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
|
/*=============================================================================
This file is part of FLINT.
FLINT is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 2 of the License, or
(at your option) any later version.
FLINT 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 General Public License for more details.
You should have received a copy of the GNU General Public License
along with FLINT; if not, write to the Free Software
Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
=============================================================================*/
/******************************************************************************
Copyright (C) 2008, 2009 William Hart
Copyright (C) 2010 Sebastian Pancratz
******************************************************************************/
#include <stdlib.h>
#include "fmpz_poly.h"
static void
__fmpz_poly_pseudo_divrem_divconquer(fmpz * Q, fmpz * R,
ulong * d, const fmpz * A, slong lenA,
const fmpz * B, slong lenB, const fmpz_preinvn_t inv)
{
if (lenB <= 16 || (lenA > 2 * lenB - 1 && lenA < 128))
{
_fmpz_poly_pseudo_divrem_basecase(Q, R, d, A, lenA, B, lenB, inv);
}
else
{
const slong n2 = lenB / 2;
const slong n1 = lenB - n2;
const fmpz * d1 = B + n2;
const fmpz * d2 = B;
const fmpz * d3 = B + n1;
const fmpz * d4 = B;
if (lenA <= lenB + n2 - 1)
{
fmpz *p1, *r1, *d2q1;
fmpz *f;
/*
Shift A right by n1, zero the bottom n2 - 1 coeffs; call this p1
*/
p1 = (fmpz *) flint_malloc((lenA - n1) * sizeof(fmpz));
{
slong i;
flint_mpn_zero((mp_ptr) p1, n2 - 1);
for (i = n2 - 1; i < lenA - n1; i++)
p1[i] = (A + n1)[i];
}
/*
Compute p1 div d3, at most a 2 n2 - 1 by n2 division, leaving
lenA - lenB + 1 <= n2 terms in the quotient
*/
r1 = R + n1;
_fmpz_poly_pseudo_divrem_divconquer(Q, r1, d, p1, lenA - n1, d3, n2, inv);
flint_free(p1);
/*
Push the relevant {n2 - 1} terms of the remainder to the
top of {R, lenA}
*/
{
slong i;
for (i = n2 - 2; i >= 0; i--)
fmpz_swap(R + lenA - (n2 - 1) + i, r1 + i);
r1 = R + lenA - (n2 - 1);
}
/*
Compute d2q1 = Q d4 of length lenA - n2, which is
at most n1 + n2 - 1 terms
*/
d2q1 = R;
_fmpz_poly_mul(d2q1, d4, n1, Q, lenA - lenB + 1);
/*
Compute R = L^d R', where R' is the terms of A we have not dealt,
of which there are at most n1 + n2 - 1; that is,
Set R to {A, n1 + n2 - 1} * f + r1 x^n1 - d2q1
*/
_fmpz_vec_neg(R, R, lenA - n2);
_fmpz_vec_add(R + n1, R + n1, R + lenA - n2 + 1, lenA - lenB);
_fmpz_vec_swap(R + lenA - n2, R + 2 * lenA - lenB + 1 - n2, n2 - (lenA - lenB + 1));
f = R + lenB - 1;
fmpz_pow_ui(f, B + (lenB - 1), *d);
_fmpz_vec_scalar_addmul_fmpz(R, A, n1 + n2 - 1, f);
}
else if (lenA > 2 * lenB - 1)
{
/*
XXX: In this case, we expect A to be modifiable
*/
ulong s1, s2;
const slong shift = lenA - 2 * lenB + 1;
fmpz * q1 = Q + shift;
fmpz * q2 = Q;
fmpz * r1 = R;
fmpz *p1, *t;
fmpz_t f;
fmpz_init(f);
/*
Shift A right until it is of length 2 lenB - 1, call this p1;
zero the bottom lenB - 1 coeffs
*/
p1 = (fmpz *) flint_malloc((2 * lenB - 1) * sizeof(fmpz));
{
slong i;
flint_mpn_zero((mp_ptr) p1, lenB - 1);
for (i = lenB - 1; i < 2*lenB - 1; i++)
p1[i] = (A + shift)[i];
}
/*
Set q1 to p1 div B, a 2 lenB - 1 by lenB division, so q1 ends up
being at most length lenB; r1 is of length at most lenB - 1
*/
_fmpz_poly_pseudo_divrem_divconquer(q1, r1, &s1, p1, 2 * lenB - 1, B, lenB, inv);
flint_free(p1);
/*
Compute t = L^s1 a2 + r1 x^shift, of length at most lenA - lenB
since r1 is of length at most lenB - 1. Here a2 is what remains
of A after the first lenR coefficients are removed
*/
t = (fmpz *) A;
fmpz_pow_ui(f, B + (lenB - 1), s1);
_fmpz_vec_scalar_mul_fmpz(t, A, lenA - lenB, f);
_fmpz_vec_add(t + shift, t + shift, r1, lenB - 1);
/*
Compute q2 = t div B; it is a smaller division than the original
since len(t) <= lenA - lenB, and r2 has length at most lenB - 1
*/
