23Define dataset class for MeasurePhotonTransferCurve task
26from astropy.table
import Table
30__all__ = [
'PhotonTransferCurveDataset']
34 """A simple class to hold the output data from the PTC task.
35 The dataset is made up of a dictionary
for each item, keyed by the
36 amplifiers
' names, which much be supplied at construction time.
37 New items cannot be added to the class to save accidentally saving to the
38 wrong property,
and the
class can be frozen if desired.
39 inputExpIdPairs records the exposures used to produce the data.
40 When fitPtc()
or fitCovariancesAstier()
is run, a mask
is built up, which
41 is by definition always the same length
as inputExpIdPairs, rawExpTimes,
42 rawMeans
and rawVars,
and is a list of bools, which are incrementally set
43 to
False as points are discarded
from the fits.
44 PTC fit parameters
for polynomials are stored
in a list
in ascending order
45 of polynomial term, i.e. par[0]*x^0 + par[1]*x + par[2]*x^2 etc
46 with the length of the list corresponding to the order of the polynomial
52 List
with the names of the amplifiers of the detector at hand.
55 Type of model fitted to the PTC:
"POLYNOMIAL",
"EXPAPPROXIMATION",
59 Maximum lag of covariances (size of square covariance matrices).
61 kwargs : `dict`, optional
62 Other keyword arguments to
pass to the parent init.
66 The stored attributes are:
68 List
with bad amplifiers names.
69 inputExpIdPairs : `dict`, [`str`, `list`]
70 Dictionary keyed by amp names containing the input exposures IDs.
71 expIdMask : `dict`, [`str`, `list`]
72 Dictionary keyed by amp names containing the mask produced after
73 outlier rejection. The mask produced by the
"FULLCOVARIANCE"
74 option may differ
from the one produced
in the other two PTC
76 rawExpTimes : `dict`, [`str`, `list`]
77 Dictionary keyed by amp names containing the unmasked exposure times.
78 rawMeans : `dict`, [`str`, `list`]
79 Dictionary keyed by amp namescontaining the unmasked average of the
80 means of the exposures
in each flat pair.
81 rawVars : `dict`, [`str`, `list`]
82 Dictionary keyed by amp names containing the variance of the
83 difference image of the exposures
in each flat pair.
84 gain : `dict`, [`str`, `list`]
85 Dictionary keyed by amp names containing the fitted gains.
86 gainErr : `dict`, [`str`, `list`]
87 Dictionary keyed by amp names containing the errors on the
89 noise : `dict`, [`str`, `list`]
90 Dictionary keyed by amp names containing the fitted noise.
91 noiseErr : `dict`, [`str`, `list`]
92 Dictionary keyed by amp names containing the errors on the fitted
94 ptcFitPars : `dict`, [`str`, `list`]
95 Dictionary keyed by amp names containing the fitted parameters of the
96 PTC model
for ptcFitTye
in [
"POLYNOMIAL",
"EXPAPPROXIMATION"].
97 ptcFitParsError : `dict`, [`str`, `list`]
98 Dictionary keyed by amp names containing the errors on the fitted
99 parameters of the PTC model
for ptcFitTye
in
100 [
"POLYNOMIAL",
"EXPAPPROXIMATION"].
101 ptcFitChiSq : `dict`, [`str`, `list`]
102 Dictionary keyed by amp names containing the reduced chi squared
103 of the fit
for ptcFitTye
in [
"POLYNOMIAL",
"EXPAPPROXIMATION"].
105 Flux value (
in ADU) where the variance of the PTC curve starts
106 decreasing consistently.
107 covariances : `dict`, [`str`, `list`]
108 Dictionary keyed by amp names containing a list of measured
109 covariances per mean flux.
110 covariancesModel : `dict`, [`str`, `list`]
111 Dictionary keyed by amp names containinging covariances model
112 (Eq. 20 of Astier+19) per mean flux.
113 covariancesSqrtWeights : `dict`, [`str`, `list`]
114 Dictionary keyed by amp names containinging sqrt. of covariances
116 aMatrix : `dict`, [`str`, `list`]
117 Dictionary keyed by amp names containing the
"a" parameters
from
118 the model
in Eq. 20 of Astier+19.
