field.py 49 KB
Newer Older
1
2
3
4
5
6
7
8
9
10
11
12
# This program 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.
#
# This program 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 this program.  If not, see <http://www.gnu.org/licenses/>.
Theo Steininger's avatar
Theo Steininger committed
13
14
15
16
17
#
# Copyright(C) 2013-2017 Max-Planck-Society
#
# NIFTy is being developed at the Max-Planck-Institut fuer Astrophysik
# and financially supported by the Studienstiftung des deutschen Volkes.
18

csongor's avatar
csongor committed
19
from __future__ import division
20

21
import ast
csongor's avatar
csongor committed
22
23
import numpy as np

Theo Steininger's avatar
Theo Steininger committed
24
25
from keepers import Versionable,\
                    Loggable
Jait Dixit's avatar
Jait Dixit committed
26

27
from d2o import distributed_data_object,\
28
    STRATEGIES as DISTRIBUTION_STRATEGIES
csongor's avatar
csongor committed
29

30
from nifty.config import nifty_configuration as gc
csongor's avatar
csongor committed
31

32
from nifty.domain_object import DomainObject
33

34
from nifty.spaces.power_space import PowerSpace
csongor's avatar
csongor committed
35

csongor's avatar
csongor committed
36
import nifty.nifty_utilities as utilities
37
38
from nifty.random import Random

csongor's avatar
csongor committed
39

Jait Dixit's avatar
Jait Dixit committed
40
class Field(Loggable, Versionable, object):
Theo Steininger's avatar
Theo Steininger committed
41
42
43
    """ The discrete representation of a continuous field over multiple spaces.

    In NIFTY, Fields are used to store data arrays and carry all the needed
44
    metainformation (i.e. the domain) for operators to be able to work on them.
Theo Steininger's avatar
Theo Steininger committed
45
46
    In addition Field has methods to work with power-spectra.

47
48
49
50
    Parameters
    ----------
    domain : DomainObject
        One of the space types NIFTY supports. RGSpace, GLSpace, HPSpace,
Theo Steininger's avatar
Theo Steininger committed
51
        LMSpace or PowerSpace. It might also be a FieldArray, which is
52
        an unstructured domain.
Theo Steininger's avatar
Theo Steininger committed
53

54
55
56
57
    val : scalar, numpy.ndarray, distributed_data_object, Field
        The values the array should contain after init. A scalar input will
        fill the whole array with this scalar. If an array is provided the
        array's dimensions must match the domain's.
Theo Steininger's avatar
Theo Steininger committed
58

59
60
    dtype : type
        A numpy.type. Most common are int, float and complex.
Theo Steininger's avatar
Theo Steininger committed
61

62
63
64
65
66
67
    distribution_strategy: optional[{'fftw', 'equal', 'not', 'freeform'}]
        Specifies which distributor will be created and used.
        'fftw'      uses the distribution strategy of pyfftw,
        'equal'     tries to  distribute the data as uniform as possible
        'not'       does not distribute the data at all
        'freeform'  distribute the data according to the given local data/shape
Theo Steininger's avatar
Theo Steininger committed
68

69
70
71
72
73
    copy: boolean

    Attributes
    ----------
    val : distributed_data_object
Theo Steininger's avatar
Theo Steininger committed
74

75
76
77
78
79
80
81
    domain : DomainObject
        See Parameters.
    domain_axes : tuple of tuples
        Enumerates the axes of the Field
    dtype : type
        Contains the datatype stored in the Field.
    distribution_strategy : string
Theo Steininger's avatar
Theo Steininger committed
82
83
        Name of the used distribution_strategy.

84
85
86
87
88
89
90
    Raise
    -----
    TypeError
        Raised if
            *the given domain contains something that is not a DomainObject
             instance
            *val is an array that has a different dimension than the domain
Theo Steininger's avatar
Theo Steininger committed
91

92
93
94
95
96
97
98
99
100
101
102
    Examples
    --------
    >>> a = Field(RGSpace([4,5]),val=2)
    >>> a.val
    <distributed_data_object>
    array([[2, 2, 2, 2, 2],
           [2, 2, 2, 2, 2],
           [2, 2, 2, 2, 2],
           [2, 2, 2, 2, 2]])
    >>> a.dtype
    dtype('int64')
Theo Steininger's avatar
Theo Steininger committed
103

104
105
106
107
108
    See Also
    --------
    distributed_data_object

    """
109

Theo Steininger's avatar
Theo Steininger committed
110
    # ---Initialization methods---
111

112
    def __init__(self, domain=None, val=None, dtype=None,
113
                 distribution_strategy=None, copy=False):
114
        self.domain = self._parse_domain(domain=domain, val=val)
115
        self.domain_axes = self._get_axes_tuple(self.domain)
csongor's avatar
csongor committed
116

Theo Steininger's avatar
Theo Steininger committed
117
        self.dtype = self._infer_dtype(dtype=dtype,
118
                                       val=val)
119

120
121
122
        self.distribution_strategy = self._parse_distribution_strategy(
                                distribution_strategy=distribution_strategy,
                                val=val)
csongor's avatar
csongor committed
123

