nifty_mpi_data.py 132 KB
Newer Older
ultimanet's avatar
ultimanet committed
1
# -*- coding: utf-8 -*-
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
# NIFTY (Numerical Information Field Theory) has been developed at the
# Max-Planck-Institute for Astrophysics.
#
# Copyright (C) 2015 Max-Planck-Society
#
# Author: Theo Steininger
# Project homepage: <http://www.mpa-garching.mpg.de/ift/nifty/>
#
# 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/>.
22
23


ultimanet's avatar
ultimanet committed
24
import numpy as np
25
from weakref import WeakValueDictionary as weakdict
ultimanet's avatar
ultimanet committed
26

27
28
29
from keepers import about,\
                    global_configuration as gc,\
                    global_dependency_injector as gdi
ultimanet's avatar
ultimanet committed
30

31
32
33
MPI = gdi[gc['mpi_module']]
h5py = gdi.get('h5py')
pyfftw = gdi.get('pyfftw')
ultimanet's avatar
ultimanet committed
34

35
36
37
38
39
40
41
42
43
_maybe_fftw = ['fftw'] if ('pyfftw' in gdi) else []
STRATEGIES = {
                'all': ['not', 'equal', 'freeform'] + _maybe_fftw,
                'global': ['not', 'equal'] + _maybe_fftw,
                'local': ['freeform'],
                'slicing': ['equal', 'freeform'] + _maybe_fftw,
                'not': ['not'],
                'hdf5': ['equal'] + _maybe_fftw,
             }
Ultima's avatar
Ultima committed
44
45
46
47
if _maybe_fftw != []:
    _default_strategy = 'fftw'
else:
    _default_strategy = 'equal'
48

49

ultimanet's avatar
ultimanet committed
50
class distributed_data_object(object):
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
    """A multidimensional array with modular MPI-based distribution schemes.

    The purpose of a distributed_data_object (d2o) is to provide the user
    with a numpy.ndarray like interface while storing the data on an arbitrary
    number of MPI nodes. The logic of a certain distribution strategy is
    implemented by an associated distributor.

    Parameters
    ----------
    global_data : array-like, at least 1-dimensional
        Used with global-type distribution strategies in order to fill the
        d2o with data during initialization.
    global_shape : tuple of ints
        Used with global-type distribution strategies. If no global_data is
        supplied, it will be used.
    dtype : {np.dtype, type}
        Used as the d2o's datatype. Overwrites the data-type of any init data.
    local_data : array-like, at least 1-dimensional
        Used with local-type distribution strategies in order to fill the
        d2o with data during initialization.
    local_shape : tuple of ints
        Used with local-type distribution strategies. If no local_data is
        supplied, local_shape will be used.
    distribution_strategy : {'fftw', 'equal', 'not', 'freeform'}, optional
        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
    hermitian : boolean
        Specifies if the given init-data is hermitian or not. The
        self.hermitian attribute will be set accordingly.
    alias : String
        Used in order to initialize the d2o from a hdf5 file.
    path : String
        Used in order to initialize the d2o from a hdf5 file. If no path is
        given, '$working_directory/alias' is used.
    comm : mpi4py.MPI.Intracomm
        The MPI communicator on which the d2o lives.
    copy : boolean
        If true it is guaranteed that the input data will be copied. If false
        copying is tried to be avoided.
    *args
        Although not directly used during the init process, further parameters
        are stored in the self.init_args attribute.
    **kwargs
        Additional keyword arguments are passed to the distributor_factory and
        furthermore get stored in the self.init_kwargs attribute.
    skip_parsing : boolean (optional keyword argument)
        If true, the distribution_factory will skip all sanity checks and
        completions of the given (keyword-)arguments. It just uses what it
        gets. Hence the user is fully responsible for supplying complete and
        consistent parameters. This can be used in order to speed up the init
        process. Also see notes section.

    Attributes
    ----------
    data : numpy.ndarray
        The numpy.ndarray in which the individual node's data is stored.
    dtype : type
        Data type of the data object.
    distribution_strategy : string
        Name of the used distribution_strategy.
    distributor : distributor
        The distributor object which takes care of all distribution and
        consolidation of the data.
    shape : tuple of int
        The global shape of the data.
    local_shape : tuple of int
        The nodes individual local shape of the stored data.
    comm : mpi4py.MPI.Intracomm
        The MPI communicator on which the d2o lives.
    hermitian : boolean
        Specfies whether the d2o's data definitely possesses hermitian
        symmetry.
    index : int
        The d2o's registration index it got from the d2o_librarian.
    init_args : list
        Any additional initialization arguments are stored here.
    init_kwargs : dict
        Any additional initialization keyword arguments are stored here.

    Raises
    ------
    ValueError
        Raised if
            * the supplied distribution strategy is not known,
            * comm is None,
            * different distribution strategies where given on the
              individual nodes,
            * different dtypes where given on the individual nodes,
            * neither a non-0-dimensional global_data nor global_shape nor
              hdf5 file supplied,
            * global_shape == (),
            * different global_shapes where given on the individual nodes,
            * neither non-0-dimensional local_data nor local_shape nor
              global d2o supplied,
            * local_shape == ()
            * the first entry of local_shape is not the same on all nodes,

    Notes
    -----
    The index is the d2o's global unique indentifier. One may use it in order
    to assemble the corresponding local d2o objects on different nodes if
    only one local object on a specific node is given.

