Hide keyboard shortcuts

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 pipe_tasks. 

2# 

3# LSST Data Management System 

4# This product includes software developed by the 

5# LSST Project (http://www.lsst.org/). 

6# See COPYRIGHT file at the top of the source tree. 

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 

23from math import ceil 

24import numpy as np 

25from scipy import ndimage 

26import lsst.geom as geom 

27import lsst.afw.image as afwImage 

28import lsst.afw.table as afwTable 

29import lsst.coadd.utils as coaddUtils 

30from lsst.daf.butler import DeferredDatasetHandle 

31from lsst.ip.diffim.dcrModel import applyDcr, calculateDcr, DcrModel 

32import lsst.meas.algorithms as measAlg 

33from lsst.meas.base import SingleFrameMeasurementTask 

34import lsst.pex.config as pexConfig 

35import lsst.pipe.base as pipeBase 

36import lsst.utils as utils 

37from .assembleCoadd import (AssembleCoaddTask, 

38 CompareWarpAssembleCoaddConfig, 

39 CompareWarpAssembleCoaddTask) 

40from .coaddBase import makeSkyInfo 

41from .measurePsf import MeasurePsfTask 

42 

43__all__ = ["DcrAssembleCoaddConnections", "DcrAssembleCoaddTask", "DcrAssembleCoaddConfig"] 

44 

45 

46class DcrAssembleCoaddConnections(pipeBase.PipelineTaskConnections, 

47 dimensions=("tract", "patch", "abstract_filter", "skymap"), 

48 defaultTemplates={"inputCoaddName": "deep", 

49 "outputCoaddName": "dcr", 

50 "warpType": "direct", 

51 "warpTypeSuffix": "", 

52 "fakesType": ""}): 

53 inputWarps = pipeBase.connectionTypes.Input( 

54 doc=("Input list of warps to be assembled i.e. stacked." 

55 "WarpType (e.g. direct, psfMatched) is controlled by the warpType config parameter"), 

56 name="{inputCoaddName}Coadd_{warpType}Warp", 

57 storageClass="ExposureF", 

58 dimensions=("tract", "patch", "skymap", "visit", "instrument"), 

59 deferLoad=True, 

60 multiple=True 

61 ) 

62 skyMap = pipeBase.connectionTypes.Input( 

63 doc="Input definition of geometry/bbox and projection/wcs for coadded exposures", 

64 name="{inputCoaddName}Coadd_skyMap", 

65 storageClass="SkyMap", 

66 dimensions=("skymap", ), 

67 ) 

68 brightObjectMask = pipeBase.connectionTypes.PrerequisiteInput( 

69 doc=("Input Bright Object Mask mask produced with external catalogs to be applied to the mask plane" 

70 " BRIGHT_OBJECT."), 

71 name="brightObjectMask", 

72 storageClass="ObjectMaskCatalog", 

73 dimensions=("tract", "patch", "skymap", "abstract_filter"), 

74 ) 

75 templateExposure = pipeBase.connectionTypes.Input( 

76 doc="Input coadded exposure, produced by previous call to AssembleCoadd", 

77 name="{fakesType}{inputCoaddName}Coadd{warpTypeSuffix}", 

78 storageClass="ExposureF", 

79 dimensions=("tract", "patch", "skymap", "abstract_filter"), 

80 ) 

81 dcrCoadds = pipeBase.connectionTypes.Output( 

82 doc="Output coadded exposure, produced by stacking input warps", 

83 name="{fakesType}{outputCoaddName}Coadd{warpTypeSuffix}", 

84 storageClass="ExposureF", 

85 dimensions=("tract", "patch", "skymap", "abstract_filter", "subfilter"), 

86 multiple=True, 

87 ) 

88 dcrNImages = pipeBase.connectionTypes.Output( 

89 doc="Output image of number of input images per pixel", 

90 name="{outputCoaddName}Coadd_nImage", 

91 storageClass="ImageU", 

92 dimensions=("tract", "patch", "skymap", "abstract_filter", "subfilter"), 

93 multiple=True, 

94 ) 

95 

96 def __init__(self, *, config=None): 

97 super().__init__(config=config) 

98 if not config.doWrite: 

99 self.outputs.remove("dcrCoadds") 

100 

101 

102class DcrAssembleCoaddConfig(CompareWarpAssembleCoaddConfig, 

103 pipelineConnections=DcrAssembleCoaddConnections): 

104 dcrNumSubfilters = pexConfig.Field( 

105 dtype=int, 

106 doc="Number of sub-filters to forward model chromatic effects to fit the supplied exposures.", 

107 default=3, 

108 ) 

109 maxNumIter = pexConfig.Field( 

110 dtype=int, 

111 optional=True, 

112 doc="Maximum number of iterations of forward modeling.", 

113 default=None, 

114 ) 

115 minNumIter = pexConfig.Field( 

116 dtype=int, 

117 optional=True, 

118 doc="Minimum number of iterations of forward modeling.", 

119 default=None, 

120 ) 

121 convergenceThreshold = pexConfig.Field( 

122 dtype=float, 

123 doc="Target relative change in convergence between iterations of forward modeling.", 

124 default=0.001, 

125 ) 

126 useConvergence = pexConfig.Field( 

127 dtype=bool, 

128 doc="Use convergence test as a forward modeling end condition?" 

129 "If not set, skips calculating convergence and runs for ``maxNumIter`` iterations", 

130 default=True, 

131 ) 

132 baseGain = pexConfig.Field( 

133 dtype=float, 

134 optional=True, 

135 doc="Relative weight to give the new solution vs. the last solution when updating the model." 

136 "A value of 1.0 gives equal weight to both solutions." 

137 "Small values imply slower convergence of the solution, but can " 

138 "help prevent overshooting and failures in the fit." 

139 "If ``baseGain`` is None, a conservative gain " 

140 "will be calculated from the number of subfilters. ", 

141 default=None, 

142 ) 

143 useProgressiveGain = pexConfig.Field( 

144 dtype=bool, 

145 doc="Use a gain that slowly increases above ``baseGain`` to accelerate convergence? " 

146 "When calculating the next gain, we use up to 5 previous gains and convergence values." 

147 "Can be set to False to force the model to change at the rate of ``baseGain``. ", 

148 default=True, 

149 ) 

150 doAirmassWeight = pexConfig.Field( 

151 dtype=bool, 

152 doc="Weight exposures by airmass? Useful if there are relatively few high-airmass observations.", 

153 default=False, 

154 ) 

155 modelWeightsWidth = pexConfig.Field( 

156 dtype=float, 

157 doc="Width of the region around detected sources to include in the DcrModel.", 

158 default=3, 

159 ) 

160 useModelWeights = pexConfig.Field( 

161 dtype=bool, 

162 doc="Width of the region around detected sources to include in the DcrModel.", 

163 default=True, 

164 ) 

165 splitSubfilters = pexConfig.Field( 

166 dtype=bool, 

167 doc="Calculate DCR for two evenly-spaced wavelengths in each subfilter." 

168 "Instead of at the midpoint", 

169 default=True, 

170 ) 

171 splitThreshold = pexConfig.Field( 

172 dtype=float, 

173 doc="Minimum DCR difference within a subfilter to use ``splitSubfilters``, in pixels." 

174 "Set to 0 to always split the subfilters.", 

175 default=0.1, 

176 ) 

177 regularizeModelIterations = pexConfig.Field( 

178 dtype=float, 

179 doc="Maximum relative change of the model allowed between iterations." 

180 "Set to zero to disable.", 

181 default=2., 

182 ) 

183 regularizeModelFrequency = pexConfig.Field( 

184 dtype=float, 

185 doc="Maximum relative change of the model allowed between subfilters." 

186 "Set to zero to disable.", 

187 default=4., 

188 ) 

189 convergenceMaskPlanes = pexConfig.ListField( 

190 dtype=str, 

191 default=["DETECTED"], 

192 doc="Mask planes to use to calculate convergence." 

193 ) 

194 regularizationWidth = pexConfig.Field( 

195 dtype=int, 

196 default=2, 

197 doc="Minimum radius of a region to include in regularization, in pixels." 

198 ) 

199 imageInterpOrder = pexConfig.Field( 

200 dtype=int, 

201 doc="The order of the spline interpolation used to shift the image plane.", 

202 default=3, 

203 ) 

204 accelerateModel = pexConfig.Field( 

205 dtype=float, 

206 doc="Factor to amplify the differences between model planes by to speed convergence.", 

207 default=3, 

208 ) 

209 doCalculatePsf = pexConfig.Field( 

210 dtype=bool, 

211 doc="Set to detect stars and recalculate the PSF from the final coadd." 