124
125
126
127
        if val is None:
            self._val = None
        else:
            self.set_val(new_val=val, copy=copy)
csongor's avatar
csongor committed
128

129
    def _parse_domain(self, domain, val=None):
130
        if domain is None:
131
132
133
134
            if isinstance(val, Field):
                domain = val.domain
            else:
                domain = ()
135
        elif isinstance(domain, DomainObject):
136
            domain = (domain,)
137
138
139
        elif not isinstance(domain, tuple):
            domain = tuple(domain)

csongor's avatar
csongor committed
140
        for d in domain:
141
            if not isinstance(d, DomainObject):
142
143
                raise TypeError(
                    "Given domain contains something that is not a "
144
                    "DomainObject instance.")
csongor's avatar
csongor committed
145
146
        return domain

Theo Steininger's avatar
Theo Steininger committed
147
148
149
150
151
152
153
154
155
156
    def _get_axes_tuple(self, things_with_shape, start=0):
        i = start
        axes_list = []
        for thing in things_with_shape:
            l = []
            for j in range(len(thing.shape)):
                l += [i]
                i += 1
            axes_list += [tuple(l)]
        return tuple(axes_list)
157

158
    def _infer_dtype(self, dtype, val):
csongor's avatar
csongor committed
159
        if dtype is None:
160
            try:
161
                dtype = val.dtype
162
            except AttributeError:
Theo Steininger's avatar
Theo Steininger committed
163
164
165
                try:
                    if val is None:
                        raise TypeError
166
                    dtype = np.result_type(val)
Theo Steininger's avatar
Theo Steininger committed
167
                except(TypeError):
168
                    dtype = np.dtype(gc['default_field_dtype'])
Theo Steininger's avatar
Theo Steininger committed
169
        else:
170
            dtype = np.dtype(dtype)
171

172
173
        dtype = np.result_type(dtype, np.float)

Theo Steininger's avatar
Theo Steininger committed
174
        return dtype
175

176
177
    def _parse_distribution_strategy(self, distribution_strategy, val):
        if distribution_strategy is None:
178
            if isinstance(val, distributed_data_object):
179
                distribution_strategy = val.distribution_strategy
180
            elif isinstance(val, Field):
181
                distribution_strategy = val.distribution_strategy
182
            else:
183
                self.logger.debug("distribution_strategy set to default!")
184
                distribution_strategy = gc['default_distribution_strategy']
185
        elif distribution_strategy not in DISTRIBUTION_STRATEGIES['global']:
186
187
188
            raise ValueError(
                    "distribution_strategy must be a global-type "
                    "strategy.")
189
        return distribution_strategy
190
191

    # ---Factory methods---
192

193
    @classmethod
194
    def from_random(cls, random_type, domain=None, dtype=None,
195
                    distribution_strategy=None, **kwargs):
196
197
198
199
200
        """ Draws a random field with the given parameters.

        Parameters
        ----------
        cls : class
Theo Steininger's avatar
Theo Steininger committed
201

202
203
204
        random_type : String
            'pm1', 'normal', 'uniform' are the supported arguments for this
            method.
Theo Steininger's avatar
Theo Steininger committed
205

206
207
        domain : DomainObject
            The domain of the output random field
Theo Steininger's avatar
Theo Steininger committed
208

209
210
        dtype : type
            The datatype of the output random field
Theo Steininger's avatar
Theo Steininger committed
211

212
213
        distribution_strategy : all supported distribution strategies
            The distribution strategy of the output random field
Theo Steininger's avatar
Theo Steininger committed
214

215
216
217
218
219
220
221
        Returns
        -------
        out : Field
            The output object.

        See Also
        --------
222
        power_synthesize
Theo Steininger's avatar
Theo Steininger committed
223

224
225

        """
Theo Steininger's avatar
Theo Steininger committed
226

227
        # create a initially empty field
228
        f = cls(domain=domain, dtype=dtype,
229
                distribution_strategy=distribution_strategy)
230
231
232
233
234
235
236

        # now use the processed input in terms of f in order to parse the
        # random arguments
        random_arguments = cls._parse_random_arguments(random_type=random_type,
                                                       f=f,
                                                       **kwargs)

Martin Reinecke's avatar
Martin Reinecke committed
237
        # extract the distributed_data_object from f and apply the appropriate
238
239
240
        # random number generator to it
        sample = f.get_val(copy=False)
        generator_function = getattr(Random, random_type)
241
242
243
244
245
246
247
248
249

        comm = sample.comm
        size = comm.size
        if (sample.distribution_strategy in DISTRIBUTION_STRATEGIES['not'] and
                size > 1):
            seed = np.random.randint(10000000)
            seed = comm.bcast(seed, root=0)
            np.random.seed(seed)

250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
        sample.apply_generator(
            lambda shape: generator_function(dtype=f.dtype,
                                             shape=shape,
                                             **random_arguments))
        return f