    In order to speed up the init process the distributor_factory checks
    if the global_configuration object gc yields gc['d2o_init_checks'] == True.
    If yes, all checks expensive checks are skipped; namely those which  need
    mpi communication. Use this in order to get a fast init speed without
    loosing d2o's init parsing logic.

    Examples
    --------
    >>> a = np.arange(16, dtype=np.float).reshape((4,4))
    >>> obj = distributed_data_object(a, dtype=np.complex)
    >>> obj
    <distributed_data_object>
    array([[  0.+0.j,   1.+0.j,   2.+0.j,   3.+0.j],
           [  4.+0.j,   5.+0.j,   6.+0.j,   7.+0.j],
           [  8.+0.j,   9.+0.j,  10.+0.j,  11.+0.j],
           [ 12.+0.j,  13.+0.j,  14.+0.j,  15.+0.j]])
ultimanet's avatar
ultimanet committed
173

174

ultimanet's avatar
ultimanet committed
175
    """
176
    def __init__(self, global_data=None, global_shape=None, dtype=None,
Ultima's avatar
Ultima committed
177
                 local_data=None, local_shape=None,
Ultima's avatar
Ultima committed
178
                 distribution_strategy=_default_strategy, hermitian=False,
179
180
181
182
183
                 alias=None, path=None, comm=MPI.COMM_WORLD,
                 copy=True, *args, **kwargs):

        # TODO: allow init with empty shape

184
185
186
187
        if isinstance(global_data, tuple) or isinstance(global_data, list):
            global_data = np.array(global_data, copy=False)
        if isinstance(local_data, tuple) or isinstance(local_data, list):
            local_data = np.array(local_data, copy=False)
188

189
        self.distributor = distributor_factory.get_distributor(
190
191
192
193
194
195
196
197
198
                                distribution_strategy=distribution_strategy,
                                comm=comm,
                                global_data=global_data,
                                global_shape=global_shape,
                                local_data=local_data,
                                local_shape=local_shape,
                                alias=alias,
                                path=path,
                                dtype=dtype,
199
                                **kwargs)
200

ultimanet's avatar
ultimanet committed
201
202
203
        self.distribution_strategy = distribution_strategy
        self.dtype = self.distributor.dtype
        self.shape = self.distributor.global_shape
Ultima's avatar
Ultima committed
204
205
        self.local_shape = self.distributor.local_shape
        self.comm = self.distributor.comm
206
207

        self.init_args = args
208
        self.init_kwargs = kwargs
209

Ultima's avatar
Ultima committed
210
        (self.data, self.hermitian) = self.distributor.initialize_data(
211
212
213
214
215
216
            global_data=global_data,
            local_data=local_data,
            alias=alias,
            path=path,
            hermitian=hermitian,
            copy=copy)
217
        self.index = d2o_librarian.register(self)
218

219
220
    @property
    def real(self):
221
222
223
224
225
226
227
228
229
        """ Returns a d2o containing the real part of the d2o's elements.

        Returns
        -------
        out : distributed_data_object
            The output object. The new datatype is the one numpy yields when
            taking the real part on the local data.
        """

230
231
232
233
234
235
236
237
238
        new_data = self.get_local_data().real
        new_dtype = new_data.dtype
        new_d2o = self.copy_empty(dtype=new_dtype)
        new_d2o.set_local_data(data=new_data,
                               hermitian=self.hermitian)
        return new_d2o

    @property
    def imag(self):
239
240
241
242
243
244
245
246
247
        """ Returns a d2o containing the imaginary part of the d2o's elements.

        Returns
        -------
        out : distributed_data_object
            The output object. The new datatype is the one numpy yields when
            taking the imaginary part on the local data.
        """

248
249
250
251
252
253
254
        new_data = self.get_local_data().imag
        new_dtype = new_data.dtype
        new_d2o = self.copy_empty(dtype=new_dtype)
        new_d2o.set_local_data(data=new_data,
                               hermitian=self.hermitian)
        return new_d2o

255
    def _fast_copy_empty(self):
256
257
258
259
260
261
        """ Make a very fast low level copy of the d2o without its data.

        This function is fast, because it uses EmptyD2o - a derived class from
        distributed_data_object and then copies the __dict__ directly. Unlike
        copy_empty, _fast_copy_empty will copy all attributes unchanged.
        """
262
263
264
265
266
267
268
269
270
271
        # make an empty d2o
        new_copy = EmptyD2o()
        # repair its class
        new_copy.__class__ = self.__class__
        # now copy everthing in the __dict__ except for the data array
        for key, value in self.__dict__.items():
            if key != 'data':
                new_copy.__dict__[key] = value
            else:
                new_copy.__dict__[key] = np.empty_like(value)
272
        # Register the new d2o at the librarian in order to get a unique index
273
        new_copy.index = d2o_librarian.register(new_copy)
274
275
        return new_copy

Ultimanet's avatar
Ultimanet committed
276
    def copy(self, dtype=None, distribution_strategy=None, **kwargs):
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
        """ Returns a full copy of the distributed data object.

        If no keyword arguments are given, the returned object will be an
        identical copy of the original d2o. By explicit specification one is
        able to define the dtype and the distribution_strategy of the returned
        d2o.

        Parameters
        ----------
        dtype : type
            The dtype that the new d2o will have. The data of the primary
            d2o will be casted.
        distribution_strategy : all supported distribution strategies
            The distribution strategy the new d2o should have. If not None and
            different from the original one, there will certainly be inter-node
            communication.
        **kwargs
            Additional keyword arguments get passed to the used copy_empty
            routine.