212 "Otherwise the PSF is estimated from a selection of the best input exposures", 

213 default=False, 

214 ) 

215 detectPsfSources = pexConfig.ConfigurableField( 

216 target=measAlg.SourceDetectionTask, 

217 doc="Task to detect sources for PSF measurement, if ``doCalculatePsf`` is set.", 

218 ) 

219 measurePsfSources = pexConfig.ConfigurableField( 

220 target=SingleFrameMeasurementTask, 

221 doc="Task to measure sources for PSF measurement, if ``doCalculatePsf`` is set." 

222 ) 

223 measurePsf = pexConfig.ConfigurableField( 

224 target=MeasurePsfTask, 

225 doc="Task to measure the PSF of the coadd, if ``doCalculatePsf`` is set.", 

226 ) 

227 

228 def setDefaults(self): 

229 CompareWarpAssembleCoaddConfig.setDefaults(self) 

230 self.assembleStaticSkyModel.retarget(CompareWarpAssembleCoaddTask) 

231 self.doNImage = True 

232 self.assembleStaticSkyModel.warpType = self.warpType 

233 # The deepCoadd and nImage files will be overwritten by this Task, so don't write them the first time 

234 self.assembleStaticSkyModel.doNImage = False 

235 self.assembleStaticSkyModel.doWrite = False 

236 self.detectPsfSources.returnOriginalFootprints = False 

237 self.detectPsfSources.thresholdPolarity = "positive" 

238 # Only use bright sources for PSF measurement 

239 self.detectPsfSources.thresholdValue = 50 

240 self.detectPsfSources.nSigmaToGrow = 2 

241 # A valid star for PSF measurement should at least fill 5x5 pixels 

242 self.detectPsfSources.minPixels = 25 

243 # Use the variance plane to calculate signal to noise 

244 self.detectPsfSources.thresholdType = "pixel_stdev" 

245 # The signal to noise limit is good enough, while the flux limit is set 

246 # in dimensionless units and may not be appropriate for all data sets. 

247 self.measurePsf.starSelector["objectSize"].doFluxLimit = False 

248 

249 

250class DcrAssembleCoaddTask(CompareWarpAssembleCoaddTask): 

251 """Assemble DCR coadded images from a set of warps. 

252 

253 Attributes 

254 ---------- 

255 bufferSize : `int` 

256 The number of pixels to grow each subregion by to allow for DCR. 

257 

258 Notes 

259 ----- 

260 As with AssembleCoaddTask, we want to assemble a coadded image from a set of 

261 Warps (also called coadded temporary exposures), including the effects of 

262 Differential Chromatic Refraction (DCR). 

263 For full details of the mathematics and algorithm, please see 

264 DMTN-037: DCR-matched template generation (https://dmtn-037.lsst.io). 

265 

266 This Task produces a DCR-corrected deepCoadd, as well as a dcrCoadd for 

267 each subfilter used in the iterative calculation. 

268 It begins by dividing the bandpass-defining filter into N equal bandwidth 

269 "subfilters", and divides the flux in each pixel from an initial coadd 

270 equally into each as a "dcrModel". Because the airmass and parallactic 

271 angle of each individual exposure is known, we can calculate the shift 

272 relative to the center of the band in each subfilter due to DCR. For each 

273 exposure we apply this shift as a linear transformation to the dcrModels 

274 and stack the results to produce a DCR-matched exposure. The matched 

275 exposures are subtracted from the input exposures to produce a set of 

276 residual images, and these residuals are reverse shifted for each 

277 exposures' subfilters and stacked. The shifted and stacked residuals are 

278 added to the dcrModels to produce a new estimate of the flux in each pixel 

279 within each subfilter. The dcrModels are solved for iteratively, which 

280 continues until the solution from a new iteration improves by less than 

281 a set percentage, or a maximum number of iterations is reached. 

282 Two forms of regularization are employed to reduce unphysical results. 

283 First, the new solution is averaged with the solution from the previous 

284 iteration, which mitigates oscillating solutions where the model 

285 overshoots with alternating very high and low values. 

286 Second, a common degeneracy when the data have a limited range of airmass or 

287 parallactic angle values is for one subfilter to be fit with very low or 

288 negative values, while another subfilter is fit with very high values. This 

289 typically appears in the form of holes next to sources in one subfilter, 

290 and corresponding extended wings in another. Because each subfilter has 

291 a narrow bandwidth we assume that physical sources that are above the noise 

292 level will not vary in flux by more than a factor of `frequencyClampFactor` 

293 between subfilters, and pixels that have flux deviations larger than that 

294 factor will have the excess flux distributed evenly among all subfilters. 

295 If `splitSubfilters` is set, then each subfilter will be further sub- 

296 divided during the forward modeling step (only). This approximates using 

297 a higher number of subfilters that may be necessary for high airmass 

298 observations, but does not increase the number of free parameters in the 

299 fit. This is needed when there are high airmass observations which would 

300 otherwise have significant DCR even within a subfilter. Because calculating 

301 the shifted images takes most of the time, splitting the subfilters is 

302 turned off by way of the `splitThreshold` option for low-airmass 

303 observations that do not suffer from DCR within a subfilter. 

304 """ 

305 

306 ConfigClass = DcrAssembleCoaddConfig 

307 _DefaultName = "dcrAssembleCoadd" 

308 

309 def __init__(self, *args, **kwargs): 

310 super().__init__(*args, **kwargs) 

311 if self.config.doCalculatePsf: 

312 self.schema = afwTable.SourceTable.makeMinimalSchema() 

313 self.makeSubtask("detectPsfSources", schema=self.schema) 

314 self.makeSubtask("measurePsfSources", schema=self.schema) 

315 self.makeSubtask("measurePsf", schema=self.schema) 

316 

317 @utils.inheritDoc(pipeBase.PipelineTask) 

318 def runQuantum(self, butlerQC, inputRefs, outputRefs): 

319 # Docstring to be formatted with info from PipelineTask.runQuantum 

320 """ 

321 Notes 

322 ----- 

323 Assemble a coadd from a set of Warps. 

324 

325 PipelineTask (Gen3) entry point to Coadd a set of Warps. 

326 Analogous to `runDataRef`, it prepares all the data products to be 

327 passed to `run`, and processes the results before returning a struct 

328 of results to be written out. AssembleCoadd cannot fit all Warps in memory. 

329 Therefore, its inputs are accessed subregion by subregion 

330 by the Gen3 `DeferredDatasetHandle` that is analagous to the Gen2 

331 `lsst.daf.persistence.ButlerDataRef`. Any updates to this method should 

332 correspond to an update in `runDataRef` while both entry points 

333 are used. 

334 """ 

335 inputData = butlerQC.get(inputRefs) 

336 

337 # Construct skyInfo expected by run 

338 # Do not remove skyMap from inputData in case makeSupplementaryDataGen3 needs it 

339 skyMap = inputData["skyMap"] 

340 outputDataId = butlerQC.quantum.dataId 

341 

342 inputData['skyInfo'] = makeSkyInfo(skyMap, 

343 tractId=outputDataId['tract'], 

344 patchId=outputDataId['patch']) 

345 

346 # Construct list of input Deferred Datasets 

347 # These quack a bit like like Gen2 DataRefs 

348 warpRefList = inputData['inputWarps'] 

349 # Perform same middle steps as `runDataRef` does 

350 inputs = self.prepareInputs(warpRefList) 

351 self.log.info("Found %d %s", len(inputs.tempExpRefList), 

352 self.getTempExpDatasetName(self.warpType)) 

353 if len(inputs.tempExpRefList) == 0: 

354 self.log.warn("No coadd temporary exposures found") 

355 return 

356 

357 supplementaryData = self.makeSupplementaryDataGen3(butlerQC, inputRefs, outputRefs) 

358 retStruct = self.run(inputData['skyInfo'], inputs.tempExpRefList, inputs.imageScalerList, 

359 inputs.weightList, supplementaryData=supplementaryData) 

360 

361 inputData.setdefault('brightObjectMask', None) 

362 for subfilter in range(self.config.dcrNumSubfilters): 

363 # Use the PSF of the stacked dcrModel, and do not recalculate the PSF for each subfilter 

364 retStruct.dcrCoadds[subfilter].setPsf(retStruct.coaddExposure.getPsf()) 

365 self.processResults(retStruct.dcrCoadds[subfilter], inputData['brightObjectMask'], outputDataId) 

366 

367 if self.config.doWrite: 

368 butlerQC.put(retStruct, outputRefs) 

369 return retStruct 

370 

371 @pipeBase.timeMethod 

372 def runDataRef(self, dataRef, selectDataList=None, warpRefList=None): 

373 """Assemble a coadd from a set of warps. 

