Coverage for python/lsst/jointcal/jointcal.py : 20%

Hot-keys on this page
r m x p toggle line displays
j k next/prev highlighted chunk
0 (zero) top of page
1 (one) first highlighted chunk
1# This file is part of jointcal.
2#
3# Developed for the LSST Data Management System.
4# This product includes software developed by the LSST Project
5# (https://www.lsst.org).
6# See the COPYRIGHT file at the top-level directory of this distribution
7# for details of code ownership.
8#
9# This program is free software: you can redistribute it and/or modify
10# it under the terms of the GNU General Public License as published by
11# the Free Software Foundation, either version 3 of the License, or
12# (at your option) any later version.
13#
14# This program is distributed in the hope that it will be useful,
15# but WITHOUT ANY WARRANTY; without even the implied warranty of
16# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
17# GNU General Public License for more details.
18#
19# You should have received a copy of the GNU General Public License
20# along with this program. If not, see <https://www.gnu.org/licenses/>.
22import dataclasses
23import collections
24import os
26import astropy.time
27import numpy as np
28import astropy.units as u
30import lsst.geom
31import lsst.utils
32import lsst.pex.config as pexConfig
33import lsst.pipe.base as pipeBase
34from lsst.afw.image import fluxErrFromABMagErr
35import lsst.pex.exceptions as pexExceptions
36import lsst.afw.cameraGeom
37import lsst.afw.table
38import lsst.log
39import lsst.meas.algorithms
40from lsst.pipe.tasks.colorterms import ColortermLibrary
41from lsst.verify import Job, Measurement
43from lsst.meas.algorithms import LoadIndexedReferenceObjectsTask, ReferenceSourceSelectorTask
44from lsst.meas.algorithms.sourceSelector import sourceSelectorRegistry
46from .dataIds import PerTractCcdDataIdContainer
48import lsst.jointcal
49from lsst.jointcal import MinimizeResult
51__all__ = ["JointcalConfig", "JointcalRunner", "JointcalTask"]
53Photometry = collections.namedtuple('Photometry', ('fit', 'model'))
54Astrometry = collections.namedtuple('Astrometry', ('fit', 'model', 'sky_to_tan_projection'))
57# TODO: move this to MeasurementSet in lsst.verify per DM-12655.
58def add_measurement(job, name, value):
59 meas = Measurement(job.metrics[name], value)
60 job.measurements.insert(meas)
63class JointcalRunner(pipeBase.ButlerInitializedTaskRunner):
64 """Subclass of TaskRunner for jointcalTask
66 jointcalTask.runDataRef() takes a number of arguments, one of which is a list of dataRefs
67 extracted from the command line (whereas most CmdLineTasks' runDataRef methods take
68 single dataRef, are are called repeatedly). This class transforms the processed
69 arguments generated by the ArgumentParser into the arguments expected by
70 Jointcal.runDataRef().
72 See pipeBase.TaskRunner for more information.
73 """
75 @staticmethod
76 def getTargetList(parsedCmd, **kwargs):
77 """
78 Return a list of tuples per tract, each containing (dataRefs, kwargs).
80 Jointcal operates on lists of dataRefs simultaneously.
81 """
82 kwargs['profile_jointcal'] = parsedCmd.profile_jointcal
83 kwargs['butler'] = parsedCmd.butler
85 # organize data IDs by tract
86 refListDict = {}
87 for ref in parsedCmd.id.refList:
88 refListDict.setdefault(ref.dataId["tract"], []).append(ref)
89 # we call runDataRef() once with each tract
90 result = [(refListDict[tract], kwargs) for tract in sorted(refListDict.keys())]
91 return result
93 def __call__(self, args):
94 """
95 Parameters
96 ----------
97 args
98 Arguments for Task.runDataRef()
100 Returns
101 -------
102 pipe.base.Struct
103 if self.doReturnResults is False:
105 - ``exitStatus``: 0 if the task completed successfully, 1 otherwise.
107 if self.doReturnResults is True:
109 - ``result``: the result of calling jointcal.runDataRef()
110 - ``exitStatus``: 0 if the task completed successfully, 1 otherwise.
111 """
112 exitStatus = 0 # exit status for shell
114 # NOTE: cannot call self.makeTask because that assumes args[0] is a single dataRef.
115 dataRefList, kwargs = args
116 butler = kwargs.pop('butler')
117 task = self.TaskClass(config=self.config, log=self.log, butler=butler)
118 result = None
119 try:
120 result = task.runDataRef(dataRefList, **kwargs)
121 exitStatus = result.exitStatus
122 job_path = butler.get('verify_job_filename')
123 result.job.write(job_path[0])
124 except Exception as e: # catch everything, sort it out later.
125 if self.doRaise:
126 raise e
127 else:
128 exitStatus = 1
129 eName = type(e).__name__
130 tract = dataRefList[0].dataId['tract']
131 task.log.fatal("Failed processing tract %s, %s: %s", tract, eName, e)
133 # Put the butler back into kwargs for the other Tasks.
134 kwargs['butler'] = butler
135 if self.doReturnResults:
136 return pipeBase.Struct(result=result, exitStatus=exitStatus)
137 else:
138 return pipeBase.Struct(exitStatus=exitStatus)
141class JointcalConfig(pexConfig.Config):
142 """Configuration for JointcalTask"""
144 doAstrometry = pexConfig.Field(
145 doc="Fit astrometry and write the fitted result.",
146 dtype=bool,
147 default=True
148 )
149 doPhotometry = pexConfig.Field(
150 doc="Fit photometry and write the fitted result.",
151 dtype=bool,
152 default=True
153 )
154 coaddName = pexConfig.Field(
155 doc="Type of coadd, typically deep or goodSeeing",
156 dtype=str,
157 default="deep"
158 )
159 positionErrorPedestal = pexConfig.Field(
160 doc="Systematic term to apply to the measured position error (pixels)",
161 dtype=float,
162 default=0.02,
163 )
164 photometryErrorPedestal = pexConfig.Field(
165 doc="Systematic term to apply to the measured error on flux or magnitude as a "
166 "fraction of source flux or magnitude delta (e.g. 0.05 is 5% of flux or +50 millimag).",
167 dtype=float,
168 default=0.0,
169 )
170 # TODO: DM-6885 matchCut should be an geom.Angle
171 matchCut = pexConfig.Field(
172 doc="Matching radius between fitted and reference stars (arcseconds)",
173 dtype=float,
174 default=3.0,
175 )
176 minMeasurements = pexConfig.Field(
177 doc="Minimum number of associated measured stars for a fitted star to be included in the fit",
178 dtype=int,
179 default=2,
180 )
181 minMeasuredStarsPerCcd = pexConfig.Field(
182 doc="Minimum number of measuredStars per ccdImage before printing warnings",
183 dtype=int,
184 default=100,
185 )
186 minRefStarsPerCcd = pexConfig.Field(
187 doc="Minimum number of measuredStars per ccdImage before printing warnings",
188 dtype=int,
189 default=30,
190 )
191 allowLineSearch = pexConfig.Field(
192 doc="Allow a line search during minimization, if it is reasonable for the model"
193 " (models with a significant non-linear component, e.g. constrainedPhotometry).",
194 dtype=bool,
195 default=False
196 )
197 astrometrySimpleOrder = pexConfig.Field(
198 doc="Polynomial order for fitting the simple astrometry model.",
199 dtype=int,
200 default=3,
201 )
202 astrometryChipOrder = pexConfig.Field(
203 doc="Order of the per-chip transform for the constrained astrometry model.",
204 dtype=int,
205 default=1,
206 )
207 astrometryVisitOrder = pexConfig.Field(
208 doc="Order of the per-visit transform for the constrained astrometry model.",
209 dtype=int,
210 default=5,
211 )
212 useInputWcs = pexConfig.Field(
213 doc="Use the input calexp WCSs to initialize a SimpleAstrometryModel.",
214 dtype=bool,
215 default=True,
216 )
217 astrometryModel = pexConfig.ChoiceField(
218 doc="Type of model to fit to astrometry",
219 dtype=str,
220 default="constrained",
221 allowed={"simple": "One polynomial per ccd",
222 "constrained": "One polynomial per ccd, and one polynomial per visit"}
223 )
224 photometryModel = pexConfig.ChoiceField(
225 doc="Type of model to fit to photometry",
226 dtype=str,
227 default="constrainedMagnitude",
228 allowed={"simpleFlux": "One constant zeropoint per ccd and visit, fitting in flux space.",
229 "constrainedFlux": "Constrained zeropoint per ccd, and one polynomial per visit,"
230 " fitting in flux space.",
231 "simpleMagnitude": "One constant zeropoint per ccd and visit,"
232 " fitting in magnitude space.",
233 "constrainedMagnitude": "Constrained zeropoint per ccd, and one polynomial per visit,"
234 " fitting in magnitude space.",
235 }
236 )
237 applyColorTerms = pexConfig.Field(
238 doc="Apply photometric color terms to reference stars?"