119 bMatrix : `dict`, [`str`, `list`]
120 Dictionary keyed by amp names containing the
"b" parameters
from
121 the model
in Eq. 20 of Astier+19.
122 covariancesModelNoB : `dict`, [`str`, `list`]
123 Dictionary keyed by amp names containing covariances model
124 (
with 'b'=0
in Eq. 20 of Astier+19)
126 aMatrixNoB : `dict`, [`str`, `list`]
127 Dictionary keyed by amp names containing the
"a" parameters
from the
128 model
in Eq. 20 of Astier+19
130 finalVars : `dict`, [`str`, `list`]
131 Dictionary keyed by amp names containing the masked variance of the
132 difference image of each flat
133 pair. If needed, each array will be right-padded
with
134 np.nan to match the length of rawExpTimes.
135 finalModelVars : `dict`, [`str`, `list`]
136 Dictionary keyed by amp names containing the masked modeled
137 variance of the difference image of each flat pair. If needed, each
138 array will be right-padded
with np.nan to match the length of
140 finalMeans : `dict`, [`str`, `list`]
141 Dictionary keyed by amp names containing the masked average of the
142 means of the exposures
in each flat pair. If needed, each array
143 will be right-padded
with np.nan to match the length of
145 photoCharge : `dict`, [`str`, `list`]
146 Dictionary keyed by amp names containing the integrated photocharge
147 for linearity calibration.
149 Version 1.1 adds the `ptcTurnoff` attribute.
153 _SCHEMA =
'Gen3 Photon Transfer Curve'
156 def __init__(self, ampNames=[], ptcFitType=None, covMatrixSide=1, **kwargs):
167 self.
rawMeans = {ampName: []
for ampName
in ampNames}
168 self.
rawVars = {ampName: []
for ampName
in ampNames}
171 self.
gain = {ampName: np.nan
for ampName
in ampNames}
172 self.
gainErr = {ampName: np.nan
for ampName
in ampNames}
173 self.
noise = {ampName: np.nan
for ampName
in ampNames}
174 self.
noiseErr = {ampName: np.nan
for ampName
in ampNames}
184 self.
aMatrix = {ampName: np.nan
for ampName
in ampNames}
185 self.
bMatrix = {ampName: np.nan
for ampName
in ampNames}
195 'rawMeans',
'rawVars',
'gain',
'gainErr',
'noise',
'noiseErr',
196 'ptcFitPars',
'ptcFitParsError',
'ptcFitChiSq',
'ptcTurnoff',
197 'aMatrixNoB',
'covariances',
'covariancesModel',
198 'covariancesSqrtWeights',
'covariancesModelNoB',
199 'aMatrix',
'bMatrix',
'finalVars',
'finalModelVars',
'finalMeans',
202 def setAmpValues(self, ampName, inputExpIdPair=[(np.nan, np.nan)], expIdMask=[np.nan],
203 rawExpTime=[np.nan], rawMean=[np.nan], rawVar=[np.nan], photoCharge=[np.nan],
204 gain=np.nan, gainErr=np.nan, noise=np.nan, noiseErr=np.nan, ptcFitPars=[np.nan],
205 ptcFitParsError=[np.nan], ptcFitChiSq=np.nan, ptcTurnoff=np.nan, covArray=[],
206 covArrayModel=[], covSqrtWeights=[], aMatrix=[], bMatrix=[], covArrayModelNoB=[],
207 aMatrixNoB=[], finalVar=[np.nan], finalModelVar=[np.nan], finalMean=[np.nan]):
208 """Function to initialize an amp of a PhotonTransferCurveDataset.
