23 __all__ = [
'PairedVisitListTaskRunner',
'SingleVisitListTaskRunner',
24 'NonexistentDatasetTaskDataIdContainer',
'parseCmdlineNumberString',
25 'countMaskedPixels',
'checkExpLengthEqual',
'ddict2dict']
29 from scipy.optimize
import leastsq
30 import numpy.polynomial.polynomial
as poly
31 from scipy.stats
import norm
43 """Correct measured sigma to account for clipping.
45 If we clip our input data and then measure sigma, then the
46 measured sigma is smaller than the true value because real
47 points beyond the clip threshold have been removed. This is a
48 small (1.5% at nSigClip=3) effect when nSigClip >~ 3, but the
49 default parameters for measure crosstalk use nSigClip=2.0.
50 This causes the measured sigma to be about 15% smaller than
51 real. This formula corrects the issue, for the symmetric case
52 (upper clip threshold equal to lower clip threshold).
57 Number of sigma the measurement was clipped by.
62 Scale factor to increase the measured sigma by.
65 varFactor = 1.0 - (2 * nSigClip * norm.pdf(nSigClip)) / (norm.cdf(nSigClip) - norm.cdf(-nSigClip))
66 return 1.0 / np.sqrt(varFactor)
70 """Calculate weighted reduced chi2.
76 List with measured data.
79 List with modeled data.
81 weightsMeasured : `list`
82 List with weights for the measured data.
85 Number of data points.
88 Number of parameters in the model.
93 redWeightedChi2 : `float`
94 Reduced weighted chi2.
97 wRes = (measured - model)*weightsMeasured
98 return ((wRes*wRes).sum())/(nData-nParsModel)
101 def makeMockFlats(expTime, gain=1.0, readNoiseElectrons=5, fluxElectrons=1000,
102 randomSeedFlat1=1984, randomSeedFlat2=666, powerLawBfParams=[],
104 """Create a pair or mock flats with isrMock.
109 Exposure time of the flats.
111 gain : `float`, optional
114 readNoiseElectrons : `float`, optional
115 Read noise rms, in electrons.
117 fluxElectrons : `float`, optional
118 Flux of flats, in electrons per second.
120 randomSeedFlat1 : `int`, optional
121 Random seed for the normal distrubutions for the mean signal and noise (flat1).
123 randomSeedFlat2 : `int`, optional
124 Random seed for the normal distrubutions for the mean signal and noise (flat2).
126 powerLawBfParams : `list`, optional
127 Parameters for `galsim.cdmodel.PowerLawCD` to simulate the brightter-fatter effect.
129 expId1 : `int`, optional
130 Exposure ID for first flat.
132 expId2 : `int`, optional
133 Exposure ID for second flat.
138 flatExp1 : `lsst.afw.image.exposure.exposure.ExposureF`
139 First exposure of flat field pair.
141 flatExp2 : `lsst.afw.image.exposure.exposure.ExposureF`
142 Second exposure of flat field pair.
146 The parameters of `galsim.cdmodel.PowerLawCD` are `n, r0, t0, rx, tx, r, t, alpha`. For more
147 information about their meaning, see the Galsim documentation
148 https://galsim-developers.github.io/GalSim/_build/html/_modules/galsim/cdmodel.html
149 and Gruen+15 (1501.02802).
151 Example: galsim.cdmodel.PowerLawCD(8, 1.1e-7, 1.1e-7, 1.0e-8, 1.0e-8, 1.0e-9, 1.0e-9, 2.0)
153 flatFlux = fluxElectrons
154 flatMean = flatFlux*expTime
155 readNoise = readNoiseElectrons
157 mockImageConfig = isrMock.IsrMock.ConfigClass()
159 mockImageConfig.flatDrop = 0.99999
160 mockImageConfig.isTrimmed =
True
162 flatExp1 = isrMock.FlatMock(config=mockImageConfig).run()
163 flatExp2 = flatExp1.clone()
164 (shapeY, shapeX) = flatExp1.getDimensions()
165 flatWidth = np.sqrt(flatMean)
167 rng1 = np.random.RandomState(randomSeedFlat1)
168 flatData1 = rng1.normal(flatMean, flatWidth, (shapeX, shapeY)) + rng1.normal(0.0, readNoise,
170 rng2 = np.random.RandomState(randomSeedFlat2)
171 flatData2 = rng2.normal(flatMean, flatWidth, (shapeX, shapeY)) + rng2.normal(0.0, readNoise,
174 if len(powerLawBfParams):
175 if not len(powerLawBfParams) == 8:
176 raise RuntimeError(
"Wrong number of parameters for `galsim.cdmodel.PowerLawCD`. "
177 f
"Expected 8; passed {len(powerLawBfParams)}.")