239 "Requires that colorterms be set to a ColortermLibrary",
240 dtype=bool,
241 default=False
242 )
243 colorterms = pexConfig.ConfigField(
244 doc="Library of photometric reference catalog name to color term dict.",
245 dtype=ColortermLibrary,
246 )
247 photometryVisitOrder = pexConfig.Field(
248 doc="Order of the per-visit polynomial transform for the constrained photometry model.",
249 dtype=int,
250 default=7,
251 )
252 photometryDoRankUpdate = pexConfig.Field(
253 doc=("Do the rank update step during minimization. "
254 "Skipping this can help deal with models that are too non-linear."),
255 dtype=bool,
256 default=True,
257 )
258 astrometryDoRankUpdate = pexConfig.Field(
259 doc=("Do the rank update step during minimization (should not change the astrometry fit). "
260 "Skipping this can help deal with models that are too non-linear."),
261 dtype=bool,
262 default=True,
263 )
264 outlierRejectSigma = pexConfig.Field(
265 doc="How many sigma to reject outliers at during minimization.",
266 dtype=float,
267 default=5.0,
268 )
269 maxPhotometrySteps = pexConfig.Field(
270 doc="Maximum number of minimize iterations to take when fitting photometry.",
271 dtype=int,
272 default=20,
273 )
274 maxAstrometrySteps = pexConfig.Field(
275 doc="Maximum number of minimize iterations to take when fitting photometry.",
276 dtype=int,
277 default=20,
278 )
279 astrometryRefObjLoader = pexConfig.ConfigurableField(
280 target=LoadIndexedReferenceObjectsTask,
281 doc="Reference object loader for astrometric fit",
282 )
283 photometryRefObjLoader = pexConfig.ConfigurableField(
284 target=LoadIndexedReferenceObjectsTask,
285 doc="Reference object loader for photometric fit",
286 )
287 sourceSelector = sourceSelectorRegistry.makeField(
288 doc="How to select sources for cross-matching",
289 default="astrometry"
290 )
291 astrometryReferenceSelector = pexConfig.ConfigurableField(
292 target=ReferenceSourceSelectorTask,
293 doc="How to down-select the loaded astrometry reference catalog.",
294 )
295 photometryReferenceSelector = pexConfig.ConfigurableField(
296 target=ReferenceSourceSelectorTask,
297 doc="How to down-select the loaded photometry reference catalog.",
298 )
299 astrometryReferenceErr = pexConfig.Field(
300 doc=("Uncertainty on reference catalog coordinates [mas] to use in place of the `coord_*Err` fields. "
301 "If None, then raise an exception if the reference catalog is missing coordinate errors. "
302 "If specified, overrides any existing `coord_*Err` values."),
303 dtype=float,
304 default=None,
305 optional=True
306 )
307 writeInitMatrix = pexConfig.Field(
308 dtype=bool,
309 doc=("Write the pre/post-initialization Hessian and gradient to text files, for debugging. "
310 "The output files will be of the form 'astrometry_preinit-mat.txt', in the current directory. "
311 "Note that these files are the dense versions of the matrix, and so may be very large."),
312 default=False
313 )
314 writeChi2FilesInitialFinal = pexConfig.Field(
315 dtype=bool,
316 doc="Write .csv files containing the contributions to chi2 for the initialization and final fit.",
317 default=False
318 )
319 writeChi2FilesOuterLoop = pexConfig.Field(
320 dtype=bool,
321 doc="Write .csv files containing the contributions to chi2 for the outer fit loop.",
322 default=False
323 )
324 writeInitialModel = pexConfig.Field(
325 dtype=bool,
326 doc=("Write the pre-initialization model to text files, for debugging."
327 " Output is written to `initial[Astro|Photo]metryModel.txt` in the current working directory."),
328 default=False
329 )
330 debugOutputPath = pexConfig.Field(
331 dtype=str,
332 default=".",
333 doc=("Path to write debug output files to. Used by "
334 "`writeInitialModel`, `writeChi2Files*`, `writeInitMatrix`.")
335 )
336 sourceFluxType = pexConfig.Field(
337 dtype=str,
338 doc="Source flux field to use in source selection and to get fluxes from the catalog.",
339 default='Calib'
340 )
342 def validate(self):
343 super().validate()
344 if self.doPhotometry and self.applyColorTerms and len(self.colorterms.data) == 0:
345 msg = "applyColorTerms=True requires the `colorterms` field be set to a ColortermLibrary."
346 raise pexConfig.FieldValidationError(JointcalConfig.colorterms, self, msg)
347 if self.doAstrometry and not self.doPhotometry and self.applyColorTerms:
348 msg = ("Only doing astrometry, but Colorterms are not applied for astrometry;"
349 "applyColorTerms=True will be ignored.")
350 lsst.log.warn(msg)
352 def setDefaults(self):
353 # Use science source selector which can filter on extendedness, SNR, and whether blended
354 self.sourceSelector.name = 'science'
355 # Use only stars because aperture fluxes of galaxies are biased and depend on seeing
356 self.sourceSelector['science'].doUnresolved = True
357 # with dependable signal to noise ratio.