        Returns
        -------
        out : distributed_data_object
            The output object. It containes the old data, possibly casted to a
            new datatype and distributed according to a new distribution
            strategy

        See Also
        --------
        copy_empty

        """
309
310
311
        temp_d2o = self.copy_empty(dtype=dtype,
                                   distribution_strategy=distribution_strategy,
                                   **kwargs)
Ultima's avatar
Ultima committed
312
        if distribution_strategy is None or \
313
                distribution_strategy == self.distribution_strategy:
Ultimanet's avatar
Ultimanet committed
314
315
            temp_d2o.set_local_data(self.get_local_data(), copy=True)
        else:
Ultima's avatar
Ultima committed
316
            temp_d2o.inject((slice(None),), self, (slice(None),))
317
        temp_d2o.hermitian = self.hermitian
318
        return temp_d2o
319
320

    def copy_empty(self, global_shape=None, local_shape=None, dtype=None,
321
                   distribution_strategy=None, **kwargs):
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
        """ Returns an empty copy of the distributed data object.

        If no keyword arguments are given, the returned object will be an
        identical copy of the original d2o containing random data. By explicit
        specification one is able to define the new dtype and
        distribution_strategy of the returned d2o and to modify the new shape.

        Parameters
        ----------
        global_shape : tuple of ints
            The global shape that the new d2o shall have. Relevant for
            global-type distribution strategies like 'equal' or 'fftw'.
        local_shape : tuple of ints
            The local shape that the new d2o shall have. Relevant for
            local-type distribution strategies like 'freeform'.
        dtype : type
            The dtype that the new d2o will have.
        distribution_strategy : all supported distribution strategies
            The distribution strategy the new d2o should have.
        **kwargs
            Additional keyword arguments get passed to the init-call if the
            full initialization of a new distributed_data_object is necessary

        Returns
        -------
        out : distributed_data_object
            The output object. It contains random data.

        See Also
        --------
        copy

        """
Ultima's avatar
Ultima committed
355
        if self.distribution_strategy == 'not' and \
356
                distribution_strategy in STRATEGIES['local'] and \
357
358
359
360
361
                local_shape is None:
            result = self.copy_empty(global_shape=global_shape,
                                     local_shape=local_shape,
                                     dtype=dtype,
                                     distribution_strategy='equal',
Ultima's avatar
Ultima committed
362
                                     **kwargs)
363
364
            return result.copy_empty(
                distribution_strategy=distribution_strategy)
365

Ultima's avatar
Ultima committed
366
        if global_shape is None:
367
            global_shape = self.shape
Ultima's avatar
Ultima committed
368
369
370
        if local_shape is None:
            local_shape = self.local_shape
        if dtype is None:
371
            dtype = self.dtype
372
373
        else:
            dtype = np.dtype(dtype)
Ultima's avatar
Ultima committed
374
        if distribution_strategy is None:
375
376
            distribution_strategy = self.distribution_strategy

377
378
379
380
381
382
383
384
        # check if all parameters remain the same -> use the _fast_copy_empty
        if (global_shape == self.shape and
                local_shape == self.local_shape and
                dtype == self.dtype and
                distribution_strategy == self.distribution_strategy and
                kwargs == self.init_kwargs):
            return self._fast_copy_empty()

385
        kwargs.update(self.init_kwargs)
386

387
388
389
390
391
392
393
394
        temp_d2o = distributed_data_object(
                                   global_shape=global_shape,
                                   local_shape=local_shape,
                                   dtype=dtype,
                                   distribution_strategy=distribution_strategy,
                                   comm=self.comm,
                                   *self.init_args,
                                   **kwargs)
395
        return temp_d2o
396

397
    def apply_scalar_function(self, function, inplace=False, dtype=None):
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
        """ Maps a scalar function on each entry of an array.

        The result of the function evaluation may be stored in the original
        array or in a new array (default). Furthermore the dtype of the
        returned array can be specified explicitly if inplace is set to False.

        Parameters
        ----------
        function : callable
            Will be applied to the array's entries. It will be the node's local
            data array into function as a whole. If this fails, the numpy
            vectorize function will be used.
        inplace : boolean
            Specifies if the result of the function evaluation should be stored
            in the original array or not.
        dtype : type
            If inplace is set to False, it is possible to specify the return
            d2o's dtype explicitly.

        Returns
        -------
        out : distributed_data_object
            Resulting d2o. This is either a newly created array or the primary
            d2o itself.
        """
423
        remember_hermitianQ = self.hermitian
424

425
        if inplace is True:
Ultimanet's avatar
Ultimanet committed
426
            temp = self
Ultima's avatar
Ultima committed
427
            if dtype is not None and self.dtype != np.dtype(dtype):
428
429
430
                about.warnings.cprint(
                    "WARNING: Inplace dtype conversion is not possible!")