178 cd = galsim.cdmodel.PowerLawCD(*powerLawBfParams)
179 tempFlatData1 = galsim.Image(flatData1)
180 temp2FlatData1 = cd.applyForward(tempFlatData1)
182 tempFlatData2 = galsim.Image(flatData2)
183 temp2FlatData2 = cd.applyForward(tempFlatData2)
185 flatExp1.image.array[:] = temp2FlatData1.array/gain
186 flatExp2.image.array[:] = temp2FlatData2.array/gain
188 flatExp1.image.array[:] = flatData1/gain
189 flatExp2.image.array[:] = flatData2/gain
194 flatExp1.getInfo().setVisitInfo(visitInfoExp1)
195 flatExp2.getInfo().setVisitInfo(visitInfoExp2)
197 return flatExp1, flatExp2
201 """Count the number of pixels in a given mask plane."""
202 maskBit = maskedIm.mask.getPlaneBitMask(maskPlane)
203 nPix = np.where(np.bitwise_and(maskedIm.mask.array, maskBit))[0].flatten().size
208 """Subclass of TaskRunner for handling intrinsically paired visits.
210 This transforms the processed arguments generated by the ArgumentParser
211 into the arguments expected by tasks which take visit pairs for their
214 Such tasks' run() methods tend to take two arguments,
215 one of which is the dataRef (as usual), and the other is the list
216 of visit-pairs, in the form of a list of tuples.
217 This list is supplied on the command line as documented,
218 and this class parses that, and passes the parsed version
221 See pipeBase.TaskRunner for more information.
226 """Parse the visit list and pass through explicitly."""
228 for visitStringPair
in parsedCmd.visitPairs:
229 visitStrings = visitStringPair.split(
",")
230 if len(visitStrings) != 2:
231 raise RuntimeError(
"Found {} visits in {} instead of 2".format(len(visitStrings),
234 visits = [int(visit)
for visit
in visitStrings]
236 raise RuntimeError(
"Could not parse {} as two integer visit numbers".format(visitStringPair))
237 visitPairs.append(visits)
239 return pipeBase.TaskRunner.getTargetList(parsedCmd, visitPairs=visitPairs, **kwargs)
243 """Parse command line numerical expression sytax and return as list of int
245 Take an input of the form "'1..5:2^123..126'" as a string, and return
246 a list of ints as [1, 3, 5, 123, 124, 125, 126]
249 for subString
in inputString.split(
"^"):
250 mat = re.search(
r"^(\d+)\.\.(\d+)(?::(\d+))?$", subString)
252 v1 = int(mat.group(1))
253 v2 = int(mat.group(2))
255 v3 = int(v3)
if v3
else 1
256 for v
in range(v1, v2 + 1, v3):
257 outList.append(int(v))
259 outList.append(int(subString))
264 """Subclass of TaskRunner for tasks requiring a list of visits per dataRef.
266 This transforms the processed arguments generated by the ArgumentParser
267 into the arguments expected by tasks which require a list of visits
268 to be supplied for each dataRef, as is common in `lsst.cp.pipe` code.
270 Such tasks' run() methods tend to take two arguments,
271 one of which is the dataRef (as usual), and the other is the list
273 This list is supplied on the command line as documented,
274 and this class parses that, and passes the parsed version
277 See `lsst.pipe.base.TaskRunner` for more information.
282 """Parse the visit list and pass through explicitly."""
285 assert len(parsedCmd.visitList) == 1,
'visitList parsing assumptions violated'
288 return pipeBase.TaskRunner.getTargetList(parsedCmd, visitList=visits, **kwargs)
292 """A DataIdContainer for the tasks for which the output does
296 """Compute refList based on idList.
298 This method must be defined as the dataset does not exist before this
304 Results of parsing the command-line.
308 Not called if ``add_id_argument`` called
309 with ``doMakeDataRefList=False``.
310 Note that this is almost a copy-and-paste of the vanilla
311 implementation, but without checking if the datasets already exist,
312 as this task exists to make them.
314 if self.datasetType
is None:
315 raise RuntimeError(
"Must call setDatasetType first")
316 butler = namespace.butler
317 for dataId
in self.idList:
318 refList = list(butler.subset(datasetType=self.datasetType, level=self.level, dataId=dataId))
322 namespace.log.warn(
"No data found for dataId=%s", dataId)
324 self.refList += refList
327 def irlsFit(initialParams, dataX, dataY, function, weightsY=None):
328 """Iteratively reweighted least squares fit.