374 

375 Coadd a set of Warps. Compute weights to be applied to each Warp and 

376 find scalings to match the photometric zeropoint to a reference Warp. 

377 Assemble the Warps using run method. 

378 Forward model chromatic effects across multiple subfilters, 

379 and subtract from the input Warps to build sets of residuals. 

380 Use the residuals to construct a new ``DcrModel`` for each subfilter, 

381 and iterate until the model converges. 

382 Interpolate over NaNs and optionally write the coadd to disk. 

383 Return the coadded exposure. 

384 

385 Parameters 

386 ---------- 

387 dataRef : `lsst.daf.persistence.ButlerDataRef` 

388 Data reference defining the patch for coaddition and the 

389 reference Warp 

390 selectDataList : `list` of `lsst.daf.persistence.ButlerDataRef` 

391 List of data references to warps. Data to be coadded will be 

392 selected from this list based on overlap with the patch defined by 

393 the data reference. 

394 

395 Returns 

396 ------- 

397 results : `lsst.pipe.base.Struct` 

398 The Struct contains the following fields: 

399 

400 - ``coaddExposure``: coadded exposure (`lsst.afw.image.Exposure`) 

401 - ``nImage``: exposure count image (`lsst.afw.image.ImageU`) 

402 - ``dcrCoadds``: `list` of coadded exposures for each subfilter 

403 - ``dcrNImages``: `list` of exposure count images for each subfilter 

404 """ 

405 if (selectDataList is None and warpRefList is None) or (selectDataList and warpRefList): 

406 raise RuntimeError("runDataRef must be supplied either a selectDataList or warpRefList") 

407 

408 skyInfo = self.getSkyInfo(dataRef) 

409 if warpRefList is None: 

410 calExpRefList = self.selectExposures(dataRef, skyInfo, selectDataList=selectDataList) 

411 if len(calExpRefList) == 0: 

412 self.log.warn("No exposures to coadd") 

413 return 

414 self.log.info("Coadding %d exposures", len(calExpRefList)) 

415 

416 warpRefList = self.getTempExpRefList(dataRef, calExpRefList) 

417 

418 inputData = self.prepareInputs(warpRefList) 

419 self.log.info("Found %d %s", len(inputData.tempExpRefList), 

420 self.getTempExpDatasetName(self.warpType)) 

421 if len(inputData.tempExpRefList) == 0: 

422 self.log.warn("No coadd temporary exposures found") 

423 return 

424 

425 supplementaryData = self.makeSupplementaryData(dataRef, warpRefList=inputData.tempExpRefList) 

426 

427 results = self.run(skyInfo, inputData.tempExpRefList, inputData.imageScalerList, 

428 inputData.weightList, supplementaryData=supplementaryData) 

429 if results is None: 

430 self.log.warn("Could not construct DcrModel for patch %s: no data to coadd.", 

431 skyInfo.patchInfo.getIndex()) 

432 return 

433 

434 if self.config.doCalculatePsf: 

435 self.measureCoaddPsf(results.coaddExposure) 

436 brightObjects = self.readBrightObjectMasks(dataRef) if self.config.doMaskBrightObjects else None 

437 for subfilter in range(self.config.dcrNumSubfilters): 

438 # Use the PSF of the stacked dcrModel, and do not recalculate the PSF for each subfilter 

439 results.dcrCoadds[subfilter].setPsf(results.coaddExposure.getPsf()) 

440 self.processResults(results.dcrCoadds[subfilter], 

441 brightObjectMasks=brightObjects, dataId=dataRef.dataId) 

442 if self.config.doWrite: 

443 self.log.info("Persisting dcrCoadd") 

444 dataRef.put(results.dcrCoadds[subfilter], "dcrCoadd", subfilter=subfilter, 

445 numSubfilters=self.config.dcrNumSubfilters) 

446 if self.config.doNImage and results.dcrNImages is not None: 

447 dataRef.put(results.dcrNImages[subfilter], "dcrCoadd_nImage", subfilter=subfilter, 

448 numSubfilters=self.config.dcrNumSubfilters) 

449 

450 return results 

451 

452 @utils.inheritDoc(AssembleCoaddTask) 

453 def makeSupplementaryDataGen3(self, butlerQC, inputRefs, outputRefs): 

454 """Load the previously-generated template coadd. 

455 

456 This can be removed entirely once we no longer support the Gen 2 butler. 

457 

458 Returns 

459 ------- 

460 templateCoadd : `lsst.pipe.base.Struct` 

461 Result struct with components: 

462 

463 - ``templateCoadd``: coadded exposure (`lsst.afw.image.ExposureF`) 

464 """ 

465 templateCoadd = butlerQC.get(inputRefs.templateExposure) 

466 

467 return pipeBase.Struct(templateCoadd=templateCoadd) 

468 

469 def measureCoaddPsf(self, coaddExposure): 

470 """Detect sources on the coadd exposure and measure the final PSF. 

471 

472 Parameters 

473 ---------- 

474 coaddExposure : `lsst.afw.image.Exposure` 

475 The final coadded exposure. 

476 """ 

477 table = afwTable.SourceTable.make(self.schema) 

478 detResults = self.detectPsfSources.run(table, coaddExposure, clearMask=False) 

479 coaddSources = detResults.sources 

480 self.measurePsfSources.run( 

481 measCat=coaddSources, 

482 exposure=coaddExposure 

483 ) 

484 # Measure the PSF on the stacked subfilter coadds if possible. 

485 # We should already have a decent estimate of the coadd PSF, however, 

486 # so in case of any errors simply log them as a warning and use the 

487 # default PSF. 

488 try: 

489 psfResults = self.measurePsf.run(coaddExposure, coaddSources) 

490 except Exception as e: 

491 self.log.warn("Unable to calculate PSF, using default coadd PSF: %s" % e) 

492 else: 

493 coaddExposure.setPsf(psfResults.psf) 

494 

495 def prepareDcrInputs(self, templateCoadd, warpRefList, weightList): 

496 """Prepare the DCR coadd by iterating through the visitInfo of the input warps. 

497 

498 Sets the property ``bufferSize``. 

499 

500 Parameters 

501 ---------- 

502 templateCoadd : `lsst.afw.image.ExposureF` 

503 The initial coadd exposure before accounting for DCR. 

504 warpRefList : `list` of `lsst.daf.butler.DeferredDatasetHandle` or 

505 `lsst.daf.persistence.ButlerDataRef` 

506 The data references to the input warped exposures. 

507 weightList : `list` of `float` 

508 The weight to give each input exposure in the coadd 

509 Will be modified in place if ``doAirmassWeight`` is set. 

510 

511 Returns 

512 ------- 

513 dcrModels : `lsst.pipe.tasks.DcrModel` 

514 Best fit model of the true sky after correcting chromatic effects. 

515 

516 Raises 

517 ------ 

518 NotImplementedError 

519 If ``lambdaMin`` is missing from the Mapper class of the obs package being used. 

520 """ 

521 sigma2fwhm = 2.*np.sqrt(2.*np.log(2.)) 

522 filterInfo = templateCoadd.getFilter() 

523 if np.isnan(filterInfo.getFilterProperty().getLambdaMin()): 

524 raise NotImplementedError("No minimum/maximum wavelength information found" 

525 " in the filter definition! Please add lambdaMin and lambdaMax" 

526 " to the Mapper class in your obs package.") 

