lsst.ip.diffim 22.0.1-28-g5706f010+b0db72719b
dipoleFitTask.py
Go to the documentation of this file.
2# LSST Data Management System
3# Copyright 2008-2016 AURA/LSST.
4#
5# This product includes software developed by the
6# LSST Project (http://www.lsst.org/).
7#
8# This program is free software: you can redistribute it and/or modify
9# it under the terms of the GNU General Public License as published by
10# the Free Software Foundation, either version 3 of the License, or
11# (at your option) any later version.
12#
13# This program is distributed in the hope that it will be useful,
14# but WITHOUT ANY WARRANTY; without even the implied warranty of
15# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
16# GNU General Public License for more details.
17#
18# You should have received a copy of the LSST License Statement and
19# the GNU General Public License along with this program. If not,
20# see <https://www.lsstcorp.org/LegalNotices/>.
21#
22
23import numpy as np
24import warnings
25
26import lsst.afw.image as afwImage
27import lsst.meas.base as measBase
28import lsst.afw.table as afwTable
29import lsst.afw.detection as afwDet
30import lsst.geom as geom
31from lsst.log import Log
32import lsst.pex.exceptions as pexExcept
33import lsst.pex.config as pexConfig
34from lsst.pipe.base import Struct
35from lsst.utils.timer import timeMethod
36
37__all__ = ("DipoleFitTask", "DipoleFitPlugin", "DipoleFitTaskConfig", "DipoleFitPluginConfig",
38 "DipoleFitAlgorithm")
39
40
41# Create a new measurement task (`DipoleFitTask`) that can handle all other SFM tasks but can
42# pass a separate pos- and neg- exposure/image to the `DipoleFitPlugin`s `run()` method.
43
44
45class DipoleFitPluginConfig(measBase.SingleFramePluginConfig):
46 """Configuration for DipoleFitPlugin
47 """
48
49 fitAllDiaSources = pexConfig.Field(
50 dtype=float, default=False,
51 doc="""Attempte dipole fit of all diaSources (otherwise just the ones consisting of overlapping
52 positive and negative footprints)""")
53
54 maxSeparation = pexConfig.Field(
55 dtype=float, default=5.,
56 doc="Assume dipole is not separated by more than maxSeparation * psfSigma")
57
58 relWeight = pexConfig.Field(
59 dtype=float, default=0.5,
60 doc="""Relative weighting of pre-subtraction images (higher -> greater influence of pre-sub.
61 images on fit)""")
62
63 tolerance = pexConfig.Field(
64 dtype=float, default=1e-7,
65 doc="Fit tolerance")
66
67 fitBackground = pexConfig.Field(
68 dtype=int, default=1,
69 doc="Set whether and how to fit for linear gradient in pre-sub. images. Possible values:"
70 "0: do not fit background at all"
71 "1 (default): pre-fit the background using linear least squares and then do not fit it as part"
72 "of the dipole fitting optimization"
73 "2: pre-fit the background using linear least squares (as in 1), and use the parameter"
74 "estimates from that fit as starting parameters for an integrated re-fit of the background")
75
76 fitSeparateNegParams = pexConfig.Field(
77 dtype=bool, default=False,
78 doc="Include parameters to fit for negative values (flux, gradient) separately from pos.")
79
80 # Config params for classification of detected diaSources as dipole or not
81 minSn = pexConfig.Field(
82 dtype=float, default=np.sqrt(2) * 5.0,
83 doc="Minimum quadrature sum of positive+negative lobe S/N to be considered a dipole")
84
85 maxFluxRatio = pexConfig.Field(
86 dtype=float, default=0.65,
87 doc="Maximum flux ratio in either lobe to be considered a dipole")
88
89 maxChi2DoF = pexConfig.Field(
90 dtype=float, default=0.05,
91 doc="""Maximum Chi2/DoF significance of fit to be considered a dipole.
92 Default value means \"Choose a chi2DoF corresponding to a significance level of at most 0.05\"
93 (note this is actually a significance, not a chi2 value).""")
94
95
96class DipoleFitTaskConfig(measBase.SingleFrameMeasurementConfig):
97 """Measurement of detected diaSources as dipoles
98
99 Currently we keep the "old" DipoleMeasurement algorithms turned on.
100 """
101
102 def setDefaults(self):
103 measBase.SingleFrameMeasurementConfig.setDefaults(self)
104
105 self.plugins.names = ["base_CircularApertureFlux",
106 "base_PixelFlags",
107 "base_SkyCoord",
108 "base_PsfFlux",
109 "base_SdssCentroid",
110 "base_SdssShape",
111 "base_GaussianFlux",
112 "base_PeakLikelihoodFlux",
113 "base_PeakCentroid",
114 "base_NaiveCentroid",
115 "ip_diffim_NaiveDipoleCentroid",
116 "ip_diffim_NaiveDipoleFlux",
117 "ip_diffim_PsfDipoleFlux",
118 "ip_diffim_ClassificationDipole",
119 ]
120
121 self.slots.calibFlux = None
122 self.slots.modelFlux = None
123 self.slots.gaussianFlux = None
124 self.slots.shape = "base_SdssShape"
125 self.slots.centroid = "ip_diffim_NaiveDipoleCentroid"
126 self.doReplaceWithNoisedoReplaceWithNoise = False
127
128
129class DipoleFitTask(measBase.SingleFrameMeasurementTask):
130 """A task that fits a dipole to a difference image, with an optional separate detection image.
131
132 Because it subclasses SingleFrameMeasurementTask, and calls
133 SingleFrameMeasurementTask.run() from its run() method, it still
134 can be used identically to a standard SingleFrameMeasurementTask.
135 """
136
137 ConfigClass = DipoleFitTaskConfig
138 _DefaultName = "ip_diffim_DipoleFit"
139
140 def __init__(self, schema, algMetadata=None, **kwargs):
141
142 measBase.SingleFrameMeasurementTask.__init__(self, schema, algMetadata, **kwargs)
143
144 dpFitPluginConfig = self.config.plugins['ip_diffim_DipoleFit']
145
146 self.dipoleFitterdipoleFitter = DipoleFitPlugin(dpFitPluginConfig, name=self._DefaultName_DefaultName,
147 schema=schema, metadata=algMetadata)
148
149 @timeMethod
150 def run(self, sources, exposure, posExp=None, negExp=None, **kwargs):
151 """Run dipole measurement and classification
152
153 Parameters
154 ----------
156 ``diaSources`` that will be measured using dipole measurement
157 exposure : `lsst.afw.image.Exposure`
158 The difference exposure on which the ``diaSources`` of the ``sources`` parameter
159 were detected. If neither ``posExp`` nor ``negExp`` are set, then the dipole is also
160 fitted directly to this difference image.
161 posExp : `lsst.afw.image.Exposure`, optional
162 "Positive" exposure, typically a science exposure, or None if unavailable
163 When `posExp` is `None`, will compute `posImage = exposure + negExp`.
164 negExp : `lsst.afw.image.Exposure`, optional
165 "Negative" exposure, typically a template exposure, or None if unavailable
166 When `negExp` is `None`, will compute `negImage = posExp - exposure`.
167 **kwargs
168 Additional keyword arguments for `lsst.meas.base.sfm.SingleFrameMeasurementTask`.
169 """
170
171 measBase.SingleFrameMeasurementTask.run(self, sources, exposure, **kwargs)
172
173 if not sources:
174 return
175
176 for source in sources:
177 self.dipoleFitterdipoleFitter.measure(source, exposure, posExp, negExp)
178
179
180class DipoleModel(object):
181 """Lightweight class containing methods for generating a dipole model for fitting
182 to sources in diffims, used by DipoleFitAlgorithm.
