35#define MUL(dst, ap, an, bp, bn) \
37 lmmp_mul_(dst, ap, an, bp, bn); \
39 lmmp_mul_(dst, bp, bn, ap, an)
42#define mul_1(dst, rn, v) \
44 mp_limb_t _c_ = lmmp_mul_1_(dst, dst, rn, v); \
62 const double ln2 = 0.69314718055994531;
67 rn =
k + (
mp_size_t)ceil(lgamma((
double)
k + 1) / ln2);
72 rn = (
mp_size_t)ceil(lgamma((
double)
n + 1) / ln2);
73 rn -=
k + (
mp_size_t)floor(lgamma((
double)
k + 1) / ln2);
98 for (
ushort i = 0; i < nfactors; i++) {
102#define MAX_T 0xffffffffffff
103 for (
uint e = 0; e < j; e++) {
122 for (
ushort i = 0; i < nfactors; ++i) {
126 fac[new_nfactors].f = f;
127 fac[new_nfactors++].j = j >> 1;
131#define MAX_T 0xffffffffffff
146 if (new_nfactors > 0) {
158 tn -= tp2[tn - 1] == 0;
159 MUL(dst, tp2, tn, mp, mpn);
161 rn -= dst[rn - 1] == 0;
168 rn -= dst[rn - 1] == 0;
200 fac[nfactors++].
j = e;
219 for (
ushort i = 1; i < primen; ++i) {
247 while (cache.
is_end == 0) {
249 for (
uint i = 0; i < cache.
size; i++) {
263 static const mp_limb_t odd_2factorial_table[17] = {1, 3, 15, 105, 945, 10395, 135135,
264 2027025, 34459425, 654729075,
265 13749310575, 316234143225,
266 7905853580625, 213458046676875,
267 6190283353629375, 191898783962510625,
268 6332659870762850625};
284 dst[shw] = odd_2factorial_table[
k];
293 dst[shw + rn] =
lmmp_shl_(dst + shw, dst + shw, rn, bits);
295 rn -= dst[rn - 1] == 0;
307 v += pi * npi * (npi + 1) / 2;
320 v += npi * N - pi * npi * (npi + 1) / 2;
328 if (
n == 0 ||
n == 1) {
333 const double logA = 0.24875447703378426;
334 const double log2 = 0.69314718055994531;
335 double n_sqr = (
uint)
n *
n;
336 double log_n = log(
n);
337 double r = 0.5 * n_sqr * log_n - 0.25 * n_sqr + 0.5 * log_n * (double)
n + 1.0 / 12.0 * log_n + logA;
345 if (
n == 0 ||
n == 1) {
351 const double zeta_diff_neg1 = -0.16542114370045093;
352 const double log2 = 0.69314718055994531;
353 const double log_2pi = 1.83787706640934548;
354 double z = (double)
n + 1;
355 double z_sqr = z * z;
356 double log_z = log(z);
357 double r = 0.5 * z_sqr * log_z - 0.75 * z_sqr + 0.5 * z * log_2pi - 1.0 / 12.0 * log_z + zeta_diff_neg1;
374 static const mp_limb_t odd_hyperfac_table[8] = {1, 1, 27, 27, 84375, 61509375, 50655615215625, 50655615215625};
375 dst[0] = odd_hyperfac_table[
n - 1];
384 for (
ushort i = 1; i < primen; ++i) {
388 fac[nfactors++].j = j;
396 dst[shw + rn] =
lmmp_shl_(dst + shw, dst + shw, rn, bits);
398 rn -= dst[rn - 1] == 0;
415 static const mp_limb_t odd_superfac_table[8] = {1, 1, 3, 9, 135, 6075, 1913625, 602791875};
416 dst[0] = odd_superfac_table[
n - 1];
425 for (
ushort i = 1; i < primen; ++i) {
429 fac[nfactors++].j = j;
437 dst[shw + rn] =
lmmp_shl_(dst + shw, dst + shw, rn, bits);
439 rn -= dst[rn - 1] == 0;
449 const double log2 = 0.69314718055994531;
450 double l = 1.000028 * (double)
n;
468 for (
ushort i = 0; i < primen; i++) {
470#define MAX_T 0xffffffffffff
503 while (cache.
