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
|
/*
Copyright (C) 2009-2015 VZLU Prague
This file is part of Octave.
Octave 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 3 of the License, or (at your
option) any later version.
Octave 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 Octave; see the file COPYING. If not, see
<http://www.gnu.org/licenses/>.
*/
#ifdef HAVE_CONFIG_H
#include <config.h>
#endif
#include "f77-fcn.h"
#include "mx-base.h"
#include "error.h"
#include "defun.h"
#include "parse.h"
extern "C"
{
F77_RET_T
F77_FUNC (ddot3, DDOT3) (const octave_idx_type&, const octave_idx_type&,
const octave_idx_type&, const double*,
const double*, double*);
F77_RET_T
F77_FUNC (sdot3, SDOT3) (const octave_idx_type&, const octave_idx_type&,
const octave_idx_type&, const float*,
const float*, float*);
F77_RET_T
F77_FUNC (zdotc3, ZDOTC3) (const octave_idx_type&, const octave_idx_type&,
const octave_idx_type&, const Complex*,
const Complex*, Complex*);
F77_RET_T
F77_FUNC (cdotc3, CDOTC3) (const octave_idx_type&, const octave_idx_type&,
const octave_idx_type&, const FloatComplex*,
const FloatComplex*, FloatComplex*);
F77_RET_T
F77_FUNC (dmatm3, DMATM3) (const octave_idx_type&, const octave_idx_type&,
const octave_idx_type&, const octave_idx_type&,
const double*, const double*, double*);
F77_RET_T
F77_FUNC (smatm3, SMATM3) (const octave_idx_type&, const octave_idx_type&,
const octave_idx_type&, const octave_idx_type&,
const float*, const float*, float*);
F77_RET_T
F77_FUNC (zmatm3, ZMATM3) (const octave_idx_type&, const octave_idx_type&,
const octave_idx_type&, const octave_idx_type&,
const Complex*, const Complex*, Complex*);
F77_RET_T
F77_FUNC (cmatm3, CMATM3) (const octave_idx_type&, const octave_idx_type&,
const octave_idx_type&, const octave_idx_type&,
const FloatComplex*, const FloatComplex*,
FloatComplex*);
}
static void
get_red_dims (const dim_vector& x, const dim_vector& y, int dim,
dim_vector& z, octave_idx_type& m, octave_idx_type& n,
octave_idx_type& k)
{
int nd = x.length ();
assert (nd == y.length ());
z = dim_vector::alloc (nd);
m = 1, n = 1, k = 1;
for (int i = 0; i < nd; i++)
{
if (i < dim)
{
z(i) = x(i);
m *= x(i);
}
else if (i > dim)
{
z(i) = x(i);
n *= x(i);
}
else
{
k = x(i);
z(i) = 1;
}
}
}
DEFUN (dot, args, ,
"-*- texinfo -*-\n\
@deftypefn {Built-in Function} {} dot (@var{x}, @var{y}, @var{dim})\n\
Compute the dot product of two vectors.\n\
\n\
If @var{x} and @var{y} are matrices, calculate the dot products along the\n\
first non-singleton dimension.\n\
\n\
If the optional argument @var{dim} is given, calculate the dot products\n\
along this dimension.\n\
\n\
This is equivalent to\n\
@code{sum (conj (@var{X}) .* @var{Y}, @var{dim})},\n\
but avoids forming a temporary array and is faster. When @var{X} and\n\
@var{Y} are column vectors, the result is equivalent to\n\
@code{@var{X}' * @var{Y}}.\n\
@seealso{cross, divergence}\n\
@end deftypefn")
{
octave_value retval;
int nargin = args.length ();
if (nargin < 2 || nargin > 3)
{
print_usage ();
return retval;
}
octave_value argx = args(0);
octave_value argy = args(1);
if (argx.is_numeric_type () && argy.is_numeric_type ())
{
dim_vector dimx = argx.dims ();
dim_vector dimy = argy.dims ();
bool match = dimx == dimy;
if (! match && nargin == 2
&& dimx.is_vector () && dimy.is_vector ())
{
// Change to column vectors.