527 tempExpName = self.getTempExpDatasetName(self.warpType) 

528 dcrShifts = [] 

529 airmassDict = {} 

530 angleDict = {} 

531 psfSizeDict = {} 

532 for visitNum, warpExpRef in enumerate(warpRefList): 

533 if isinstance(warpExpRef, DeferredDatasetHandle): 

534 # Gen 3 API 

535 visitInfo = warpExpRef.get(component="visitInfo") 

536 psf = warpExpRef.get(component="psf") 

537 else: 

538 # Gen 2 API. Delete this when Gen 2 retired 

539 visitInfo = warpExpRef.get(tempExpName + "_visitInfo") 

540 psf = warpExpRef.get(tempExpName).getPsf() 

541 visit = warpExpRef.dataId["visit"] 

542 psfSize = psf.computeShape().getDeterminantRadius()*sigma2fwhm 

543 airmass = visitInfo.getBoresightAirmass() 

544 parallacticAngle = visitInfo.getBoresightParAngle().asDegrees() 

545 airmassDict[visit] = airmass 

546 angleDict[visit] = parallacticAngle 

547 psfSizeDict[visit] = psfSize 

548 if self.config.doAirmassWeight: 

549 weightList[visitNum] *= airmass 

550 dcrShifts.append(np.max(np.abs(calculateDcr(visitInfo, templateCoadd.getWcs(), 

551 filterInfo, self.config.dcrNumSubfilters)))) 

552 self.log.info("Selected airmasses:\n%s", airmassDict) 

553 self.log.info("Selected parallactic angles:\n%s", angleDict) 

554 self.log.info("Selected PSF sizes:\n%s", psfSizeDict) 

555 self.bufferSize = int(np.ceil(np.max(dcrShifts)) + 1) 

556 psf = self.selectCoaddPsf(templateCoadd, warpRefList) 

557 dcrModels = DcrModel.fromImage(templateCoadd.maskedImage, 

558 self.config.dcrNumSubfilters, 

559 filterInfo=filterInfo, 

560 psf=psf) 

561 return dcrModels 

562 

563 def run(self, skyInfo, warpRefList, imageScalerList, weightList, 

564 supplementaryData=None): 

565 """Assemble the coadd. 

566 

567 Requires additional inputs Struct ``supplementaryData`` to contain a 

568 ``templateCoadd`` that serves as the model of the static sky. 

569 

570 Find artifacts and apply them to the warps' masks creating a list of 

571 alternative masks with a new "CLIPPED" plane and updated "NO_DATA" plane 

572 Then pass these alternative masks to the base class's assemble method. 

573 

574 Divide the ``templateCoadd`` evenly between each subfilter of a 

575 ``DcrModel`` as the starting best estimate of the true wavelength- 

576 dependent sky. Forward model the ``DcrModel`` using the known 

577 chromatic effects in each subfilter and calculate a convergence metric 

578 based on how well the modeled template matches the input warps. If 

579 the convergence has not yet reached the desired threshold, then shift 

580 and stack the residual images to build a new ``DcrModel``. Apply 

581 conditioning to prevent oscillating solutions between iterations or 

582 between subfilters. 

583 

584 Once the ``DcrModel`` reaches convergence or the maximum number of 

585 iterations has been reached, fill the metadata for each subfilter 

586 image and make them proper ``coaddExposure``s. 

587 

588 Parameters 

589 ---------- 

590 skyInfo : `lsst.pipe.base.Struct` 

591 Patch geometry information, from getSkyInfo 

592 warpRefList : `list` of `lsst.daf.butler.DeferredDatasetHandle` or 

593 `lsst.daf.persistence.ButlerDataRef` 

594 The data references to the input warped exposures. 

595 imageScalerList : `list` of `lsst.pipe.task.ImageScaler` 

596 The image scalars correct for the zero point of the exposures. 

597 weightList : `list` of `float` 

598 The weight to give each input exposure in the coadd 

599 supplementaryData : `lsst.pipe.base.Struct` 

600 Result struct returned by ``makeSupplementaryData`` with components: 

601 

602 - ``templateCoadd``: coadded exposure (`lsst.afw.image.Exposure`) 

603 

604 Returns 

605 ------- 

606 result : `lsst.pipe.base.Struct` 

607 Result struct with components: 

608 

609 - ``coaddExposure``: coadded exposure (`lsst.afw.image.Exposure`) 

610 - ``nImage``: exposure count image (`lsst.afw.image.ImageU`) 

611 - ``dcrCoadds``: `list` of coadded exposures for each subfilter 

612 - ``dcrNImages``: `list` of exposure count images for each subfilter 

613 """ 

614 minNumIter = self.config.minNumIter or self.config.dcrNumSubfilters 

615 maxNumIter = self.config.maxNumIter or self.config.dcrNumSubfilters*3 

616 templateCoadd = supplementaryData.templateCoadd 

617 baseMask = templateCoadd.mask.clone() 

618 # The variance plane is for each subfilter 

619 # and should be proportionately lower than the full-band image 

620 baseVariance = templateCoadd.variance.clone() 

621 baseVariance /= self.config.dcrNumSubfilters 

622 spanSetMaskList = self.findArtifacts(templateCoadd, warpRefList, imageScalerList) 

623 # Note that the mask gets cleared in ``findArtifacts``, but we want to preserve the mask. 

624 templateCoadd.setMask(baseMask) 

625 badMaskPlanes = self.config.badMaskPlanes[:] 

626 # Note that is important that we do not add "CLIPPED" to ``badMaskPlanes`` 

627 # This is because pixels in observations that are significantly affect by DCR 

628 # are likely to have many pixels that are both "DETECTED" and "CLIPPED", 

629 # but those are necessary to constrain the DCR model. 

630 badPixelMask = templateCoadd.mask.getPlaneBitMask(badMaskPlanes) 

631 

632 stats = self.prepareStats(mask=badPixelMask) 

633 dcrModels = self.prepareDcrInputs(templateCoadd, warpRefList, weightList) 

634 if self.config.doNImage: 

635 dcrNImages, dcrWeights = self.calculateNImage(dcrModels, skyInfo.bbox, warpRefList, 

636 spanSetMaskList, stats.ctrl) 

637 nImage = afwImage.ImageU(skyInfo.bbox) 

638 # Note that this nImage will be a factor of dcrNumSubfilters higher than 

639 # the nImage returned by assembleCoadd for most pixels. This is because each 

640 # subfilter may have a different nImage, and fractional values are not allowed. 

641 for dcrNImage in dcrNImages: 

642 nImage += dcrNImage 

643 else: 

644 dcrNImages = None 

645 

646 subregionSize = geom.Extent2I(*self.config.subregionSize) 

647 nSubregions = (ceil(skyInfo.bbox.getHeight()/subregionSize[1]) * 

648 ceil(skyInfo.bbox.getWidth()/subregionSize[0])) 

649 subIter = 0 

650 for subBBox in self._subBBoxIter(skyInfo.bbox, subregionSize): 

651 modelIter = 0 

652 subIter += 1 

653 self.log.info("Computing coadd over patch %s subregion %s of %s: %s", 

654 skyInfo.patchInfo.getIndex(), subIter, nSubregions, subBBox) 

655 dcrBBox = geom.Box2I(subBBox) 

656 dcrBBox.grow(self.bufferSize) 

657 dcrBBox.clip(dcrModels.bbox) 

658 modelWeights = self.calculateModelWeights(dcrModels, dcrBBox) 

659 subExposures = self.loadSubExposures(dcrBBox, stats.ctrl, warpRefList, 

660 imageScalerList, spanSetMaskList) 

661 convergenceMetric = self.calculateConvergence(dcrModels, subExposures, subBBox, 

662 warpRefList, weightList, stats.ctrl) 

663 self.log.info("Initial convergence : %s", convergenceMetric) 

664 convergenceList = [convergenceMetric] 

665 gainList = [] 

666 convergenceCheck = 1. 

667 refImage = templateCoadd.image 

668 while (convergenceCheck > self.config.convergenceThreshold or modelIter <= minNumIter): 

669 gain = self.calculateGain(convergenceList, gainList) 

670 self.dcrAssembleSubregion(dcrModels, subExposures, subBBox, dcrBBox, warpRefList, 

671 stats.ctrl, convergenceMetric, gain, 

672 modelWeights, refImage, dcrWeights) 

673 if self.config.useConvergence: 

674 convergenceMetric = self.calculateConvergence(dcrModels, subExposures, subBBox, 

675 warpRefList, weightList, stats.ctrl) 

676 if convergenceMetric == 0: 

677 self.log.warn("Coadd patch %s subregion %s had convergence metric of 0.0 which is " 

678 "most likely due to there being no valid data in the region.", 

679 skyInfo.patchInfo.getIndex(), subIter) 

680 break 

681 convergenceCheck = (convergenceList[-1] - convergenceMetric)/convergenceMetric 

682 if (convergenceCheck < 0) & (modelIter > minNumIter): 

683 self.log.warn("Coadd patch %s subregion %s diverged before reaching maximum " 

684 "iterations or desired convergence improvement of %s." 

685 " Divergence: %s", 

686 skyInfo.patchInfo.getIndex(), subIter, 

687 self.config.convergenceThreshold, convergenceCheck) 

688 break 

689 convergenceList.append(convergenceMetric) 

690 if modelIter > maxNumIter: 

691 if self.config.useConvergence: 

692 self.log.warn("Coadd patch %s subregion %s reached maximum iterations " 

693 "before reaching desired convergence improvement of %s." 

