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
|
/* mpn_rootrem(rootp,remp,ap,an,nth) -- Compute the nth root of {ap,an}, and
store the truncated integer part at rootp and the remainder at remp.
THE FUNCTIONS IN THIS FILE ARE INTERNAL FUNCTIONS WITH MUTABLE
INTERFACES. IT IS ONLY SAFE TO REACH THEM THROUGH DOCUMENTED INTERFACES.
IN FACT, IT IS ALMOST GUARANTEED THAT THEY'LL CHANGE OR DISAPPEAR IN A
FUTURE GNU MP RELEASE.
Copyright 2002 Free Software Foundation, Inc.
This file is part of the GNU MP Library.
The GNU MP Library is free software; you can redistribute it and/or modify
it under the terms of the GNU Lesser General Public License as published by
the Free Software Foundation; either version 2.1 of the License, or (at your
option) any later version.
The GNU MP Library is distributed in the hope that it will be useful, but
WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public
License for more details.
You should have received a copy of the GNU Lesser General Public License
along with the GNU MP Library; see the file COPYING.LIB. If not, write to
the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
MA 02111-1307, USA. */
/*
We use Newton's method to compute the root of a:
n
f(x) := x - a
n - 1
f'(x) := x n
n-1 n-1 n-1
x - a/x a/x - x a/x + (n-1)x
new x = x - f(x)/f'(x) = x - ---------- = x + --------- = --------------
n n n
*/
#include <stdio.h>
#include <stdlib.h>
#include "gmp.h"
#include "gmp-impl.h"
#include "longlong.h"
mp_size_t
mpn_rootrem (mp_ptr rootp, mp_ptr remp,
mp_srcptr up, mp_size_t un, mp_limb_t nth)
{
mp_ptr pp, qp, xp;
mp_size_t pn, xn, qn;
unsigned long int unb, xnb, bit;
unsigned int cnt;
mp_size_t i;
unsigned long int n_valid_bits, adj;
TMP_DECL (marker);
TMP_MARK (marker);
pp = TMP_ALLOC_LIMBS (un + 2);
count_leading_zeros (cnt, up[un - 1]);
unb = un * GMP_NUMB_BITS - cnt + GMP_NAIL_BITS;
xnb = (unb + nth - 1) / nth;
if (xnb == 1)
{
if (remp == NULL)
remp = pp;
mpn_sub_1 (remp, up, un, (mp_limb_t) 1);
MPN_NORMALIZE (remp, un);
rootp[0] = 1;
TMP_FREE (marker);
return un;
}
xn = (xnb + GMP_NUMB_BITS - 1) / GMP_NUMB_BITS;
qp = TMP_ALLOC_LIMBS (un); /* should try and reduce this allocation */
xp = TMP_ALLOC_LIMBS (xn + 1);
/* Set initial root to only ones. This is an overestimate of the actual root
by less than a factor of 2. */
for (i = 0; i < xn; i++)
xp[i] = GMP_NUMB_MAX;
xp[xnb / GMP_NUMB_BITS] = ((mp_limb_t) 1 << (xnb % GMP_NUMB_BITS)) - 1;
/* Improve the initial approximation, one bit at a time. Keep the
approximations >= root(U,nth). */
bit = xnb - 2;
n_valid_bits = 0;
for (i = 0; (nth >> i) != 0; i++)
{
mp_limb_t xl = xp[bit / GMP_NUMB_BITS];
xp[bit / GMP_NUMB_BITS] = xl ^ (mp_limb_t) 1 << bit % GMP_NUMB_BITS;
pn = mpn_pow_1 (pp, xp, xn, nth, qp);
/* If the new root approximation is too small, restore old value. */
if (! (un < pn || (un == pn && mpn_cmp (up, pp, pn) < 0)))
xp[bit / GMP_NUMB_BITS] = xl; /* restore old value */
n_valid_bits += 1;
if (bit == 0)
goto done;
bit--;
}
adj = n_valid_bits - 1;
/* Newton loop. Converges downwards towards root(U,nth). Currently we use
full precision from iteration 1. Clearly, we should use just n_valid_bits
of precision in each step, and thus save most of the computations. */
while (n_valid_bits <= xnb)
{
mp_limb_t cy;
pn = mpn_pow_1 (pp, xp, xn, nth - 1, qp);
mpn_tdiv_qr (qp, pp, (mp_size_t) 0, up, un, pp, pn); /* junk remainder */
qn = un - pn + 1;
cy = mpn_addmul_1 (qp, xp, xn, nth - 1);
if (qn == xn + 1)
{
cy += qp[xn];
if (cy == nth)
{
for (i = xn - 1; i >= 0; i--)
qp[i] = GMP_NUMB_MAX;
cy = nth - 1;
}
}
qp[xn] = cy;
qn = xn + (cy != 0);
mpn_divrem_1 (xp, (mp_size_t) 0, qp, qn, nth);
n_valid_bits = n_valid_bits * 2 - adj;
}
/* The computed result might be one unit too large. Adjust as necessary. */
done:
pn = mpn_pow_1 (pp, xp, xn, nth, qp);
if (un < pn || (un == pn && mpn_cmp (up, pp, pn) < 0))
{
mpn_decr_u (xp, 1);
pn = mpn_pow_1 (pp, xp, xn, nth, qp);
ASSERT_ALWAYS (! (un < pn || (un == pn && mpn_cmp (up, pp, pn) < 0)));
}
if (remp == NULL)
remp = pp;
mpn_sub (remp, up, un, pp, pn);
MPN_NORMALIZE (remp, un);
MPN_COPY (rootp, xp, xn);
TMP_FREE (marker);
return un;
}
|