Ultimanet's avatar
Ultimanet committed
431
        else:
432
            temp = self.copy_empty(dtype=dtype)
Ultimanet's avatar
Ultimanet committed
433

Ultima's avatar
Ultima committed
434
        if np.prod(self.local_shape) != 0:
435
            try:
Ultima's avatar
Ultima committed
436
437
                temp.data[:] = function(self.data)
            except:
438
439
                about.warnings.cprint(
                    "WARNING: Trying to use np.vectorize!")
Ultima's avatar
Ultima committed
440
441
                temp.data[:] = np.vectorize(function)(self.data)
        else:
442
443
            # Noting to do here. The value-empty array
            # is also geometrically empty
Ultima's avatar
Ultima committed
444
            pass
445

446
447
448
449
        if function in (np.exp, np.log):
            temp.hermitian = remember_hermitianQ
        else:
            temp.hermitian = False
Ultimanet's avatar
Ultimanet committed
450
        return temp
451

452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
    def apply_generator(self, generator, copy=False):
        """ Evaluates generator(local_shape) and stores the result locally.

        Parameters
        ----------
        generator : callable
            This function must be able to process the node's local data shape
            and return a numpy.ndarray of this very shape. This array is then 
            stored as the local data array on each node.
        copy : boolean
            Specifies whether the self.set_local_data method is instructed to 
            copy the result from generator or not.

        Notes
        -----
        The generator function yields node-local results. Therefore it is 
        assumed that the resulting overall d2o does not possess hermitian 
        symmetry anymore. Therefore self.hermitian is set to False.
                        
        """
        self.set_local_data(generator(self.distributor.local_shape), copy=copy)
Ultimanet's avatar
Ultimanet committed
473
        self.hermitian = False
474

ultimanet's avatar
ultimanet committed
475
    def __str__(self):
476
        """ x.__str__() <==> str(x)"""
ultimanet's avatar
ultimanet committed
477
        return self.data.__str__()
478

ultimanet's avatar
ultimanet committed
479
    def __repr__(self):
480
        """ x.__repr__() <==> repr(x)"""
481
        return '<distributed_data_object>\n' + self.data.__repr__()
482

483
    def _compare_helper(self, other, op):
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
        """ _compare_helper is used for <, <=, ==, !=, >= and >.
        
        It checks the class of `other` and then utilizes the appropriate 
        methods of self. If `other` is not a scalar, numpy.ndarray or 
        distributed_data_object this method will use numpy casting.
        
        Parameters
        ----------
        other : scalar, numpy.ndarray, distributed_data_object, array_like
            This is the object that will be compared to self.
        op : string
            The name of the comparison function, e.g. '__ne__'.
        
        Returns
        -------
        result : boolean, distributed_data_object
            If `other` was None, False will be returned. This follows the 
            behaviour of numpy but will changed as soon as numpy changed their 
            convention. In every other case a distributed_data_object with 
            element-wise comparison results will be returned.

        """
        
        if other is not None:
            result = self.copy_empty(dtype=np.bool_)

510
        # Case 1: 'other' is a scalar
511
        # -> make element-wise comparison
Ultimanet's avatar
Ultimanet committed
512
        if np.isscalar(other):
513
            result.set_local_data(
514
                getattr(self.get_local_data(copy=False), op)(other))
515
            return result
Ultimanet's avatar
Ultimanet committed
516

517
        # Case 2: 'other' is a numpy array or a distributed_data_object
518
        # -> extract the local data and make element-wise comparison
Ultimanet's avatar
Ultimanet committed
519
        elif isinstance(other, np.ndarray) or\
520
                isinstance(other, distributed_data_object):
Ultimanet's avatar
Ultimanet committed
521
            temp_data = self.distributor.extract_local_data(other)
522
523
            result.set_local_data(
                getattr(self.get_local_data(copy=False), op)(temp_data))
Ultimanet's avatar
Ultimanet committed
524
            return result
525
526

        # Case 3: 'other' is None
Ultima's avatar
Ultima committed
527
        elif other is None:
Ultimanet's avatar
Ultimanet committed
528
            return False
529
530
531

        # Case 4: 'other' is something different
        # -> make a numpy casting and make a recursive call
Ultimanet's avatar
Ultimanet committed
532
533
        else:
            temp_other = np.array(other)
534
            return getattr(self, op)(temp_other)
535

536
    def __ne__(self, other):
537
538
539
540
541
542
543
        """ x.__ne__(y) <==> x != y

        See Also
        --------
        _compare_helper

        """
544
        return self._compare_helper(other, '__ne__')
545

546
    def __lt__(self, other):
547
548
549
550
551
552
553
554
        """ x.__lt__(y) <==> x < y

        See Also
        --------
        _compare_helper

        """

555
        return self._compare_helper(other, '__lt__')
556

557
    def __le__(self, other):
558
559
560
561
562
563
564
565
        """ x.__le__(y) <==> x <= y

        See Also
        --------
        _compare_helper

        """

566
567
568
        return self._compare_helper(other, '__le__')

    def __eq__(self, other):
569
570
571
572
573
574
575
        """ x.__eq__(y) <==> x == y

        See Also
        --------
        _compare_helper

        """
576
577

        return self._compare_helper(other, '__eq__')
578

579
    def __ge__(self, other):
580
581
582
583
584
585
586
587
        """ x.__ge__(y) <==> x >= y

        See Also
        --------
        _compare_helper

        """

588
589
590
        return self._compare_helper(other, '__ge__')

    def __gt__(self, other):
591
592
593
594
595
596
597
598
        """ x.__gt__(y) <==> x > y

        See Also
        --------
        _compare_helper

        """

599
600
        return self._compare_helper(other, '__gt__')

Ultima's avatar
Ultima committed
601
    def __iter__(self):
602
603
604
605
606
607
608
609
610
        """ x.__iter__() <==> iter(x)

        The __iter__ call returns an iterator it got from self.distributor.

