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

427

428

429

430

431

432

433

434

435

436

437

438

439

440

441

442

443

444

445

446

447

448

449

450

451

452

453

454

455

456

457

458

459

460

461

462

463

464

465

466

467

468

469

470

471

472

473

474

475

476

477

478

479

480

481

482

483

484

485

486

487

488

489

490

491

492

493

494

495

496

497

498

499

500

501

502

503

504

505

506

507

508

509

510

511

512

513

514

515

516

517

518

519

520

521

522

523

524

525

526

527

528

529

530

531

532

533

534

535

536

537

538

"""Dictionary Of Keys based matrix""" 

 

from __future__ import division, print_function, absolute_import 

 

__docformat__ = "restructuredtext en" 

 

__all__ = ['dok_matrix', 'isspmatrix_dok'] 

 

import functools 

import operator 

import itertools 

 

import numpy as np 

 

from scipy._lib.six import zip as izip, xrange, iteritems, iterkeys, itervalues 

 

from .base import spmatrix, isspmatrix 

from .sputils import (isdense, getdtype, isshape, isintlike, isscalarlike, 

upcast, upcast_scalar, IndexMixin, get_index_dtype, 

check_shape) 

 

try: 

from operator import isSequenceType as _is_sequence 

except ImportError: 

def _is_sequence(x): 

return (hasattr(x, '__len__') or hasattr(x, '__next__') 

or hasattr(x, 'next')) 

 

 

class dok_matrix(spmatrix, IndexMixin, dict): 

""" 

Dictionary Of Keys based sparse matrix. 

 

This is an efficient structure for constructing sparse 

matrices incrementally. 

 

This can be instantiated in several ways: 

dok_matrix(D) 

with a dense matrix, D 

 

dok_matrix(S) 

with a sparse matrix, S 

 

dok_matrix((M,N), [dtype]) 

create the matrix with initial shape (M,N) 

dtype is optional, defaulting to dtype='d' 

 

Attributes 

---------- 

dtype : dtype 

Data type of the matrix 

shape : 2-tuple 

Shape of the matrix 

ndim : int 

Number of dimensions (this is always 2) 

nnz 

Number of nonzero elements 

 

Notes 

----- 

 

Sparse matrices can be used in arithmetic operations: they support 

addition, subtraction, multiplication, division, and matrix power. 

 

Allows for efficient O(1) access of individual elements. 

Duplicates are not allowed. 

Can be efficiently converted to a coo_matrix once constructed. 

 

Examples 

-------- 

>>> import numpy as np 

>>> from scipy.sparse import dok_matrix 

>>> S = dok_matrix((5, 5), dtype=np.float32) 

>>> for i in range(5): 

... for j in range(5): 

... S[i, j] = i + j # Update element 

 

""" 

format = 'dok' 

 

def __init__(self, arg1, shape=None, dtype=None, copy=False): 

dict.__init__(self) 

spmatrix.__init__(self) 

 

self.dtype = getdtype(dtype, default=float) 

if isinstance(arg1, tuple) and isshape(arg1): # (M,N) 

M, N = arg1 

self._shape = check_shape((M, N)) 

elif isspmatrix(arg1): # Sparse ctor 

if isspmatrix_dok(arg1) and copy: 

arg1 = arg1.copy() 

else: 

arg1 = arg1.todok() 

 

if dtype is not None: 

arg1 = arg1.astype(dtype) 

 

dict.update(self, arg1) 

self._shape = check_shape(arg1.shape) 

self.dtype = arg1.dtype 

else: # Dense ctor 

try: 

arg1 = np.asarray(arg1) 

except: 

raise TypeError('Invalid input format.') 

 

if len(arg1.shape) != 2: 

raise TypeError('Expected rank <=2 dense array or matrix.') 

 

from .coo import coo_matrix 

d = coo_matrix(arg1, dtype=dtype).todok() 

dict.update(self, d) 

self._shape = check_shape(arg1.shape) 

self.dtype = d.dtype 

 

def update(self, val): 

# Prevent direct usage of update 

raise NotImplementedError("Direct modification to dok_matrix element " 

"is not allowed.") 

 

def _update(self, data): 

"""An update method for dict data defined for direct access to 

`dok_matrix` data. Main purpose is to be used for effcient conversion 

from other spmatrix classes. Has no checking if `data` is valid.""" 

return dict.update(self, data) 

 

def set_shape(self, shape): 

new_matrix = self.reshape(shape, copy=False).asformat(self.format) 

self.__dict__ = new_matrix.__dict__ 

dict.clear(self) 

dict.update(self, new_matrix) 

 

shape = property(fget=spmatrix.get_shape, fset=set_shape) 

 

def getnnz(self, axis=None): 

if axis is not None: 

raise NotImplementedError("getnnz over an axis is not implemented " 

"for DOK format.") 

return dict.__len__(self) 

 

def count_nonzero(self): 

return sum(x != 0 for x in itervalues(self)) 

 

getnnz.__doc__ = spmatrix.getnnz.__doc__ 

count_nonzero.__doc__ = spmatrix.count_nonzero.__doc__ 

 

def __len__(self): 

return dict.__len__(self) 

 

def get(self, key, default=0.): 

"""This overrides the dict.get method, providing type checking 

but otherwise equivalent functionality. 