183
184 See also:
185 `DMTN-007: Dipole characterization for image differencing <https://dmtn-007.lsst.io>`_.
186 """
187
188 def __init__(self):
189 import lsstDebug
190 self.debugdebug = lsstDebug.Info(__name__).debug
191 self.loglog = Log.getLogger(__name__)
192
193 def makeBackgroundModel(self, in_x, pars=None):
194 """Generate gradient model (2-d array) with up to 2nd-order polynomial
195
196 Parameters
197 ----------
198 in_x : `numpy.array`
199 (2, w, h)-dimensional `numpy.array`, containing the
200 input x,y meshgrid providing the coordinates upon which to
201 compute the gradient. This will typically be generated via
202 `_generateXYGrid()`. `w` and `h` correspond to the width and
203 height of the desired grid.
204 pars : `list` of `float`, optional
205 Up to 6 floats for up
206 to 6 2nd-order 2-d polynomial gradient parameters, in the
207 following order: (intercept, x, y, xy, x**2, y**2). If `pars`
208 is emtpy or `None`, do nothing and return `None` (for speed).
209
210 Returns
211 -------
212 result : `None` or `numpy.array`
213 return None, or 2-d numpy.array of width/height matching
214 input bbox, containing computed gradient values.
215 """
216
217 # Don't fit for other gradient parameters if the intercept is not included.
218 if (pars is None) or (len(pars) <= 0) or (pars[0] is None):
219 return
220
221 y, x = in_x[0, :], in_x[1, :]
222 gradient = np.full_like(x, pars[0], dtype='float64')
223 if len(pars) > 1 and pars[1] is not None:
224 gradient += pars[1] * x
225 if len(pars) > 2 and pars[2] is not None:
226 gradient += pars[2] * y
227 if len(pars) > 3 and pars[3] is not None:
228 gradient += pars[3] * (x * y)
229 if len(pars) > 4 and pars[4] is not None:
230 gradient += pars[4] * (x * x)
231 if len(pars) > 5 and pars[5] is not None:
232 gradient += pars[5] * (y * y)
233
234 return gradient
235
236 def _generateXYGrid(self, bbox):
237 """Generate a meshgrid covering the x,y coordinates bounded by bbox
238
239 Parameters
240 ----------
241 bbox : `lsst.geom.Box2I`
242 input Bounding Box defining the coordinate limits
243
244 Returns
245 -------
246 in_x : `numpy.array`
247 (2, w, h)-dimensional numpy array containing the grid indexing over x- and
248 y- coordinates
249 """
250
251 x, y = np.mgrid[bbox.getBeginY():bbox.getEndY(), bbox.getBeginX():bbox.getEndX()]
252 in_x = np.array([y, x]).astype(np.float64)
253 in_x[0, :] -= np.mean(in_x[0, :])
254 in_x[1, :] -= np.mean(in_x[1, :])
255 return in_x
256
257 def _getHeavyFootprintSubimage(self, fp, badfill=np.nan, grow=0):
258 """Extract the image from a ``~lsst.afw.detection.HeavyFootprint``
259 as an `lsst.afw.image.ImageF`.
260
261 Parameters
262 ----------
264 HeavyFootprint to use to generate the subimage
265 badfill : `float`, optional
266 Value to fill in pixels in extracted image that are outside the footprint
267 grow : `int`
268 Optionally grow the footprint by this amount before extraction
269
270 Returns
271 -------
272 subim2 : `lsst.afw.image.ImageF`
273 An `~lsst.afw.image.ImageF` containing the subimage.
274 """
275 bbox = fp.getBBox()
276 if grow > 0:
277 bbox.grow(grow)
278
279 subim2 = afwImage.ImageF(bbox, badfill)
280 fp.getSpans().unflatten(subim2.getArray(), fp.getImageArray(), bbox.getCorners()[0])
281 return subim2
282
283 def fitFootprintBackground(self, source, posImage, order=1):
284 """Fit a linear (polynomial) model of given order (max 2) to the background of a footprint.
285
286 Only fit the pixels OUTSIDE of the footprint, but within its bounding box.
287
288 Parameters
289 ----------
291 SourceRecord, the footprint of which is to be fit
292 posImage : `lsst.afw.image.Exposure`
293 The exposure from which to extract the footprint subimage
294 order : `int`
295 Polynomial order of background gradient to fit.
296
297 Returns
298 -------
299 pars : `tuple` of `float`
300 `tuple` of length (1 if order==0; 3 if order==1; 6 if order == 2),
301 containing the resulting fit parameters
302 """
303
304 # TODO look into whether to use afwMath background methods -- see
305 # http://lsst-web.ncsa.illinois.edu/doxygen/x_masterDoxyDoc/_background_example.html
306 fp = source.getFootprint()
307 bbox = fp.getBBox()
308 bbox.grow(3)
309 posImg = afwImage.ImageF(posImage.getMaskedImage().getImage(), bbox, afwImage.PARENT)
310
311 # This code constructs the footprint image so that we can identify the pixels that are
312 # outside the footprint (but within the bounding box). These are the pixels used for
313 # fitting the background.
314 posHfp = afwDet.HeavyFootprintF(fp, posImage.getMaskedImage())
315 posFpImg = self._getHeavyFootprintSubimage_getHeavyFootprintSubimage(posHfp, grow=3)
316
317 isBg = np.isnan(posFpImg.getArray()).ravel()
318
319 data = posImg.getArray().ravel()
320 data = data[isBg]
321 B = data
322
323 x, y = np.mgrid[bbox.getBeginY():bbox.getEndY(), bbox.getBeginX():bbox.getEndX()]
324 x = x.astype(np.float64).ravel()
325 x -= np.mean(x)
326 x = x[isBg]
327 y = y.astype(np.float64).ravel()
328 y -= np.mean(y)
329 y = y[isBg]
330 b = np.ones_like(x, dtype=np.float64)
331
332 M = np.vstack([b]).T # order = 0
333 if order == 1:
334 M = np.vstack([b, x, y]).T
335 elif order == 2:
336 M = np.vstack([b, x, y, x**2., y**2., x*y]).T
337
338 pars = np.linalg.lstsq(M, B, rcond=-1)[0]
339 return pars
340
341 def makeStarModel(self, bbox, psf, xcen, ycen, flux):
342 """Generate a 2D image model of a single PDF centered at the given coordinates.
343
344 Parameters
345 ----------
346 bbox : `lsst.geom.Box`
347 Bounding box marking pixel coordinates for generated model
348 psf : TODO: DM-17458
349 Psf model used to generate the 'star'
350 xcen : `float`
351 Desired x-centroid of the 'star'
352 ycen : `float`
353 Desired y-centroid of the 'star'
354 flux : `float`
355 Desired flux of the 'star'
356
357 Returns
358 -------
359 p_Im : `lsst.afw.image.Image`
360 2-d stellar image of width/height matching input ``bbox``,
361 containing PSF with given centroid and flux
362 """
363
364 # Generate the psf image, normalize to flux
365 psf_img = psf.computeImage(geom.Point2D(xcen, ycen)).convertF()
366 psf_img_sum = np.nansum(psf_img.getArray())
367 psf_img *= (flux/psf_img_sum)
368
369 # Clip the PSF image bounding box to fall within the footprint bounding box
370 psf_box = psf_img.getBBox()
371 psf_box.clip(bbox)
372 psf_img = afwImage.ImageF(psf_img, psf_box, afwImage.PARENT)
373
374 # Then actually crop the psf image.