        See Also
        --------
        distributor.get_iter

        """
Ultima's avatar
Ultima committed
611
612
        return self.distributor.get_iter(self)

Ultimanet's avatar
Ultimanet committed
613
    def equal(self, other):
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
        """  Checks if `other` and `self` are structurally the same. 
        
        In contrast to the element-wise comparison with `__eq__`, `equal` 
        checks more than only the equality of the array data. 
        It checks the equality of
            * shape
            * dtype
            * init_args
            * init_kwargs
            * distribution_strategy
            * node's local data

        Parameters
        ----------
        other : object
            The object that will be compared to `self`.

        Returns
        -------
        result : boolean
            True if above conditions are met, False otherwise. 

        """

Ultimanet's avatar
Ultimanet committed
638
639
640
641
642
643
644
645
646
        if other is None:
            return False
        try:
            assert(self.dtype == other.dtype)
            assert(self.shape == other.shape)
            assert(self.init_args == other.init_args)
            assert(self.init_kwargs == other.init_kwargs)
            assert(self.distribution_strategy == other.distribution_strategy)
            assert(np.all(self.data == other.data))
Ultimanet's avatar
Ultimanet committed
647
        except(AssertionError, AttributeError):
Ultimanet's avatar
Ultimanet committed
648
649
650
651
            return False
        else:
            return True

652
    def __pos__(self):
653
654
655
656
657
        """ x.__pos__() <==> +x 
        
        Returns a (positive) copy of `self`.
        """
        
658
        temp_d2o = self.copy_empty()
659
660
        temp_d2o.set_local_data(data=self.get_local_data().__pos__(), 
                                copy=False)
661
        return temp_d2o
662

ultimanet's avatar
ultimanet committed
663
    def __neg__(self):
664
665
666
667
668
        """ x.__neg__() <==> -x 
        
        Returns a negative copy of `self`.
        """
        
669
        temp_d2o = self.copy_empty()
670
        temp_d2o.set_local_data(data=self.get_local_data().__neg__(),
671
                                copy=False)
ultimanet's avatar
ultimanet committed
672
        return temp_d2o
673

674
    def __abs__(self):
675
676
677
678
679
        """ x.__abs__() <==> abs(x)

        Returns an absolute valued copy of `self`.
        """

680
        # translate complex dtypes
681
682
683
684
685
686
        if self.dtype == np.dtype('complex64'):
            new_dtype = np.dtype('float32')
        elif self.dtype == np.dtype('complex128'):
            new_dtype = np.dtype('float64')
        elif issubclass(self.dtype.type, np.complexfloating):
            new_dtype = np.dtype('float')
Ultimanet's avatar
Ultimanet committed
687
688
        else:
            new_dtype = self.dtype
689
690
        temp_d2o = self.copy_empty(dtype=new_dtype)
        temp_d2o.set_local_data(data=self.get_local_data().__abs__(),
691
                                copy=False)
692
        return temp_d2o
693

Ultima's avatar
Ultima committed
694
    def _builtin_helper(self, operator, other, inplace=False):
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
        """ Used for various binary operations like +, -, *, /, **, *=, +=,...

        _builtin_helper checks whether `other` is a scalar or an array and 
        based on that extracts the locally relevant data from it. If `self`
        is hermitian, _builtin_helper tries to conserve this flag; but without 
        checking hermitianity explicitly. 

        Parameters
        ----------
        operator : callable

        other : scalar, array-like

        inplace : boolean
            If the result shall be saved in the data array of `self`. Used for 
            +=, -=, etc...
        Returns
        -------
        out : distributed_data_object
            The distributed_data_object containing the computation's result.
            Equals `self` if `inplace is True`.

        """
718
        # Case 1: other is not a scalar
Ultimanet's avatar
Ultimanet committed
719
        if not (np.isscalar(other) or np.shape(other) == (1,)):
720
            try:
721
                hermitian_Q = (other.hermitian and self.hermitian)
722
723
            except(AttributeError):
                hermitian_Q = False
724
            # extract the local data from the 'other' object
Ultimanet's avatar
Ultimanet committed
725
726
            temp_data = self.distributor.extract_local_data(other)
            temp_data = operator(temp_data)
727

728
        # Case 2: other is a scalar
729
        else:
730
            # if other is a scalar packed in a d2o, extract its value.
731
            if isinstance(other, distributed_data_object):
732
                other = other[0]
733

734
            if np.isrealobj(other): 
735
736
737
738
                hermitian_Q = self.hermitian
            else:
                hermitian_Q = False

739
            temp_data = operator(other)
740
741
        
        # select the return-distributed_data_object
742
        if inplace is True:
743
744
            temp_d2o = self
        else:
745
            # use common datatype for self and other
746
            new_dtype = np.dtype(np.find_common_type((self.dtype,),
747
                                                     (temp_data.dtype,)))
748
            temp_d2o = self.copy_empty(dtype=new_dtype)
749
750
        
        # write the new data into the return-distributed_data_object
751
        temp_d2o.set_local_data(data=temp_data, copy=False)
752
        temp_d2o.hermitian = hermitian_Q
ultimanet's avatar
ultimanet committed
753
        return temp_d2o
754

ultimanet's avatar
ultimanet committed
755
    def __add__(self, other):
756
757
758
759
760
761
762
        """ x.__add__(y) <==> x+y

