Coverage Report

Created: 2026-04-01 06:08

next uncovered line (L), next uncovered region (R), next uncovered branch (B)
/src/fftw3/dft/rader.c
Line
Count
Source
1
/*
2
 * Copyright (c) 2003, 2007-14 Matteo Frigo
3
 * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology
4
 *
5
 * This program is free software; you can redistribute it and/or modify
6
 * it under the terms of the GNU General Public License as published by
7
 * the Free Software Foundation; either version 2 of the License, or
8
 * (at your option) any later version.
9
 *
10
 * This program is distributed in the hope that it will be useful,
11
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
13
 * GNU General Public License for more details.
14
 *
15
 * You should have received a copy of the GNU General Public License
16
 * along with this program; if not, write to the Free Software
17
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
18
 *
19
 */
20
21
#include "dft/dft.h"
22
23
/*
24
 * Compute transforms of prime sizes using Rader's trick: turn them
25
 * into convolutions of size n - 1, which you then perform via a pair
26
 * of FFTs.
27
 */
28
29
typedef struct {
30
     solver super;
31
} S;
32
33
typedef struct {
34
     plan_dft super;
35
36
     plan *cld1, *cld2;
37
     R *omega;
38
     INT n, g, ginv;
39
     INT is, os;
40
     plan *cld_omega;
41
} P;
42
43
static rader_tl *omegas = 0;
44
45
static R *mkomega(enum wakefulness wakefulness, plan *p_, INT n, INT ginv)
46
43
{
47
43
     plan_dft *p = (plan_dft *) p_;
48
43
     R *omega;
49
43
     INT i, gpower;
50
43
     trigreal scale;
51
43
     triggen *t;
52
53
43
     if ((omega = X(rader_tl_find)(n, n, ginv, omegas)))
54
0
    return omega;
55
56
43
     omega = (R *)MALLOC(sizeof(R) * (n - 1) * 2, TWIDDLES);
57
58
43
     scale = n - 1.0; /* normalization for convolution */
59
60
43
     t = X(mktriggen)(wakefulness, n);
61
3.90k
     for (i = 0, gpower = 1; i < n-1; ++i, gpower = MULMOD(gpower, ginv, n)) {
62
3.86k
    trigreal w[2];
63
3.86k
    t->cexpl(t, gpower, w);
64
3.86k
    omega[2*i] = w[0] / scale;
65
3.86k
    omega[2*i+1] = FFT_SIGN * w[1] / scale;
66
3.86k
     }
67
43
     X(triggen_destroy)(t);
68
43
     A(gpower == 1);
69
70
43
     p->apply(p_, omega, omega + 1, omega, omega + 1);
71
72
43
     X(rader_tl_insert)(n, n, ginv, omega, &omegas);
73
43
     return omega;
74
43
}
75
76
static void free_omega(R *omega)
77
43
{
78
43
     X(rader_tl_delete)(omega, &omegas);
79
43
}
80
81
82
/***************************************************************************/
83
84
/* Below, we extensively use the identity that fft(x*)* = ifft(x) in
85
   order to share data between forward and backward transforms and to
86
   obviate the necessity of having separate forward and backward
87
   plans.  (Although we often compute separate plans these days anyway
88
   due to the differing strides, etcetera.)