358 self.sourceSelector['science'].doSignalToNoise = True
359 # Min SNR must be > 0 because jointcal cannot handle negative fluxes,
360 # and S/N > 10 to use sources that are not too faint, and thus better measured.
361 self.sourceSelector['science'].signalToNoise.minimum = 10.
362 # Base SNR on CalibFlux because that is the flux jointcal that fits and must be positive
363 fluxField = f"slot_{self.sourceFluxType}Flux_instFlux"
364 self.sourceSelector['science'].signalToNoise.fluxField = fluxField
365 self.sourceSelector['science'].signalToNoise.errField = fluxField + "Err"
366 # Do not trust blended sources' aperture fluxes which also depend on seeing
367 self.sourceSelector['science'].doIsolated = True
368 # Do not trust either flux or centroid measurements with flags,
369 # chosen from the usual QA flags for stars)
370 self.sourceSelector['science'].doFlags = True
371 badFlags = ['base_PixelFlags_flag_edge', 'base_PixelFlags_flag_saturated',
372 'base_PixelFlags_flag_interpolatedCenter', 'base_SdssCentroid_flag',
373 'base_PsfFlux_flag', 'base_PixelFlags_flag_suspectCenter']
374 self.sourceSelector['science'].flags.bad = badFlags
376 # Default to Gaia-DR2 (with proper motions) for astrometry and
377 # PS1-DR1 for photometry, with a reasonable initial filterMap.
378 self.astrometryRefObjLoader.ref_dataset_name = "gaia_dr2_20200414"
379 self.astrometryRefObjLoader.requireProperMotion = True
380 self.astrometryRefObjLoader.anyFilterMapsToThis = 'phot_g_mean'
381 self.photometryRefObjLoader.ref_dataset_name = "ps1_pv3_3pi_20170110"
384def writeModel(model, filename, log):
385 """Write model to outfile."""
386 with open(filename, "w") as file:
387 file.write(repr(model))
388 log.info("Wrote %s to file: %s", model, filename)
391@dataclasses.dataclass
392class JointcalInputData:
393 """The input data jointcal needs for each detector/visit."""
394 visit: int
395 """The visit identifier of this exposure."""
396 catalog: lsst.afw.table.SourceCatalog
397 """The catalog derived from this exposure."""
398 visitInfo: lsst.afw.image.VisitInfo
399 """The VisitInfo of this exposure."""
400 detector: lsst.afw.cameraGeom.Detector
401 """The detector of this exposure."""
402 photoCalib: lsst.afw.image.PhotoCalib
403 """The photometric calibration of this exposure."""
404 wcs: lsst.afw.geom.skyWcs
405 """The WCS of this exposure."""
406 bbox: lsst.geom.Box2I
407 """The bounding box of this exposure."""
408 filter: lsst.afw.image.FilterLabel
409 """The filter of this exposure."""
412class JointcalTask(pipeBase.PipelineTask, pipeBase.CmdLineTask):
413 """Astrometricly and photometricly calibrate across multiple visits of the
414 same field.
416 Parameters
417 ----------
418 butler : `lsst.daf.persistence.Butler`
419 The butler is passed to the refObjLoader constructor in case it is
420 needed. Ignored if the refObjLoader argument provides a loader directly.
421 Used to initialize the astrometry and photometry refObjLoaders.
422 profile_jointcal : `bool`
423 Set to True to profile different stages of this jointcal run.
424 """
426 ConfigClass = JointcalConfig
427 RunnerClass = JointcalRunner
428 _DefaultName = "jointcal"
430 def __init__(self, butler=None, profile_jointcal=False, **kwargs):
431 super().__init__(**kwargs)
432 self.profile_jointcal = profile_jointcal
433 self.makeSubtask("sourceSelector")
434 if self.config.doAstrometry:
435 self.makeSubtask('astrometryRefObjLoader', butler=butler)
436 self.makeSubtask("astrometryReferenceSelector")
437 else:
438 self.astrometryRefObjLoader = None
439 if self.config.doPhotometry:
440 self.makeSubtask('photometryRefObjLoader', butler=butler)
441 self.makeSubtask("photometryReferenceSelector")
442 else:
443 self.photometryRefObjLoader = None
445 # To hold various computed metrics for use by tests
446 self.job = Job.load_metrics_package(subset='jointcal')
448 # We don't currently need to persist the metadata.
449 # If we do in the future, we will have to add appropriate dataset templates
450 # to each obs package (the metadata template should look like `jointcal_wcs`).
451 def _getMetadataName(self):
452 return None
454 @classmethod
455 def _makeArgumentParser(cls):
456 """Create an argument parser"""
457 parser = pipeBase.ArgumentParser(name=cls._DefaultName)
458 parser.add_argument("--profile_jointcal", default=False, action="store_true",
459 help="Profile steps of jointcal separately.")
460 parser.add_id_argument("--id", "calexp", help="data ID, e.g. --id visit=6789 ccd=0..9",
461 ContainerClass=PerTractCcdDataIdContainer)
462 return parser
464 def _build_ccdImage(self, data, associations, jointcalControl):
465 """
466 Extract the necessary things from this dataRef to add a new ccdImage.
468 Parameters
469 ----------
470 data : `JointcalInputData`
471 The loaded input data.
472 associations : `lsst.jointcal.Associations`
473 Object to add the info to, to construct a new CcdImage
474 jointcalControl : `jointcal.JointcalControl`
475 Control object for associations management
477 Returns
478 ------
479 namedtuple
480 ``wcs``
481 The TAN WCS of this image, read from the calexp
482 (`lsst.afw.geom.SkyWcs`).
483 ``key``
484 A key to identify this dataRef by its visit and ccd ids
485 (`namedtuple`).
486 ``band``
487 This calexp's filter band (`str`) (used to e.g. load refcats)
488 """
489 goodSrc = self.sourceSelector.run(data.catalog)
491 if len(goodSrc.sourceCat) == 0:
492 self.log.warn("No sources selected in visit %s ccd %s", data.visit, data.detector.getId())
493 else:
494 self.log.info("%d sources selected in visit %d ccd %d", len(goodSrc.sourceCat),
495 data.visit,
496 data.detector.getId())
497 associations.createCcdImage(goodSrc.sourceCat,
498 data.wcs,
499 data.visitInfo,
500 data.bbox,
501 data.filter.physicalLabel,
502 data.photoCalib,
503 data.detector,
504 data.visit,
505 data.detector.getId(),
506 jointcalControl)
508 Result = collections.namedtuple('Result_from_build_CcdImage', ('wcs', 'key', 'band'))
509 Key = collections.namedtuple('Key', ('visit', 'ccd'))
510 return Result(data.wcs, Key(data.visit, data.detector.getId()), data.filter.bandLabel)
512 def _readDataId(self, butler, dataId):
513 """Read all of the data for one dataId from the butler. (gen2 version)"""
514 # Not all instruments have `visit` in their dataIds.