        See Also
        --------
        _builtin_helper
        """

Ultima's avatar
Ultima committed
763
        return self._builtin_helper(self.get_local_data().__add__, other)
ultimanet's avatar
ultimanet committed
764
765

    def __radd__(self, other):
766
767
768
769
770
771
772
        """ x.__radd__(y) <==> y+x

        See Also
        --------
        _builtin_helper
        """

Ultima's avatar
Ultima committed
773
        return self._builtin_helper(self.get_local_data().__radd__, other)
Ultimanet's avatar
Ultimanet committed
774
775

    def __iadd__(self, other):
776
777
778
779
780
781
782
        """ x.__iadd__(y) <==> x+=y

        See Also
        --------
        _builtin_helper
        """

783
        return self._builtin_helper(self.get_local_data().__iadd__,
784
785
                                    other,
                                    inplace=True)
Ultimanet's avatar
Ultimanet committed
786

ultimanet's avatar
ultimanet committed
787
    def __sub__(self, other):
788
789
790
791
792
793
794
        """ x.__sub__(y) <==> x-y

        See Also
        --------
        _builtin_helper
        """

Ultima's avatar
Ultima committed
795
        return self._builtin_helper(self.get_local_data().__sub__, other)
796

ultimanet's avatar
ultimanet committed
797
    def __rsub__(self, other):
798
799
800
801
802
803
804
        """ x.__rsub__(y) <==> y-x

        See Also
        --------
        _builtin_helper
        """

Ultima's avatar
Ultima committed
805
        return self._builtin_helper(self.get_local_data().__rsub__, other)
806

ultimanet's avatar
ultimanet committed
807
    def __isub__(self, other):
808
809
810
811
812
813
814
        """ x.__isub__(y) <==> x-=y

        See Also
        --------
        _builtin_helper
        """

815
        return self._builtin_helper(self.get_local_data().__isub__,
816
817
                                    other,
                                    inplace=True)
818

ultimanet's avatar
ultimanet committed
819
    def __div__(self, other):
820
821
822
823
824
825
826
        """ x.__div__(y) <==> x/y

        See Also
        --------
        _builtin_helper
        """

Ultima's avatar
Ultima committed
827
        return self._builtin_helper(self.get_local_data().__div__, other)
828

829
    def __truediv__(self, other):
830
831
832
833
834
835
836
        """ x.__truediv__(y) <==> x/y

        See Also
        --------
        _builtin_helper
        """

837
        return self.__div__(other)
838

ultimanet's avatar
ultimanet committed
839
    def __rdiv__(self, other):
840
841
842
843
844
845
846
        """ x.__rdiv__(y) <==> y/x

        See Also
        --------
        _builtin_helper
        """

Ultima's avatar
Ultima committed
847
        return self._builtin_helper(self.get_local_data().__rdiv__, other)
848

849
    def __rtruediv__(self, other):
850
851
852
853
854
855
856
        """ x.__rtruediv__(y) <==> y/x

        See Also
        --------
        _builtin_helper
        """

857
        return self.__rdiv__(other)
ultimanet's avatar
ultimanet committed
858

Ultimanet's avatar
Ultimanet committed
859
    def __idiv__(self, other):
860
861
862
863
864
865
866
        """ x.__idiv__(y) <==> x/=y

        See Also
        --------
        _builtin_helper
        """

867
        return self._builtin_helper(self.get_local_data().__idiv__,
868
869
870
                                    other,
                                    inplace=True)

871
    def __itruediv__(self, other):
872
873
874
875
876
877
878
        """ x.__itruediv__(y) <==> x/=y

        See Also
        --------
        _builtin_helper
        """

879
        return self.__idiv__(other)
880

ultimanet's avatar
ultimanet committed
881
    def __floordiv__(self, other):
882
883
884
885
886
887
888
        """ x.__floordiv__(y) <==> x//y

        See Also
        --------
        _builtin_helper
        """

889
        return self._builtin_helper(self.get_local_data().__floordiv__,
890
891
                                    other)

ultimanet's avatar
ultimanet committed
892
    def __rfloordiv__(self, other):
893
        return self._builtin_helper(self.get_local_data().__rfloordiv__,
894
895
                                    other)

Ultimanet's avatar
Ultimanet committed
896
    def __ifloordiv__(self, other):
Ultima's avatar
Ultima committed
897
        return self._builtin_helper(
898
899
            self.get_local_data().__ifloordiv__, other,
            inplace=True)
900

ultimanet's avatar
ultimanet committed
901
    def __mul__(self, other):
Ultima's avatar
Ultima committed
902
        return self._builtin_helper(self.get_local_data().__mul__, other)
903

ultimanet's avatar
ultimanet committed
904
    def __rmul__(self, other):
Ultima's avatar
Ultima committed
905
        return self._builtin_helper(self.get_local_data().__rmul__, other)
ultimanet's avatar
ultimanet committed
906
907

    def __imul__(self, other):
908
        return self._builtin_helper(self.get_local_data().__imul__,
909
910
                                    other,
                                    inplace=True)
Ultimanet's avatar
Ultimanet committed
911

ultimanet's avatar
ultimanet committed
912
    def __pow__(self, other):
Ultima's avatar
Ultima committed
913
        return self._builtin_helper(self.get_local_data().__pow__, other)
914

ultimanet's avatar
ultimanet committed
915
    def __rpow__(self, other):
Ultima's avatar
Ultima committed
916
        return self._builtin_helper(self.get_local_data().__rpow__, other)
ultimanet's avatar
ultimanet committed
917
918

    def __ipow__(self, other):
919
        return self._builtin_helper(self.get_local_data().__ipow__,
920
921
922
                                    other,
                                    inplace=True)