330 This uses the `lsst.cp.pipe.utils.fitLeastSq`, but applies
331 weights based on the Cauchy distribution to the fitter. See
332 e.g. Holland and Welsch, 1977, doi:10.1080/03610927708827533
336 initialParams : `list` [`float`]
338 dataX : `numpy.array` [`float`]
340 dataY : `numpy.array` [`float`]
344 weightsY : `numpy.array` [`float`]
345 Weights to apply to the data.
349 polyFit : `list` [`float`]
350 Final best fit parameters.
351 polyFitErr : `list` [`float`]
352 Final errors on fit parameters.
355 weightsY : `list` [`float`]
356 Final weights used for each point.
360 weightsY = np.ones_like(dataX)
362 polyFit, polyFitErr, chiSq =
fitLeastSq(initialParams, dataX, dataY, function, weightsY=weightsY)
363 for iteration
in range(10):
365 resid = np.abs(dataY - function(polyFit, dataX)) / np.sqrt(dataY)
366 weightsY = 1.0 / (1.0 + np.sqrt(resid / 2.385))
367 polyFit, polyFitErr, chiSq =
fitLeastSq(initialParams, dataX, dataY, function, weightsY=weightsY)
369 return polyFit, polyFitErr, chiSq, weightsY
372 def fitLeastSq(initialParams, dataX, dataY, function, weightsY=None):
373 """Do a fit and estimate the parameter errors using using scipy.optimize.leastq.
375 optimize.leastsq returns the fractional covariance matrix. To estimate the
376 standard deviation of the fit parameters, multiply the entries of this matrix
377 by the unweighted reduced chi squared and take the square root of the diagonal elements.
381 initialParams : `list` of `float`
382 initial values for fit parameters. For ptcFitType=POLYNOMIAL, its length
383 determines the degree of the polynomial.
385 dataX : `numpy.array` of `float`
386 Data in the abscissa axis.
388 dataY : `numpy.array` of `float`
389 Data in the ordinate axis.
391 function : callable object (function)
392 Function to fit the data with.
394 weightsY : `numpy.array` of `float`
395 Weights of the data in the ordinate axis.
399 pFitSingleLeastSquares : `list` of `float`
400 List with fitted parameters.
402 pErrSingleLeastSquares : `list` of `float`
403 List with errors for fitted parameters.
405 reducedChiSqSingleLeastSquares : `float`
406 Reduced chi squared, unweighted if weightsY is not provided.
409 weightsY = np.ones(len(dataX))
411 def errFunc(p, x, y, weightsY=None):
413 weightsY = np.ones(len(x))
414 return (function(p, x) - y)*weightsY
416 pFit, pCov, infoDict, errMessage, success = leastsq(errFunc, initialParams,
417 args=(dataX, dataY, weightsY), full_output=1,
420 if (len(dataY) > len(initialParams))
and pCov
is not None:
425 pCov = np.zeros((len(initialParams), len(initialParams)))
427 reducedChiSq = np.nan
430 for i
in range(len(pFit)):
431 errorVec.append(np.fabs(pCov[i][i])**0.5)
433 pFitSingleLeastSquares = pFit
434 pErrSingleLeastSquares = np.array(errorVec)
436 return pFitSingleLeastSquares, pErrSingleLeastSquares, reducedChiSq
439 def fitBootstrap(initialParams, dataX, dataY, function, weightsY=None, confidenceSigma=1.):
440 """Do a fit using least squares and bootstrap to estimate parameter errors.
442 The bootstrap error bars are calculated by fitting 100 random data sets.
446 initialParams : `list` of `float`
447 initial values for fit parameters. For ptcFitType=POLYNOMIAL, its length
448 determines the degree of the polynomial.
450 dataX : `numpy.array` of `float`
451 Data in the abscissa axis.
453 dataY : `numpy.array` of `float`
454 Data in the ordinate axis.
456 function : callable object (function)
457 Function to fit the data with.
459 weightsY : `numpy.array` of `float`, optional.
460 Weights of the data in the ordinate axis.
462 confidenceSigma : `float`, optional.