515 if "visit" in dataId.keys():
516 visit = dataId["visit"]
517 else:
518 visit = butler.getButler().queryMetadata("calexp", ("visit"), butler.dataId)[0]
520 catalog = butler.get('src',
521 flags=lsst.afw.table.SOURCE_IO_NO_FOOTPRINTS,
522 dataId=dataId)
523 return JointcalInputData(visit=visit,
524 catalog=catalog,
525 visitInfo=butler.get('calexp_visitInfo', dataId=dataId),
526 detector=butler.get('calexp_detector', dataId=dataId),
527 photoCalib=butler.get('calexp_photoCalib', dataId=dataId),
528 wcs=butler.get('calexp_wcs', dataId=dataId),
529 bbox=butler.get('calexp_bbox', dataId=dataId),
530 filter=butler.get('calexp_filterLabel', dataId=dataId))
532 def loadData(self, dataRefs, associations, jointcalControl, profile_jointcal=False):
533 """Read the data that jointcal needs to run. (Gen2 version)"""
534 visit_ccd_to_dataRef = {}
535 oldWcsList = []
536 bands = []
537 load_cat_prof_file = 'jointcal_loadData.prof' if profile_jointcal else ''
538 with pipeBase.cmdLineTask.profile(load_cat_prof_file):
539 # Need the bounding-box of the focal plane (the same for all visits) for photometry visit models
540 camera = dataRefs[0].get('camera', immediate=True)
541 self.focalPlaneBBox = camera.getFpBBox()
542 for dataRef in dataRefs:
543 data = self._readDataId(dataRef.getButler(), dataRef.dataId)
544 result = self._build_ccdImage(data, associations, jointcalControl)
545 oldWcsList.append(result.wcs)
546 visit_ccd_to_dataRef[result.key] = dataRef
547 bands.append(result.band)
548 bands = collections.Counter(bands)
550 return oldWcsList, bands, visit_ccd_to_dataRef
552 def _getDebugPath(self, filename):
553 """Constructs a path to filename using the configured debug path.
554 """
555 return os.path.join(self.config.debugOutputPath, filename)
557 def _prep_sky(self, associations, bands):
558 """Prepare on-sky and other data that must be computed after data has
559 been read.
560 """
561 associations.computeCommonTangentPoint()
563 boundingCircle = associations.computeBoundingCircle()
564 center = lsst.geom.SpherePoint(boundingCircle.getCenter())
565 radius = lsst.geom.Angle(boundingCircle.getOpeningAngle().asRadians(), lsst.geom.radians)
567 self.log.info(f"Data has center={center} with radius={radius.asDegrees()} degrees.")
569 # Determine a default filter band associated with the catalog. See DM-9093
570 defaultBand = bands.most_common(1)[0][0]
571 self.log.debug("Using '%s' filter band for reference flux", defaultBand)
573 return boundingCircle, center, radius, defaultBand
575 @pipeBase.timeMethod
576 def runDataRef(self, dataRefs, profile_jointcal=False):
577 """
578 Jointly calibrate the astrometry and photometry across a set of images.
580 NOTE: this is for gen2 middleware only.
582 Parameters
583 ----------
584 dataRefs : `list` of `lsst.daf.persistence.ButlerDataRef`
585 List of data references to the exposures to be fit.
586 profile_jointcal : `bool`
587 Profile the individual steps of jointcal.
589 Returns
590 -------
591 result : `lsst.pipe.base.Struct`
592 Struct of metadata from the fit, containing:
594 ``dataRefs``
595 The provided data references that were fit (with updated WCSs)
596 ``oldWcsList``
597 The original WCS from each dataRef
598 ``metrics``
599 Dictionary of internally-computed metrics for testing/validation.
600 """
601 if len(dataRefs) == 0:
602 raise ValueError('Need a non-empty list of data references!')
604 exitStatus = 0 # exit status for shell
606 sourceFluxField = "slot_%sFlux" % (self.config.sourceFluxType,)
607 jointcalControl = lsst.jointcal.JointcalControl(sourceFluxField)
608 associations = lsst.jointcal.Associations()
610 oldWcsList, bands, visit_ccd_to_dataRef = self.loadData(dataRefs,
611 associations,
612 jointcalControl,
613 profile_jointcal=profile_jointcal)
615 boundingCircle, center, radius, defaultBand = self._prep_sky(associations, bands)
617 tract = dataRefs[0].dataId['tract']
619 if self.config.doAstrometry:
620 astrometry = self._do_load_refcat_and_fit(associations, defaultBand, center, radius,
621 name="astrometry",
622 refObjLoader=self.astrometryRefObjLoader,
623 referenceSelector=self.astrometryReferenceSelector,
624 fit_function=self._fit_astrometry,
625 profile_jointcal=profile_jointcal,
626 tract=tract)
627 self._write_astrometry_results(associations, astrometry.model, visit_ccd_to_dataRef)
628 else:
629 astrometry = Astrometry(None, None, None)
631 if self.config.doPhotometry:
632 photometry = self._do_load_refcat_and_fit(associations, defaultBand, center, radius,
633 name="photometry",
634 refObjLoader=self.photometryRefObjLoader,
635 referenceSelector=self.photometryReferenceSelector,
636 fit_function=self._fit_photometry,
637 profile_jointcal=profile_jointcal,
638 tract=tract,
639 reject_bad_fluxes=True)
640 self._write_photometry_results(associations, photometry.model, visit_ccd_to_dataRef)
641 else:
642 photometry = Photometry(None, None)
644 return pipeBase.Struct(dataRefs=dataRefs,
645 oldWcsList=oldWcsList,
646 job=self.job,
647 astrometryRefObjLoader=self.astrometryRefObjLoader,
648 photometryRefObjLoader=self.photometryRefObjLoader,
649 defaultBand=defaultBand,
650 exitStatus=exitStatus)
652 def _get_refcat_coordinate_error_override(self, refCat, name):
653 """Check whether we should override the refcat coordinate errors, and
654 return the overridden error if necessary.
656 Parameters
657 ----------
658 refCat : `lsst.afw.table.SimpleCatalog`
659 The reference catalog to check for a ``coord_raErr`` field.
660 name : `str`
661 Whether we are doing "astrometry" or "photometry".
663 Returns
664 -------
665 refCoordErr : `float`
666 The refcat coordinate error to use, or NaN if we are not overriding
667 those fields.
669 Raises
670 ------
671 lsst.pex.config.FieldValidationError
672 Raised if the refcat does not contain coordinate errors and
673 ``config.astrometryReferenceErr`` is not set.
674 """
675 # This value doesn't matter for photometry, so just set something to
676 # keep old refcats from causing problems.
677 if name.lower() == "photometry":
678 if 'coord_raErr' not in refCat.schema:
679 return 100
680 else:
681 return float('nan')
683 if self.config.astrometryReferenceErr is None and 'coord_raErr' not in refCat.schema:
684 msg = ("Reference catalog does not contain coordinate errors, "
685 "and config.astrometryReferenceErr not supplied.")