    @staticmethod
    def _parse_random_arguments(random_type, f, **kwargs):
        if random_type == "pm1":
            random_arguments = {}

        elif random_type == "normal":
            mean = kwargs.get('mean', 0)
            std = kwargs.get('std', 1)
            random_arguments = {'mean': mean,
                                'std': std}

        elif random_type == "uniform":
            low = kwargs.get('low', 0)
            high = kwargs.get('high', 1)
            random_arguments = {'low': low,
                                'high': high}

csongor's avatar
csongor committed
273
        else:
274
275
            raise KeyError(
                "unsupported random key '" + str(random_type) + "'.")
csongor's avatar
csongor committed
276

277
        return random_arguments
csongor's avatar
csongor committed
278

279
280
    # ---Powerspectral methods---

Martin Reinecke's avatar
Martin Reinecke committed
281
    def power_analyze(self, spaces=None, logarithmic=None, nbin=None,
282
                      binbounds=None, keep_phase_information=False):
Theo Steininger's avatar
Theo Steininger committed
283
        """ Computes the square root power spectrum for a subspace of `self`.
Theo Steininger's avatar
Theo Steininger committed
284

Theo Steininger's avatar
Theo Steininger committed
285
286
287
        Creates a PowerSpace for the space addressed by `spaces` with the given
        binning and computes the power spectrum as a Field over this
        PowerSpace. This can only be done if the subspace to  be analyzed is a
288
        harmonic space. The resulting field has the same units as the initial
Theo Steininger's avatar
Theo Steininger committed
289
        field, corresponding to the square root of the power spectrum.
290
291
292

        Parameters
        ----------
Theo Steininger's avatar
Theo Steininger committed
293
294
295
296
297
        spaces : int *optional*
            The subspace for which the powerspectrum shall be computed
            (default : None).
        logarithmic : boolean *optional*
            True if the output PowerSpace should use logarithmic binning.
Martin Reinecke's avatar
Martin Reinecke committed
298
            {default : None}
Theo Steininger's avatar
Theo Steininger committed
299
300
301
302
303
304
        nbin : int *optional*
            The number of bins the resulting PowerSpace shall have
            (default : None).
            if nbin==None : maximum number of bins is used
        binbounds : array-like *optional*
            Inner bounds of the bins (default : None).
Martin Reinecke's avatar
Martin Reinecke committed
305
306
            Overrides nbin and logarithmic.
            if binbounds==None : bins are inferred.
307
308
309
310
311
312
313
314
315
316
        keep_phase_information : boolean, *optional*
            If False, return a real-valued result containing the power spectrum
            of the input Field.
            If True, return a complex-valued result whose real component
            contains the power spectrum computed from the real part of the
            input Field, and whose imaginary component contains the power
            spectrum computed from the imaginary part of the input Field.
            The absolute value of this result should be identical to the output
            of power_analyze with keep_phase_information=False.
            (default : False).
Theo Steininger's avatar
Theo Steininger committed
317

318
319
320
321
        Raise
        -----
        ValueError
            Raised if
Theo Steininger's avatar
Theo Steininger committed
322
323
                *len(domain) is != 1 when spaces==None
                *len(spaces) is != 1 if not None
324
                *the analyzed space is not harmonic
Theo Steininger's avatar
Theo Steininger committed
325

326
327
        Returns
        -------
Theo Steininger's avatar
Theo Steininger committed
328
        out : Field
329
330
331
332
333
334
            The output object. It's domain is a PowerSpace and it contains
            the power spectrum of 'self's field.

        See Also
        --------
        power_synthesize, PowerSpace
Theo Steininger's avatar
Theo Steininger committed
335

336
        """
Theo Steininger's avatar
Theo Steininger committed
337

Theo Steininger's avatar
Theo Steininger committed
338
        # check if all spaces in `self.domain` are either harmonic or
339
340
341
        # power_space instances
        for sp in self.domain:
            if not sp.harmonic and not isinstance(sp, PowerSpace):
Theo Steininger's avatar
Theo Steininger committed
342
                self.logger.info(
343
                    "Field has a space in `domain` which is neither "
344
345
346
                    "harmonic nor a PowerSpace.")

        # check if the `spaces` input is valid
347
348
        spaces = utilities.cast_axis_to_tuple(spaces, len(self.domain))
        if spaces is None:
349
            spaces = range(len(self.domain))
350
351

        if len(spaces) == 0:
352
353
            raise ValueError(
                "No space for analysis specified.")
354

355
356
357
358
359
360
361
362
363
364
365
366
367
        if keep_phase_information:
            parts_val = self._hermitian_decomposition(
                                              domain=self.domain,
                                              val=self.val,
                                              spaces=spaces,
                                              domain_axes=self.domain_axes,
                                              preserve_gaussian_variance=False)
            parts = [self.copy_empty().set_val(part_val, copy=False)
                     for part_val in parts_val]
        else:
            parts = [self]

        parts = [abs(part)**2 for part in parts]
368
369

        for space_index in spaces:
370
371
            parts = [self._single_power_analyze(
                                work_field=part,
372
373
374
                                space_index=space_index,
                                logarithmic=logarithmic,
                                nbin=nbin,
375
376
                                binbounds=binbounds)
                     for part in parts]
377