375 # Usually not necessary, but if the dipole is near the edge of the image...
376 # Would be nice if we could compare original pos_box with clipped pos_box and
377 # see if it actually was clipped.
378 p_Im = afwImage.ImageF(bbox)
379 tmpSubim = afwImage.ImageF(p_Im, psf_box, afwImage.PARENT)
380 tmpSubim += psf_img
381
382 return p_Im
383
384 def makeModel(self, x, flux, xcenPos, ycenPos, xcenNeg, ycenNeg, fluxNeg=None,
385 b=None, x1=None, y1=None, xy=None, x2=None, y2=None,
386 bNeg=None, x1Neg=None, y1Neg=None, xyNeg=None, x2Neg=None, y2Neg=None,
387 **kwargs):
388 """Generate dipole model with given parameters.
389
390 This is the function whose sum-of-squared difference from data
391 is minimized by `lmfit`.
392
393 x : TODO: DM-17458
394 Input independent variable. Used here as the grid on
395 which to compute the background gradient model.
396 flux : `float`
397 Desired flux of the positive lobe of the dipole
398 xcenPos : `float`
399 Desired x-centroid of the positive lobe of the dipole
400 ycenPos : `float`
401 Desired y-centroid of the positive lobe of the dipole
402 xcenNeg : `float`
403 Desired x-centroid of the negative lobe of the dipole
404 ycenNeg : `float`
405 Desired y-centroid of the negative lobe of the dipole
406 fluxNeg : `float`, optional
407 Desired flux of the negative lobe of the dipole, set to 'flux' if None
408 b, x1, y1, xy, x2, y2 : `float`
409 Gradient parameters for positive lobe.
410 bNeg, x1Neg, y1Neg, xyNeg, x2Neg, y2Neg : `float`, optional
411 Gradient parameters for negative lobe.
412 They are set to the corresponding positive values if None.
413
414 **kwargs
415 Keyword arguments passed through ``lmfit`` and
416 used by this function. These must include:
417
418 - ``psf`` Psf model used to generate the 'star'
419 - ``rel_weight`` Used to signify least-squares weighting of posImage/negImage
420 relative to diffim. If ``rel_weight == 0`` then posImage/negImage are ignored.
421 - ``bbox`` Bounding box containing region to be modelled
422
423 Returns
424 -------
425 zout : `numpy.array`
426 Has width and height matching the input bbox, and
427 contains the dipole model with given centroids and flux(es). If
428 ``rel_weight`` = 0, this is a 2-d array with dimensions matching
429 those of bbox; otherwise a stack of three such arrays,
430 representing the dipole (diffim), positive and negative images
431 respectively.
432 """
433
434 psf = kwargs.get('psf')
435 rel_weight = kwargs.get('rel_weight') # if > 0, we're including pre-sub. images
436 fp = kwargs.get('footprint')
437 bbox = fp.getBBox()
438
439 if fluxNeg is None:
440 fluxNeg = flux
441
442 if self.debugdebug:
443 self.loglog.debug('%.2f %.2f %.2f %.2f %.2f %.2f',
444 flux, fluxNeg, xcenPos, ycenPos, xcenNeg, ycenNeg)
445 if x1 is not None:
446 self.loglog.debug(' %.2f %.2f %.2f', b, x1, y1)
447 if xy is not None:
448 self.loglog.debug(' %.2f %.2f %.2f', xy, x2, y2)
449
450 posIm = self.makeStarModelmakeStarModel(bbox, psf, xcenPos, ycenPos, flux)
451 negIm = self.makeStarModelmakeStarModel(bbox, psf, xcenNeg, ycenNeg, fluxNeg)
452
453 in_x = x
454 if in_x is None: # use the footprint to generate the input grid
455 y, x = np.mgrid[bbox.getBeginY():bbox.getEndY(), bbox.getBeginX():bbox.getEndX()]
456 in_x = np.array([x, y]) * 1.
457 in_x[0, :] -= in_x[0, :].mean() # center it!
458 in_x[1, :] -= in_x[1, :].mean()
459
460 if b is not None:
461 gradient = self.makeBackgroundModelmakeBackgroundModel(in_x, (b, x1, y1, xy, x2, y2))
462
463 # If bNeg is None, then don't fit the negative background separately
464 if bNeg is not None:
465 gradientNeg = self.makeBackgroundModelmakeBackgroundModel(in_x, (bNeg, x1Neg, y1Neg, xyNeg, x2Neg, y2Neg))
466 else:
467 gradientNeg = gradient
468
469 posIm.getArray()[:, :] += gradient
470 negIm.getArray()[:, :] += gradientNeg
471
472 # Generate the diffIm model
473 diffIm = afwImage.ImageF(bbox)
474 diffIm += posIm
475 diffIm -= negIm
476
477 zout = diffIm.getArray()
478 if rel_weight > 0.:
479 zout = np.append([zout], [posIm.getArray(), negIm.getArray()], axis=0)
480
481 return zout
482
483
484class DipoleFitAlgorithm(object):
485 """Fit a dipole model using an image difference.
486
487 See also:
488 `DMTN-007: Dipole characterization for image differencing <https://dmtn-007.lsst.io>`_.
489 """
490
491 # This is just a private version number to sync with the ipython notebooks that I have been
492 # using for algorithm development.
493 _private_version_ = '0.0.5'
494
495 # Below is a (somewhat incomplete) list of improvements
496 # that would be worth investigating, given the time:
497
498 # todo 1. evaluate necessity for separate parameters for pos- and neg- images
499 # todo 2. only fit background OUTSIDE footprint (DONE) and dipole params INSIDE footprint (NOT DONE)?
500 # todo 3. correct normalization of least-squares weights based on variance planes
501 # todo 4. account for PSFs that vary across the exposures (should be happening by default?)
502 # todo 5. correctly account for NA/masks (i.e., ignore!)
503 # todo 6. better exception handling in the plugin
504 # todo 7. better classification of dipoles (e.g. by comparing chi2 fit vs. monopole?)
505 # todo 8. (DONE) Initial fast estimate of background gradient(s) params -- perhaps using numpy.lstsq
506 # todo 9. (NOT NEEDED - see (2)) Initial fast test whether a background gradient needs to be fit
507 # todo 10. (DONE) better initial estimate for flux when there's a strong gradient
508 # todo 11. (DONE) requires a new package `lmfit` -- investiate others? (astropy/scipy/iminuit?)
509
510 def __init__(self, diffim, posImage=None, negImage=None):
511 """Algorithm to run dipole measurement on a diaSource
512
513 Parameters
514 ----------
515 diffim : `lsst.afw.image.Exposure`
516 Exposure on which the diaSources were detected
517 posImage : `lsst.afw.image.Exposure`
518 "Positive" exposure from which the template was subtracted
519 negImage : `lsst.afw.image.Exposure`
520 "Negative" exposure which was subtracted from the posImage
521 """
522
523 self.diffimdiffim = diffim
524 self.posImageposImage = posImage
525 self.negImagenegImage = negImage
526 self.psfSigmapsfSigma = None
527 if diffim is not None:
528 self.psfSigmapsfSigma = diffim.getPsf().computeShape().getDeterminantRadius()
529
530 self.loglog = Log.getLogger(__name__)
531
532 import lsstDebug
533 self.debugdebug = lsstDebug.Info(__name__).debug
534
535 def fitDipoleImpl(self, source, tol=1e-7, rel_weight=0.5,
536 fitBackground=1, bgGradientOrder=1, maxSepInSigma=5.,
537 separateNegParams=True, verbose=False):
538 """Fit a dipole model to an input difference image.