""" 

try: 

i, j = key 

assert isintlike(i) and isintlike(j) 

except (AssertionError, TypeError, ValueError): 

raise IndexError('Index must be a pair of integers.') 

if (i < 0 or i >= self.shape[0] or j < 0 or j >= self.shape[1]): 

raise IndexError('Index out of bounds.') 

return dict.get(self, key, default) 

 

def __getitem__(self, index): 

"""If key=(i, j) is a pair of integers, return the corresponding 

element. If either i or j is a slice or sequence, return a new sparse 

matrix with just these elements. 

""" 

zero = self.dtype.type(0) 

i, j = self._unpack_index(index) 

 

i_intlike = isintlike(i) 

j_intlike = isintlike(j) 

 

if i_intlike and j_intlike: 

i = int(i) 

j = int(j) 

if i < 0: 

i += self.shape[0] 

if i < 0 or i >= self.shape[0]: 

raise IndexError('Index out of bounds.') 

if j < 0: 

j += self.shape[1] 

if j < 0 or j >= self.shape[1]: 

raise IndexError('Index out of bounds.') 

return dict.get(self, (i,j), zero) 

elif ((i_intlike or isinstance(i, slice)) and 

(j_intlike or isinstance(j, slice))): 

# Fast path for slicing very sparse matrices 

i_slice = slice(i, i+1) if i_intlike else i 

j_slice = slice(j, j+1) if j_intlike else j 

i_indices = i_slice.indices(self.shape[0]) 

j_indices = j_slice.indices(self.shape[1]) 

i_seq = xrange(*i_indices) 

j_seq = xrange(*j_indices) 

newshape = (len(i_seq), len(j_seq)) 

newsize = _prod(newshape) 

 

if len(self) < 2*newsize and newsize != 0: 

# Switch to the fast path only when advantageous 

# (count the iterations in the loops, adjust for complexity) 

# 

# We also don't handle newsize == 0 here (if 

# i/j_intlike, it can mean index i or j was out of 

# bounds) 

return self._getitem_ranges(i_indices, j_indices, newshape) 

 

i, j = self._index_to_arrays(i, j) 

 

if i.size == 0: 

return dok_matrix(i.shape, dtype=self.dtype) 

 

min_i = i.min() 

if min_i < -self.shape[0] or i.max() >= self.shape[0]: 

raise IndexError('Index (%d) out of range -%d to %d.' % 

(i.min(), self.shape[0], self.shape[0]-1)) 

if min_i < 0: 

i = i.copy() 

i[i < 0] += self.shape[0] 

 

min_j = j.min() 

if min_j < -self.shape[1] or j.max() >= self.shape[1]: 

raise IndexError('Index (%d) out of range -%d to %d.' % 

(j.min(), self.shape[1], self.shape[1]-1)) 

if min_j < 0: 

j = j.copy() 

j[j < 0] += self.shape[1] 

 

newdok = dok_matrix(i.shape, dtype=self.dtype) 

 

for key in itertools.product(xrange(i.shape[0]), xrange(i.shape[1])): 

v = dict.get(self, (i[key], j[key]), zero) 

if v: 

dict.__setitem__(newdok, key, v) 

 

return newdok 

 

def _getitem_ranges(self, i_indices, j_indices, shape): 

# performance golf: we don't want Numpy scalars here, they are slow 

i_start, i_stop, i_stride = map(int, i_indices) 

j_start, j_stop, j_stride = map(int, j_indices) 

 

newdok = dok_matrix(shape, dtype=self.dtype) 

 

for (ii, jj) in iterkeys(self): 

# ditto for numpy scalars 

ii = int(ii) 

jj = int(jj) 

a, ra = divmod(ii - i_start, i_stride) 

if a < 0 or a >= shape[0] or ra != 0: 

continue 

b, rb = divmod(jj - j_start, j_stride) 

if b < 0 or b >= shape[1] or rb != 0: 

continue 

dict.__setitem__(newdok, (a, b), 

dict.__getitem__(self, (ii, jj))) 

return newdok 

 

def __setitem__(self, index, x): 

if isinstance(index, tuple) and len(index) == 2: 

# Integer index fast path 

i, j = index 

if (isintlike(i) and isintlike(j) and 0 <= i < self.shape[0] 

and 0 <= j < self.shape[1]): 

v = np.asarray(x, dtype=self.dtype) 

if v.ndim == 0 and v != 0: 

dict.__setitem__(self, (int(i), int(j)), v[()]) 

return 

 

i, j = self._unpack_index(index) 

i, j = self._index_to_arrays(i, j) 

 

if isspmatrix(x): 

x = x.toarray() 

 

# Make x and i into the same shape 

x = np.asarray(x, dtype=self.dtype) 

x, _ = np.broadcast_arrays(x, i) 

 

if x.shape != i.shape: 

raise ValueError("Shape mismatch in assignment.") 