378
379
380
381
382
383
        if keep_phase_information:
            result_field = parts[0] + 1j*parts[1]
        else:
            result_field = parts[0]

        return result_field
384
385
386

    @classmethod
    def _single_power_analyze(cls, work_field, space_index, logarithmic, nbin,
387
                              binbounds):
388

389
        if not work_field.domain[space_index].harmonic:
390
391
            raise ValueError(
                "The analyzed space must be harmonic.")
392

393
394
395
396
397
398
        # Create the target PowerSpace instance:
        # If the associated signal-space field was real, we extract the
        # hermitian and anti-hermitian parts of `self` and put them
        # into the real and imaginary parts of the power spectrum.
        # If it was complex, all the power is put into a real power spectrum.

399
        distribution_strategy = \
400
401
            work_field.val.get_axes_local_distribution_strategy(
                work_field.domain_axes[space_index])
402

403
        harmonic_domain = work_field.domain[space_index]
404
        power_domain = PowerSpace(harmonic_partner=harmonic_domain,
405
                                  distribution_strategy=distribution_strategy,
Theo Steininger's avatar
Theo Steininger committed
406
407
                                  logarithmic=logarithmic, nbin=nbin,
                                  binbounds=binbounds)
408
409
        power_spectrum = cls._calculate_power_spectrum(
                                field_val=work_field.val,
Martin Reinecke's avatar
Martin Reinecke committed
410
                                pdomain=power_domain,
411
                                axes=work_field.domain_axes[space_index])
412
413

        # create the result field and put power_spectrum into it
414
        result_domain = list(work_field.domain)
415
        result_domain[space_index] = power_domain
416
        result_dtype = power_spectrum.dtype
417

418
        result_field = work_field.copy_empty(
419
                   domain=result_domain,
420
                   dtype=result_dtype,
421
                   distribution_strategy=power_spectrum.distribution_strategy)
422
423
424
425
        result_field.set_val(new_val=power_spectrum, copy=False)

        return result_field

426
    @classmethod
Martin Reinecke's avatar
Martin Reinecke committed
427
    def _calculate_power_spectrum(cls, field_val, pdomain, axes=None):
428

Martin Reinecke's avatar
Martin Reinecke committed
429
430
431
        pindex = pdomain.pindex
        # MR FIXME: how about iterating over slices, instead of replicating
        # pindex? Would save memory and probably isn't slower.
432
        if axes is not None:
433
434
435
436
437
            pindex = cls._shape_up_pindex(
                            pindex=pindex,
                            target_shape=field_val.shape,
                            target_strategy=field_val.distribution_strategy,
                            axes=axes)
Theo Steininger's avatar
Theo Steininger committed
438

439
        power_spectrum = pindex.bincount(weights=field_val,
440
                                         axis=axes)
Martin Reinecke's avatar
Martin Reinecke committed
441
        rho = pdomain.rho
442
443
444
445
446
447
448
449
        if axes is not None:
            new_rho_shape = [1, ] * len(power_spectrum.shape)
            new_rho_shape[axes[0]] = len(rho)
            rho = rho.reshape(new_rho_shape)
        power_spectrum /= rho

        return power_spectrum

450
451
    @staticmethod
    def _shape_up_pindex(pindex, target_shape, target_strategy, axes):
452
453
        if pindex.distribution_strategy not in \
                DISTRIBUTION_STRATEGIES['global']:
454
            raise ValueError("pindex's distribution strategy must be "
455
456
457
458
459
460
                             "global-type")

        if pindex.distribution_strategy in DISTRIBUTION_STRATEGIES['slicing']:
            if ((0 not in axes) or
                    (target_strategy is not pindex.distribution_strategy)):
                raise ValueError(
461
                    "A slicing distributor shall not be reshaped to "
462
463
                    "something non-sliced.")

Theo Steininger's avatar
Theo Steininger committed
464
        semiscaled_local_shape = [1, ] * len(target_shape)
Theo Steininger's avatar
Theo Steininger committed
465
466
        for i in range(len(axes)):
            semiscaled_local_shape[axes[i]] = pindex.local_shape[i]
467
        local_data = pindex.get_local_data(copy=False)
Theo Steininger's avatar
Theo Steininger committed
468
        semiscaled_local_data = local_data.reshape(semiscaled_local_shape)
469
470
        result_obj = pindex.copy_empty(global_shape=target_shape,
                                       distribution_strategy=target_strategy)
Theo Steininger's avatar
Theo Steininger committed
471
        result_obj.data[:] = semiscaled_local_data
472
473
474

        return result_obj

475
    def power_synthesize(self, spaces=None, real_power=True, real_signal=True,
476
                         mean=None, std=None, distribution_strategy=None):
Theo Steininger's avatar
Theo Steininger committed
477
        """ Yields a sampled field with `self`**2 as its power spectrum.
Theo Steininger's avatar
Theo Steininger committed
478