694 " Final convergence improvement: %s", 

695 skyInfo.patchInfo.getIndex(), subIter, 

696 self.config.convergenceThreshold, convergenceCheck) 

697 break 

698 

699 if self.config.useConvergence: 

700 self.log.info("Iteration %s with convergence metric %s, %.4f%% improvement (gain: %.2f)", 

701 modelIter, convergenceMetric, 100.*convergenceCheck, gain) 

702 modelIter += 1 

703 else: 

704 if self.config.useConvergence: 

705 self.log.info("Coadd patch %s subregion %s finished with " 

706 "convergence metric %s after %s iterations", 

707 skyInfo.patchInfo.getIndex(), subIter, convergenceMetric, modelIter) 

708 else: 

709 self.log.info("Coadd patch %s subregion %s finished after %s iterations", 

710 skyInfo.patchInfo.getIndex(), subIter, modelIter) 

711 if self.config.useConvergence and convergenceMetric > 0: 

712 self.log.info("Final convergence improvement was %.4f%% overall", 

713 100*(convergenceList[0] - convergenceMetric)/convergenceMetric) 

714 

715 dcrCoadds = self.fillCoadd(dcrModels, skyInfo, warpRefList, weightList, 

716 calibration=self.scaleZeroPoint.getPhotoCalib(), 

717 coaddInputs=templateCoadd.getInfo().getCoaddInputs(), 

718 mask=baseMask, 

719 variance=baseVariance) 

720 coaddExposure = self.stackCoadd(dcrCoadds) 

721 return pipeBase.Struct(coaddExposure=coaddExposure, nImage=nImage, 

722 dcrCoadds=dcrCoadds, dcrNImages=dcrNImages) 

723 

724 def calculateNImage(self, dcrModels, bbox, warpRefList, spanSetMaskList, statsCtrl): 

725 """Calculate the number of exposures contributing to each subfilter. 

726 

727 Parameters 

728 ---------- 

729 dcrModels : `lsst.pipe.tasks.DcrModel` 

730 Best fit model of the true sky after correcting chromatic effects. 

731 bbox : `lsst.geom.box.Box2I` 

732 Bounding box of the patch to coadd. 

733 warpRefList : `list` of `lsst.daf.butler.DeferredDatasetHandle` or 

734 `lsst.daf.persistence.ButlerDataRef` 

735 The data references to the input warped exposures. 

736 spanSetMaskList : `list` of `dict` containing spanSet lists, or None 

737 Each element of the `dict` contains the new mask plane name 

738 (e.g. "CLIPPED and/or "NO_DATA") as the key, 

739 and the list of SpanSets to apply to the mask. 

740 statsCtrl : `lsst.afw.math.StatisticsControl` 

741 Statistics control object for coadd 

742 

743 Returns 

744 ------- 

745 dcrNImages : `list` of `lsst.afw.image.ImageU` 

746 List of exposure count images for each subfilter 

747 dcrWeights : `list` of `lsst.afw.image.ImageF` 

748 Per-pixel weights for each subfilter. 

749 Equal to 1/(number of unmasked images contributing to each pixel). 

750 """ 

751 dcrNImages = [afwImage.ImageU(bbox) for subfilter in range(self.config.dcrNumSubfilters)] 

752 dcrWeights = [afwImage.ImageF(bbox) for subfilter in range(self.config.dcrNumSubfilters)] 

753 tempExpName = self.getTempExpDatasetName(self.warpType) 

754 for warpExpRef, altMaskSpans in zip(warpRefList, spanSetMaskList): 

755 if isinstance(warpExpRef, DeferredDatasetHandle): 

756 # Gen 3 API 

757 exposure = warpExpRef.get(parameters={'bbox': bbox}) 

758 else: 

759 # Gen 2 API. Delete this when Gen 2 retired 

760 exposure = warpExpRef.get(tempExpName + "_sub", bbox=bbox) 

761 visitInfo = exposure.getInfo().getVisitInfo() 

762 wcs = exposure.getInfo().getWcs() 

763 mask = exposure.mask 

764 if altMaskSpans is not None: 

765 self.applyAltMaskPlanes(mask, altMaskSpans) 

766 weightImage = np.zeros_like(exposure.image.array) 

767 weightImage[(mask.array & statsCtrl.getAndMask()) == 0] = 1. 

768 # The weights must be shifted in exactly the same way as the residuals, 

769 # because they will be used as the denominator in the weighted average of residuals. 

770 weightsGenerator = self.dcrResiduals(weightImage, visitInfo, wcs, dcrModels.filter) 

771 for shiftedWeights, dcrNImage, dcrWeight in zip(weightsGenerator, dcrNImages, dcrWeights): 

772 dcrNImage.array += np.rint(shiftedWeights).astype(dcrNImage.array.dtype) 

773 dcrWeight.array += shiftedWeights 

774 # Exclude any pixels that don't have at least one exposure contributing in all subfilters 

775 weightsThreshold = 1. 

776 goodPix = dcrWeights[0].array > weightsThreshold 

777 for weights in dcrWeights[1:]: 

778 goodPix = (weights.array > weightsThreshold) & goodPix 

779 for subfilter in range(self.config.dcrNumSubfilters): 

780 dcrWeights[subfilter].array[goodPix] = 1./dcrWeights[subfilter].array[goodPix] 

781 dcrWeights[subfilter].array[~goodPix] = 0. 

782 dcrNImages[subfilter].array[~goodPix] = 0 

783 return (dcrNImages, dcrWeights) 

784 

785 def dcrAssembleSubregion(self, dcrModels, subExposures, bbox, dcrBBox, warpRefList, 

786 statsCtrl, convergenceMetric, 

787 gain, modelWeights, refImage, dcrWeights): 

788 """Assemble the DCR coadd for a sub-region. 

789 

790 Build a DCR-matched template for each input exposure, then shift the 

791 residuals according to the DCR in each subfilter. 

792 Stack the shifted residuals and apply them as a correction to the 

793 solution from the previous iteration. 

794 Restrict the new model solutions from varying by more than a factor of 

795 `modelClampFactor` from the last solution, and additionally restrict the 

796 individual subfilter models from varying by more than a factor of 

797 `frequencyClampFactor` from their average. 

798 Finally, mitigate potentially oscillating solutions by averaging the new 

799 solution with the solution from the previous iteration, weighted by 

800 their convergence metric. 

801 

802 Parameters 

803 ---------- 

804 dcrModels : `lsst.pipe.tasks.DcrModel` 

805 Best fit model of the true sky after correcting chromatic effects. 

806 subExposures : `dict` of `lsst.afw.image.ExposureF` 

807 The pre-loaded exposures for the current subregion. 

808 bbox : `lsst.geom.box.Box2I` 

809 Bounding box of the subregion to coadd. 

810 dcrBBox : `lsst.geom.box.Box2I` 

811 Sub-region of the coadd which includes a buffer to allow for DCR. 

812 warpRefList : `list` of `lsst.daf.butler.DeferredDatasetHandle` or 

813 `lsst.daf.persistence.ButlerDataRef` 

814 The data references to the input warped exposures. 

815 statsCtrl : `lsst.afw.math.StatisticsControl` 

816 Statistics control object for coadd 

817 convergenceMetric : `float` 

818 Quality of fit metric for the matched templates of the input images. 

819 gain : `float`, optional 

820 Relative weight to give the new solution when updating the model. 

821 modelWeights : `numpy.ndarray` or `float` 

822 A 2D array of weight values that tapers smoothly to zero away from detected sources. 

823 Set to a placeholder value of 1.0 if ``self.config.useModelWeights`` is False. 

824 refImage : `lsst.afw.image.Image` 

825 A reference image used to supply the default pixel values. 

826 dcrWeights : `list` of `lsst.afw.image.Image` 

827 Per-pixel weights for each subfilter. 

828 Equal to 1/(number of unmasked images contributing to each pixel). 

829 """ 

830 residualGeneratorList = [] 

831 

832 for warpExpRef in warpRefList: 

833 visit = warpExpRef.dataId["visit"] 

834 exposure = subExposures[visit] 

835 visitInfo = exposure.getInfo().getVisitInfo() 

836 wcs = exposure.getInfo().getWcs() 

837 templateImage = dcrModels.buildMatchedTemplate(exposure=exposure, 

838 order=self.config.imageInterpOrder, 

839 splitSubfilters=self.config.splitSubfilters, 

840 splitThreshold=self.config.splitThreshold, 

841 amplifyModel=self.config.accelerateModel) 

842 residual = exposure.image.array - templateImage.array 

843 # Note that the variance plane here is used to store weights, not the actual variance 

844 residual *= exposure.variance.array 

845 # The residuals are stored as a list of generators. 