89
90
   Of course, since the new FFTW gives us separate pointers to
91
   the real and imaginary parts, we could have instead used the
92
   fft(r,i) = ifft(i,r) form of this identity, but it was easier to
93
   reuse the code from our old version. */
94
95
static void apply(const plan *ego_, R *ri, R *ii, R *ro, R *io)
96
77
{
97
77
     const P *ego = (const P *) ego_;
98
77
     INT is, os;
99
77
     INT k, gpower, g, r;
100
77
     R *buf;
101
77
     R r0 = ri[0], i0 = ii[0];
102
103
77
     r = ego->n; is = ego->is; os = ego->os; g = ego->g; 
104
77
     buf = (R *) MALLOC(sizeof(R) * (r - 1) * 2, BUFFERS);
105
106
     /* First, permute the input, storing in buf: */
107
5.63k
     for (gpower = 1, k = 0; k < r - 1; ++k, gpower = MULMOD(gpower, g, r)) {
108
5.56k
    R rA, iA;
109
5.56k
    rA = ri[gpower * is];
110
5.56k
    iA = ii[gpower * is];
111
5.56k
    buf[2*k] = rA; buf[2*k + 1] = iA;
112
5.56k
     }
113
77
     /* gpower == g^(r-1) mod r == 1 */;
114
115
116
     /* compute DFT of buf, storing in output (except DC): */
117
77
     {
118
77
      plan_dft *cld = (plan_dft *) ego->cld1;
119
77
      cld->apply(ego->cld1, buf, buf+1, ro+os, io+os);
120
77
     }
121
122
     /* set output DC component: */
123
77
     {
124
77
    ro[0] = r0 + ro[os];
125
77
    io[0] = i0 + io[os];
126
77
     }
127
128
     /* now, multiply by omega: */
129
77
     {
130
77
    const R *omega = ego->omega;
131
5.63k
    for (k = 0; k < r - 1; ++k) {
132
5.56k
         E rB, iB, rW, iW;
133
5.56k
         rW = omega[2*k];
134
5.56k
         iW = omega[2*k+1];
135
5.56k
         rB = ro[(k+1)*os];
136
5.56k
         iB = io[(k+1)*os];
137
5.56k
         ro[(k+1)*os] = rW * rB - iW * iB;
138
5.56k
         io[(k+1)*os] = -(rW * iB + iW * rB);
139
5.56k
    }
140
77
     }
141
     
142
     /* this will add input[0] to all of the outputs after the ifft */
143
77
     ro[os] += r0;
144
77
     io[os] -= i0;
145
146
     /* inverse FFT: */
147
77
     {
148
77
      plan_dft *cld = (plan_dft *) ego->cld2;
149
77
      cld->apply(ego->cld2, ro+os, io+os, buf, buf+1);
150
77
     }
151
     
152
     /* finally, do inverse permutation to unshuffle the output: */
153
77
     {
154
77
    INT ginv = ego->ginv;
155
77
    gpower = 1;
156
5.63k
    for (k = 0; k < r - 1; ++k, gpower = MULMOD(gpower, ginv, r)) {
157
5.56k
         ro[gpower * os] = buf[2*k];
158
5.56k
         io[gpower * os] = -buf[2*k+1];
159
5.56k
    }
160
77
    A(gpower == 1);
161
77
     }
162
163
164
77
     X(ifree)(buf);
165
77
}
166
167
/***************************************************************************/
168
169
static void awake(plan *ego_, enum wakefulness wakefulness)
170
86
{
171
86
     P *ego = (P *) ego_;
172
173
86
     X(plan_awake)(ego->cld1, wakefulness);
174
86
     X(plan_awake)(ego->cld2, wakefulness);
175
86
     X(plan_awake)(ego->cld_omega, wakefulness);
176
177
86
     switch (wakefulness) {
178
43
   case SLEEPY:
179
43
        free_omega(ego->omega);
180
43
        ego->omega = 0;
181
43
        break;
182
43
   default:
183
43
        ego->g = X(find_generator)(ego->n);
184
43
        ego->ginv = X(power_mod)(ego->g, ego->n - 2, ego->n);
185
43
        A(MULMOD(ego->g, ego->ginv, ego->n) == 1);
186
187
43
        ego->omega = mkomega(wakefulness,
188
43
           ego->cld_omega, ego->n, ego->ginv);
189
43
        break;
190
86
     }
191
86
}
192
193
static void destroy(plan *ego_)
194
92
{
195
92
     P *ego = (P *) ego_;
196
92
     X(plan_destroy_internal)(ego->cld_omega);
197
92
     X(plan_destroy_internal)(ego->cld2);
198
92
     X(plan_destroy_internal)(ego->cld1);
199
92
}
200
201
static void print(const plan *ego_, printer *p)
202
0
{
203
0
     const P *ego = (const P *)ego_;
204
0
     p->print(p, "(dft-rader-%D%ois=%oos=%(%p%)",
205
0
              ego->n, ego->is, ego->os, ego->cld1);
206
0
     if (ego->cld2 != ego->cld1)
207
0
          p->print(p, "%(%p%)", ego->cld2);
208
0
     if (ego->cld_omega != ego->cld1 && ego->cld_omega != ego->cld2)
209
0
          p->print(p, "%(%p%)", ego->cld_omega);
210
0
     p->putchr(p, ')');
211
0
}
212
213
static int applicable(const solver *ego_, const problem *p_,
214
          const planner *plnr)
215
1.16k
{
216
1.16k
     const problem_dft *p = (const problem_dft *) p_;
217
1.16k
     UNUSED(ego_);
218
1.16k
     return (1
219
1.16k
       && p->sz->rnk == 1
220
836
       && p->vecsz->rnk == 0
221
321
       && CIMPLIES(NO_SLOWP(plnr), p->sz->dims[0].n > RADER_MAX_SLOW)
222
228
       && X(is_prime)(p->sz->dims[0].n)
223
224
       /* proclaim the solver SLOW if p-1 is not easily factorizable.