Theo Steininger's avatar
Theo Steininger committed
479
480
        This method draws a Gaussian random field in the harmonic partner
        domain of this fields domains, using this field as power spectrum.
Theo Steininger's avatar
Theo Steininger committed
481

482
483
484
        Parameters
        ----------
        spaces : {tuple, int, None} *optional*
Theo Steininger's avatar
Theo Steininger committed
485
486
487
            Specifies the subspace containing all the PowerSpaces which
            should be converted (default : None).
            if spaces==None : Tries to convert the whole domain.
488
        real_power : boolean *optional*
Theo Steininger's avatar
Theo Steininger committed
489
490
            Determines whether the power spectrum is treated as intrinsically
            real or complex (default : True).
491
        real_signal : boolean *optional*
Theo Steininger's avatar
Theo Steininger committed
492
493
494
495
496
497
            True will result in a purely real signal-space field
            (default : True).
        mean : float *optional*
            The mean of the Gaussian noise field which is used for the Field
            synthetization (default : None).
            if mean==None : mean will be set to 0
498
        std : float *optional*
Theo Steininger's avatar
Theo Steininger committed
499
500
501
            The standard deviation of the Gaussian noise field which is used
            for the Field synthetization (default : None).
            if std==None : std will be set to 1
Theo Steininger's avatar
Theo Steininger committed
502

503
504
505
506
        Returns
        -------
        out : Field
            The output object. A random field created with the power spectrum
Theo Steininger's avatar
Theo Steininger committed
507
            stored in the `spaces` in `self`.
508

Theo Steininger's avatar
Theo Steininger committed
509
510
511
512
513
514
        Notes
        -----
        For this the spaces specified by `spaces` must be a PowerSpace.
        This expects this field to be the square root of a power spectrum, i.e.
        to have the unit of the field to be sampled.

515
516
517
        See Also
        --------
        power_analyze
Theo Steininger's avatar
Theo Steininger committed
518
519
520
521
522

        Raises
        ------
        ValueError : If domain specified by `spaces` is not a PowerSpace.

523
        """
Theo Steininger's avatar
Theo Steininger committed
524

525
526
527
        # check if the `spaces` input is valid
        spaces = utilities.cast_axis_to_tuple(spaces, len(self.domain))

Theo Steininger's avatar
Theo Steininger committed
528
529
530
        if spaces is None:
            spaces = range(len(self.domain))

531
532
533
534
535
        for power_space_index in spaces:
            power_space = self.domain[power_space_index]
            if not isinstance(power_space, PowerSpace):
                raise ValueError("A PowerSpace is needed for field "
                                 "synthetization.")
536
537
538

        # create the result domain
        result_domain = list(self.domain)
539
540
        for power_space_index in spaces:
            power_space = self.domain[power_space_index]
541
            harmonic_domain = power_space.harmonic_partner
542
            result_domain[power_space_index] = harmonic_domain
543
544
545

        # create random samples: one or two, depending on whether the
        # power spectrum is real or complex
546
        if real_power:
547
            result_list = [None]
548
549
        else:
            result_list = [None, None]
550

551
552
553
        if distribution_strategy is None:
            distribution_strategy = gc['default_distribution_strategy']

554
555
        result_list = [self.__class__.from_random(
                             'normal',
556
557
558
                             mean=mean,
                             std=std,
                             domain=result_domain,
559
                             dtype=np.complex,
560
                             distribution_strategy=distribution_strategy)
561
562
563
564
565
566
                       for x in result_list]

        # from now on extract the values from the random fields for further
        # processing without killing the fields.
        # if the signal-space field should be real, hermitianize the field
        # components
567
568

        spec = self.val.get_full_data()
569
570
        spec = np.sqrt(spec)

571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
        for power_space_index in spaces:
            spec = self._spec_to_rescaler(spec, result_list, power_space_index)
        local_rescaler = spec

        result_val_list = [x.val for x in result_list]

        # apply the rescaler to the random fields
        result_val_list[0].apply_scalar_function(
                                            lambda x: x * local_rescaler.real,
                                            inplace=True)

        if not real_power:
            result_val_list[1].apply_scalar_function(
                                            lambda x: x * local_rescaler.imag,
                                            inplace=True)

587
        if real_signal:
588
            result_val_list = [self._hermitian_decomposition(
589
590
591
592
593
                                            result_domain,
                                            result_val,
                                            spaces,
                                            result_list[0].domain_axes,
                                            preserve_gaussian_variance=True)[0]
594
                               for result_val in result_val_list]
595
596
597
598
599
600
601