539
540 Actually, fits the subimage bounded by the input source's
541 footprint) and optionally constrain the fit using the
542 pre-subtraction images posImage and negImage.
543
544 Parameters
545 ----------
546 source : TODO: DM-17458
547 TODO: DM-17458
548 tol : float, optional
549 TODO: DM-17458
550 rel_weight : `float`, optional
551 TODO: DM-17458
552 fitBackground : `int`, optional
553 TODO: DM-17458
554 bgGradientOrder : `int`, optional
555 TODO: DM-17458
556 maxSepInSigma : `float`, optional
557 TODO: DM-17458
558 separateNegParams : `bool`, optional
559 TODO: DM-17458
560 verbose : `bool`, optional
561 TODO: DM-17458
562
563 Returns
564 -------
565 result : `lmfit.MinimizerResult`
566 return `lmfit.MinimizerResult` object containing the fit
567 parameters and other information.
568 """
569
570 # Only import lmfit if someone wants to use the new DipoleFitAlgorithm.
571 import lmfit
572
573 fp = source.getFootprint()
574 bbox = fp.getBBox()
575 subim = afwImage.MaskedImageF(self.diffimdiffim.getMaskedImage(), bbox=bbox, origin=afwImage.PARENT)
576
577 z = diArr = subim.getArrays()[0]
578 weights = 1. / subim.getArrays()[2] # get the weights (=1/variance)
579
580 if rel_weight > 0. and ((self.posImageposImage is not None) or (self.negImagenegImage is not None)):
581 if self.negImagenegImage is not None:
582 negSubim = afwImage.MaskedImageF(self.negImagenegImage.getMaskedImage(), bbox, origin=afwImage.PARENT)
583 if self.posImageposImage is not None:
584 posSubim = afwImage.MaskedImageF(self.posImageposImage.getMaskedImage(), bbox, origin=afwImage.PARENT)
585 if self.posImageposImage is None: # no science image provided; generate it from diffim + negImage
586 posSubim = subim.clone()
587 posSubim += negSubim
588 if self.negImagenegImage is None: # no template provided; generate it from the posImage - diffim
589 negSubim = posSubim.clone()
590 negSubim -= subim
591
592 z = np.append([z], [posSubim.getArrays()[0],
593 negSubim.getArrays()[0]], axis=0)
594 # Weight the pos/neg images by rel_weight relative to the diffim
595 weights = np.append([weights], [1. / posSubim.getArrays()[2] * rel_weight,
596 1. / negSubim.getArrays()[2] * rel_weight], axis=0)
597 else:
598 rel_weight = 0. # a short-cut for "don't include the pre-subtraction data"
599
600 # It seems that `lmfit` requires a static functor as its optimized method, which eliminates
601 # the ability to pass a bound method or other class method. Here we write a wrapper which
602 # makes this possible.
603 def dipoleModelFunctor(x, flux, xcenPos, ycenPos, xcenNeg, ycenNeg, fluxNeg=None,
604 b=None, x1=None, y1=None, xy=None, x2=None, y2=None,
605 bNeg=None, x1Neg=None, y1Neg=None, xyNeg=None, x2Neg=None, y2Neg=None,
606 **kwargs):
607 """Generate dipole model with given parameters.
608
609 It simply defers to `modelObj.makeModel()`, where `modelObj` comes
610 out of `kwargs['modelObj']`.
611 """
612 modelObj = kwargs.pop('modelObj')
613 return modelObj.makeModel(x, flux, xcenPos, ycenPos, xcenNeg, ycenNeg, fluxNeg=fluxNeg,
614 b=b, x1=x1, y1=y1, xy=xy, x2=x2, y2=y2,
615 bNeg=bNeg, x1Neg=x1Neg, y1Neg=y1Neg, xyNeg=xyNeg,
616 x2Neg=x2Neg, y2Neg=y2Neg, **kwargs)
617
618 dipoleModel = DipoleModel()
619
620 modelFunctor = dipoleModelFunctor # dipoleModel.makeModel does not work for now.
621 # Create the lmfit model (lmfit uses scipy 'leastsq' option by default - Levenberg-Marquardt)
622 # Note we can also tell it to drop missing values from the data.
623 gmod = lmfit.Model(modelFunctor, verbose=verbose, missing='drop')
624 # independent_vars=independent_vars) #, param_names=param_names)
625
626 # Add the constraints for centroids, fluxes.
627 # starting constraint - near centroid of footprint
628 fpCentroid = np.array([fp.getCentroid().getX(), fp.getCentroid().getY()])
629 cenNeg = cenPos = fpCentroid
630
631 pks = fp.getPeaks()
632
633 if len(pks) >= 1:
634 cenPos = pks[0].getF() # if individual (merged) peaks were detected, use those
635 if len(pks) >= 2: # peaks are already sorted by centroid flux so take the most negative one
636 cenNeg = pks[-1].getF()
637
638 # For close/faint dipoles the starting locs (min/max) might be way off, let's help them a bit.
639 # First assume dipole is not separated by more than 5*psfSigma.
640 maxSep = self.psfSigmapsfSigma * maxSepInSigma
641
642 # As an initial guess -- assume the dipole is close to the center of the footprint.
643 if np.sum(np.sqrt((np.array(cenPos) - fpCentroid)**2.)) > maxSep:
644 cenPos = fpCentroid
645 if np.sum(np.sqrt((np.array(cenNeg) - fpCentroid)**2.)) > maxSep:
646 cenPos = fpCentroid
647
648 # parameter hints/constraints: https://lmfit.github.io/lmfit-py/model.html#model-param-hints-section
649 # might make sense to not use bounds -- see http://lmfit.github.io/lmfit-py/bounds.html
650 # also see this discussion -- https://github.com/scipy/scipy/issues/3129
651 gmod.set_param_hint('xcenPos', value=cenPos[0],
652 min=cenPos[0]-maxSep, max=cenPos[0]+maxSep)
653 gmod.set_param_hint('ycenPos', value=cenPos[1],
654 min=cenPos[1]-maxSep, max=cenPos[1]+maxSep)
655 gmod.set_param_hint('xcenNeg', value=cenNeg[0],
656 min=cenNeg[0]-maxSep, max=cenNeg[0]+maxSep)
657 gmod.set_param_hint('ycenNeg', value=cenNeg[1],
658 min=cenNeg[1]-maxSep, max=cenNeg[1]+maxSep)
659
660 # Use the (flux under the dipole)*5 for an estimate.
661 # Lots of testing showed that having startingFlux be too high was better than too low.
662 startingFlux = np.nansum(np.abs(diArr) - np.nanmedian(np.abs(diArr))) * 5.
663 posFlux = negFlux = startingFlux
664
665 # TBD: set max. flux limit?
666 gmod.set_param_hint('flux', value=posFlux, min=0.1)
667
668 if separateNegParams:
669 # TBD: set max negative lobe flux limit?
670 gmod.set_param_hint('fluxNeg', value=np.abs(negFlux), min=0.1)
671
672 # Fixed parameters (don't fit for them if there are no pre-sub images or no gradient fit requested):
673 # Right now (fitBackground == 1), we fit a linear model to the background and then subtract
674 # it from the data and then don't fit the background again (this is faster).
