rg_space.py 12.8 KB
Newer Older
1
2
3
4
# 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.
5
#
6
7
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
8
9
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU General Public License for more details.
#
11
# You should have received a copy of the GNU General Public License
12
# 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.
Marco Selig's avatar
Marco Selig committed
18
19
20
21
22
23
24
25
26
27

"""
    ..                  __   ____   __
    ..                /__/ /   _/ /  /_
    ..      __ ___    __  /  /_  /   _/  __   __
    ..    /   _   | /  / /   _/ /  /   /  / /  /
    ..   /  / /  / /  / /  /   /  /_  /  /_/  /
    ..  /__/ /__/ /__/ /__/    \___/  \___   /  rg
    ..                               /______/

Marco Selig's avatar
Marco Selig committed
28
    NIFTY submodule for regular Cartesian grids.
Marco Selig's avatar
Marco Selig committed
29
30
31

"""
from __future__ import division
Martin Reinecke's avatar
Martin Reinecke committed
32
from builtins import range
Martin Reinecke's avatar
Martin Reinecke committed
33
from functools import reduce
Ultimanet's avatar
Ultimanet committed
34

Marco Selig's avatar
Marco Selig committed
35
import numpy as np
Ultimanet's avatar
Ultimanet committed
36

37
38
from d2o import distributed_data_object,\
                STRATEGIES as DISTRIBUTION_STRATEGIES
39

Martin Reinecke's avatar
Martin Reinecke committed
40
from ..space import Space
Martin Reinecke's avatar
Martin Reinecke committed
41
from ...config import nifty_configuration
csongor's avatar
csongor committed
42

Marco Selig's avatar
Marco Selig committed
43

Theo Steininger's avatar
Theo Steininger committed
44
class RGSpace(Space):
Marco Selig's avatar
Marco Selig committed
45
46
47
48
49
50
51
52
53
    """
        ..      _____   _______
        ..    /   __/ /   _   /
        ..   /  /    /  /_/  /
        ..  /__/     \____  /  space class
        ..          /______/

        NIFTY subclass for spaces of regular Cartesian grids.

Theo Steininger's avatar
Theo Steininger committed
54
55
56
57
        Parameters
        ----------
        shape : {int, numpy.ndarray}
            Number of grid points or numbers of gridpoints along each axis.
58
59
60
61
        zerocenter : {bool, numpy.ndarray} *optional*
            Whether x==0 (or k==0, respectively) is located in the center of
            the grid (or the center of each axis speparately) or not.
            (default: False).
Theo Steininger's avatar
Theo Steininger committed
62
63
64
65
66
67
68
69
70
71
        distances : {float, numpy.ndarray}, *optional*
            Distance between two grid points along each axis
            (default: None).
            If distances==None:
                if harmonic==True, all distances will be set to 1
                if harmonic==False, the distance along each axis will be
                  set to the inverse of the number of points along that
                  axis.
        harmonic : bool, *optional*
        Whether the space represents a grid in position or harmonic space.
Theo Steininger's avatar
Theo Steininger committed
72
            (default: False).
Marco Selig's avatar
Marco Selig committed
73
74
75

        Attributes
        ----------
Martin Reinecke's avatar
Martin Reinecke committed
76
        harmonic : bool
Theo Steininger's avatar
Theo Steininger committed
77
78
79
80
81
            Whether or not the grid represents a position or harmonic space.
        zerocenter : tuple of bool
            Whether x==0 (or k==0, respectively) is located in the center of
            the grid (or the center of each axis speparately) or not.
        distances : tuple of floats
82
83
84
85
86
87
88
89
90
            Distance between two grid points along the correponding axis.
        dim : np.int
            Total number of dimensionality, i.e. the number of pixels.
        harmonic : bool
            Specifies whether the space is a signal or harmonic space.
        total_volume : np.float
            The total volume of the space.
        shape : tuple of np.ints
            The shape of the space's data array.
Theo Steininger's avatar
Theo Steininger committed
91

Marco Selig's avatar
Marco Selig committed
92
93
    """