        # store the result into the fields
        [x.set_val(new_val=y, copy=False) for x, y in
            zip(result_list, result_val_list)]

        if real_power:
            result = result_list[0]
602
603
604
            if not issubclass(result_val_list[0].dtype.type,
                              np.complexfloating):
                result = result.real
605
        else:
606
607
608
609
            result = result_list[0] + 1j*result_list[1]

        return result

610
    @staticmethod
611
612
    def _hermitian_decomposition(domain, val, spaces, domain_axes,
                                 preserve_gaussian_variance=False):
613
614
615
616
617
618

        flipped_val = val
        for space in spaces:
            flipped_val = domain[space].hermitianize_inverter(
                                                    x=flipped_val,
                                                    axes=domain_axes[space])
619
620
        # if no flips at all where performed `h` is a real field.
        # if all spaces use the default implementation of doing nothing when
Theo Steininger's avatar
Theo Steininger committed
621
        # no flips are applied, one can use `is` to infer this case.
622
623
624
625
626
627
628
629

        if flipped_val is val:
            h = flipped_val.real
            a = 1j * flipped_val.imag
        else:
            flipped_val = flipped_val.conjugate()
            h = (val + flipped_val)/2.
            a = val - h
630
631

        # correct variance
632
        if preserve_gaussian_variance:
Martin Reinecke's avatar
Martin Reinecke committed
633
634
            assert issubclass(val.dtype.type, np.complexfloating),\
                    "complex input field is needed here"
635
636
637
            h *= np.sqrt(2)
            a *= np.sqrt(2)

638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
#            The code below should not be needed in practice, since it would
#            only ever be called when hermitianizing a purely real field.
#            However it might be of educational use and keep us from forgetting
#            how these things are done ...

#            if not issubclass(val.dtype.type, np.complexfloating):
#                # in principle one must not correct the variance for the fixed
#                # points of the hermitianization. However, for a complex field
#                # the input field loses half of its power at its fixed points
#                # in the `hermitian` part. Hence, here a factor of sqrt(2) is
#                # also necessary!
#                # => The hermitianization can be done on a space level since
#                # either nothing must be done (LMSpace) or ALL points need a
#                # factor of sqrt(2)
#                # => use the preserve_gaussian_variance flag in the
#                # hermitian_decomposition method above.
#
#                # This code is for educational purposes:
#                fixed_points = [domain[i].hermitian_fixed_points()
#                                for i in spaces]
#                fixed_points = [[fp] if fp is None else fp
#                                for fp in fixed_points]
#
#                for product_point in itertools.product(*fixed_points):
#                    slice_object = np.array((slice(None), )*len(val.shape),
#                                            dtype=np.object)
#                    for i, sp in enumerate(spaces):
#                        point_component = product_point[i]
#                        if point_component is None:
#                            point_component = slice(None)
#                        slice_object[list(domain_axes[sp])] = point_component
#
#                    slice_object = tuple(slice_object)
#                    h[slice_object] /= np.sqrt(2)
#                    a[slice_object] /= np.sqrt(2)

674
675
        return (h, a)

676
677
    def _spec_to_rescaler(self, spec, result_list, power_space_index):
        power_space = self.domain[power_space_index]
678
679
680

        # weight the random fields with the power spectrum
        # therefore get the pindex from the power space
681
        pindex = power_space.pindex
682
683
684
685
686
687
688
        # take the local data from pindex. This data must be compatible to the
        # local data of the field given the slice of the PowerSpace
        local_distribution_strategy = \
            result_list[0].val.get_axes_local_distribution_strategy(
                result_list[0].domain_axes[power_space_index])

        if pindex.distribution_strategy is not local_distribution_strategy:
689
            raise AttributeError(
Martin Reinecke's avatar
Martin Reinecke committed
690
                "The distribution_strategy of pindex does not fit the "
691
692
693
694
695
696
697
                "slice_local distribution strategy of the synthesized field.")

        # Now use numpy advanced indexing in order to put the entries of the
        # power spectrum into the appropriate places of the pindex array.
        # Do this for every 'pindex-slice' in parallel using the 'slice(None)'s
        local_pindex = pindex.get_local_data(copy=False)

698
699
700
701
702
        local_blow_up = [slice(None)]*len(spec.shape)
        # it is important to count from behind, since spec potentially grows
        # with every iteration
        index = self.domain_axes[power_space_index][0]-len(self.shape)
        local_blow_up[index] = local_pindex
703
        # here, the power_spectrum is distributed into the new shape
704
705
        local_rescaler = spec[local_blow_up]
        return local_rescaler
706

Theo Steininger's avatar
Theo Steininger committed
707
    # ---Properties---
708

Theo Steininger's avatar
Theo Steininger committed
709
    def set_val(self, new_val=None, copy=False):
Theo Steininger's avatar
Theo Steininger committed
710
        """ Sets the fields distributed_data_object.
711
712
713

        Parameters
        ----------
Theo Steininger's avatar
Theo Steininger committed
714
        new_val : scalar, array-like, Field, None *optional*
715
716
            The values to be stored in the field.
            {default : None}
Theo Steininger's avatar
Theo Steininger committed
717

718
        copy : boolean, *optional*
Theo Steininger's avatar
Theo Steininger committed
719
720
            If False, Field tries to not copy the input data but use it
            directly.
721
722
723
724
725
726
            {default : False}
        See Also
        --------
        val

        """
Theo Steininger's avatar
Theo Steininger committed
727

728
729
        new_val = self.cast(new_val)
        if copy:
Theo Steininger's avatar
Theo Steininger committed
730
731
            new_val = new_val.copy()
        self._val = new_val
732
        return self
csongor's avatar
csongor committed
733