675 # A slower alternative (fitBackground == 2) is to use the estimated background parameters as
676 # starting points in the integrated model fit. That is currently not performed by default,
677 # but might be desirable in some cases.
678 bgParsPos = bgParsNeg = (0., 0., 0.)
679 if ((rel_weight > 0.) and (fitBackground != 0) and (bgGradientOrder >= 0)):
680 pbg = 0.
681 bgFitImage = self.posImageposImage if self.posImageposImage is not None else self.negImagenegImage
682 # Fit the gradient to the background (linear model)
683 bgParsPos = bgParsNeg = dipoleModel.fitFootprintBackground(source, bgFitImage,
684 order=bgGradientOrder)
685
686 # Generate the gradient and subtract it from the pre-subtraction image data
687 if fitBackground == 1:
688 in_x = dipoleModel._generateXYGrid(bbox)
689 pbg = dipoleModel.makeBackgroundModel(in_x, tuple(bgParsPos))
690 z[1, :] -= pbg
691 z[1, :] -= np.nanmedian(z[1, :])
692 posFlux = np.nansum(z[1, :])
693 gmod.set_param_hint('flux', value=posFlux*1.5, min=0.1)
694
695 if separateNegParams and self.negImagenegImage is not None:
696 bgParsNeg = dipoleModel.fitFootprintBackground(source, self.negImagenegImage,
697 order=bgGradientOrder)
698 pbg = dipoleModel.makeBackgroundModel(in_x, tuple(bgParsNeg))
699 z[2, :] -= pbg
700 z[2, :] -= np.nanmedian(z[2, :])
701 if separateNegParams:
702 negFlux = np.nansum(z[2, :])
703 gmod.set_param_hint('fluxNeg', value=negFlux*1.5, min=0.1)
704
705 # Do not subtract the background from the images but include the background parameters in the fit
706 if fitBackground == 2:
707 if bgGradientOrder >= 0:
708 gmod.set_param_hint('b', value=bgParsPos[0])
709 if separateNegParams:
710 gmod.set_param_hint('bNeg', value=bgParsNeg[0])
711 if bgGradientOrder >= 1:
712 gmod.set_param_hint('x1', value=bgParsPos[1])
713 gmod.set_param_hint('y1', value=bgParsPos[2])
714 if separateNegParams:
715 gmod.set_param_hint('x1Neg', value=bgParsNeg[1])
716 gmod.set_param_hint('y1Neg', value=bgParsNeg[2])
717 if bgGradientOrder >= 2:
718 gmod.set_param_hint('xy', value=bgParsPos[3])
719 gmod.set_param_hint('x2', value=bgParsPos[4])
720 gmod.set_param_hint('y2', value=bgParsPos[5])
721 if separateNegParams:
722 gmod.set_param_hint('xyNeg', value=bgParsNeg[3])
723 gmod.set_param_hint('x2Neg', value=bgParsNeg[4])
724 gmod.set_param_hint('y2Neg', value=bgParsNeg[5])
725
726 y, x = np.mgrid[bbox.getBeginY():bbox.getEndY(), bbox.getBeginX():bbox.getEndX()]
727 in_x = np.array([x, y]).astype(np.float)
728 in_x[0, :] -= in_x[0, :].mean() # center it!
729 in_x[1, :] -= in_x[1, :].mean()
730
731 # Instead of explicitly using a mask to ignore flagged pixels, just set the ignored pixels'
732 # weights to 0 in the fit. TBD: need to inspect mask planes to set this mask.
733 mask = np.ones_like(z, dtype=bool) # TBD: set mask values to False if the pixels are to be ignored
734
735 # I'm not sure about the variance planes in the diffim (or convolved pre-sub. images
736 # for that matter) so for now, let's just do an un-weighted least-squares fit
737 # (override weights computed above).
738 weights = mask.astype(np.float64)
739 if self.posImageposImage is not None and rel_weight > 0.:
740 weights = np.array([np.ones_like(diArr), np.ones_like(diArr)*rel_weight,
741 np.ones_like(diArr)*rel_weight])
742
743 # Set the weights to zero if mask is False
744 if np.any(~mask):
745 weights[~mask] = 0.
746
747 # Note that although we can, we're not required to set initial values for params here,
748 # since we set their param_hint's above.
749 # Can add "method" param to not use 'leastsq' (==levenberg-marquardt), e.g. "method='nelder'"
750 with warnings.catch_warnings():
751 warnings.simplefilter("ignore") # temporarily turn off silly lmfit warnings
752 result = gmod.fit(z, weights=weights, x=in_x,
753 verbose=verbose,
754 fit_kws={'ftol': tol, 'xtol': tol, 'gtol': tol,
755 'maxfev': 250}, # see scipy docs
756 psf=self.diffimdiffim.getPsf(), # hereon: kwargs that get passed to genDipoleModel()
757 rel_weight=rel_weight,
758 footprint=fp,
759 modelObj=dipoleModel)
760
761 if verbose: # the ci_report() seems to fail if neg params are constrained -- TBD why.
762 # Never wanted in production - this takes a long time (longer than the fit!)
763 # This is how to get confidence intervals out:
764 # https://lmfit.github.io/lmfit-py/confidence.html and
765 # http://cars9.uchicago.edu/software/python/lmfit/model.html
766 print(result.fit_report(show_correl=False))
767 if separateNegParams:
768 print(result.ci_report())
769
770 return result
771
772 def fitDipole(self, source, tol=1e-7, rel_weight=0.1,
773 fitBackground=1, maxSepInSigma=5., separateNegParams=True,
774 bgGradientOrder=1, verbose=False, display=False):
775 """Fit a dipole model to an input ``diaSource`` (wraps `fitDipoleImpl`).
776
777 Actually, fits the subimage bounded by the input source's
778 footprint) and optionally constrain the fit using the
779 pre-subtraction images self.posImageposImage (science) and
780 self.negImagenegImage (template). Wraps the output into a
781 `pipeBase.Struct` named tuple after computing additional
782 statistics such as orientation and SNR.
783
784 Parameters
785 ----------
787 Record containing the (merged) dipole source footprint detected on the diffim
788 tol : `float`, optional
789 Tolerance parameter for scipy.leastsq() optimization
790 rel_weight : `float`, optional
791 Weighting of posImage/negImage relative to the diffim in the fit
792 fitBackground : `int`, {0, 1, 2}, optional
793 How to fit linear background gradient in posImage/negImage
794
795 - 0: do not fit background at all
796 - 1 (default): pre-fit the background using linear least squares and then do not fit it
797 as part of the dipole fitting optimization
798 - 2: pre-fit the background using linear least squares (as in 1), and use the parameter
799 estimates from that fit as starting parameters for an integrated "re-fit" of the
800 background as part of the overall dipole fitting optimization.
801 maxSepInSigma : `float`, optional
802 Allowed window of centroid parameters relative to peak in input source footprint
803 separateNegParams : `bool`, optional
804 Fit separate parameters to the flux and background gradient in
805 bgGradientOrder : `int`, {0, 1, 2}, optional
806 Desired polynomial order of background gradient
807 verbose: `bool`, optional
808 Be verbose
809 display
810 Display input data, best fit model(s) and residuals in a matplotlib window.
811
812 Returns
813 -------
814 result : `struct`
815 `pipeBase.Struct` object containing the fit parameters and other information.