846 # This allows the residual for a given subfilter and exposure to be created 

847 # on the fly, instead of needing to store them all in memory. 

848 residualGeneratorList.append(self.dcrResiduals(residual, visitInfo, wcs, dcrModels.filter)) 

849 

850 dcrSubModelOut = self.newModelFromResidual(dcrModels, residualGeneratorList, dcrBBox, statsCtrl, 

851 gain=gain, 

852 modelWeights=modelWeights, 

853 refImage=refImage, 

854 dcrWeights=dcrWeights) 

855 dcrModels.assign(dcrSubModelOut, bbox) 

856 

857 def dcrResiduals(self, residual, visitInfo, wcs, filterInfo): 

858 """Prepare a residual image for stacking in each subfilter by applying the reverse DCR shifts. 

859 

860 Parameters 

861 ---------- 

862 residual : `numpy.ndarray` 

863 The residual masked image for one exposure, 

864 after subtracting the matched template 

865 visitInfo : `lsst.afw.image.VisitInfo` 

866 Metadata for the exposure. 

867 wcs : `lsst.afw.geom.SkyWcs` 

868 Coordinate system definition (wcs) for the exposure. 

869 filterInfo : `lsst.afw.image.Filter` 

870 The filter definition, set in the current instruments' obs package. 

871 Required for any calculation of DCR, including making matched templates. 

872 

873 Yields 

874 ------ 

875 residualImage : `numpy.ndarray` 

876 The residual image for the next subfilter, shifted for DCR. 

877 """ 

878 # Pre-calculate the spline-filtered residual image, so that step can be 

879 # skipped in the shift calculation in `applyDcr`. 

880 filteredResidual = ndimage.spline_filter(residual, order=self.config.imageInterpOrder) 

881 # Note that `splitSubfilters` is always turned off in the reverse direction. 

882 # This option introduces additional blurring if applied to the residuals. 

883 dcrShift = calculateDcr(visitInfo, wcs, filterInfo, self.config.dcrNumSubfilters, 

884 splitSubfilters=False) 

885 for dcr in dcrShift: 

886 yield applyDcr(filteredResidual, dcr, useInverse=True, splitSubfilters=False, 

887 doPrefilter=False, order=self.config.imageInterpOrder) 

888 

889 def newModelFromResidual(self, dcrModels, residualGeneratorList, dcrBBox, statsCtrl, 

890 gain, modelWeights, refImage, dcrWeights): 

891 """Calculate a new DcrModel from a set of image residuals. 

892 

893 Parameters 

894 ---------- 

895 dcrModels : `lsst.pipe.tasks.DcrModel` 

896 Current model of the true sky after correcting chromatic effects. 

897 residualGeneratorList : `generator` of `numpy.ndarray` 

898 The residual image for the next subfilter, shifted for DCR. 

899 dcrBBox : `lsst.geom.box.Box2I` 

900 Sub-region of the coadd which includes a buffer to allow for DCR. 

901 statsCtrl : `lsst.afw.math.StatisticsControl` 

902 Statistics control object for coadd 

903 gain : `float` 

904 Relative weight to give the new solution when updating the model. 

905 modelWeights : `numpy.ndarray` or `float` 

906 A 2D array of weight values that tapers smoothly to zero away from detected sources. 

907 Set to a placeholder value of 1.0 if ``self.config.useModelWeights`` is False. 

908 refImage : `lsst.afw.image.Image` 

909 A reference image used to supply the default pixel values. 

910 dcrWeights : `list` of `lsst.afw.image.Image` 

911 Per-pixel weights for each subfilter. 

912 Equal to 1/(number of unmasked images contributing to each pixel). 

913 

914 Returns 

915 ------- 

916 dcrModel : `lsst.pipe.tasks.DcrModel` 

917 New model of the true sky after correcting chromatic effects. 

918 """ 

919 newModelImages = [] 

920 for subfilter, model in enumerate(dcrModels): 

921 residualsList = [next(residualGenerator) for residualGenerator in residualGeneratorList] 

922 residual = np.sum(residualsList, axis=0) 

923 residual *= dcrWeights[subfilter][dcrBBox].array 

924 # `MaskedImage`s only support in-place addition, so rename for readability 

925 newModel = model[dcrBBox].clone() 

926 newModel.array += residual 

927 # Catch any invalid values 

928 badPixels = ~np.isfinite(newModel.array) 

929 newModel.array[badPixels] = model[dcrBBox].array[badPixels] 

930 if self.config.regularizeModelIterations > 0: 

931 dcrModels.regularizeModelIter(subfilter, newModel, dcrBBox, 

932 self.config.regularizeModelIterations, 

933 self.config.regularizationWidth) 

934 newModelImages.append(newModel) 

935 if self.config.regularizeModelFrequency > 0: 

936 dcrModels.regularizeModelFreq(newModelImages, dcrBBox, statsCtrl, 

937 self.config.regularizeModelFrequency, 

938 self.config.regularizationWidth) 

939 dcrModels.conditionDcrModel(newModelImages, dcrBBox, gain=gain) 

940 self.applyModelWeights(newModelImages, refImage[dcrBBox], modelWeights) 

941 return DcrModel(newModelImages, dcrModels.filter, dcrModels.psf, 

942 dcrModels.mask, dcrModels.variance) 

943 

944 def calculateConvergence(self, dcrModels, subExposures, bbox, warpRefList, weightList, statsCtrl): 

945 """Calculate a quality of fit metric for the matched templates. 

946 

947 Parameters 

948 ---------- 

949 dcrModels : `lsst.pipe.tasks.DcrModel` 

950 Best fit model of the true sky after correcting chromatic effects. 

951 subExposures : `dict` of `lsst.afw.image.ExposureF` 

952 The pre-loaded exposures for the current subregion. 

953 bbox : `lsst.geom.box.Box2I` 

954 Sub-region to coadd 

955 warpRefList : `list` of `lsst.daf.butler.DeferredDatasetHandle` or 

956 `lsst.daf.persistence.ButlerDataRef` 

957 The data references to the input warped exposures. 

958 weightList : `list` of `float` 

959 The weight to give each input exposure in the coadd 

960 statsCtrl : `lsst.afw.math.StatisticsControl` 

961 Statistics control object for coadd 

962 

963 Returns 

964 ------- 

965 convergenceMetric : `float` 

966 Quality of fit metric for all input exposures, within the sub-region 

967 """ 

968 significanceImage = np.abs(dcrModels.getReferenceImage(bbox)) 

969 nSigma = 3. 

970 significanceImage += nSigma*dcrModels.calculateNoiseCutoff(dcrModels[1], statsCtrl, 

971 bufferSize=self.bufferSize) 

972 if np.max(significanceImage) == 0: 

973 significanceImage += 1. 

974 weight = 0 

975 metric = 0. 

976 metricList = {} 

977 for warpExpRef, expWeight in zip(warpRefList, weightList): 

978 visit = warpExpRef.dataId["visit"] 

979 exposure = subExposures[visit][bbox] 

980 singleMetric = self.calculateSingleConvergence(dcrModels, exposure, significanceImage, statsCtrl) 

981 metric += singleMetric 

982 metricList[visit] = singleMetric 

983 weight += 1. 

984 self.log.info("Individual metrics:\n%s", metricList) 

985 return 1.0 if weight == 0.0 else metric/weight 

986 

987 def calculateSingleConvergence(self, dcrModels, exposure, significanceImage, statsCtrl): 

988 """Calculate a quality of fit metric for a single matched template. 

989 

990 Parameters 

991 ---------- 

992 dcrModels : `lsst.pipe.tasks.DcrModel` 

993 Best fit model of the true sky after correcting chromatic effects. 

994 exposure : `lsst.afw.image.ExposureF` 

995 The input warped exposure to evaluate. 

996 significanceImage : `numpy.ndarray` 

997 Array of weights for each pixel corresponding to its significance 

998 for the convergence calculation. 