734
    def get_val(self, copy=False):
Theo Steininger's avatar
Theo Steininger committed
735
        """ Returns the distributed_data_object associated with this Field.
736
737
738
739

        Parameters
        ----------
        copy : boolean
Theo Steininger's avatar
Theo Steininger committed
740
741
            If true, a copy of the Field's underlying distributed_data_object
            is returned.
Theo Steininger's avatar
Theo Steininger committed
742

743
744
745
746
747
748
749
750
751
        Returns
        -------
        out : distributed_data_object

        See Also
        --------
        val

        """
Theo Steininger's avatar
Theo Steininger committed
752

753
754
755
        if self._val is None:
            self.set_val(None)

756
        if copy:
Theo Steininger's avatar
Theo Steininger committed
757
            return self._val.copy()
758
        else:
Theo Steininger's avatar
Theo Steininger committed
759
            return self._val
csongor's avatar
csongor committed
760

Theo Steininger's avatar
Theo Steininger committed
761
762
    @property
    def val(self):
Theo Steininger's avatar
Theo Steininger committed
763
        """ Returns the distributed_data_object associated with this Field.
Theo Steininger's avatar
Theo Steininger committed
764

765
766
767
768
769
770
771
772
773
        Returns
        -------
        out : distributed_data_object

        See Also
        --------
        get_val

        """
Theo Steininger's avatar
Theo Steininger committed
774

775
        return self.get_val(copy=False)
csongor's avatar
csongor committed
776

Theo Steininger's avatar
Theo Steininger committed
777
778
    @val.setter
    def val(self, new_val):
779
        self.set_val(new_val=new_val, copy=False)
csongor's avatar
csongor committed
780

781
782
    @property
    def shape(self):
Theo Steininger's avatar
Theo Steininger committed
783
        """ Returns the total shape of the Field's data array.
Theo Steininger's avatar
Theo Steininger committed
784

785
786
787
        Returns
        -------
        out : tuple
Martin Reinecke's avatar
Martin Reinecke committed
788
            The output object. The tuple contains the dimensions of the spaces
789
790
791
792
793
794
795
            in domain.

        See Also
        --------
        dim

        """
Theo Steininger's avatar
Theo Steininger committed
796
797
798
799
800
801
802
803
        if not hasattr(self, '_shape'):
            shape_tuple = tuple(sp.shape for sp in self.domain)
            try:
                global_shape = reduce(lambda x, y: x + y, shape_tuple)
            except TypeError:
                global_shape = ()
            self._shape = global_shape
        return self._shape
csongor's avatar
csongor committed
804

805
806
    @property
    def dim(self):
Theo Steininger's avatar
Theo Steininger committed
807
        """ Returns the total number of pixel-dimensions the field has.
Theo Steininger's avatar
Theo Steininger committed
808

Theo Steininger's avatar
Theo Steininger committed
809
        Effectively, all values from shape are multiplied.
Theo Steininger's avatar
Theo Steininger committed
810

811
812
813
814
815
816
817
818
819
820
        Returns
        -------
        out : int
            The dimension of the Field.

        See Also
        --------
        shape

        """
Theo Steininger's avatar
Theo Steininger committed
821

822
        dim_tuple = tuple(sp.dim for sp in self.domain)
Theo Steininger's avatar
Theo Steininger committed
823
824
825
826
        try:
            return reduce(lambda x, y: x * y, dim_tuple)
        except TypeError:
            return 0
csongor's avatar
csongor committed
827

828
829
    @property
    def dof(self):
Theo Steininger's avatar
Theo Steininger committed
830
831
832
833
834
835
        """ Returns the total number of degrees of freedom the Field has. For
        real Fields this is equal to `self.dim`. For complex Fields it is
        2*`self.dim`.

        """

Theo Steininger's avatar
Theo Steininger committed
836
837
838
839
840
841
842
        dof = self.dim
        if issubclass(self.dtype.type, np.complexfloating):
            dof *= 2
        return dof

    @property
    def total_volume(self):
Theo Steininger's avatar
Theo Steininger committed
843
844
845
        """ Returns the total volume of all spaces in the domain.
        """

Theo Steininger's avatar
Theo Steininger committed
846
        volume_tuple = tuple(sp.total_volume for sp in self.domain)
847
        try:
Theo Steininger's avatar
Theo Steininger committed
848
            return reduce(lambda x, y: x * y, volume_tuple)
849
        except TypeError:
Theo Steininger's avatar
Theo Steininger committed
850
            return 0.
851

Theo Steininger's avatar
Theo Steininger committed
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
    @property
    def real(self):
        """ The real part of the field (data is not copied).
        """
        real_part = self.val.real
        result = self.copy_empty(dtype=real_part.dtype)
        result.set_val(new_val=real_part, copy=False)
        return result

    @property
    def imag(self):
        """ The imaginary part of the field (data is not copied).
        """
        real_part = self.val.imag
        result = self.copy_empty(dtype=real_part.dtype)
        result.set_val(new_val=real_part, copy=False)
        return result