686 raise pexConfig.FieldValidationError(JointcalConfig.astrometryReferenceErr,
687 self.config,
688 msg)
690 if self.config.astrometryReferenceErr is not None and 'coord_raErr' in refCat.schema:
691 self.log.warn("Overriding reference catalog coordinate errors with %f/coordinate [mas]",
692 self.config.astrometryReferenceErr)
694 if self.config.astrometryReferenceErr is None:
695 return float('nan')
696 else:
697 return self.config.astrometryReferenceErr
699 def _compute_proper_motion_epoch(self, ccdImageList):
700 """Return the proper motion correction epoch of the provided images.
702 Parameters
703 ----------
704 ccdImageList : `list` [`lsst.jointcal.CcdImage`]
705 The images to compute the appropriate epoch for.
707 Returns
708 -------
709 epoch : `astropy.time.Time`
710 The date to use for proper motion corrections.
711 """
712 mjds = [ccdImage.getMjd() for ccdImage in ccdImageList]
713 return astropy.time.Time(np.mean(mjds), format='mjd', scale="tai")
715 def _do_load_refcat_and_fit(self, associations, defaultBand, center, radius,
716 tract="", profile_jointcal=False, match_cut=3.0,
717 reject_bad_fluxes=False, *,
718 name="", refObjLoader=None, referenceSelector=None,
719 fit_function=None):
720 """Load reference catalog, perform the fit, and return the result.
722 Parameters
723 ----------
724 associations : `lsst.jointcal.Associations`
725 The star/reference star associations to fit.
726 defaultBand : `str`
727 filter to load from reference catalog.
728 center : `lsst.geom.SpherePoint`
729 ICRS center of field to load from reference catalog.
730 radius : `lsst.geom.Angle`
731 On-sky radius to load from reference catalog.
732 name : `str`
733 Name of thing being fit: "astrometry" or "photometry".
734 refObjLoader : `lsst.meas.algorithms.LoadReferenceObjectsTask`
735 Reference object loader to use to load a reference catalog.
736 referenceSelector : `lsst.meas.algorithms.ReferenceSourceSelectorTask`
737 Selector to use to pick objects from the loaded reference catalog.
738 fit_function : callable
739 Function to call to perform fit (takes Associations object).
740 tract : `str`, optional
741 Name of tract currently being fit.
742 profile_jointcal : `bool`, optional
743 Separately profile the fitting step.
744 match_cut : `float`, optional
745 Radius in arcseconds to find cross-catalog matches to during
746 associations.associateCatalogs.
747 reject_bad_fluxes : `bool`, optional
748 Reject refCat sources with NaN/inf flux or NaN/0 fluxErr.
750 Returns
751 -------
752 result : `Photometry` or `Astrometry`
753 Result of `fit_function()`
754 """
755 self.log.info("====== Now processing %s...", name)
756 # TODO: this should not print "trying to invert a singular transformation:"
757 # if it does that, something's not right about the WCS...
758 associations.associateCatalogs(match_cut)
759 add_measurement(self.job, 'jointcal.associated_%s_fittedStars' % name,
760 associations.fittedStarListSize())
762 applyColorterms = False if name.lower() == "astrometry" else self.config.applyColorTerms
763 epoch = self._compute_proper_motion_epoch(associations.getCcdImageList())
764 refCat, fluxField = self._load_reference_catalog(refObjLoader, referenceSelector,
765 center, radius, defaultBand,
766 applyColorterms=applyColorterms,
767 epoch=epoch)
768 refCoordErr = self._get_refcat_coordinate_error_override(refCat, name)
770 associations.collectRefStars(refCat,
771 self.config.matchCut*lsst.geom.arcseconds,
772 fluxField,
773 refCoordinateErr=refCoordErr,
774 rejectBadFluxes=reject_bad_fluxes)
775 add_measurement(self.job, 'jointcal.collected_%s_refStars' % name,
776 associations.refStarListSize())
778 associations.prepareFittedStars(self.config.minMeasurements)
780 self._check_star_lists(associations, name)
781 add_measurement(self.job, 'jointcal.selected_%s_refStars' % name,
782 associations.nFittedStarsWithAssociatedRefStar())
783 add_measurement(self.job, 'jointcal.selected_%s_fittedStars' % name,
784 associations.fittedStarListSize())
785 add_measurement(self.job, 'jointcal.selected_%s_ccdImages' % name,
786 associations.nCcdImagesValidForFit())
788 load_cat_prof_file = 'jointcal_fit_%s.prof'%name if profile_jointcal else ''
789 dataName = "{}_{}".format(tract, defaultBand)
790 with pipeBase.cmdLineTask.profile(load_cat_prof_file):
791 result = fit_function(associations, dataName)
792 # TODO DM-12446: turn this into a "butler save" somehow.
793 # Save reference and measurement chi2 contributions for this data
794 if self.config.writeChi2FilesInitialFinal:
795 baseName = self._getDebugPath(f"{name}_final_chi2-{dataName}")
796 result.fit.saveChi2Contributions(baseName+"{type}")
797 self.log.info("Wrote chi2 contributions files: %s", baseName)
799 return result
801 def _load_reference_catalog(self, refObjLoader, referenceSelector, center, radius, filterName,
802 applyColorterms=False, epoch=None):
803 """Load the necessary reference catalog sources, convert fluxes to
804 correct units, and apply color term corrections if requested.
806 Parameters
807 ----------
808 refObjLoader : `lsst.meas.algorithms.LoadReferenceObjectsTask`
809 The reference catalog loader to use to get the data.
810 referenceSelector : `lsst.meas.algorithms.ReferenceSourceSelectorTask`
811 Source selector to apply to loaded reference catalog.
812 center : `lsst.geom.SpherePoint`
813 The center around which to load sources.
814 radius : `lsst.geom.Angle`
815 The radius around ``center`` to load sources in.
816 filterName : `str`
817 The name of the camera filter to load fluxes for.
818 applyColorterms : `bool`
819 Apply colorterm corrections to the refcat for ``filterName``?
820 epoch : `astropy.time.Time`, optional
821 Epoch to which to correct refcat proper motion and parallax,
822 or `None` to not apply such corrections.
824 Returns
825 -------
826 refCat : `lsst.afw.table.SimpleCatalog`
827 The loaded reference catalog.
828 fluxField : `str`
829 The name of the reference catalog flux field appropriate for ``filterName``.
830 """
831 skyCircle = refObjLoader.loadSkyCircle(center,
832 radius,
833 filterName,
834 epoch=epoch)
836 selected = referenceSelector.run(skyCircle.refCat)
837 # Need memory contiguity to get reference filters as a vector.