_fmpz_poly_pseudo_divrem_divconquer(q2, R, &s2, t, lenA - lenB, B, lenB, inv);
/*
Write out Q = L^s2 q1 x^shift + q2, of length at most
lenB + shift. Note q2 has length at most shift since it is at
most an lenA - lenB by lenB division; q1 cannot have length zero
since we are doing pseudo division
*/
fmpz_pow_ui(f, B + (lenB - 1), s2);
_fmpz_vec_scalar_mul_fmpz(q1, q1, lenB, f);
*d = s1 + s2;
fmpz_clear(f);
}
else /* n1 + 2 n2 - 1 < lenA <= 2 lenB - 1 */
{
fmpz * q1 = Q + n2;
fmpz * q2 = Q;
fmpz * r1 = R;
fmpz * d2q1 = R + (n1 - 1);
fmpz *p1, *t;
fmpz_t f;
ulong s1, s2;
fmpz_init(f);
/*
Set p1 to the top lenA - 2 n2 coeffs of A, clearing the bottom
n1 - 1 coeffs
*/
p1 = (fmpz *) flint_malloc((lenA - 2 * n2) * sizeof(fmpz));
{
slong i;
flint_mpn_zero((mp_ptr) p1, n1 - 1);
for (i = n1 - 1; i < lenA - 2 * n2; i++)
p1[i] = (A + 2 * n2)[i];
}
/*
Set q1 to p1 div d1, at most a 2 n1 - 1 by n1 division, so q1 ends
up being of length at most n1; r1 is of length n1 - 1
*/
_fmpz_poly_pseudo_divrem_divconquer(q1, r1, &s1, p1, lenA - 2 * n2, d1, n1, inv);
flint_free(p1);
/*
Compute d2q1 = d2q1, of length lenA - lenB
Note lenA - lenB <= lenB - 1 <= 2 n2 and lenA - (n1 - 1) > 2 n2,
so we can store d2q1 in the top 2 n2 coeffs of R
*/
if (n2 >= lenA - n1 - 2 * n2 + 1)
_fmpz_poly_mul(d2q1, d2, n2, q1, lenA - (n1 + 2 * n2 - 1));
else
_fmpz_poly_mul(d2q1, q1, lenA - (n1 + 2 * n2 - 1), d2, n2);
/*
Compute
t = L^s1 * (a2 x^{n1 + n2 - 1} + a3)
+ r1 x^{2 n2} - d2q1 x^n2
of length at most lenB + n2 - 1, since r1 is of length at most
n1 - 1 and d2q1 is of length at most n1 + n2 - 1
*/
t = _fmpz_vec_init(n1 + 2 * n2 - 1);
fmpz_pow_ui(f, B + (lenB - 1), s1);
_fmpz_vec_scalar_mul_fmpz(t, A, n1 + 2 * n2 - 1, f);
_fmpz_vec_add(t + 2 * n2, t + 2 * n2, r1, n1 - 1);
_fmpz_vec_sub(t + n2, t + n2, d2q1, lenA - lenB);
/*
Compute q2 = t div B and set R to the remainder, at most a
lenB + n2 - 1 by lenB division, so q2 is of length at most n2
*/
_fmpz_poly_pseudo_divrem_divconquer(q2, R, &s2, t, lenB + n2 - 1, B, lenB, inv);
_fmpz_vec_clear(t, n1 + 2 * n2 - 1);
/*
Write Q = L^s2 q1 x^n2 + q2; note len(q1) is non-zero since
we are performing pseudo division
*/
fmpz_pow_ui(f, B + (lenB - 1), s2);
_fmpz_vec_scalar_mul_fmpz(q1, q1, lenA - lenB + 1 - n2, f);
*d = s1 + s2;
fmpz_clear(f);
}
}
}
void
_fmpz_poly_pseudo_divrem_divconquer(fmpz * Q, fmpz * R,
ulong * d, const fmpz * A, slong lenA,
const fmpz * B, slong lenB, const fmpz_preinvn_t inv)
{
if (lenA <= 2 * lenB - 1)
{
__fmpz_poly_pseudo_divrem_divconquer(Q, R, d, A, lenA, B, lenB, inv);
}
else /* lenA > 2 * lenB - 1 */
{
fmpz *S = _fmpz_vec_init(lenA);
_fmpz_vec_set(S, A, lenA);
__fmpz_poly_pseudo_divrem_divconquer(Q, R, d, S, lenA, B, lenB, inv);
_fmpz_vec_clear(S, lenA);
}
}
void
fmpz_poly_pseudo_divrem_divconquer(fmpz_poly_t Q, fmpz_poly_t R,
ulong * d, const fmpz_poly_t A,
const fmpz_poly_t B)
{
slong lenq, lenr;
fmpz *q, *r;
if (B->length == 0)
{
flint_printf("Exception (fmpz_poly_pseudo_divrem_divconquer). Division by zero.\n");
abort();
}
if (Q == R)
{
flint_printf("Exception (fmpz_poly_pseudo_divrem_divconquer). \n"
"Output arguments Q and R may not be aliased.\n");
abort();
}
if (A->length < B->length)
{
fmpz_poly_zero(Q);
fmpz_poly_set(R, A);
*d = 0;
return;
}
lenq = A->length - B->length + 1;
lenr = A->length;
if (Q == A || Q == B)
q = _fmpz_vec_init(lenq);
else
{
fmpz_poly_fit_length(Q, lenq);
q = Q->coeffs;
}
if (R == A || R == B)
r = _fmpz_vec_init(lenr);
else
{
fmpz_poly_fit_length(R, lenr);
r = R->coeffs;
}
_fmpz_poly_pseudo_divrem_divconquer(q, r, d, A->coeffs, A->length,
B->coeffs, B->length, NULL);
lenr = B->length - 1;
FMPZ_VEC_NORM(r, lenr);
if (Q == A || Q == B)
{
_fmpz_vec_clear(Q->coeffs, Q->alloc);
Q->coeffs = q;
Q->alloc = lenq;
Q->length = lenq;
}
else
_fmpz_poly_set_length(Q, lenq);
if (R == A || R == B)
{
_fmpz_vec_clear(R->coeffs, R->alloc);
R->coeffs = r;
R->alloc = A->length;
R->length = lenr;
}
else
_fmpz_poly_set_length(R, lenr);
}
|