999 statsCtrl : `lsst.afw.math.StatisticsControl` 

1000 Statistics control object for coadd 

1001 

1002 Returns 

1003 ------- 

1004 convergenceMetric : `float` 

1005 Quality of fit metric for one exposure, within the sub-region. 

1006 """ 

1007 convergeMask = exposure.mask.getPlaneBitMask(self.config.convergenceMaskPlanes) 

1008 templateImage = dcrModels.buildMatchedTemplate(exposure=exposure, 

1009 order=self.config.imageInterpOrder, 

1010 splitSubfilters=self.config.splitSubfilters, 

1011 splitThreshold=self.config.splitThreshold, 

1012 amplifyModel=self.config.accelerateModel) 

1013 diffVals = np.abs(exposure.image.array - templateImage.array)*significanceImage 

1014 refVals = np.abs(exposure.image.array + templateImage.array)*significanceImage/2. 

1015 

1016 finitePixels = np.isfinite(diffVals) 

1017 goodMaskPixels = (exposure.mask.array & statsCtrl.getAndMask()) == 0 

1018 convergeMaskPixels = exposure.mask.array & convergeMask > 0 

1019 usePixels = finitePixels & goodMaskPixels & convergeMaskPixels 

1020 if np.sum(usePixels) == 0: 

1021 metric = 0. 

1022 else: 

1023 diffUse = diffVals[usePixels] 

1024 refUse = refVals[usePixels] 

1025 metric = np.sum(diffUse/np.median(diffUse))/np.sum(refUse/np.median(diffUse)) 

1026 return metric 

1027 

1028 def stackCoadd(self, dcrCoadds): 

1029 """Add a list of sub-band coadds together. 

1030 

1031 Parameters 

1032 ---------- 

1033 dcrCoadds : `list` of `lsst.afw.image.ExposureF` 

1034 A list of coadd exposures, each exposure containing 

1035 the model for one subfilter. 

1036 

1037 Returns 

1038 ------- 

1039 coaddExposure : `lsst.afw.image.ExposureF` 

1040 A single coadd exposure that is the sum of the sub-bands. 

1041 """ 

1042 coaddExposure = dcrCoadds[0].clone() 

1043 for coadd in dcrCoadds[1:]: 

1044 coaddExposure.maskedImage += coadd.maskedImage 

1045 return coaddExposure 

1046 

1047 def fillCoadd(self, dcrModels, skyInfo, warpRefList, weightList, calibration=None, coaddInputs=None, 

1048 mask=None, variance=None): 

1049 """Create a list of coadd exposures from a list of masked images. 

1050 

1051 Parameters 

1052 ---------- 

1053 dcrModels : `lsst.pipe.tasks.DcrModel` 

1054 Best fit model of the true sky after correcting chromatic effects. 

1055 skyInfo : `lsst.pipe.base.Struct` 

1056 Patch geometry information, from getSkyInfo 

1057 warpRefList : `list` of `lsst.daf.butler.DeferredDatasetHandle` or 

1058 `lsst.daf.persistence.ButlerDataRef` 

1059 The data references to the input warped exposures. 

1060 weightList : `list` of `float` 

1061 The weight to give each input exposure in the coadd 

1062 calibration : `lsst.afw.Image.PhotoCalib`, optional 

1063 Scale factor to set the photometric calibration of an exposure. 

1064 coaddInputs : `lsst.afw.Image.CoaddInputs`, optional 

1065 A record of the observations that are included in the coadd. 

1066 mask : `lsst.afw.image.Mask`, optional 

1067 Optional mask to override the values in the final coadd. 

1068 variance : `lsst.afw.image.Image`, optional 

1069 Optional variance plane to override the values in the final coadd. 

1070 

1071 Returns 

1072 ------- 

1073 dcrCoadds : `list` of `lsst.afw.image.ExposureF` 

1074 A list of coadd exposures, each exposure containing 

1075 the model for one subfilter. 

1076 """ 

1077 dcrCoadds = [] 

1078 refModel = dcrModels.getReferenceImage() 

1079 for model in dcrModels: 

1080 if self.config.accelerateModel > 1: 

1081 model.array = (model.array - refModel)*self.config.accelerateModel + refModel 

1082 coaddExposure = afwImage.ExposureF(skyInfo.bbox, skyInfo.wcs) 

1083 if calibration is not None: 

1084 coaddExposure.setPhotoCalib(calibration) 

1085 if coaddInputs is not None: 

1086 coaddExposure.getInfo().setCoaddInputs(coaddInputs) 

1087 # Set the metadata for the coadd, including PSF and aperture corrections. 

1088 self.assembleMetadata(coaddExposure, warpRefList, weightList) 

1089 # Overwrite the PSF 

1090 coaddExposure.setPsf(dcrModels.psf) 

1091 coaddUtils.setCoaddEdgeBits(dcrModels.mask[skyInfo.bbox], dcrModels.variance[skyInfo.bbox]) 

1092 maskedImage = afwImage.MaskedImageF(dcrModels.bbox) 

1093 maskedImage.image = model 

1094 maskedImage.mask = dcrModels.mask 

1095 maskedImage.variance = dcrModels.variance 

1096 coaddExposure.setMaskedImage(maskedImage[skyInfo.bbox]) 

1097 coaddExposure.setPhotoCalib(self.scaleZeroPoint.getPhotoCalib()) 

1098 if mask is not None: 

1099 coaddExposure.setMask(mask) 

1100 if variance is not None: 

1101 coaddExposure.setVariance(variance) 

1102 dcrCoadds.append(coaddExposure) 

1103 return dcrCoadds 

1104 

1105 def calculateGain(self, convergenceList, gainList): 

1106 """Calculate the gain to use for the current iteration. 

1107 

1108 After calculating a new DcrModel, each value is averaged with the 

1109 value in the corresponding pixel from the previous iteration. This 

1110 reduces oscillating solutions that iterative techniques are plagued by, 

1111 and speeds convergence. By far the biggest changes to the model 

1112 happen in the first couple iterations, so we can also use a more 

1113 aggressive gain later when the model is changing slowly. 

1114 

1115 Parameters 

1116 ---------- 

1117 convergenceList : `list` of `float` 

1118 The quality of fit metric from each previous iteration. 

1119 gainList : `list` of `float` 

1120 The gains used in each previous iteration: appended with the new 

1121 gain value. 

1122 Gains are numbers between ``self.config.baseGain`` and 1. 

1123 

1124 Returns 

1125 ------- 

1126 gain : `float` 

1127 Relative weight to give the new solution when updating the model. 

1128 A value of 1.0 gives equal weight to both solutions. 

1129 

1130 Raises 

1131 ------ 

1132 ValueError 

1133 If ``len(convergenceList) != len(gainList)+1``. 

1134 """ 

1135 nIter = len(convergenceList) 

1136 if nIter != len(gainList) + 1: 

1137 raise ValueError("convergenceList (%d) must be one element longer than gainList (%d)." 

1138 % (len(convergenceList), len(gainList))) 

1139 

1140 if self.config.baseGain is None: 

1141 # If ``baseGain`` is not set, calculate it from the number of DCR subfilters 

1142 # The more subfilters being modeled, the lower the gain should be. 

1143 baseGain = 1./(self.config.dcrNumSubfilters - 1) 

1144 else: 

1145 baseGain = self.config.baseGain 

1146 

1147 if self.config.useProgressiveGain and nIter > 2: 

1148 # To calculate the best gain to use, compare the past gains that have been used 

1149 # with the resulting convergences to estimate the best gain to use. 

1150 # Algorithmically, this is a Kalman filter. 

1151 # If forward modeling proceeds perfectly, the convergence metric should 

1152 # asymptotically approach a final value. 

1153 # We can estimate that value from the measured changes in convergence 

1154 # weighted by the gains used in each previous iteration. 

1155 estFinalConv = [((1 + gainList[i])*convergenceList[i + 1] - convergenceList[i])/gainList[i] 

1156 for i in range(nIter - 1)] 

1157 # The convergence metric is strictly positive, so if the estimated final convergence is 

1158 # less than zero, force it to zero. 

1159 estFinalConv = np.array(estFinalConv) 

1160 estFinalConv[estFinalConv < 0] = 0 

1161 # Because the estimate may slowly change over time, only use the most recent measurements. 

1162 estFinalConv = np.median(estFinalConv[max(nIter - 5, 0):]) 

1163 lastGain = gainList[-1] 

1164 lastConv = convergenceList[-2] 

1165 newConv = convergenceList[-1] 

1166 # The predicted convergence is the value we would get if the new model calculated 

1167 # in the previous iteration was perfect. Recall that the updated model that is 

1168 # actually used is the gain-weighted average of the new and old model, 

1169 # so the convergence would be similarly weighted. 