dimx = dimx.redim (1);
argx = argx.reshape (dimx);
dimy = dimy.redim (1);
argy = argy.reshape (dimy);
match = ! error_state && (dimx == dimy);
}
if (match)
{
int dim;
if (nargin == 2)
dim = dimx.first_non_singleton ();
else
dim = args(2).int_value (true) - 1;
if (error_state)
;
else if (dim < 0)
error ("dot: DIM must be a valid dimension");
else
{
octave_idx_type m, n, k;
dim_vector dimz;
if (argx.is_complex_type () || argy.is_complex_type ())
{
if (argx.is_single_type () || argy.is_single_type ())
{
FloatComplexNDArray x = argx.float_complex_array_value ();
FloatComplexNDArray y = argy.float_complex_array_value ();
get_red_dims (dimx, dimy, dim, dimz, m, n, k);
FloatComplexNDArray z(dimz);
if (! error_state)
F77_XFCN (cdotc3, CDOTC3, (m, n, k,
x.data (), y.data (),
z.fortran_vec ()));
retval = z;
}
else
{
ComplexNDArray x = argx.complex_array_value ();
ComplexNDArray y = argy.complex_array_value ();
get_red_dims (dimx, dimy, dim, dimz, m, n, k);
ComplexNDArray z(dimz);
if (! error_state)
F77_XFCN (zdotc3, ZDOTC3, (m, n, k,
x.data (), y.data (),
z.fortran_vec ()));
retval = z;
}
}
else if (argx.is_float_type () && argy.is_float_type ())
{
if (argx.is_single_type () || argy.is_single_type ())
{
FloatNDArray x = argx.float_array_value ();
FloatNDArray y = argy.float_array_value ();
get_red_dims (dimx, dimy, dim, dimz, m, n, k);
FloatNDArray z(dimz);
if (! error_state)
F77_XFCN (sdot3, SDOT3, (m, n, k, x.data (), y.data (),
z.fortran_vec ()));
retval = z;
}
else
{
NDArray x = argx.array_value ();
NDArray y = argy.array_value ();
get_red_dims (dimx, dimy, dim, dimz, m, n, k);
NDArray z(dimz);
if (! error_state)
F77_XFCN (ddot3, DDOT3, (m, n, k, x.data (), y.data (),
z.fortran_vec ()));
retval = z;
}
}
else
{
// Non-optimized evaluation.
octave_value_list tmp;
tmp(1) = dim + 1;
tmp(0) = do_binary_op (octave_value::op_el_mul, argx, argy);
if (! error_state)
{
tmp = feval ("sum", tmp, 1);
if (! tmp.empty ())
retval = tmp(0);
}
}
}
}
else
error ("dot: sizes of X and Y must match");
}
else
error ("dot: X and Y must be numeric");
return retval;
}
/*
%!assert (dot ([1, 2], [2, 3]), 8)
%!test
%! x = [2, 1; 2, 1];
%! y = [-0.5, 2; 0.5, -2];
%! assert (dot (x, y), [0 0]);
%! assert (dot (single (x), single (y)), single ([0 0]));
%!test
%! x = [1+i, 3-i; 1-i, 3-i];
%! assert (dot (x, x), [4, 20]);
%! assert (dot (single (x), single (x)), single ([4, 20]));
%!test
%! x = int8 ([1 2]);
%! y = int8 ([2 3]);
%! assert (dot (x, y), 8);
%!test
%! x = int8 ([1 2; 3 4]);
%! y = int8 ([5 6; 7 8]);
%! assert (dot (x, y), [26 44]);
%! assert (dot (x, y, 2), [17; 53]);
%! assert (dot (x, y, 3), [5 12; 21 32]);
%% Test input validation
%!error dot ()
%!error dot (1)
%!error dot (1,2,3,4)
%!error <X and Y must be numeric> dot ({1,2}, [3,4])
%!error <X and Y must be numeric> dot ([1,2], {3,4})
%!error <sizes of X and Y must match> dot ([1 2], [1 2 3])
%!error <sizes of X and Y must match> dot ([1 2]', [1 2 3]')
%!error <sizes of X and Y must match> dot (ones (2,2), ones (2,3))
%!error <DIM must be a valid dimension> dot ([1 2], [1 2], 0)
*/
DEFUN (blkmm, args, ,
"-*- texinfo -*-\n\
@deftypefn {Built-in Function} {} blkmm (@var{A}, @var{B})\n\
Compute products of matrix blocks.\n\