838 if not selected.sourceCat.isContiguous():
839 refCat = selected.sourceCat.copy(deep=True)
840 else:
841 refCat = selected.sourceCat
843 if applyColorterms:
844 refCatName = refObjLoader.ref_dataset_name
845 self.log.info("Applying color terms for filterName=%r reference catalog=%s",
846 filterName, refCatName)
847 colorterm = self.config.colorterms.getColorterm(
848 filterName=filterName, photoCatName=refCatName, doRaise=True)
850 refMag, refMagErr = colorterm.getCorrectedMagnitudes(refCat, filterName)
851 refCat[skyCircle.fluxField] = u.Magnitude(refMag, u.ABmag).to_value(u.nJy)
852 # TODO: I didn't want to use this, but I'll deal with it in DM-16903
853 refCat[skyCircle.fluxField+'Err'] = fluxErrFromABMagErr(refMagErr, refMag) * 1e9
855 return refCat, skyCircle.fluxField
857 def _check_star_lists(self, associations, name):
858 # TODO: these should be len(blah), but we need this properly wrapped first.
859 if associations.nCcdImagesValidForFit() == 0:
860 raise RuntimeError('No images in the ccdImageList!')
861 if associations.fittedStarListSize() == 0:
862 raise RuntimeError('No stars in the {} fittedStarList!'.format(name))
863 if associations.refStarListSize() == 0:
864 raise RuntimeError('No stars in the {} reference star list!'.format(name))
866 def _logChi2AndValidate(self, associations, fit, model, chi2Label, writeChi2Name=None):
867 """Compute chi2, log it, validate the model, and return chi2.
869 Parameters
870 ----------
871 associations : `lsst.jointcal.Associations`
872 The star/reference star associations to fit.
873 fit : `lsst.jointcal.FitterBase`
874 The fitter to use for minimization.
875 model : `lsst.jointcal.Model`
876 The model being fit.
877 chi2Label : `str`
878 Label to describe the chi2 (e.g. "Initialized", "Final").
879 writeChi2Name : `str`, optional
880 Filename prefix to write the chi2 contributions to.
881 Do not supply an extension: an appropriate one will be added.
883 Returns
884 -------
885 chi2: `lsst.jointcal.Chi2Accumulator`
886 The chi2 object for the current fitter and model.
888 Raises
889 ------
890 FloatingPointError
891 Raised if chi2 is infinite or NaN.
892 ValueError
893 Raised if the model is not valid.
894 """
895 if writeChi2Name is not None:
896 fullpath = self._getDebugPath(writeChi2Name)
897 fit.saveChi2Contributions(fullpath+"{type}")
898 self.log.info("Wrote chi2 contributions files: %s", fullpath)
900 chi2 = fit.computeChi2()
901 self.log.info("%s %s", chi2Label, chi2)
902 self._check_stars(associations)
903 if not np.isfinite(chi2.chi2):
904 raise FloatingPointError(f'{chi2Label} chi2 is invalid: {chi2}')
905 if not model.validate(associations.getCcdImageList(), chi2.ndof):
906 raise ValueError("Model is not valid: check log messages for warnings.")
907 return chi2
909 def _fit_photometry(self, associations, dataName=None):
910 """
911 Fit the photometric data.
913 Parameters
914 ----------
915 associations : `lsst.jointcal.Associations`
916 The star/reference star associations to fit.
917 dataName : `str`
918 Name of the data being processed (e.g. "1234_HSC-Y"), for
919 identifying debugging files.
921 Returns
922 -------
923 fit_result : `namedtuple`
924 fit : `lsst.jointcal.PhotometryFit`
925 The photometric fitter used to perform the fit.
926 model : `lsst.jointcal.PhotometryModel`
927 The photometric model that was fit.
928 """
929 self.log.info("=== Starting photometric fitting...")
931 # TODO: should use pex.config.RegistryField here (see DM-9195)
932 if self.config.photometryModel == "constrainedFlux":
933 model = lsst.jointcal.ConstrainedFluxModel(associations.getCcdImageList(),
934 self.focalPlaneBBox,
935 visitOrder=self.config.photometryVisitOrder,
936 errorPedestal=self.config.photometryErrorPedestal)
937 # potentially nonlinear problem, so we may need a line search to converge.
938 doLineSearch = self.config.allowLineSearch
939 elif self.config.photometryModel == "constrainedMagnitude":
940 model = lsst.jointcal.ConstrainedMagnitudeModel(associations.getCcdImageList(),
941 self.focalPlaneBBox,
942 visitOrder=self.config.photometryVisitOrder,
943 errorPedestal=self.config.photometryErrorPedestal)
944 # potentially nonlinear problem, so we may need a line search to converge.
945 doLineSearch = self.config.allowLineSearch
946 elif self.config.photometryModel == "simpleFlux":
947 model = lsst.jointcal.SimpleFluxModel(associations.getCcdImageList(),
948 errorPedestal=self.config.photometryErrorPedestal)
949 doLineSearch = False # purely linear in model parameters, so no line search needed
950 elif self.config.photometryModel == "simpleMagnitude":
951 model = lsst.jointcal.SimpleMagnitudeModel(associations.getCcdImageList(),
952 errorPedestal=self.config.photometryErrorPedestal)
953 doLineSearch = False # purely linear in model parameters, so no line search needed
955 fit = lsst.jointcal.PhotometryFit(associations, model)
956 # TODO DM-12446: turn this into a "butler save" somehow.
957 # Save reference and measurement chi2 contributions for this data
958 if self.config.writeChi2FilesInitialFinal:
959 baseName = f"photometry_initial_chi2-{dataName}"
960 else:
961 baseName = None
962 if self.config.writeInitialModel:
963 fullpath = self._getDebugPath("initialPhotometryModel.txt")
964 writeModel(model, fullpath, self.log)
965 self._logChi2AndValidate(associations, fit, model, "Initialized", writeChi2Name=baseName)
967 def getChi2Name(whatToFit):
968 if self.config.writeChi2FilesOuterLoop:
969 return f"photometry_init-%s_chi2-{dataName}" % whatToFit
970 else:
971 return None
973 # The constrained model needs the visit transform fit first; the chip
974 # transform is initialized from the singleFrame PhotoCalib, so it's close.
975 dumpMatrixFile = self._getDebugPath("photometry_preinit") if self.config.writeInitMatrix else ""
976 if self.config.photometryModel.startswith("constrained"):
977 # no line search: should be purely (or nearly) linear,
978 # and we want a large step size to initialize with.
979 fit.minimize("ModelVisit", dumpMatrixFile=dumpMatrixFile)
980 self._logChi2AndValidate(associations, fit, model, "Initialize ModelVisit",
981 writeChi2Name=getChi2Name("ModelVisit"))
982 dumpMatrixFile = "" # so we don't redo the output on the next step
984 fit.minimize("Model", doLineSearch=doLineSearch, dumpMatrixFile=dumpMatrixFile)
985 self._logChi2AndValidate(associations, fit, model, "Initialize Model",
986 writeChi2Name=getChi2Name("Model"))
988 fit.minimize("Fluxes") # no line search: always purely linear.
989 self._logChi2AndValidate(associations, fit, model, "Initialize Fluxes",
990 writeChi2Name=getChi2Name("Fluxes"))
992 fit.minimize("Model Fluxes", doLineSearch=doLineSearch)
993 self._logChi2AndValidate(associations, fit, model, "Initialize ModelFluxes",
994 writeChi2Name=getChi2Name("ModelFluxes"))
996 model.freezeErrorTransform()
997 self.log.debug("Photometry error scales are frozen.")