463 Number of sigmas that determine confidence interval for the bootstrap errors.
467 pFitBootstrap : `list` of `float`
468 List with fitted parameters.
470 pErrBootstrap : `list` of `float`
471 List with errors for fitted parameters.
473 reducedChiSqBootstrap : `float`
474 Reduced chi squared, unweighted if weightsY is not provided.
477 weightsY = np.ones(len(dataX))
479 def errFunc(p, x, y, weightsY):
481 weightsY = np.ones(len(x))
482 return (function(p, x) - y)*weightsY
485 pFit, _ = leastsq(errFunc, initialParams, args=(dataX, dataY, weightsY), full_output=0)
488 residuals = errFunc(pFit, dataX, dataY, weightsY)
492 randomDelta = np.random.normal(0., np.fabs(residuals), len(dataY))
493 randomDataY = dataY + randomDelta
494 randomFit, _ = leastsq(errFunc, initialParams,
495 args=(dataX, randomDataY, weightsY), full_output=0)
496 pars.append(randomFit)
497 pars = np.array(pars)
498 meanPfit = np.mean(pars, 0)
501 errPfit = confidenceSigma*np.std(pars, 0)
502 pFitBootstrap = meanPfit
503 pErrBootstrap = errPfit
507 return pFitBootstrap, pErrBootstrap, reducedChiSq
511 """Polynomial function definition
515 Polynomial coefficients. Its length determines the polynomial order.
522 Ordinate array after evaluating polynomial of order len(pars)-1 at `x`.
524 return poly.polyval(x, [*pars])
528 """Single brighter-fatter parameter model for PTC; Equation 16 of Astier+19.
533 Parameters of the model: a00 (brightter-fatter), gain (e/ADU), and noise (e^2).
540 C_00 (variance) in ADU^2.
542 a00, gain, noise = pars
543 return 0.5/(a00*gain*gain)*(np.exp(2*a00*x*gain)-1) + noise/(gain*gain)
547 """Arrange exposures by exposure time.
551 exposureList : `list`[`lsst.afw.image.exposure.exposure.ExposureF`]
552 Input list of exposures.
556 flatsAtExpTime : `dict` [`float`,
557 `list`[`lsst.afw.image.exposure.exposure.ExposureF`]]
558 Dictionary that groups flat-field exposures that have the same
559 exposure time (seconds).
562 for exp
in exposureList:
564 expTime = tempFlat.getInfo().getVisitInfo().getExposureTime()
565 listAtExpTime = flatsAtExpTime.setdefault(expTime, [])
566 listAtExpTime.append(tempFlat)
568 return flatsAtExpTime
572 """Arrange exposures by exposure ID.
574 There is no guarantee that this will properly group exposures, but
575 allows a sequence of flats that have different illumination
576 (despite having the same exposure time) to be processed.
580 exposureList : `list`[`lsst.afw.image.exposure.exposure.ExposureF`]
581 Input list of exposures.
585 flatsAtExpId : `dict` [`float`,
586 `list`[`lsst.afw.image.exposure.exposure.ExposureF`]]
587 Dictionary that groups flat-field exposures sequentially by
593 This algorithm sorts the input exposures by their exposure id, and
594 then assigns each pair of exposures (exp_j, exp_{j+1}) to pair k,
595 such that 2*k = j, where j is the python index of one of the
596 exposures (starting from zero). By checking for the IndexError
597 while appending, we can ensure that there will only ever be fully
601 sortedExposures = sorted(exposureList, key=
lambda exp: exp.getInfo().getVisitInfo().getExposureId())
603 for jPair, exp
in enumerate(sortedExposures):
606 listAtExpId = flatsAtExpId.setdefault(kPair, [])
608 listAtExpId.append(exp)
609 listAtExpId.append(sortedExposures[jPair + 1])
617 """Check the exposure lengths of two exposures are equal.
621 exp1 : `lsst.afw.image.exposure.ExposureF`
622 First exposure to check
623 exp2 : `lsst.afw.image.exposure.ExposureF`
624 Second exposure to check
625 v1 : `int` or `str`, optional
626 First visit of the visit pair
627 v2 : `int` or `str`, optional
628 Second visit of the visit pair
629 raiseWithMessage : `bool`
630 If True, instead of returning a bool, raise a RuntimeError if exposure
631 times are not equal, with a message about which visits mismatch if the
632 information is available.
637 Raised if the exposure lengths of the two exposures are not equal
639 expTime1 = exp1.getInfo().getVisitInfo().getExposureTime()
640 expTime2 = exp2.getInfo().getVisitInfo().getExposureTime()
641 if expTime1 != expTime2:
643 msg =
"Exposure lengths for visit pairs must be equal. " + \
644 "Found %s and %s" % (expTime1, expTime2)
646 msg +=
" for visit pair %s, %s" % (v1, v2)
647 raise RuntimeError(msg)
653 def validateIsrConfig(isrTask, mandatory=None, forbidden=None, desirable=None, undesirable=None,
654 checkTrim=True, logName=None):
655 """Check that appropriate ISR settings have been selected for the task.