816
817 result : `callable`
818 `lmfit.MinimizerResult` object for debugging and error estimation, etc.
819
820 Notes
821 -----
822 Parameter `fitBackground` has three options, thus it is an integer:
823
824 """
825
826 fitResult = self.fitDipoleImplfitDipoleImpl(
827 source, tol=tol, rel_weight=rel_weight, fitBackground=fitBackground,
828 maxSepInSigma=maxSepInSigma, separateNegParams=separateNegParams,
829 bgGradientOrder=bgGradientOrder, verbose=verbose)
830
831 # Display images, model fits and residuals (currently uses matplotlib display functions)
832 if display:
833 fp = source.getFootprint()
834 self.displayFitResultsdisplayFitResults(fp, fitResult)
835
836 fitParams = fitResult.best_values
837 if fitParams['flux'] <= 1.: # usually around 0.1 -- the minimum flux allowed -- i.e. bad fit.
838 out = Struct(posCentroidX=np.nan, posCentroidY=np.nan,
839 negCentroidX=np.nan, negCentroidY=np.nan,
840 posFlux=np.nan, negFlux=np.nan, posFluxErr=np.nan, negFluxErr=np.nan,
841 centroidX=np.nan, centroidY=np.nan, orientation=np.nan,
842 signalToNoise=np.nan, chi2=np.nan, redChi2=np.nan)
843 return out, fitResult
844
845 centroid = ((fitParams['xcenPos'] + fitParams['xcenNeg']) / 2.,
846 (fitParams['ycenPos'] + fitParams['ycenNeg']) / 2.)
847 dx, dy = fitParams['xcenPos'] - fitParams['xcenNeg'], fitParams['ycenPos'] - fitParams['ycenNeg']
848 angle = np.arctan2(dy, dx) / np.pi * 180. # convert to degrees (should keep as rad?)
849
850 # Exctract flux value, compute signalToNoise from flux/variance_within_footprint
851 # Also extract the stderr of flux estimate.
852 def computeSumVariance(exposure, footprint):
853 box = footprint.getBBox()
854 subim = afwImage.MaskedImageF(exposure.getMaskedImage(), box, origin=afwImage.PARENT)
855 return np.sqrt(np.nansum(subim.getArrays()[1][:, :]))
856
857 fluxVal = fluxVar = fitParams['flux']
858 fluxErr = fluxErrNeg = fitResult.params['flux'].stderr
859 if self.posImageposImage is not None:
860 fluxVar = computeSumVariance(self.posImageposImage, source.getFootprint())
861 else:
862 fluxVar = computeSumVariance(self.diffimdiffim, source.getFootprint())
863
864 fluxValNeg, fluxVarNeg = fluxVal, fluxVar
865 if separateNegParams:
866 fluxValNeg = fitParams['fluxNeg']
867 fluxErrNeg = fitResult.params['fluxNeg'].stderr
868 if self.negImagenegImage is not None:
869 fluxVarNeg = computeSumVariance(self.negImagenegImage, source.getFootprint())
870
871 try:
872 signalToNoise = np.sqrt((fluxVal/fluxVar)**2 + (fluxValNeg/fluxVarNeg)**2)
873 except ZeroDivisionError: # catch divide by zero - should never happen.
874 signalToNoise = np.nan
875
876 out = Struct(posCentroidX=fitParams['xcenPos'], posCentroidY=fitParams['ycenPos'],
877 negCentroidX=fitParams['xcenNeg'], negCentroidY=fitParams['ycenNeg'],
878 posFlux=fluxVal, negFlux=-fluxValNeg, posFluxErr=fluxErr, negFluxErr=fluxErrNeg,
879 centroidX=centroid[0], centroidY=centroid[1], orientation=angle,
880 signalToNoise=signalToNoise, chi2=fitResult.chisqr, redChi2=fitResult.redchi)
881
882 # fitResult may be returned for debugging
883 return out, fitResult
884
885 def displayFitResults(self, footprint, result):
886 """Display data, model fits and residuals (currently uses matplotlib display functions).
887
888 Parameters
889 ----------
890 footprint : TODO: DM-17458
891 Footprint containing the dipole that was fit
892 result : `lmfit.MinimizerResult`
893 `lmfit.MinimizerResult` object returned by `lmfit` optimizer
894
895 Returns
896 -------
897 fig : `matplotlib.pyplot.plot`
898 """
899 try:
900 import matplotlib.pyplot as plt
901 except ImportError as err:
902 self.loglog.warning('Unable to import matplotlib: %s', err)
903 raise err
904
905 def display2dArray(arr, title='Data', extent=None):
906 """Use `matplotlib.pyplot.imshow` to display a 2-D array with a given coordinate range.
907 """
908 fig = plt.imshow(arr, origin='lower', interpolation='none', cmap='gray', extent=extent)
909 plt.title(title)
910 plt.colorbar(fig, cmap='gray')
911 return fig
912
913 z = result.data
914 fit = result.best_fit
915 bbox = footprint.getBBox()
916 extent = (bbox.getBeginX(), bbox.getEndX(), bbox.getBeginY(), bbox.getEndY())
917 if z.shape[0] == 3:
918 fig = plt.figure(figsize=(8, 8))
919 for i in range(3):
920 plt.subplot(3, 3, i*3+1)
921 display2dArray(z[i, :], 'Data', extent=extent)
922 plt.subplot(3, 3, i*3+2)
923 display2dArray(fit[i, :], 'Model', extent=extent)
924 plt.subplot(3, 3, i*3+3)
925 display2dArray(z[i, :] - fit[i, :], 'Residual', extent=extent)
926 return fig
927 else:
928 fig = plt.figure(figsize=(8, 2.5))
929 plt.subplot(1, 3, 1)
930 display2dArray(z, 'Data', extent=extent)
931 plt.subplot(1, 3, 2)
932 display2dArray(fit, 'Model', extent=extent)
933 plt.subplot(1, 3, 3)
934 display2dArray(z - fit, 'Residual', extent=extent)
935 return fig
936
937 plt.show()
938
939
940@measBase.register("ip_diffim_DipoleFit")
941class DipoleFitPlugin(measBase.SingleFramePlugin):
942 """A single frame measurement plugin that fits dipoles to all merged (two-peak) ``diaSources``.
943
944 This measurement plugin accepts up to three input images in
945 its `measure` method. If these are provided, it includes data
946 from the pre-subtraction posImage (science image) and optionally
947 negImage (template image) to constrain the fit. The meat of the
948 fitting routines are in the class `~lsst.module.name.DipoleFitAlgorithm`.
949
950 Notes
951 -----
952 The motivation behind this plugin and the necessity for including more than
953 one exposure are documented in DMTN-007 (http://dmtn-007.lsst.io).
954
955 This class is named `ip_diffim_DipoleFit` so that it may be used alongside
956 the existing `ip_diffim_DipoleMeasurement` classes until such a time as those
957 are deemed to be replaceable by this.
958 """
959
960 ConfigClass = DipoleFitPluginConfig
961 DipoleFitAlgorithmClass = DipoleFitAlgorithm # Pointer to the class that performs the fit
962
963 FAILURE_EDGE = 1 # too close to the edge
964 FAILURE_FIT = 2 # failure in the fitting
965 FAILURE_NOT_DIPOLE = 4 # input source is not a putative dipole to begin with
966
967 @classmethod
969 """Set execution order to `FLUX_ORDER`.
