1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
|
"""
Classes to perform aggregations over cubes
"""
from __future__ import absolute_import, division, print_function
try:
from itertools import izip
except ImportError: # python3
izip = zip
from functools import wraps
import numpy as np
from glue.external.six.moves import range as xrange
def check_empty(func):
@wraps(func)
def wrapper(self, *args, **kwargs):
if self.empty_slice:
return np.zeros(self.shape) * np.nan
return func(self, *args, **kwargs)
return wrapper
class Aggregate(object):
"""
Collapse >=3D datasets into 2D images, using different
aggregation methods
"""
def __init__(self, data, attribute, zax, slc, zlim):
"""
Parameters
----------
data : :class:`~glue.core.data.Data`
attribute : :class:`~glue.core.component_id.ComponentID`
zax : int
Which axis to collapse over.
slc : tuple of int
Describes the
current 2D slice through the image. Used to
define the orientation, as well as axis values
for remaining dimensions of >3D cubes
zlim : tuple of float
Float values of [lo, hi), describing the limits
of the slab to collapse over
"""
self.data = data
self.attribute = attribute
self.zax = zax
self.slc = slc
self.zlim = min(zlim), max(zlim)
@property
def shape(self):
"""
The shape of the 2D aggregated array
"""
s = self.data.shape
return s[self.slc.index('y')], s[self.slc.index('x')]
@property
def empty_slice(self):
"""
True if the slice is empty
"""
return self.zlim[0] == self.zlim[1]
def _subslice(self):
view = [slice(None, None) for _ in self.data.shape]
ax_collapse = self.zax
for i, s in enumerate(self.slc):
if s not in ['x', 'y'] and i != self.zax:
view[i] = s
if i < self.zax:
ax_collapse -= 1
view[self.zax] = slice(*self.zlim)
return view, ax_collapse
def _prepare_cube(self, attribute=None):
view, ax_collapse = self._subslice()
att = attribute or self.attribute
cube = self.data[att, view]
return cube, ax_collapse
def _iter_slice(self, attribute=None):
# iterate through the uncollapsed slab one plane at a time
view, ax_collapse = self._subslice()
att = attribute or self.attribute
for z in xrange(*self.zlim):
view[self.zax] = z
plane = self.data[att, view]
yield np.nan_to_num(plane)
def _iter_slice_index(self):
"""Loop over slices of the target attribute and its world coordinate"""
att = self.data.get_world_component_id(self.zax)
loop = izip(self._iter_slice(), self._iter_slice(att))
return loop
def _finalize(self, cube):
if self.slc.index('x') < self.slc.index('y'):
cube = cube.T
return cube
def collapse_using(self, function):
"""
Produce a collapsed image using a numpy aggregation function
"""
cube, ax = self._prepare_cube()
result = function(cube, axis=ax)
return self._finalize(result)
def _to_world(self, idx):
args = [None] * self.data.ndim
y, x = np.mgrid[:idx.shape[0], :idx.shape[1]]
for i, s in enumerate(self.slc):
if s not in ['x', 'y']:
args[i] = np.ones(idx.size) * s
args[self.slc.index('y')] = y.ravel()
args[self.slc.index('x')] = x.ravel()
args[self.zax] = idx.ravel()
att = self.data.get_world_component_id(self.zax)
return self.data[att, args].reshape(idx.shape)
@staticmethod
def all_operators():
return (Aggregate.sum,
Aggregate.mean,
Aggregate.max,
Aggregate.argmax,
Aggregate.argmin,
Aggregate.mom1,
Aggregate.mom2,
Aggregate.median)
@staticmethod
def _mean(cube, axis):
s = np.nansum(cube, axis)
ct = np.isfinite(cube).sum(axis)
return 1. * s / ct
@check_empty
def sum(self):
return self.collapse_using(np.nansum)
@check_empty
def mean(self):
return self.collapse_using(self._mean)
@check_empty
def max(self):
return self.collapse_using(np.nanmax)
@check_empty
def median(self):
# NOTE: nans are treated as infinity in this case
return self.collapse_using(np.median)
@check_empty
def argmax(self):
"""
Location of peak value, in world coords
"""
idx = self.collapse_using(np.nanargmax)
return self._to_world(idx)
@check_empty
def argmin(self):
"""
Location of minimum value, in world coords
"""
idx = self.collapse_using(np.nanargmin)
return self._to_world(idx)
@check_empty
def mom1(self):
"""
Intensity-weighted coordinate. Pixel units.
"""
# build up slice-by-slice, to avoid big temporary cubes
loop = self._iter_slice_index()
val, loc = next(loop)
val = np.maximum(val, 0)
w, result = val, loc * val
for val, loc in loop:
val = np.maximum(val, 0)
result += val * loc
w += val
return self._finalize(result / w)
@check_empty
def mom2(self):
"""
Intensity-weighted coordinate dispersion. Pixel units.
"""
loop = self._iter_slice_index()
val, loc = next(loop)
val = np.maximum(val, 0)
w, x, x2 = val, val * loc, val * loc * loc
for val, loc in loop:
val = np.maximum(val, 0)
w += val
x += loc * val
x2 += loc ** 2 * val
return self._finalize(np.sqrt(x2 / w - (x / w) ** 2))
|