\n\
The blocks are given as 2-dimensional subarrays of the arrays @var{A},\n\
@var{B}. The size of @var{A} must have the form @code{[m,k,@dots{}]} and\n\
size of @var{B} must be @code{[k,n,@dots{}]}. The result is then of size\n\
@code{[m,n,@dots{}]} and is computed as follows:\n\
\n\
@example\n\
@group\n\
for i = 1:prod (size (@var{A})(3:end))\n\
@var{C}(:,:,i) = @var{A}(:,:,i) * @var{B}(:,:,i)\n\
endfor\n\
@end group\n\
@end example\n\
@end deftypefn")
{
octave_value retval;
int nargin = args.length ();
if (nargin != 2)
{
print_usage ();
return retval;
}
octave_value argx = args(0);
octave_value argy = args(1);
if (argx.is_numeric_type () && argy.is_numeric_type ())
{
const dim_vector dimx = argx.dims ();
const dim_vector dimy = argy.dims ();
int nd = dimx.length ();
octave_idx_type m = dimx(0);
octave_idx_type k = dimx(1);
octave_idx_type n = dimy(1);
octave_idx_type np = 1;
bool match = dimy(0) == k && nd == dimy.length ();
dim_vector dimz = dim_vector::alloc (nd);
dimz(0) = m;
dimz(1) = n;
for (int i = 2; match && i < nd; i++)
{
match = match && dimx(i) == dimy(i);
dimz(i) = dimx(i);
np *= dimz(i);
}
if (match)
{
if (argx.is_complex_type () || argy.is_complex_type ())
{
if (argx.is_single_type () || argy.is_single_type ())
{
FloatComplexNDArray x = argx.float_complex_array_value ();
FloatComplexNDArray y = argy.float_complex_array_value ();
FloatComplexNDArray z(dimz);
if (! error_state)
F77_XFCN (cmatm3, CMATM3, (m, n, k, np,
x.data (), y.data (),
z.fortran_vec ()));
retval = z;
}
else
{
ComplexNDArray x = argx.complex_array_value ();
ComplexNDArray y = argy.complex_array_value ();
ComplexNDArray z(dimz);
if (! error_state)
F77_XFCN (zmatm3, ZMATM3, (m, n, k, np,
x.data (), y.data (),
z.fortran_vec ()));
retval = z;
}
}
else
{
if (argx.is_single_type () || argy.is_single_type ())
{
FloatNDArray x = argx.float_array_value ();
FloatNDArray y = argy.float_array_value ();
FloatNDArray z(dimz);
if (! error_state)
F77_XFCN (smatm3, SMATM3, (m, n, k, np,
x.data (), y.data (),
z.fortran_vec ()));
retval = z;
}
else
{
NDArray x = argx.array_value ();
NDArray y = argy.array_value ();
NDArray z(dimz);
if (! error_state)
F77_XFCN (dmatm3, DMATM3, (m, n, k, np,
x.data (), y.data (),
z.fortran_vec ()));
retval = z;
}
}
}
else
error ("blkmm: A and B dimensions don't match: (%s) and (%s)",
dimx.str ().c_str (), dimy.str ().c_str ());
}
else
error ("blkmm: A and B must be numeric");
return retval;
}
/*
%!test
%! x(:,:,1) = [1 2; 3 4];
%! x(:,:,2) = [1 1; 1 1];
%! z(:,:,1) = [7 10; 15 22];
%! z(:,:,2) = [2 2; 2 2];
%! assert (blkmm (x,x), z);
%! assert (blkmm (single (x), single (x)), single (z));
%! assert (blkmm (x, single (x)), single (z));
%!test
%! x(:,:,1) = [1 2; 3 4];
%! x(:,:,2) = [1i 1i; 1i 1i];
%! z(:,:,1) = [7 10; 15 22];
%! z(:,:,2) = [-2 -2; -2 -2];
%! assert (blkmm (x,x), z);
%! assert (blkmm (single (x), single (x)), single (z));
%! assert (blkmm (x, single (x)), single (z));
%% Test input validation
%!error blkmm ()
%!error blkmm (1)
%!error blkmm (1,2,3)
%!error <A and B dimensions don't match> blkmm (ones (2,2), ones (3,3))
%!error <A and B must be numeric> blkmm ({1,2}, [3,4])
%!error <A and B must be numeric> blkmm ([3,4], {1,2})
*/
|