212 The parameters are all documented in `init`.
215 if len(covArray) == 0:
216 covArray = [nanMatrix]
217 if len(covArrayModel) == 0:
218 covArrayModel = [nanMatrix]
219 if len(covSqrtWeights) == 0:
220 covSqrtWeights = [nanMatrix]
221 if len(covArrayModelNoB) == 0:
222 covArrayModelNoB = [nanMatrix]
223 if len(aMatrix) == 0:
225 if len(bMatrix) == 0:
227 if len(aMatrixNoB) == 0:
228 aMatrixNoB = nanMatrix
236 self.
gain[ampName] = gain
237 self.
gainErr[ampName] = gainErr
238 self.
noise[ampName] = noise
248 self.
aMatrix[ampName] = aMatrix
249 self.
bMatrix[ampName] = bMatrix
259 """Update calibration metadata.
260 This calls the base class's method after ensuring the required
261 calibration keywords will be saved.
264 setDate : `bool`, optional
265 Update the CALIBDATE fields in the metadata to the current
266 time. Defaults to
False.
268 Other keyword parameters to set
in the metadata.
276 """Construct a calibration from a dictionary of properties.
277 Must be implemented by the specific calibration subclasses.
281 Dictionary of properties.
284 calib : `lsst.ip.isr.CalibType`
285 Constructed calibration.
289 Raised if the supplied dictionary
is for a different
293 if calib._OBSTYPE != dictionary[
'metadata'][
'OBSTYPE']:
294 raise RuntimeError(f
"Incorrect Photon Transfer Curve dataset supplied. "
295 f
"Expected {calib._OBSTYPE}, found {dictionary['metadata']['OBSTYPE']}")
296 calib.setMetadata(dictionary[
'metadata'])
297 calib.ptcFitType = dictionary[
'ptcFitType']
298 calib.covMatrixSide = dictionary[
'covMatrixSide']
299 calib.badAmps = np.array(dictionary[
'badAmps'],
'str').tolist()
301 covMatrixSide = calib.covMatrixSide
303 covDimensionsProduct = len(np.array(list(dictionary[
'covariances'].values())[0]).ravel())
304 nSignalPoints = int(covDimensionsProduct/(covMatrixSide*covMatrixSide))
306 for ampName
in dictionary[
'ampNames']:
307 covsAmp = np.array(dictionary[
'covariances'][ampName]).reshape((nSignalPoints, covMatrixSide,
316 maskCovsAmp = np.array([~np.isnan(entry).all()
for entry
in covsAmp])
317 maskAmp = ~np.isnan(np.array(dictionary[
'finalMeans'][ampName]))
319 maskCovsAmp = np.array([
True])
320 maskAmp = np.array([
True])
322 calib.ampNames.append(ampName)
323 calib.inputExpIdPairs[ampName] = np.array(dictionary[
'inputExpIdPairs'][ampName]).tolist()
324 calib.expIdMask[ampName] = np.array(dictionary[
'expIdMask'][ampName]).tolist()
325 calib.rawExpTimes[ampName] = np.array(dictionary[
'rawExpTimes'][ampName]).tolist()
326 calib.rawMeans[ampName] = np.array(dictionary[
'rawMeans'][ampName]).tolist()
327 calib.rawVars[ampName] = np.array(dictionary[
'rawVars'][ampName]).tolist()
328 calib.gain[ampName] = np.array(dictionary[
'gain'][ampName]).tolist()
329 calib.gainErr[ampName] = np.array(dictionary[
'gainErr'][ampName]).tolist()
330 calib.noise[ampName] = np.array(dictionary[
'noise'][ampName]).tolist()
331 calib.noiseErr[ampName] = np.array(dictionary[
'noiseErr'][ampName]).tolist()
332 calib.ptcFitPars[ampName] = np.array(dictionary[
'ptcFitPars'][ampName]).tolist()
333 calib.ptcFitParsError[ampName] = np.array(dictionary[
'ptcFitParsError'][ampName]).tolist()
334 calib.ptcFitChiSq[ampName] = np.array(dictionary[
'ptcFitChiSq'][ampName]).tolist()
335 calib.ptcTurnoff[ampName] = np.array(dictionary[
'ptcTurnoff'][ampName]).tolist()
336 calib.covariances[ampName] = covsAmp[maskCovsAmp].tolist()
337 calib.covariancesModel[ampName] = np.array(