225
    Bluestein should take care of this case. */
226
132
       && CIMPLIES(NO_SLOWP(plnr), X(factors_into_small_primes)(p->sz->dims[0].n - 1))
227
1.16k
    );
228
1.16k
}
229
230
static int mkP(P *pln, INT n, INT is, INT os, R *ro, R *io,
231
         planner *plnr)
232
92
{
233
92
     plan *cld1 = (plan *) 0;
234
92
     plan *cld2 = (plan *) 0;
235
92
     plan *cld_omega = (plan *) 0;
236
92
     R *buf = (R *) 0;
237
238
     /* initial allocation for the purpose of planning */
239
92
     buf = (R *) MALLOC(sizeof(R) * (n - 1) * 2, BUFFERS);
240
241
92
     cld1 = X(mkplan_f_d)(plnr, 
242
92
        X(mkproblem_dft_d)(X(mktensor_1d)(n - 1, 2, os),
243
92
               X(mktensor_1d)(1, 0, 0),
244
92
               buf, buf + 1, ro + os, io + os),
245
92
        NO_SLOW, 0, 0);
246
92
     if (!cld1) goto nada;
247
248
92
     cld2 = X(mkplan_f_d)(plnr, 
249
92
        X(mkproblem_dft_d)(X(mktensor_1d)(n - 1, os, 2),
250
92
               X(mktensor_1d)(1, 0, 0),
251
92
               ro + os, io + os, buf, buf + 1),
252
92
        NO_SLOW, 0, 0);
253
254
92
     if (!cld2) goto nada;
255
256
     /* plan for omega array */
257
92
     cld_omega = X(mkplan_f_d)(plnr, 
258
92
             X(mkproblem_dft_d)(X(mktensor_1d)(n - 1, 2, 2),
259
92
              X(mktensor_1d)(1, 0, 0),
260
92
              buf, buf + 1, buf, buf + 1),
261
92
             NO_SLOW, ESTIMATE, 0);
262
92
     if (!cld_omega) goto nada;
263
264
     /* deallocate buffers; let awake() or apply() allocate them for real */
265
92
     X(ifree)(buf);
266
92
     buf = 0;
267
268
92
     pln->cld1 = cld1;
269
92
     pln->cld2 = cld2;
270
92
     pln->cld_omega = cld_omega;
271
92
     pln->omega = 0;
272
92
     pln->n = n;
273
92
     pln->is = is;
274
92
     pln->os = os;
275
276
92
     X(ops_add)(&cld1->ops, &cld2->ops, &pln->super.super.ops);
277
92
     pln->super.super.ops.other += (n - 1) * (4 * 2 + 6) + 6;
278
92
     pln->super.super.ops.add += (n - 1) * 2 + 4;
279
92
     pln->super.super.ops.mul += (n - 1) * 4;
280
281
92
     return 1;
282
283
0
 nada:
284
0
     X(ifree0)(buf);
285
0
     X(plan_destroy_internal)(cld_omega);
286
0
     X(plan_destroy_internal)(cld2);
287
0
     X(plan_destroy_internal)(cld1);
288
0
     return 0;
289
92
}
290
291
static plan *mkplan(const solver *ego, const problem *p_, planner *plnr)
292
1.16k
{
293
1.16k
     const problem_dft *p = (const problem_dft *) p_;
294
1.16k
     P *pln;
295
1.16k
     INT n;
296
1.16k
     INT is, os;
297
298
1.16k
     static const plan_adt padt = {
299
1.16k
    X(dft_solve), awake, print, destroy
300
1.16k
     };
301
302
1.16k
     if (!applicable(ego, p_, plnr))
303
1.07k
    return (plan *) 0;
304
305
92
     n = p->sz->dims[0].n;
306
92
     is = p->sz->dims[0].is;
307
92
     os = p->sz->dims[0].os;
308
309
92
     pln = MKPLAN_DFT(P, &padt, apply);
310
92
     if (!mkP(pln, n, is, os, p->ro, p->io, plnr)) {
311
0
    X(ifree)(pln);
312
0
    return (plan *) 0;
313
0
     }
314
92
     return &(pln->super.super);
315
92
}
316
317
static solver *mksolver(void)
318
1
{
319
1
     static const solver_adt sadt = { PROBLEM_DFT, mkplan, 0 };
320
1
     S *slv = MKSOLVER(S, &sadt);
321
1
     return &(slv->super);
322
1
}
323
324
void X(dft_rader_register)(planner *p)
325
1
{
326
1
     REGISTER_SOLVER(p, mksolver());
327
1
}