94
95
    # ---Overwritten properties and methods---

96
    def __init__(self, shape, zerocenter=False, distances=None,
Martin Reinecke's avatar
Martin Reinecke committed
97
                 harmonic=False):
98
99
        self._harmonic = bool(harmonic)

Martin Reinecke's avatar
Martin Reinecke committed
100
        super(RGSpace, self).__init__()
101

102
103
104
        self._shape = self._parse_shape(shape)
        self._distances = self._parse_distances(distances)
        self._zerocenter = self._parse_zerocenter(zerocenter)
Marco Selig's avatar
Marco Selig committed
105

106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
# This code is unused but may be useful to keep around if it is ever needed
# again in the future ...

#    def hermitian_fixed_points(self):
#        dimensions = len(self.shape)
#        mid_index = np.array(self.shape)//2
#        ndlist = [1]*dimensions
#        for k in range(dimensions):
#            if self.shape[k] % 2 == 0:
#                ndlist[k] = 2
#        ndlist = tuple(ndlist)
#        fixed_points = []
#        for index in np.ndindex(ndlist):
#            for k in range(dimensions):
#                if self.shape[k] % 2 != 0 and self.zerocenter[k]:
#                    index = list(index)
#                    index[k] = 1
#                    index = tuple(index)
#            fixed_points += [tuple(index * mid_index)]
#        return fixed_points
126

127
    def hermitianize_inverter(self, x, axes):
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
        if nifty_configuration['harmonic_rg_base'] == 'real':
            return x
        else:
            # calculate the number of dimensions the input array has
            dimensions = len(x.shape)
            # prepare the slicing object which will be used for mirroring
            slice_primitive = [slice(None), ] * dimensions
            # copy the input data
            y = x.copy()

            # flip in the desired directions
            for k in range(len(axes)):
                i = axes[k]
                slice_picker = slice_primitive[:]
                slice_inverter = slice_primitive[:]
                if (not self.zerocenter[k]) or self.shape[k] % 2 == 0:
                    slice_picker[i] = slice(1, None, None)
                    slice_inverter[i] = slice(None, 0, -1)
                else:
                    slice_picker[i] = slice(None)
                    slice_inverter[i] = slice(None, None, -1)
                slice_picker = tuple(slice_picker)
                slice_inverter = tuple(slice_inverter)

                try:
                    y.set_data(to_key=slice_picker, data=y,
                               from_key=slice_inverter)
                except(AttributeError):
                    y[slice_picker] = y[slice_inverter]
            return y
158

159
160
    # ---Mandatory properties and methods---

161
162
163
164
    def __repr__(self):
        return ("RGSpace(shape=%r, zerocenter=%r, distances=%r, harmonic=%r)"
                % (self.shape, self.zerocenter, self.distances, self.harmonic))

165
166
167
168
169
170
171
172
173
174
    @property
    def harmonic(self):
        return self._harmonic

    @property
    def shape(self):
        return self._shape

    @property
    def dim(self):
Martin Reinecke's avatar
Martin Reinecke committed
175
        return int(reduce(lambda x, y: x*y, self.shape))
176
177
178
179
180
181
182
183
184

    @property
    def total_volume(self):
        return self.dim * reduce(lambda x, y: x*y, self.distances)

    def copy(self):
        return self.__class__(shape=self.shape,
                              zerocenter=self.zerocenter,
                              distances=self.distances,
Martin Reinecke's avatar
Martin Reinecke committed
185
                              harmonic=self.harmonic)
186
187

    def weight(self, x, power=1, axes=None, inplace=False):
188
        weight = reduce(lambda x, y: x*y, self.distances) ** np.float(power)
189
190
191
192
193
194
195
        if inplace:
            x *= weight
            result_x = x
        else:
            result_x = x*weight
        return result_x

196
    def get_distance_array(self, distribution_strategy):
Theo Steininger's avatar
Theo Steininger committed
197
198
        """ Calculates an n-dimensional array with its entries being the
        lengths of the vectors from the zero point of the grid.
theos's avatar
theos committed
199