Ultima's avatar
Ultima committed
923
924
    def __mod__(self, other):
        return self._builtin_helper(self.get_local_data().__mod__, other)
925

Ultima's avatar
Ultima committed
926
    def __rmod__(self, other):
927
        return self._builtin_helper(self.get_local_data().__rmod__, other)
928

Ultima's avatar
Ultima committed
929
    def __imod__(self, other):
930
        return self._builtin_helper(self.get_local_data().__imod__,
931
932
933
                                    other,
                                    inplace=True)

934
935
    def __len__(self):
        return self.shape[0]
936

937
    def get_dim(self):
938
        return np.prod(self.shape)
939

940
    def vdot(self, other):
941
        other = self.distributor.extract_local_data(other)
942
943
944
945
946
947
948
949
950
        local_vdot = np.array([np.vdot(self.get_local_data(), other)])
        global_vdot = np.empty_like(local_vdot)
        self.distributor._Allreduce_sum(sendbuf=local_vdot,
                                        recvbuf=global_vdot)

#        local_vdot = np.vdot(self.get_local_data(), other)
#        local_vdot_list = self.distributor._allgather(local_vdot)
#        global_vdot = np.result_type(self.dtype,
#                                     other.dtype).type(np.sum(local_vdot_list))
951
        return global_vdot[0]
Ultimanet's avatar
Ultimanet committed
952

ultimanet's avatar
ultimanet committed
953
    def __getitem__(self, key):
Ultima's avatar
Ultima committed
954
        return self.get_data(key)
955

ultimanet's avatar
ultimanet committed
956
957
    def __setitem__(self, key, data):
        self.set_data(data, key)
958

959
    def _contraction_helper(self, function, **kwargs):
960
961
962
        if self.shape == (0,):
            raise ValueError("ERROR: Zero-size array to reduction operation " +
                             "which has no identity")
Ultima's avatar
Ultima committed
963
964
965
966
967
968
969
        if np.prod(self.data.shape) == 0:
            local = 0
            include = False
        else:
            local = function(self.data, **kwargs)
            include = True

970
        local_list = self.distributor._allgather(local)
971
        local_list = np.array(local_list, dtype=np.dtype(local_list[0]))
Ultima's avatar
Ultima committed
972
973
974
        include_list = np.array(self.distributor._allgather(include))
        work_list = local_list[include_list]
        if work_list.shape[0] == 0:
975
            raise ValueError("ERROR: Zero-size array to reduction operation " +
Ultima's avatar
Ultima committed
976
                             "which has no identity")
977
        else:
Ultima's avatar
Ultima committed
978
979
            result = function(work_list, axis=0)
            return result
980

981
982
983
    def min(self, **kwargs):
        return self.amin(**kwargs)

984
    def amin(self, **kwargs):
985
        return self._contraction_helper(np.amin, **kwargs)
986
987

    def nanmin(self, **kwargs):
988
        return self._contraction_helper(np.nanmin, **kwargs)
989

990
991
992
    def max(self, **kwargs):
        return self.amax(**kwargs)

993
    def amax(self, **kwargs):
994
        return self._contraction_helper(np.amax, **kwargs)
995

996
    def nanmax(self, **kwargs):
997
        return self._contraction_helper(np.nanmax, **kwargs)
998

999
    def sum(self, **kwargs):
1000
1001
        if self.shape == (0,):
            return self.dtype.type(0)
1002
1003
1004
        return self._contraction_helper(np.sum, **kwargs)

    def prod(self, **kwargs):
1005
1006
        if self.shape == (0,):
            return self.dtype.type(1)
1007
1008
        return self._contraction_helper(np.prod, **kwargs)

1009
    def mean(self, power=1):
1010
1011
        if self.shape == (0,):
            return np.mean(np.array([], dtype=self.dtype))
1012
        # compute the local means and the weights for the mean-mean.
Ultima's avatar
Ultima committed
1013
1014
1015
1016
1017
1018
        if np.prod(self.data.shape) == 0:
            local_mean = 0
            include = False
        else:
            local_mean = np.mean(self.data**power)
            include = True
1019

1020
        local_weight = np.prod(self.data.shape)
1021
        # collect the local means and cast the result to a ndarray
Ultima's avatar
Ultima committed
1022
1023
        local_mean_list = self.distributor._allgather(local_mean)
        local_weight_list = self.distributor._allgather(local_weight)
1024