is_end == 0) {
505 for (
uint i = 0; i < cache.
size; i++) {
mp_size_t lmmp_elem_mul_ulong_(mp_ptr restrict dst, const ulongp restrict limbs, mp_size_t n, mp_ptr restrict tp)
mp_size_t lmmp_factors_mul_ushort_(mp_ptr dst, mp_size_t rn, fac_ptr fac, ushort nfactors)
计算因子的累乘,并将结果放入dst中
mp_size_t lmmp_factors_mul_(mp_ptr dst, mp_size_t rn, fac_ptr fac, uint nfactors)
计算因子的累乘,并将结果放入dst中
#define MUL(dst, ap, an, bp, bn)
Copyright (C) 2026 HJimmyK (Jericho Knox)
#define mul_1(dst, rn, v)
#define lmmp_limb_popcnt_
#define lmmp_copy(dst, src, n)
#define lmmp_zero(dst, n)
#define lmmp_debug_assert(x)
#define lmmp_param_assert(x)
mp_limb_t lmmp_shl_(mp_ptr dst, mp_srcptr numa, mp_size_t na, mp_size_t shl)
左移操作 [dst,na] = [numa,na]<<shl,dst的低shl位填充0
#define _udiv32by32_q_preinv(q, n0, dinv)
#define FACTORS_MUL_N_THRESHOLD
mp_size_t lmmp_odd_factorial_uint_(mp_ptr dst, mp_size_t rn, uint n)
计算 n!
mp_size_t lmmp_odd_nPr_ushort_(mp_ptr dst, mp_size_t rn, ulong n, ulong r)
计算 nPr 排列数的奇数部分
mp_size_t lmmp_superfac_size_(ushort n, mp_bitcnt_t *bits)
计算super阶乘的 limb 缓冲区长度
mp_size_t lmmp_2factorial_(mp_ptr dst, mp_bitcnt_t bits, mp_size_t rn, uint n)
计算 n!
mp_size_t lmmp_primefac_size_(uint n)
计算质数阶乘的 limb 缓冲区长度
mp_size_t lmmp_hyperfac_(mp_ptr dst, mp_bitcnt_t bits, mp_size_t rn, ushort n)
计算hyper阶乘(k^k累乘至n)
mp_size_t lmmp_2factorial_size_(uint n, mp_bitcnt_t *bits)
计算 n!
mp_size_t lmmp_hyperfac_size_(ushort n, mp_bitcnt_t *bits)
计算hyper阶乘的 limb 缓冲区长度
mp_size_t lmmp_primefac_(mp_ptr dst, mp_size_t rn, uint n)
计算质数阶乘(不超过n的质数累乘)
mp_size_t lmmp_superfac_(mp_ptr dst, mp_bitcnt_t bits, mp_size_t rn, ushort n)
计算super阶乘(k!
void lmmp_prime_cache_free_(prime_cache_t *cache)
释放素数表缓存
void lmmp_prime_cache_next_(prime_cache_t *cache)
素数表缓存更新(从小到大遍历全局质数表)
static ulong lmmp_prime_size_(ulong n)
估计 n 范围内的素数数量
const ushort prime_short_table[6542]
Copyright (C) 2026 HJimmyK (Jericho Knox)
void lmmp_prime_int_table_init_(uint n)
初始化全局素数表
ushort lmmp_prime_cnt16_(ushort n)
计算小于等于 n 的素数数量
void lmmp_prime_cache_init_(prime_cache_t *cache, uint n)
初始化素数表缓存
#define restrict
Copyright (C) 2026 HJimmyK (Jericho Knox)
#define SALLOC_TYPE(n, type)
#define TALLOC_TYPE(n, type)
#define BALLOC_TYPE(n, type)