999 chi2 = self._iterate_fit(associations,
1000 fit,
1001 self.config.maxPhotometrySteps,
1002 "photometry",
1003 "Model Fluxes",
1004 doRankUpdate=self.config.photometryDoRankUpdate,
1005 doLineSearch=doLineSearch,
1006 dataName=dataName)
1008 add_measurement(self.job, 'jointcal.photometry_final_chi2', chi2.chi2)
1009 add_measurement(self.job, 'jointcal.photometry_final_ndof', chi2.ndof)
1010 return Photometry(fit, model)
1012 def _fit_astrometry(self, associations, dataName=None):
1013 """
1014 Fit the astrometric data.
1016 Parameters
1017 ----------
1018 associations : `lsst.jointcal.Associations`
1019 The star/reference star associations to fit.
1020 dataName : `str`
1021 Name of the data being processed (e.g. "1234_HSC-Y"), for
1022 identifying debugging files.
1024 Returns
1025 -------
1026 fit_result : `namedtuple`
1027 fit : `lsst.jointcal.AstrometryFit`
1028 The astrometric fitter used to perform the fit.
1029 model : `lsst.jointcal.AstrometryModel`
1030 The astrometric model that was fit.
1031 sky_to_tan_projection : `lsst.jointcal.ProjectionHandler`
1032 The model for the sky to tangent plane projection that was used in the fit.
1033 """
1035 self.log.info("=== Starting astrometric fitting...")
1037 associations.deprojectFittedStars()
1039 # NOTE: need to return sky_to_tan_projection so that it doesn't get garbage collected.
1040 # TODO: could we package sky_to_tan_projection and model together so we don't have to manage
1041 # them so carefully?
1042 sky_to_tan_projection = lsst.jointcal.OneTPPerVisitHandler(associations.getCcdImageList())
1044 if self.config.astrometryModel == "constrained":
1045 model = lsst.jointcal.ConstrainedAstrometryModel(associations.getCcdImageList(),
1046 sky_to_tan_projection,
1047 chipOrder=self.config.astrometryChipOrder,
1048 visitOrder=self.config.astrometryVisitOrder)
1049 elif self.config.astrometryModel == "simple":
1050 model = lsst.jointcal.SimpleAstrometryModel(associations.getCcdImageList(),
1051 sky_to_tan_projection,
1052 self.config.useInputWcs,
1053 nNotFit=0,
1054 order=self.config.astrometrySimpleOrder)
1056 fit = lsst.jointcal.AstrometryFit(associations, model, self.config.positionErrorPedestal)
1057 # TODO DM-12446: turn this into a "butler save" somehow.
1058 # Save reference and measurement chi2 contributions for this data
1059 if self.config.writeChi2FilesInitialFinal:
1060 baseName = f"astrometry_initial_chi2-{dataName}"
1061 else:
1062 baseName = None
1063 if self.config.writeInitialModel:
1064 fullpath = self._getDebugPath("initialAstrometryModel.txt")
1065 writeModel(model, fullpath, self.log)
1066 self._logChi2AndValidate(associations, fit, model, "Initial", writeChi2Name=baseName)
1068 def getChi2Name(whatToFit):
1069 if self.config.writeChi2FilesOuterLoop:
1070 return f"astrometry_init-%s_chi2-{dataName}" % whatToFit
1071 else:
1072 return None
1074 dumpMatrixFile = self._getDebugPath("astrometry_preinit") if self.config.writeInitMatrix else ""
1075 # The constrained model needs the visit transform fit first; the chip
1076 # transform is initialized from the detector's cameraGeom, so it's close.
1077 if self.config.astrometryModel == "constrained":
1078 fit.minimize("DistortionsVisit", dumpMatrixFile=dumpMatrixFile)
1079 self._logChi2AndValidate(associations, fit, model, "Initialize DistortionsVisit",
1080 writeChi2Name=getChi2Name("DistortionsVisit"))
1081 dumpMatrixFile = "" # so we don't redo the output on the next step
1083 fit.minimize("Distortions", dumpMatrixFile=dumpMatrixFile)
1084 self._logChi2AndValidate(associations, fit, model, "Initialize Distortions",
1085 writeChi2Name=getChi2Name("Distortions"))
1087 fit.minimize("Positions")
1088 self._logChi2AndValidate(associations, fit, model, "Initialize Positions",
1089 writeChi2Name=getChi2Name("Positions"))
1091 fit.minimize("Distortions Positions")
1092 self._logChi2AndValidate(associations, fit, model, "Initialize DistortionsPositions",
1093 writeChi2Name=getChi2Name("DistortionsPositions"))
1095 chi2 = self._iterate_fit(associations,
1096 fit,
1097 self.config.maxAstrometrySteps,
1098 "astrometry",
1099 "Distortions Positions",
1100 doRankUpdate=self.config.astrometryDoRankUpdate,
1101 dataName=dataName)
1103 add_measurement(self.job, 'jointcal.astrometry_final_chi2', chi2.chi2)
1104 add_measurement(self.job, 'jointcal.astrometry_final_ndof', chi2.ndof)
1106 return Astrometry(fit, model, sky_to_tan_projection)
1108 def _check_stars(self, associations):
1109 """Count measured and reference stars per ccd and warn/log them."""
1110 for ccdImage in associations.getCcdImageList():
1111 nMeasuredStars, nRefStars = ccdImage.countStars()
1112 self.log.debug("ccdImage %s has %s measured and %s reference stars",
1113 ccdImage.getName(), nMeasuredStars, nRefStars)
1114 if nMeasuredStars < self.config.minMeasuredStarsPerCcd:
1115 self.log.warn("ccdImage %s has only %s measuredStars (desired %s)",
1116 ccdImage.getName(), nMeasuredStars, self.config.minMeasuredStarsPerCcd)
1117 if nRefStars < self.config.minRefStarsPerCcd:
1118 self.log.warn("ccdImage %s has only %s RefStars (desired %s)",
1119 ccdImage.getName(), nRefStars, self.config.minRefStarsPerCcd)
1121 def _iterate_fit(self, associations, fitter, max_steps, name, whatToFit,
1122 dataName="",
1123 doRankUpdate=True,
1124 doLineSearch=False):
1125 """Run fitter.minimize up to max_steps times, returning the final chi2.
1127 Parameters
1128 ----------
1129 associations : `lsst.jointcal.Associations`
1130 The star/reference star associations to fit.
1131 fitter : `lsst.jointcal.FitterBase`
1132 The fitter to use for minimization.
1133 max_steps : `int`
1134 Maximum number of steps to run outlier rejection before declaring
1135 convergence failure.
1136 name : {'photometry' or 'astrometry'}
1137 What type of data are we fitting (for logs and debugging files).
1138 whatToFit : `str`
1139 Passed to ``fitter.minimize()`` to define the parameters to fit.
1140 dataName : `str`, optional
1141 Descriptive name for this dataset (e.g. tract and filter),
1142 for debugging.