1170 predictedConv = (estFinalConv*lastGain + lastConv)/(1. + lastGain) 

1171 # If the measured and predicted convergence are very close, that indicates 

1172 # that our forward model is accurate and we can use a more aggressive gain 

1173 # If the measured convergence is significantly worse (or better!) than predicted, 

1174 # that indicates that the model is not converging as expected and 

1175 # we should use a more conservative gain. 

1176 delta = (predictedConv - newConv)/((lastConv - estFinalConv)/(1 + lastGain)) 

1177 newGain = 1 - abs(delta) 

1178 # Average the gains to prevent oscillating solutions. 

1179 newGain = (newGain + lastGain)/2. 

1180 gain = max(baseGain, newGain) 

1181 else: 

1182 gain = baseGain 

1183 gainList.append(gain) 

1184 return gain 

1185 

1186 def calculateModelWeights(self, dcrModels, dcrBBox): 

1187 """Build an array that smoothly tapers to 0 away from detected sources. 

1188 

1189 Parameters 

1190 ---------- 

1191 dcrModels : `lsst.pipe.tasks.DcrModel` 

1192 Best fit model of the true sky after correcting chromatic effects. 

1193 dcrBBox : `lsst.geom.box.Box2I` 

1194 Sub-region of the coadd which includes a buffer to allow for DCR. 

1195 

1196 Returns 

1197 ------- 

1198 weights : `numpy.ndarray` or `float` 

1199 A 2D array of weight values that tapers smoothly to zero away from detected sources. 

1200 Set to a placeholder value of 1.0 if ``self.config.useModelWeights`` is False. 

1201 

1202 Raises 

1203 ------ 

1204 ValueError 

1205 If ``useModelWeights`` is set and ``modelWeightsWidth`` is negative. 

1206 """ 

1207 if not self.config.useModelWeights: 

1208 return 1.0 

1209 if self.config.modelWeightsWidth < 0: 

1210 raise ValueError("modelWeightsWidth must not be negative if useModelWeights is set") 

1211 convergeMask = dcrModels.mask.getPlaneBitMask(self.config.convergenceMaskPlanes) 

1212 convergeMaskPixels = dcrModels.mask[dcrBBox].array & convergeMask > 0 

1213 weights = np.zeros_like(dcrModels[0][dcrBBox].array) 

1214 weights[convergeMaskPixels] = 1. 

1215 weights = ndimage.filters.gaussian_filter(weights, self.config.modelWeightsWidth) 

1216 weights /= np.max(weights) 

1217 return weights 

1218 

1219 def applyModelWeights(self, modelImages, refImage, modelWeights): 

1220 """Smoothly replace model pixel values with those from a 

1221 reference at locations away from detected sources. 

1222 

1223 Parameters 

1224 ---------- 

1225 modelImages : `list` of `lsst.afw.image.Image` 

1226 The new DCR model images from the current iteration. 

1227 The values will be modified in place. 

1228 refImage : `lsst.afw.image.MaskedImage` 

1229 A reference image used to supply the default pixel values. 

1230 modelWeights : `numpy.ndarray` or `float` 

1231 A 2D array of weight values that tapers smoothly to zero away from detected sources. 

1232 Set to a placeholder value of 1.0 if ``self.config.useModelWeights`` is False. 

1233 """ 

1234 if self.config.useModelWeights: 

1235 for model in modelImages: 

1236 model.array *= modelWeights 

1237 model.array += refImage.array*(1. - modelWeights)/self.config.dcrNumSubfilters 

1238 

1239 def loadSubExposures(self, bbox, statsCtrl, warpRefList, imageScalerList, spanSetMaskList): 

1240 """Pre-load sub-regions of a list of exposures. 

1241 

1242 Parameters 

1243 ---------- 

1244 bbox : `lsst.geom.box.Box2I` 

1245 Sub-region to coadd 

1246 statsCtrl : `lsst.afw.math.StatisticsControl` 

1247 Statistics control object for coadd 

1248 warpRefList : `list` of `lsst.daf.butler.DeferredDatasetHandle` or 

1249 `lsst.daf.persistence.ButlerDataRef` 

1250 The data references to the input warped exposures. 

1251 imageScalerList : `list` of `lsst.pipe.task.ImageScaler` 

1252 The image scalars correct for the zero point of the exposures. 

1253 spanSetMaskList : `list` of `dict` containing spanSet lists, or None 

1254 Each element is dict with keys = mask plane name to add the spans to 

1255 

1256 Returns 

1257 ------- 

1258 subExposures : `dict` 

1259 The `dict` keys are the visit IDs, 

1260 and the values are `lsst.afw.image.ExposureF` 

1261 The pre-loaded exposures for the current subregion. 

1262 The variance plane contains weights, and not the variance 

1263 """ 

1264 tempExpName = self.getTempExpDatasetName(self.warpType) 

1265 zipIterables = zip(warpRefList, imageScalerList, spanSetMaskList) 

1266 subExposures = {} 

1267 for warpExpRef, imageScaler, altMaskSpans in zipIterables: 

1268 if isinstance(warpExpRef, DeferredDatasetHandle): 

1269 exposure = warpExpRef.get(parameters={'bbox': bbox}) 

1270 else: 

1271 exposure = warpExpRef.get(tempExpName + "_sub", bbox=bbox) 

1272 visit = warpExpRef.dataId["visit"] 

1273 if altMaskSpans is not None: 

1274 self.applyAltMaskPlanes(exposure.mask, altMaskSpans) 

1275 imageScaler.scaleMaskedImage(exposure.maskedImage) 

1276 # Note that the variance plane here is used to store weights, not the actual variance 

1277 exposure.variance.array[:, :] = 0. 

1278 # Set the weight of unmasked pixels to 1. 

1279 exposure.variance.array[(exposure.mask.array & statsCtrl.getAndMask()) == 0] = 1. 

1280 # Set the image value of masked pixels to zero. 

1281 # This eliminates needing the mask plane when stacking images in ``newModelFromResidual`` 

1282 exposure.image.array[(exposure.mask.array & statsCtrl.getAndMask()) > 0] = 0. 

1283 subExposures[visit] = exposure 

1284 return subExposures 

1285 

1286 def selectCoaddPsf(self, templateCoadd, warpRefList): 

1287 """Compute the PSF of the coadd from the exposures with the best seeing. 

1288 

1289 Parameters 

1290 ---------- 

1291 templateCoadd : `lsst.afw.image.ExposureF` 

1292 The initial coadd exposure before accounting for DCR. 

1293 warpRefList : `list` of `lsst.daf.butler.DeferredDatasetHandle` or 

1294 `lsst.daf.persistence.ButlerDataRef` 

1295 The data references to the input warped exposures. 

1296 

1297 Returns 

1298 ------- 

1299 psf : `lsst.meas.algorithms.CoaddPsf` 

1300 The average PSF of the input exposures with the best seeing. 

1301 """ 

1302 sigma2fwhm = 2.*np.sqrt(2.*np.log(2.)) 

1303 tempExpName = self.getTempExpDatasetName(self.warpType) 

1304 # Note: ``ccds`` is a `lsst.afw.table.ExposureCatalog` with one entry per ccd and per visit 

1305 # If there are multiple ccds, it will have that many times more elements than ``warpExpRef`` 

1306 ccds = templateCoadd.getInfo().getCoaddInputs().ccds 

1307 psfRefSize = templateCoadd.getPsf().computeShape().getDeterminantRadius()*sigma2fwhm 

1308 psfSizes = np.zeros(len(ccds)) 

1309 ccdVisits = np.array(ccds["visit"]) 

1310 for warpExpRef in warpRefList: 

1311 if isinstance(warpExpRef, DeferredDatasetHandle): 

1312 # Gen 3 API 

1313 psf = warpExpRef.get(component="psf") 

1314 else: 

1315 # Gen 2 API. Delete this when Gen 2 retired 

1316 psf = warpExpRef.get(tempExpName).getPsf() 

1317 visit = warpExpRef.dataId["visit"] 

1318 psfSize = psf.computeShape().getDeterminantRadius()*sigma2fwhm 

1319 psfSizes[ccdVisits == visit] = psfSize 

1320 # Note that the input PSFs include DCR, which should be absent from the DcrCoadd 

1321 # The selected PSFs are those that have a FWHM less than or equal to the smaller 

1322 # of the mean or median FWHM of the input exposures. 

1323 sizeThreshold = min(np.median(psfSizes), psfRefSize) 

1324 goodPsfs = psfSizes <= sizeThreshold 

1325 psf = measAlg.CoaddPsf(ccds[goodPsfs], templateCoadd.getWcs(), 

1326 self.config.coaddPsf.makeControl()) 

1327 return psf