970
971 This includes algorithms that require both `getShape()` and `getCentroid()`,
972 in addition to a Footprint and its Peaks.
973 """
974 return cls.FLUX_ORDER
975
976 def __init__(self, config, name, schema, metadata):
977 measBase.SingleFramePlugin.__init__(self, config, name, schema, metadata)
978
979 self.loglog = Log.getLogger(name)
980
981 self._setupSchema_setupSchema(config, name, schema, metadata)
982
983 def _setupSchema(self, config, name, schema, metadata):
984 # Get a FunctorKey that can quickly look up the "blessed" centroid value.
985 self.centroidKeycentroidKey = afwTable.Point2DKey(schema["slot_Centroid"])
986
987 # Add some fields for our outputs, and save their Keys.
988 # Use setattr() to programmatically set the pos/neg named attributes to values, e.g.
989 # self.posCentroidKeyX = 'ip_diffim_DipoleFit_pos_centroid_x'
990
991 for pos_neg in ['pos', 'neg']:
992
993 key = schema.addField(
994 schema.join(name, pos_neg, "instFlux"), type=float, units="count",
995 doc="Dipole {0} lobe flux".format(pos_neg))
996 setattr(self, ''.join((pos_neg, 'FluxKey')), key)
997
998 key = schema.addField(
999 schema.join(name, pos_neg, "instFluxErr"), type=float, units="count",
1000 doc="1-sigma uncertainty for {0} dipole flux".format(pos_neg))
1001 setattr(self, ''.join((pos_neg, 'FluxErrKey')), key)
1002
1003 for x_y in ['x', 'y']:
1004 key = schema.addField(
1005 schema.join(name, pos_neg, "centroid", x_y), type=float, units="pixel",
1006 doc="Dipole {0} lobe centroid".format(pos_neg))
1007 setattr(self, ''.join((pos_neg, 'CentroidKey', x_y.upper())), key)
1008
1009 for x_y in ['x', 'y']:
1010 key = schema.addField(
1011 schema.join(name, "centroid", x_y), type=float, units="pixel",
1012 doc="Dipole centroid")
1013 setattr(self, ''.join(('centroidKey', x_y.upper())), key)
1014
1015 self.fluxKeyfluxKey = schema.addField(
1016 schema.join(name, "instFlux"), type=float, units="count",
1017 doc="Dipole overall flux")
1018
1019 self.orientationKeyorientationKey = schema.addField(
1020 schema.join(name, "orientation"), type=float, units="deg",
1021 doc="Dipole orientation")
1022
1023 self.separationKeyseparationKey = schema.addField(
1024 schema.join(name, "separation"), type=float, units="pixel",
1025 doc="Pixel separation between positive and negative lobes of dipole")
1026
1027 self.chi2dofKeychi2dofKey = schema.addField(
1028 schema.join(name, "chi2dof"), type=float,
1029 doc="Chi2 per degree of freedom of dipole fit")
1030
1031 self.signalToNoiseKeysignalToNoiseKey = schema.addField(
1032 schema.join(name, "signalToNoise"), type=float,
1033 doc="Estimated signal-to-noise of dipole fit")
1034
1035 self.classificationFlagKeyclassificationFlagKey = schema.addField(
1036 schema.join(name, "flag", "classification"), type="Flag",
1037 doc="Flag indicating diaSource is classified as a dipole")
1038
1039 self.classificationAttemptedFlagKeyclassificationAttemptedFlagKey = schema.addField(
1040 schema.join(name, "flag", "classificationAttempted"), type="Flag",
1041 doc="Flag indicating diaSource was attempted to be classified as a dipole")
1042
1043 self.flagKeyflagKey = schema.addField(
1044 schema.join(name, "flag"), type="Flag",
1045 doc="General failure flag for dipole fit")
1046
1047 self.edgeFlagKeyedgeFlagKey = schema.addField(
1048 schema.join(name, "flag", "edge"), type="Flag",
1049 doc="Flag set when dipole is too close to edge of image")
1050
1051 def measure(self, measRecord, exposure, posExp=None, negExp=None):
1052 """Perform the non-linear least squares minimization on the putative dipole source.
1053
1054 Parameters
1055 ----------
1056 measRecord : `lsst.afw.table.SourceRecord`
1057 diaSources that will be measured using dipole measurement
1058 exposure : `lsst.afw.image.Exposure`
1059 Difference exposure on which the diaSources were detected; `exposure = posExp-negExp`
1060 If both `posExp` and `negExp` are `None`, will attempt to fit the
1061 dipole to just the `exposure` with no constraint.
1062 posExp : `lsst.afw.image.Exposure`, optional
1063 "Positive" exposure, typically a science exposure, or None if unavailable
1064 When `posExp` is `None`, will compute `posImage = exposure + negExp`.
1065 negExp : `lsst.afw.image.Exposure`, optional
1066 "Negative" exposure, typically a template exposure, or None if unavailable
1067 When `negExp` is `None`, will compute `negImage = posExp - exposure`.
1068
1069 Notes
1070 -----
1071 The main functionality of this routine was placed outside of
1072 this plugin (into `DipoleFitAlgorithm.fitDipole()`) so that
1073 `DipoleFitAlgorithm.fitDipole()` can be called separately for
1074 testing (@see `tests/testDipoleFitter.py`)
1075
1076 Returns
1077 -------
1078 result : TODO: DM-17458
1079 TODO: DM-17458
1080 """
1081
1082 result = None
1083 pks = measRecord.getFootprint().getPeaks()
1084
1085 # Check if the footprint consists of a putative dipole - else don't fit it.
1086 if (
1087 (len(pks) <= 1) # one peak in the footprint - not a dipole
1088 or (len(pks) > 1 and (np.sign(pks[0].getPeakValue())
1089 == np.sign(pks[-1].getPeakValue()))) # peaks are same sign - not a dipole
1090 ):
1091 measRecord.set(self.classificationFlagKeyclassificationFlagKey, False)
1092 measRecord.set(self.classificationAttemptedFlagKeyclassificationAttemptedFlagKey, False)
1093 self.failfail(measRecord, measBase.MeasurementError('not a dipole', self.FAILURE_NOT_DIPOLEFAILURE_NOT_DIPOLE))
1094 if not self.config.fitAllDiaSources:
1095 return result
1096
1097 try:
1098 alg = self.DipoleFitAlgorithmClassDipoleFitAlgorithmClass(exposure, posImage=posExp, negImage=negExp)
1099 result, _ = alg.fitDipole(
1100 measRecord, rel_weight=self.config.relWeight,
1101 tol=self.config.tolerance,
1102 maxSepInSigma=self.config.maxSeparation,
1103 fitBackground=self.config.fitBackground,
1104 separateNegParams=self.config.fitSeparateNegParams,
1105 verbose=False, display=False)
1106 except pexExcept.LengthError:
1107 self.failfail(measRecord, measBase.MeasurementError('edge failure', self.FAILURE_EDGEFAILURE_EDGE))
1108 except Exception:
1109 self.failfail(measRecord, measBase.MeasurementError('dipole fit failure', self.FAILURE_FITFAILURE_FIT))
1110
1111 if result is None:
1112 measRecord.set(self.classificationFlagKeyclassificationFlagKey, False)
1113 measRecord.set(self.classificationAttemptedFlagKeyclassificationAttemptedFlagKey, False)
1114 return result
1115
1116 self.loglog.debug("Dipole fit result: %d %s", measRecord.getId(), str(result))
1117
1118 if result.posFlux <= 1.: # usually around 0.1 -- the minimum flux allowed -- i.e. bad fit.