338 dictionary[
'covariancesModel'][ampName]).reshape(
339 (nSignalPoints, covMatrixSide, covMatrixSide))[maskCovsAmp].tolist()
340 calib.covariancesSqrtWeights[ampName] = np.array(
341 dictionary[
'covariancesSqrtWeights'][ampName]).reshape(
342 (nSignalPoints, covMatrixSide, covMatrixSide))[maskCovsAmp].tolist()
343 calib.aMatrix[ampName] = np.array(dictionary[
'aMatrix'][ampName]).reshape(
344 (covMatrixSide, covMatrixSide)).tolist()
345 calib.bMatrix[ampName] = np.array(dictionary[
'bMatrix'][ampName]).reshape(
346 (covMatrixSide, covMatrixSide)).tolist()
347 calib.covariancesModelNoB[ampName] = np.array(
348 dictionary[
'covariancesModelNoB'][ampName]).reshape(
349 (nSignalPoints, covMatrixSide, covMatrixSide))[maskCovsAmp].tolist()
350 calib.aMatrixNoB[ampName] = np.array(
351 dictionary[
'aMatrixNoB'][ampName]).reshape((covMatrixSide, covMatrixSide)).tolist()
352 calib.finalVars[ampName] = np.array(dictionary[
'finalVars'][ampName])[maskAmp].tolist()
353 calib.finalModelVars[ampName] = np.array(dictionary[
'finalModelVars'][ampName])[maskAmp].tolist()
354 calib.finalMeans[ampName] = np.array(dictionary[
'finalMeans'][ampName])[maskAmp].tolist()
355 calib.photoCharge[ampName] = np.array(dictionary[
'photoCharge'][ampName]).tolist()
356 calib.updateMetadata()
360 """Return a dictionary containing the calibration properties.
361 The dictionary should be able to be round-tripped through
366 Dictionary of properties.
372 outDict['metadata'] = metadata
377 outDict[
'badAmps'] = self.
badAmps
382 outDict[
'rawVars'] = self.
rawVars
383 outDict[
'gain'] = self.
gain
384 outDict[
'gainErr'] = self.
gainErr
385 outDict[
'noise'] = self.
noise
394 outDict[
'aMatrix'] = self.
aMatrix
395 outDict[
'bMatrix'] = self.
bMatrix
407 """Construct calibration from a list of tables.
408 This method uses the `fromDict` method to create the
409 calibration, after constructing an appropriate dictionary from
413 tableList : `list` [`lsst.afw.table.Table`]
414 List of tables to use to construct the datasetPtc.
417 calib : `lsst.cp.pipe.`
418 The calibration defined
in the tables.
420 ptcTable = tableList[0]
422 metadata = ptcTable.meta
424 inDict['metadata'] = metadata
425 inDict[
'ampNames'] = []
426 inDict[
'ptcFitType'] = []
427 inDict[
'covMatrixSide'] = []
428 inDict[
'inputExpIdPairs'] = dict()
429 inDict[
'expIdMask'] = dict()
430 inDict[
'rawExpTimes'] = dict()
431 inDict[
'rawMeans'] = dict()
432 inDict[
'rawVars'] = dict()
433 inDict[
'gain'] = dict()
434 inDict[
'gainErr'] = dict()
435 inDict[
'noise'] = dict()
436 inDict[
'noiseErr'] = dict()
437 inDict[
'ptcFitPars'] = dict()
438 inDict[
'ptcFitParsError'] = dict()
439 inDict[
'ptcFitChiSq'] = dict()
440 inDict[
'ptcTurnoff'] = dict()
441 inDict[
'covariances'] = dict()
442 inDict[
'covariancesModel'] = dict()
443 inDict[
'covariancesSqrtWeights'] = dict()
444 inDict[
'aMatrix'] = dict()
445 inDict[
'bMatrix'] = dict()
446 inDict[
'covariancesModelNoB'] = dict()
447 inDict[
'aMatrixNoB'] = dict()
448 inDict[
'finalVars'] = dict()
449 inDict[
'finalModelVars'] = dict()
450 inDict[
'finalMeans'] = dict()
451 inDict[
'badAmps'] = []
452 inDict[
'photoCharge'] = dict()
454 calibVersion = metadata[
'PTC_VERSION']
455 if calibVersion == 1.0:
456 cls().log.warning(f
"Previous version found for PTC dataset: {calibVersion}. "
457 f
"Setting 'ptcTurnoff' in all amps to last value in 'finalMeans'.")