1025
1026
        local_mean_list = np.array(local_mean_list,
                                   dtype=np.dtype(local_mean_list[0]))
1027
1028
        local_weight_list = np.array(local_weight_list)
        # extract the parts from the non-empty nodes
Ultima's avatar
Ultima committed
1029
1030
1031
1032
1033
        include_list = np.array(self.distributor._allgather(include))
        work_mean_list = local_mean_list[include_list]
        work_weight_list = local_weight_list[include_list]
        if work_mean_list.shape[0] == 0:
            raise ValueError("ERROR:  Mean of empty slice.")
1034
1035
        else:
            # compute the denominator for the weighted mean-mean
Ultima's avatar
Ultima committed
1036
            global_weight = np.sum(work_weight_list)
1037
            # compute the numerator
Ultima's avatar
Ultima committed
1038
            numerator = np.sum(work_mean_list * work_weight_list)
1039
            global_mean = numerator / global_weight
Ultima's avatar
Ultima committed
1040
            return global_mean
1041
1042

    def var(self):
1043
1044
        if self.shape == (0,):
            return np.var(np.array([], dtype=self.dtype))
1045
1046
1047
1048
1049
1050
1051

        if issubclass(self.dtype.type, np.complexfloating):
            mean_of_the_square = abs(self**2).mean()
            square_of_the_mean = abs(self.mean())**2
        else:
            mean_of_the_square = self.mean(power=2)
            square_of_the_mean = self.mean()**2
1052
        return mean_of_the_square - square_of_the_mean
1053

1054
    def std(self):
1055
1056
1057
1058
        if self.shape == (0,):
            return np.std(np.array([], dtype=self.dtype))
        if self.shape == (0,):
            return np.nan
1059
        return np.sqrt(self.var())
1060

Ultima's avatar
Ultima committed
1061
    def argmin(self):
1062
1063
1064
        if self.shape == (0,):
            raise ValueError(
                "ERROR: attempt to get argmin of an empty object")
Ultima's avatar
Ultima committed
1065
1066
1067
1068
1069
1070
        if np.prod(self.data.shape) == 0:
            local_argmin = np.nan
            local_argmin_value = np.nan
            globalized_local_argmin = np.nan
        else:
            local_argmin = np.argmin(self.data)
1071
            local_argmin_value = self.data[np.unravel_index(local_argmin,
Ultima's avatar
Ultima committed
1072
                                                            self.data.shape)]
1073

Ultima's avatar
Ultima committed
1074
            globalized_local_argmin = self.distributor.globalize_flat_index(
1075
1076
1077
1078
                local_argmin)
        local_argmin_list = self.distributor._allgather(
                                                    (local_argmin_value,
                                                     globalized_local_argmin))
1079
        local_argmin_list = np.array(local_argmin_list, dtype=[
1080
1081
            ('value', np.dtype('complex128')),
            ('index', np.dtype('float'))])
1082
1083
        local_argmin_list = np.sort(local_argmin_list,
                                    order=['value', 'index'])
Ultima's avatar
Ultima committed
1084
        return np.int(local_argmin_list[0][1])
1085

Ultima's avatar
Ultima committed
1086
    def argmax(self):
1087
1088
1089
        if self.shape == (0,):
            raise ValueError(
                "ERROR: attempt to get argmax of an empty object")
Ultima's avatar
Ultima committed
1090
1091
1092
1093
1094
1095
        if np.prod(self.data.shape) == 0:
            local_argmax = np.nan
            local_argmax_value = np.nan
            globalized_local_argmax = np.nan
        else:
            local_argmax = np.argmax(self.data)
1096
            local_argmax_value = -self.data[np.unravel_index(local_argmax,
1097
                                                             self.data.shape)]
Ultima's avatar
Ultima committed
1098
            globalized_local_argmax = self.distributor.globalize_flat_index(
1099
1100
1101
1102
                local_argmax)
        local_argmax_list = self.distributor._allgather(
                                                  (local_argmax_value,
                                                   globalized_local_argmax))
1103
        local_argmax_list = np.array(local_argmax_list, dtype=[
1104
1105
            ('value', np.dtype('complex128')),
            ('index', np.dtype('float'))])
1106
1107
        local_argmax_list = np.sort(local_argmax_list,
                                    order=['value', 'index'])
Ultima's avatar
Ultima committed
1108
        return np.int(local_argmax_list[0][1])
1109

1110
    def argmin_nonflat(self):
Ultima's avatar
Ultima committed
1111
        return np.unravel_index(self.argmin(), self.shape)
1112

Ultima's avatar
Ultima committed
1113
1114
    def argmax_nonflat(self):
        return np.unravel_index(self.argmax(), self.shape)
1115

1116
1117
1118
1119
    def conjugate(self):
        temp_d2o = self.copy_empty()
        temp_data = np.conj(self.get_local_data())
        temp_d2o.set_local_data(temp_data)
Ultima's avatar
Ultima committed
1120
        temp_d2o.hermitian = self.hermitian
1121
1122
1123
        return temp_d2o

    def conj(self):
1124
1125
        return self.conjugate()

1126
    def median(self):
1127
        about.warnings.cprint(
1128
1129
1130
            "WARNING: The current implementation of median is very expensive!")
        median = np.median(self.get_full_data())
        return median
1131

1132
    def _is_helper(self, function):
Ultima's avatar
Ultima committed
1133
        temp_d2o = self.copy_empty(dtype=np.dtype('bool'))
1134
        temp_d2o.set_local_data(function(self.data))
1135
        return temp_d2o
1136

1137
1138
1139
    def iscomplex(self):
        return self._is_helper(np.iscomplex)

1140
    def isreal(self):
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
        return self._is_helper(np.isreal)

    def isnan(self):
        return self._is_helper(np.isnan)

    def isinf(self):
        return self._is_helper(np.isinf)

    def isfinite(self):
        return self._is_helper(np.isfinite)

    def nan_to_num(self):
        temp_d2o = self.copy_empty()
        temp_d2o.set_local_data(np.nan_to_num(self.data))