1143 doRankUpdate : `bool`, optional
1144 Do an Eigen rank update during minimization, or recompute the full
1145 matrix and gradient?
1146 doLineSearch : `bool`, optional
1147 Do a line search for the optimum step during minimization?
1149 Returns
1150 -------
1151 chi2: `lsst.jointcal.Chi2Statistic`
1152 The final chi2 after the fit converges, or is forced to end.
1154 Raises
1155 ------
1156 FloatingPointError
1157 Raised if the fitter fails with a non-finite value.
1158 RuntimeError
1159 Raised if the fitter fails for some other reason;
1160 log messages will provide further details.
1161 """
1162 dumpMatrixFile = self._getDebugPath(f"{name}_postinit") if self.config.writeInitMatrix else ""
1163 oldChi2 = lsst.jointcal.Chi2Statistic()
1164 oldChi2.chi2 = float("inf")
1165 for i in range(max_steps):
1166 if self.config.writeChi2FilesOuterLoop:
1167 writeChi2Name = f"{name}_iterate_{i}_chi2-{dataName}"
1168 else:
1169 writeChi2Name = None
1170 result = fitter.minimize(whatToFit,
1171 self.config.outlierRejectSigma,
1172 doRankUpdate=doRankUpdate,
1173 doLineSearch=doLineSearch,
1174 dumpMatrixFile=dumpMatrixFile)
1175 dumpMatrixFile = "" # clear it so we don't write the matrix again.
1176 chi2 = self._logChi2AndValidate(associations, fitter, fitter.getModel(),
1177 f"Fit iteration {i}", writeChi2Name=writeChi2Name)
1179 if result == MinimizeResult.Converged:
1180 if doRankUpdate:
1181 self.log.debug("fit has converged - no more outliers - redo minimization "
1182 "one more time in case we have lost accuracy in rank update.")
1183 # Redo minimization one more time in case we have lost accuracy in rank update
1184 result = fitter.minimize(whatToFit, self.config.outlierRejectSigma)
1185 chi2 = self._logChi2AndValidate(associations, fitter, fitter.getModel(), "Fit completed")
1187 # log a message for a large final chi2, TODO: DM-15247 for something better
1188 if chi2.chi2/chi2.ndof >= 4.0:
1189 self.log.error("Potentially bad fit: High chi-squared/ndof.")
1191 break
1192 elif result == MinimizeResult.Chi2Increased:
1193 self.log.warn("Still some outliers remaining but chi2 increased - retry")
1194 # Check whether the increase was large enough to cause trouble.
1195 chi2Ratio = chi2.chi2 / oldChi2.chi2
1196 if chi2Ratio > 1.5:
1197 self.log.warn('Significant chi2 increase by a factor of %.4g / %.4g = %.4g',
1198 chi2.chi2, oldChi2.chi2, chi2Ratio)
1199 # Based on a variety of HSC jointcal logs (see DM-25779), it
1200 # appears that chi2 increases more than a factor of ~2 always
1201 # result in the fit diverging rapidly and ending at chi2 > 1e10.
1202 # Using 10 as the "failure" threshold gives some room between
1203 # leaving a warning and bailing early.
1204 if chi2Ratio > 10:
1205 msg = ("Large chi2 increase between steps: fit likely cannot converge."
1206 " Try setting one or more of the `writeChi2*` config fields and looking"
1207 " at how individual star chi2-values evolve during the fit.")
1208 raise RuntimeError(msg)
1209 oldChi2 = chi2
1210 elif result == MinimizeResult.NonFinite:
1211 filename = self._getDebugPath("{}_failure-nonfinite_chi2-{}.csv".format(name, dataName))
1212 # TODO DM-12446: turn this into a "butler save" somehow.
1213 fitter.saveChi2Contributions(filename+"{type}")
1214 msg = "Nonfinite value in chi2 minimization, cannot complete fit. Dumped star tables to: {}"
1215 raise FloatingPointError(msg.format(filename))
1216 elif result == MinimizeResult.Failed:
1217 raise RuntimeError("Chi2 minimization failure, cannot complete fit.")
1218 else:
1219 raise RuntimeError("Unxepected return code from minimize().")
1220 else:
1221 self.log.error("%s failed to converge after %d steps"%(name, max_steps))
1223 return chi2
1225 def _write_astrometry_results(self, associations, model, visit_ccd_to_dataRef):
1226 """
1227 Write the fitted astrometric results to a new 'jointcal_wcs' dataRef.
1229 Parameters
1230 ----------
1231 associations : `lsst.jointcal.Associations`
1232 The star/reference star associations to fit.
1233 model : `lsst.jointcal.AstrometryModel`
1234 The astrometric model that was fit.
1235 visit_ccd_to_dataRef : `dict` of Key: `lsst.daf.persistence.ButlerDataRef`
1236 Dict of ccdImage identifiers to dataRefs that were fit.
1237 """
1239 ccdImageList = associations.getCcdImageList()
1240 for ccdImage in ccdImageList:
1241 # TODO: there must be a better way to identify this ccdImage than a visit,ccd pair?
1242 ccd = ccdImage.ccdId
1243 visit = ccdImage.visit
1244 dataRef = visit_ccd_to_dataRef[(visit, ccd)]
1245 self.log.info("Updating WCS for visit: %d, ccd: %d", visit, ccd)
1246 skyWcs = model.makeSkyWcs(ccdImage)
1247 try:
1248 dataRef.put(skyWcs, 'jointcal_wcs')
1249 except pexExceptions.Exception as e:
1250 self.log.fatal('Failed to write updated Wcs: %s', str(e))
1251 raise e
1253 def _write_photometry_results(self, associations, model, visit_ccd_to_dataRef):
1254 """
1255 Write the fitted photometric results to a new 'jointcal_photoCalib' dataRef.
1257 Parameters
1258 ----------
1259 associations : `lsst.jointcal.Associations`
1260 The star/reference star associations to fit.
1261 model : `lsst.jointcal.PhotometryModel`
1262 The photoometric model that was fit.
1263 visit_ccd_to_dataRef : `dict` of Key: `lsst.daf.persistence.ButlerDataRef`
1264 Dict of ccdImage identifiers to dataRefs that were fit.
1265 """
1267 ccdImageList = associations.getCcdImageList()
1268 for ccdImage in ccdImageList:
1269 # TODO: there must be a better way to identify this ccdImage than a visit,ccd pair?
1270 ccd = ccdImage.ccdId
1271 visit = ccdImage.visit
1272 dataRef = visit_ccd_to_dataRef[(visit, ccd)]
1273 self.log.info("Updating PhotoCalib for visit: %d, ccd: %d", visit, ccd)
1274 photoCalib = model.toPhotoCalib(ccdImage)
1275 try:
1276 dataRef.put(photoCalib, 'jointcal_photoCalib')
1277 except pexExceptions.Exception as e:
1278 self.log.fatal('Failed to write updated PhotoCalib: %s', str(e))
1279 raise e