 

if np.size(x) == 0: 

return 

 

min_i = i.min() 

if min_i < -self.shape[0] or i.max() >= self.shape[0]: 

raise IndexError('Index (%d) out of range -%d to %d.' % 

(i.min(), self.shape[0], self.shape[0]-1)) 

if min_i < 0: 

i = i.copy() 

i[i < 0] += self.shape[0] 

 

min_j = j.min() 

if min_j < -self.shape[1] or j.max() >= self.shape[1]: 

raise IndexError('Index (%d) out of range -%d to %d.' % 

(j.min(), self.shape[1], self.shape[1]-1)) 

if min_j < 0: 

j = j.copy() 

j[j < 0] += self.shape[1] 

 

dict.update(self, izip(izip(i.flat, j.flat), x.flat)) 

 

if 0 in x: 

zeroes = x == 0 

for key in izip(i[zeroes].flat, j[zeroes].flat): 

if dict.__getitem__(self, key) == 0: 

# may have been superseded by later update 

del self[key] 

 

def __add__(self, other): 

if isscalarlike(other): 

res_dtype = upcast_scalar(self.dtype, other) 

new = dok_matrix(self.shape, dtype=res_dtype) 

# Add this scalar to every element. 

M, N = self.shape 

for key in itertools.product(xrange(M), xrange(N)): 

aij = dict.get(self, (key), 0) + other 

if aij: 

new[key] = aij 

# new.dtype.char = self.dtype.char 

elif isspmatrix_dok(other): 

if other.shape != self.shape: 

raise ValueError("Matrix dimensions are not equal.") 

# We could alternatively set the dimensions to the largest of 

# the two matrices to be summed. Would this be a good idea? 

res_dtype = upcast(self.dtype, other.dtype) 

new = dok_matrix(self.shape, dtype=res_dtype) 

dict.update(new, self) 

with np.errstate(over='ignore'): 

dict.update(new, 

((k, new[k] + other[k]) for k in iterkeys(other))) 

elif isspmatrix(other): 

csc = self.tocsc() 

new = csc + other 

elif isdense(other): 

new = self.todense() + other 

else: 

return NotImplemented 

return new 

 

def __radd__(self, other): 

if isscalarlike(other): 

new = dok_matrix(self.shape, dtype=self.dtype) 

M, N = self.shape 

for key in itertools.product(xrange(M), xrange(N)): 

aij = dict.get(self, (key), 0) + other 

if aij: 

new[key] = aij 

elif isspmatrix_dok(other): 

if other.shape != self.shape: 

raise ValueError("Matrix dimensions are not equal.") 

new = dok_matrix(self.shape, dtype=self.dtype) 

dict.update(new, self) 

dict.update(new, 

((k, self[k] + other[k]) for k in iterkeys(other))) 

elif isspmatrix(other): 

csc = self.tocsc() 

new = csc + other 

elif isdense(other): 

new = other + self.todense() 

else: 

return NotImplemented 

return new 

 

def __neg__(self): 

if self.dtype.kind == 'b': 

raise NotImplementedError('Negating a sparse boolean matrix is not' 

' supported.') 

new = dok_matrix(self.shape, dtype=self.dtype) 

dict.update(new, ((k, -self[k]) for k in iterkeys(self))) 

return new 

 

def _mul_scalar(self, other): 

res_dtype = upcast_scalar(self.dtype, other) 

# Multiply this scalar by every element. 

new = dok_matrix(self.shape, dtype=res_dtype) 

dict.update(new, ((k, v * other) for k, v in iteritems(self))) 

return new 

 

def _mul_vector(self, other): 

# matrix * vector 

result = np.zeros(self.shape[0], dtype=upcast(self.dtype, other.dtype)) 

for (i, j), v in iteritems(self): 

result[i] += v * other[j] 

return result 

 

def _mul_multivector(self, other): 