657 Note that this checks that the task itself is configured correctly rather
658 than checking a config.
662 isrTask : `lsst.ip.isr.IsrTask`
663 The task whose config is to be validated
665 mandatory : `iterable` of `str`
666 isr steps that must be set to True. Raises if False or missing
668 forbidden : `iterable` of `str`
669 isr steps that must be set to False. Raises if True, warns if missing
671 desirable : `iterable` of `str`
672 isr steps that should probably be set to True. Warns is False, info if
675 undesirable : `iterable` of `str`
676 isr steps that should probably be set to False. Warns is True, info if
680 Check to ensure the isrTask's assembly subtask is trimming the images.
681 This is a separate config as it is very ugly to do this within the
682 normal configuration lists as it is an option of a sub task.
687 Raised if ``mandatory`` config parameters are False,
688 or if ``forbidden`` parameters are True.
691 Raised if parameter ``isrTask`` is an invalid type.
695 Logs warnings using an isrValidation logger for desirable/undesirable
696 options that are of the wrong polarity or if keys are missing.
698 if not isinstance(isrTask, ipIsr.IsrTask):
699 raise TypeError(f
'Must supply an instance of lsst.ip.isr.IsrTask not {type(isrTask)}')
701 configDict = isrTask.config.toDict()
703 if logName
and isinstance(logName, str):
704 log = lsst.log.getLogger(logName)
706 log = lsst.log.getLogger(
"isrValidation")
709 for configParam
in mandatory:
710 if configParam
not in configDict:
711 raise RuntimeError(f
"Mandatory parameter {configParam} not found in the isr configuration.")
712 if configDict[configParam]
is False:
713 raise RuntimeError(f
"Must set config.isr.{configParam} to True for this task.")
716 for configParam
in forbidden:
717 if configParam
not in configDict:
718 log.warn(f
"Failed to find forbidden key {configParam} in the isr config. The keys in the"
719 " forbidden list should each have an associated Field in IsrConfig:"
720 " check that there is not a typo in this case.")
722 if configDict[configParam]
is True:
723 raise RuntimeError(f
"Must set config.isr.{configParam} to False for this task.")
726 for configParam
in desirable:
727 if configParam
not in configDict:
728 log.info(f
"Failed to find key {configParam} in the isr config. You probably want"
729 " to set the equivalent for your obs_package to True.")
731 if configDict[configParam]
is False:
732 log.warn(f
"Found config.isr.{configParam} set to False for this task."
733 " The cp_pipe Config recommends setting this to True.")
735 for configParam
in undesirable:
736 if configParam
not in configDict:
737 log.info(f
"Failed to find key {configParam} in the isr config. You probably want"
738 " to set the equivalent for your obs_package to False.")
740 if configDict[configParam]
is True:
741 log.warn(f
"Found config.isr.{configParam} set to True for this task."
742 " The cp_pipe Config recommends setting this to False.")
745 if not isrTask.assembleCcd.config.doTrim:
746 raise RuntimeError(
"Must trim when assembling CCDs. Set config.isr.assembleCcd.doTrim to True")
750 """Convert nested default dictionaries to regular dictionaries.
752 This is needed to prevent yaml persistence issues.
757 A possibly nested set of `defaultdict`.
762 A possibly nested set of `dict`.
764 for k, v
in d.items():
765 if isinstance(v, dict):
def makeDataRefList(self, namespace)
def getTargetList(parsedCmd, **kwargs)
def getTargetList(parsedCmd, **kwargs)
def countMaskedPixels(maskedIm, maskPlane)
def validateIsrConfig(isrTask, mandatory=None, forbidden=None, desirable=None, undesirable=None, checkTrim=True, logName=None)
def sigmaClipCorrection(nSigClip)
def fitBootstrap(initialParams, dataX, dataY, function, weightsY=None, confidenceSigma=1.)
def fitLeastSq(initialParams, dataX, dataY, function, weightsY=None)
def arrangeFlatsByExpId(exposureList)
def parseCmdlineNumberString(inputString)
def arrangeFlatsByExpTime(exposureList)
def checkExpLengthEqual(exp1, exp2, v1=None, v2=None, raiseWithMessage=False)
def makeMockFlats(expTime, gain=1.0, readNoiseElectrons=5, fluxElectrons=1000, randomSeedFlat1=1984, randomSeedFlat2=666, powerLawBfParams=[], expId1=0, expId2=1)
def calculateWeightedReducedChi2(measured, model, weightsMeasured, nData, nParsModel)
def irlsFit(initialParams, dataX, dataY, function, weightsY=None)
def funcPolynomial(pars, x)