
if (which_strip % 2 == 0)
which_strip++;
else
which_strip--;
break;
case 1:
/*
* If we're in an even numbered strip, move to the left,
* and if we pass under the singularity at k,
* subtract pi from the estimated_argument.
* If we're in an odd numbered strip, move to the right,
* and if we pass under the singularity at k,
* add pi to the estimated_argument.
*/
if (which_strip % 2 == 0)
{
which_strip--;
if (which_strip == 2*k - 1)
estimated_argument -= PI;
}
else
{
which_strip++;
if (which_strip == 2*k)
estimated_argument += PI;
}
break;
case 2:
/*
* If we're in an even numbered strip, move to the left,
* and if we pass over the singularity at k,
* add pi to the estimated_argument.
* If we're in an odd numbered strip, move to the right,
* and if we pass over the singularity at k,
* subtract pi from the estimated_argument.
*/
if (which_strip % 2 == 0)
{
which_strip--;
if (which_strip == 2*k - 1)
estimated_argument += PI;
}
else
{
which_strip++;
if (which_strip == 2*k)
estimated_argument -= PI;
}
break;
default:
uFatalError("log_w_minus_k_with_history", "chern_simons");
}
/*
* Compute log(w - k) using the estimated_argument.
*/
return (
complex_log(
complex_minus(
w,
complex_real_mult((double)k, One)
),
estimated_argument
)
);
}
static int get_history_length(
ShapeInversion *z_history)
{
int length;
length = 0;
while (z_history != NULL)
{
length++;
z_history = z_history->next;
}
return length;
}
static int get_wide_angle(
ShapeInversion *z_history,
int requested_index)
{
while (--requested_index >= 0)
z_history = z_history->next;
return z_history->wide_angle;
}
|