# matrix * multivector 

result_shape = (self.shape[0], other.shape[1]) 

result_dtype = upcast(self.dtype, other.dtype) 

result = np.zeros(result_shape, dtype=result_dtype) 

for (i, j), v in iteritems(self): 

result[i,:] += v * other[j,:] 

return result 

 

def __imul__(self, other): 

if isscalarlike(other): 

dict.update(self, ((k, v * other) for k, v in iteritems(self))) 

return self 

return NotImplemented 

 

def __truediv__(self, other): 

if isscalarlike(other): 

res_dtype = upcast_scalar(self.dtype, other) 

new = dok_matrix(self.shape, dtype=res_dtype) 

dict.update(new, ((k, v / other) for k, v in iteritems(self))) 

return new 

return self.tocsr() / other 

 

def __itruediv__(self, other): 

if isscalarlike(other): 

dict.update(self, ((k, v / other) for k, v in iteritems(self))) 

return self 

return NotImplemented 

 

def __reduce__(self): 

# this approach is necessary because __setstate__ is called after 

# __setitem__ upon unpickling and since __init__ is not called there 

# is no shape attribute hence it is not possible to unpickle it. 

return dict.__reduce__(self) 

 

# What should len(sparse) return? For consistency with dense matrices, 

# perhaps it should be the number of rows? For now it returns the number 

# of non-zeros. 

 

def transpose(self, axes=None, copy=False): 

if axes is not None: 

raise ValueError("Sparse matrices do not support " 

"an 'axes' parameter because swapping " 

"dimensions is the only logical permutation.") 

 

M, N = self.shape 

new = dok_matrix((N, M), dtype=self.dtype, copy=copy) 

dict.update(new, (((right, left), val) 

for (left, right), val in iteritems(self))) 

return new 

 

transpose.__doc__ = spmatrix.transpose.__doc__ 

 

def conjtransp(self): 

"""Return the conjugate transpose.""" 

M, N = self.shape 

new = dok_matrix((N, M), dtype=self.dtype) 

dict.update(new, (((right, left), np.conj(val)) 

for (left, right), val in iteritems(self))) 

return new 

 

def copy(self): 

new = dok_matrix(self.shape, dtype=self.dtype) 

dict.update(new, self) 

return new 

 

copy.__doc__ = spmatrix.copy.__doc__ 

 

def getrow(self, i): 

"""Returns the i-th row as a (1 x n) DOK matrix.""" 

new = dok_matrix((1, self.shape[1]), dtype=self.dtype) 

dict.update(new, (((0, j), self[i, j]) for j in xrange(self.shape[1]))) 

return new 

 

def getcol(self, j): 

"""Returns the j-th column as a (m x 1) DOK matrix.""" 

new = dok_matrix((self.shape[0], 1), dtype=self.dtype) 

dict.update(new, (((i, 0), self[i, j]) for i in xrange(self.shape[0]))) 

return new 

 

def tocoo(self, copy=False): 

from .coo import coo_matrix 

if self.nnz == 0: 

return coo_matrix(self.shape, dtype=self.dtype) 

 

idx_dtype = get_index_dtype(maxval=max(self.shape)) 

data = np.fromiter(itervalues(self), dtype=self.dtype, count=self.nnz) 

row = np.fromiter((i for i, _ in iterkeys(self)), dtype=idx_dtype, count=self.nnz) 

col = np.fromiter((j for _, j in iterkeys(self)), dtype=idx_dtype, count=self.nnz) 

A = coo_matrix((data, (row, col)), shape=self.shape, dtype=self.dtype) 

A.has_canonical_format = True 

return A 

 

tocoo.__doc__ = spmatrix.tocoo.__doc__ 

 

def todok(self, copy=False): 

if copy: 

return self.copy() 

return self 

 

todok.__doc__ = spmatrix.todok.__doc__ 

 

def tocsc(self, copy=False): 

return self.tocoo(copy=False).tocsc(copy=copy) 

 

tocsc.__doc__ = spmatrix.tocsc.__doc__ 

 

def resize(self, *shape): 

shape = check_shape(shape) 

newM, newN = shape 

M, N = self.shape 

if newM < M or newN < N: 

# Remove all elements outside new dimensions 

for (i, j) in list(iterkeys(self)): 

if i >= newM or j >= newN: 

del self[i, j] 

self._shape = shape 

 

resize.__doc__ = spmatrix.resize.__doc__ 

 

 

def isspmatrix_dok(x): 

"""Is x of dok_matrix type? 

 

Parameters 

---------- 

x 

object to check for being a dok matrix 

 

Returns 

------- 

bool 

True if x is a dok matrix, False otherwise 

 

Examples 

-------- 

>>> from scipy.sparse import dok_matrix, isspmatrix_dok 

>>> isspmatrix_dok(dok_matrix([[5]])) 

True 

 

>>> from scipy.sparse import dok_matrix, csr_matrix, isspmatrix_dok 

>>> isspmatrix_dok(csr_matrix([[5]])) 

False 

""" 

return isinstance(x, dok_matrix) 

 

 

def _prod(x): 

"""Product of a list of numbers; ~40x faster vs np.prod for Python tuples""" 

if len(x) == 0: 

return 1 

return functools.reduce(operator.mul, x)