Theo Steininger's avatar
Theo Steininger committed
200
201
        Parameters
        ----------
Theo Steininger's avatar
Theo Steininger committed
202
203
204
        distribution_strategy : str
            The distribution_strategy which shall be used the returned
            distributed_data_object.
theos's avatar
theos committed
205

Theo Steininger's avatar
Theo Steininger committed
206
207
        Returns
        -------
Theo Steininger's avatar
Theo Steininger committed
208
        distributed_data_object
Theo Steininger's avatar
Theo Steininger committed
209
210
            A d2o containing the distances.

theos's avatar
theos committed
211
        """
Theo Steininger's avatar
Theo Steininger committed
212

Martin Reinecke's avatar
Martin Reinecke committed
213
        if (not self.harmonic): raise NotImplementedError
theos's avatar
theos committed
214
215
216
        shape = self.shape
        # prepare the distributed_data_object
        nkdict = distributed_data_object(
Martin Reinecke's avatar
Martin Reinecke committed
217
                        global_shape=shape, dtype=np.float64,
theos's avatar
theos committed
218
219
220
221
222
223
224
225
226
                        distribution_strategy=distribution_strategy)

        if distribution_strategy in DISTRIBUTION_STRATEGIES['slicing']:
            # get the node's individual slice of the first dimension
            slice_of_first_dimension = slice(
                                    *nkdict.distributor.local_slice[0:2])
        elif distribution_strategy in DISTRIBUTION_STRATEGIES['not']:
            slice_of_first_dimension = slice(0, shape[0])
        else:
227
228
            raise ValueError(
                "Unsupported distribution strategy")
theos's avatar
theos committed
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
        dists = self._distance_array_helper(slice_of_first_dimension)
        nkdict.set_local_data(dists)

        return nkdict

    def _distance_array_helper(self, slice_of_first_dimension):
        dk = self.distances
        shape = self.shape

        inds = []
        for a in shape:
            inds += [slice(0, a)]

        cords = np.ogrid[inds]

244
245
        dists = (cords[0] - shape[0]//2)*dk[0]
        dists *= dists
theos's avatar
theos committed
246
        # apply zerocenterQ shift
247
248
        if not self.zerocenter[0]:
            dists = np.fft.ifftshift(dists)
theos's avatar
theos committed
249
250
251
        # only save the individual slice
        dists = dists[slice_of_first_dimension]
        for ii in range(1, len(shape)):
252
253
            temp = (cords[ii] - shape[ii] // 2) * dk[ii]
            temp *= temp
254
            if not self.zerocenter[ii]:
Martin Reinecke's avatar
Martin Reinecke committed
255
                temp = np.fft.ifftshift(temp)
theos's avatar
theos committed
256
257
258
259
            dists = dists + temp
        dists = np.sqrt(dists)
        return dists

Martin Reinecke's avatar
Martin Reinecke committed
260
    def get_unique_distances(self):
Martin Reinecke's avatar
Martin Reinecke committed
261
        if (not self.harmonic): raise NotImplementedError
Martin Reinecke's avatar
Martin Reinecke committed
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
        dimensions = len(self.shape)
        if dimensions == 1:  # extra easy
            maxdist = self.shape[0]//2
            return np.arange(maxdist+1, dtype=np.float64) * self.distances[0]
        if np.all(self.distances == self.distances[0]):  # shortcut
            maxdist = np.asarray(self.shape)//2
            tmp = np.sum(maxdist*maxdist)
            tmp = np.zeros(tmp+1, dtype=np.bool)
            t2 = np.arange(maxdist[0]+1, dtype=np.int64)
            t2 *= t2
            for i in range(1, dimensions):
                t3 = np.arange(maxdist[i]+1, dtype=np.int64)
                t3 *= t3
                t2 = np.add.outer(t2, t3)
            tmp[t2] = True
            return np.sqrt(np.nonzero(tmp)[0])*self.distances[0]
        else:  # do it the hard way
            tmp = self.get_distance_array('not').unique()  # expensive!
            tol = 1e-12*tmp[-1]
            # remove all points that are closer than tol to their right
            # neighbors.
            # I'm appending the last value*2 to the array to treat the
            # rightmost point correctly.
            return tmp[np.diff(np.r_[tmp, 2*tmp[-1]]) > tol]

    def get_natural_binbounds(self):
Martin Reinecke's avatar
Martin Reinecke committed
288
        if (not self.harmonic): raise NotImplementedError
Martin Reinecke's avatar
Martin Reinecke committed
289
290
291
        tmp = self.get_unique_distances()
        return 0.5*(tmp[:-1]+tmp[1:])