1119 self.failfail(measRecord, measBase.MeasurementError('dipole fit failure', self.FAILURE_FITFAILURE_FIT))
1120
1121 # add chi2, coord/flux uncertainties (TBD), dipole classification
1122 # Add the relevant values to the measRecord
1123 measRecord[self.posFluxKey] = result.posFlux
1124 measRecord[self.posFluxErrKey] = result.signalToNoise # to be changed to actual sigma!
1125 measRecord[self.posCentroidKeyX] = result.posCentroidX
1126 measRecord[self.posCentroidKeyY] = result.posCentroidY
1127
1128 measRecord[self.negFluxKey] = result.negFlux
1129 measRecord[self.negFluxErrKey] = result.signalToNoise # to be changed to actual sigma!
1130 measRecord[self.negCentroidKeyX] = result.negCentroidX
1131 measRecord[self.negCentroidKeyY] = result.negCentroidY
1132
1133 # Dia source flux: average of pos+neg
1134 measRecord[self.fluxKeyfluxKey] = (abs(result.posFlux) + abs(result.negFlux))/2.
1135 measRecord[self.orientationKeyorientationKey] = result.orientation
1136 measRecord[self.separationKeyseparationKey] = np.sqrt((result.posCentroidX - result.negCentroidX)**2.
1137 + (result.posCentroidY - result.negCentroidY)**2.)
1138 measRecord[self.centroidKeyX] = result.centroidX
1139 measRecord[self.centroidKeyY] = result.centroidY
1140
1141 measRecord[self.signalToNoiseKeysignalToNoiseKey] = result.signalToNoise
1142 measRecord[self.chi2dofKeychi2dofKey] = result.redChi2
1143
1144 self.doClassifydoClassify(measRecord, result.chi2)
1145
1146 def doClassify(self, measRecord, chi2val):
1147 """Classify a source as a dipole.
1148
1149 Parameters
1150 ----------
1151 measRecord : TODO: DM-17458
1152 TODO: DM-17458
1153 chi2val : TODO: DM-17458
1154 TODO: DM-17458
1155
1156 Notes
1157 -----
1158 Sources are classified as dipoles, or not, according to three criteria:
1159
1160 1. Does the total signal-to-noise surpass the ``minSn``?
1161 2. Are the pos/neg fluxes greater than 1.0 and no more than 0.65 (``maxFluxRatio``)
1162 of the total flux? By default this will never happen since ``posFlux == negFlux``.
1163 3. Is it a good fit (``chi2dof`` < 1)? (Currently not used.)
1164 """
1165
1166 # First, does the total signal-to-noise surpass the minSn?
1167 passesSn = measRecord[self.signalToNoiseKeysignalToNoiseKey] > self.config.minSn
1168
1169 # Second, are the pos/neg fluxes greater than 1.0 and no more than 0.65 (param maxFluxRatio)
1170 # of the total flux? By default this will never happen since posFlux = negFlux.
1171 passesFluxPos = (abs(measRecord[self.posFluxKey])
1172 / (measRecord[self.fluxKeyfluxKey]*2.)) < self.config.maxFluxRatio
1173 passesFluxPos &= (abs(measRecord[self.posFluxKey]) >= 1.0)
1174 passesFluxNeg = (abs(measRecord[self.negFluxKey])
1175 / (measRecord[self.fluxKeyfluxKey]*2.)) < self.config.maxFluxRatio
1176 passesFluxNeg &= (abs(measRecord[self.negFluxKey]) >= 1.0)
1177 allPass = (passesSn and passesFluxPos and passesFluxNeg) # and passesChi2)
1178
1179 # Third, is it a good fit (chi2dof < 1)?
1180 # Use scipy's chi2 cumulative distrib to estimate significance
1181 # This doesn't really work since I don't trust the values in the variance plane (which
1182 # affects the least-sq weights, which affects the resulting chi2).
1183 # But I'm going to keep this here for future use.
1184 if False:
1185 from scipy.stats import chi2
1186 ndof = chi2val / measRecord[self.chi2dofKeychi2dofKey]
1187 significance = chi2.cdf(chi2val, ndof)
1188 passesChi2 = significance < self.config.maxChi2DoF
1189 allPass = allPass and passesChi2
1190
1191 measRecord.set(self.classificationAttemptedFlagKeyclassificationAttemptedFlagKey, True)
1192
1193 if allPass: # Note cannot pass `allPass` into the `measRecord.set()` call below...?
1194 measRecord.set(self.classificationFlagKeyclassificationFlagKey, True)
1195 else:
1196 measRecord.set(self.classificationFlagKeyclassificationFlagKey, False)
1197
1198 def fail(self, measRecord, error=None):
1199 """Catch failures and set the correct flags.
1200 """
1201
1202 measRecord.set(self.flagKeyflagKey, True)
1203 if error is not None:
1204 if error.getFlagBit() == self.FAILURE_EDGEFAILURE_EDGE:
1205 self.loglog.warning('DipoleFitPlugin not run on record %d: %s', measRecord.getId(), str(error))
1206 measRecord.set(self.edgeFlagKeyedgeFlagKey, True)
1207 if error.getFlagBit() == self.FAILURE_FITFAILURE_FIT:
1208 self.loglog.warning('DipoleFitPlugin failed on record %d: %s', measRecord.getId(), str(error))
1209 measRecord.set(self.flagKeyflagKey, True)
1210 if error.getFlagBit() == self.FAILURE_NOT_DIPOLEFAILURE_NOT_DIPOLE:
1211 self.loglog.debug('DipoleFitPlugin not run on record %d: %s',
1212 measRecord.getId(), str(error))
1213 measRecord.set(self.classificationAttemptedFlagKeyclassificationAttemptedFlagKey, False)
1214 measRecord.set(self.flagKeyflagKey, True)
1215 else:
1216 self.loglog.warning('DipoleFitPlugin failed on record %d', measRecord.getId())
def fitDipoleImpl(self, source, tol=1e-7, rel_weight=0.5, fitBackground=1, bgGradientOrder=1, maxSepInSigma=5., separateNegParams=True, verbose=False)
def fitDipole(self, source, tol=1e-7, rel_weight=0.1, fitBackground=1, maxSepInSigma=5., separateNegParams=True, bgGradientOrder=1, verbose=False, display=False)
def displayFitResults(self, footprint, result)
def __init__(self, diffim, posImage=None, negImage=None)
def fail(self, measRecord, error=None)
def measure(self, measRecord, exposure, posExp=None, negExp=None)
def doClassify(self, measRecord, chi2val)
def _setupSchema(self, config, name, schema, metadata)
def __init__(self, config, name, schema, metadata)
def __init__(self, schema, algMetadata=None, **kwargs)
def run(self, sources, exposure, posExp=None, negExp=None, **kwargs)
def _getHeavyFootprintSubimage(self, fp, badfill=np.nan, grow=0)
def makeModel(self, x, flux, xcenPos, ycenPos, xcenNeg, ycenNeg, fluxNeg=None, b=None, x1=None, y1=None, xy=None, x2=None, y2=None, bNeg=None, x1Neg=None, y1Neg=None, xyNeg=None, x2Neg=None, y2Neg=None, **kwargs)
def makeBackgroundModel(self, in_x, pars=None)
def fitFootprintBackground(self, source, posImage, order=1)
def makeStarModel(self, bbox, psf, xcen, ycen, flux)