Theo Steininger's avatar
Theo Steininger committed
870
    # ---Special unary/binary operations---
871

csongor's avatar
csongor committed
872
    def cast(self, x=None, dtype=None):
Theo Steininger's avatar
Theo Steininger committed
873
        """ Transforms x to a d2o with the correct dtype and shape.
Theo Steininger's avatar
Theo Steininger committed
874

875
876
        Parameters
        ----------
Theo Steininger's avatar
Theo Steininger committed
877
        x : scalar, d2o, Field, array_like
878
879
            The input that shall be casted on a d2o of the same shape like the
            domain.
Theo Steininger's avatar
Theo Steininger committed
880

881
        dtype : type
Theo Steininger's avatar
Theo Steininger committed
882
883
            The datatype the output shall have. This can be used to override
            the fields dtype.
Theo Steininger's avatar
Theo Steininger committed
884

885
886
887
888
889
890
891
892
893
894
        Returns
        -------
        out : distributed_data_object
            The output object.

        See Also
        --------
        _actual_cast

        """
csongor's avatar
csongor committed
895
896
        if dtype is None:
            dtype = self.dtype
897
898
        else:
            dtype = np.dtype(dtype)
899

900
901
        casted_x = x

902
        for ind, sp in enumerate(self.domain):
903
            casted_x = sp.pre_cast(casted_x,
904
905
906
                                   axes=self.domain_axes[ind])

        casted_x = self._actual_cast(casted_x, dtype=dtype)
907
908

        for ind, sp in enumerate(self.domain):
909
910
            casted_x = sp.post_cast(casted_x,
                                    axes=self.domain_axes[ind])
911

912
        return casted_x
csongor's avatar
csongor committed
913

Theo Steininger's avatar
Theo Steininger committed
914
    def _actual_cast(self, x, dtype=None):
915
        if isinstance(x, Field):
csongor's avatar
csongor committed
916
917
918
919
920
            x = x.get_val()

        if dtype is None:
            dtype = self.dtype

921
        return_x = distributed_data_object(
922
923
924
                            global_shape=self.shape,
                            dtype=dtype,
                            distribution_strategy=self.distribution_strategy)
925
926
        return_x.set_full_data(x, copy=False)
        return return_x
Theo Steininger's avatar
Theo Steininger committed
927

928
    def copy(self, domain=None, dtype=None, distribution_strategy=None):
929
        """ Returns a full copy of the Field.
Theo Steininger's avatar
Theo Steininger committed
930

931
932
933
934
935
936
937
938
939
        If no keyword arguments are given, the returned object will be an
        identical copy of the original Field. By explicit specification one is
        able to define the domain, the dtype and the distribution_strategy of
        the returned Field.

        Parameters
        ----------
        domain : DomainObject
            The new domain the Field shall have.
Theo Steininger's avatar
Theo Steininger committed
940

941
942
        dtype : type
            The new dtype the Field shall have.
Theo Steininger's avatar
Theo Steininger committed
943

944
        distribution_strategy : all supported distribution strategies
Theo Steininger's avatar
Theo Steininger committed
945
946
            The new distribution strategy the Field shall have.

947
948
949
950
951
952
953
954
955
956
        Returns
        -------
        out : Field
            The output object. An identical copy of 'self'.

        See Also
        --------
        copy_empty

        """
Theo Steininger's avatar
Theo Steininger committed
957

Theo Steininger's avatar
Theo Steininger committed
958
        copied_val = self.get_val(copy=True)
959
960
961
962
        new_field = self.copy_empty(
                                domain=domain,
                                dtype=dtype,
                                distribution_strategy=distribution_strategy)
Theo Steininger's avatar
Theo Steininger committed
963
964
        new_field.set_val(new_val=copied_val, copy=False)
        return new_field
csongor's avatar
csongor committed
965

966
    def copy_empty(self, domain=None, dtype=None, distribution_strategy=None):
967
968
969
        """ Returns an empty copy of the Field.

        If no keyword arguments are given, the returned object will be an
Theo Steininger's avatar
Theo Steininger committed
970
971
972
973
974
        identical copy of the original Field. The memory for the data array
        is only allocated but not actively set to any value
        (c.f. numpy.ndarray.copy_empty). By explicit specification one is able
        to change the domain, the dtype and the distribution_strategy of the
        returned Field.
Theo Steininger's avatar
Theo Steininger committed
975

976
977
978
979
        Parameters
        ----------
        domain : DomainObject
            The new domain the Field shall have.
Theo Steininger's avatar
Theo Steininger committed
980

981
982
        dtype : type
            The new dtype the Field shall have.
Theo Steininger's avatar
Theo Steininger committed
983

Theo Steininger's avatar
Theo Steininger committed
984
        distribution_strategy : string, all supported distribution strategies
985
            The distribution strategy the new Field should have.
Theo Steininger's avatar
Theo Steininger committed
986

987
988
989
        Returns
        -------
        out : Field
Theo Steininger's avatar
Theo Steininger committed
990
            The output object.
991
992
993