292
    def get_fft_smoothing_kernel_function(self, sigma):
Martin Reinecke's avatar
Martin Reinecke committed
293
        if (not self.harmonic): raise NotImplementedError
Theo Steininger's avatar
Theo Steininger committed
294
        return lambda x: np.exp(-2. * np.pi*np.pi * x*x * sigma*sigma)
theos's avatar
theos committed
295

296
297
298
299
    # ---Added properties and methods---

    @property
    def distances(self):
Theo Steininger's avatar
Theo Steininger committed
300
301
302
        """Distance between two grid points along each axis. It is a tuple
        of positive floating point numbers with the n-th entry giving the
        distances of grid points along the n-th dimension.
303
        """
Theo Steininger's avatar
Theo Steininger committed
304

305
306
307
308
        return self._distances

    @property
    def zerocenter(self):
309
        """Returns True if grid points lie symmetrically around zero.
Theo Steininger's avatar
Theo Steininger committed
310

311
312
        Returns
        -------
Theo Steininger's avatar
Theo Steininger committed
313
314
315
316
317
        bool
            True if the grid points are centered around the 0 grid point. This
            option is most common for harmonic spaces (where both conventions
            are used) but may be used for position spaces, too.

318
        """
Theo Steininger's avatar
Theo Steininger committed
319

320
321
322
323
324
325
326
327
328
329
330
331
        return self._zerocenter

    def _parse_shape(self, shape):
        if np.isscalar(shape):
            shape = (shape,)
        temp = np.empty(len(shape), dtype=np.int)
        temp[:] = shape
        return tuple(temp)

    def _parse_distances(self, distances):
        if distances is None:
            if self.harmonic:
Martin Reinecke's avatar
Martin Reinecke committed
332
                temp = np.ones_like(self.shape, dtype=np.float64)
333
            else:
Martin Reinecke's avatar
Martin Reinecke committed
334
                temp = 1 / np.array(self.shape, dtype=np.float64)
335
        else:
Martin Reinecke's avatar
Martin Reinecke committed
336
            temp = np.empty(len(self.shape), dtype=np.float64)
337
338
339
340
341
342
            temp[:] = distances
        return tuple(temp)

    def _parse_zerocenter(self, zerocenter):
        temp = np.empty(len(self.shape), dtype=bool)
        temp[:] = zerocenter
343
344
        if np.any(np.logical_and(temp, np.array(self.shape) % 2)):
            raise ValueError("All zerocentered axis must have even length!")
345
        return tuple(temp)
346
347
348
349

    # ---Serialization---

    def _to_hdf5(self, hdf5_group):
Jait Dixit's avatar
Jait Dixit committed
350
351
352
        hdf5_group['shape'] = self.shape
        hdf5_group['zerocenter'] = self.zerocenter
        hdf5_group['distances'] = self.distances
353
        hdf5_group['harmonic'] = self.harmonic
Jait Dixit's avatar
Jait Dixit committed
354

355
356
357
        return None

    @classmethod
Theo Steininger's avatar
Theo Steininger committed
358
    def _from_hdf5(cls, hdf5_group, repository):
359
        result = cls(
Jait Dixit's avatar
Jait Dixit committed
360
361
362
            shape=hdf5_group['shape'][:],
            zerocenter=hdf5_group['zerocenter'][:],
            distances=hdf5_group['distances'][:],
363
            harmonic=hdf5_group['harmonic'][()],
Jait Dixit's avatar
Jait Dixit committed
364
            )
365
        return result