458 for record
in ptcTable:
459 ampName = record[
'AMPLIFIER_NAME']
461 inDict[
'ptcFitType'] = record[
'PTC_FIT_TYPE']
462 inDict[
'covMatrixSide'] = record[
'COV_MATRIX_SIDE']
463 inDict[
'ampNames'].append(ampName)
464 inDict[
'inputExpIdPairs'][ampName] = record[
'INPUT_EXP_ID_PAIRS']
465 inDict[
'expIdMask'][ampName] = record[
'EXP_ID_MASK']
466 inDict[
'rawExpTimes'][ampName] = record[
'RAW_EXP_TIMES']
467 inDict[
'rawMeans'][ampName] = record[
'RAW_MEANS']
468 inDict[
'rawVars'][ampName] = record[
'RAW_VARS']
469 inDict[
'gain'][ampName] = record[
'GAIN']
470 inDict[
'gainErr'][ampName] = record[
'GAIN_ERR']
471 inDict[
'noise'][ampName] = record[
'NOISE']
472 inDict[
'noiseErr'][ampName] = record[
'NOISE_ERR']
473 inDict[
'ptcFitPars'][ampName] = record[
'PTC_FIT_PARS']
474 inDict[
'ptcFitParsError'][ampName] = record[
'PTC_FIT_PARS_ERROR']
475 inDict[
'ptcFitChiSq'][ampName] = record[
'PTC_FIT_CHI_SQ']
476 inDict[
'covariances'][ampName] = record[
'COVARIANCES']
477 inDict[
'covariancesModel'][ampName] = record[
'COVARIANCES_MODEL']
478 inDict[
'covariancesSqrtWeights'][ampName] = record[
'COVARIANCES_SQRT_WEIGHTS']
479 inDict[
'aMatrix'][ampName] = record[
'A_MATRIX']
480 inDict[
'bMatrix'][ampName] = record[
'B_MATRIX']
481 inDict[
'covariancesModelNoB'][ampName] = record[
'COVARIANCES_MODEL_NO_B']
482 inDict[
'aMatrixNoB'][ampName] = record[
'A_MATRIX_NO_B']
483 inDict[
'finalVars'][ampName] = record[
'FINAL_VARS']
484 inDict[
'finalModelVars'][ampName] = record[
'FINAL_MODEL_VARS']
485 inDict[
'finalMeans'][ampName] = record[
'FINAL_MEANS']
486 inDict[
'badAmps'] = record[
'BAD_AMPS']
487 inDict[
'photoCharge'][ampName] = record[
'PHOTO_CHARGE']
488 if calibVersion == 1.0:
489 mask = record[
'FINAL_MEANS'].mask
490 inDict[
'ptcTurnoff'][ampName] = record[
'FINAL_MEANS'][~mask][-1]
492 inDict[
'ptcTurnoff'][ampName] = record[
'PTC_TURNOFF']
496 """Construct a list of tables containing the information in this
499 The list of tables should create an identical calibration
500 after being passed to this class's fromTable method.
503 tableList : `list` [`astropy.table.Table`]
504 List of tables containing the linearity calibration
510 for i, ampName
in enumerate(self.
ampNames):
511 nPoints.append(len(list(self.
covariances.values())[i]))
512 nSignalPoints = max(nPoints)
514 for i, ampName
in enumerate(self.
ampNames):
515 nPadPoints[ampName] = nSignalPoints - len(list(self.
covariances.values())[i])
518 catalog = Table([{
'AMPLIFIER_NAME': ampName,
522 if len(self.
expIdMask[ampName])
else np.nan,
524 if len(self.
expIdMask[ampName])
else np.nan,
525 'RAW_EXP_TIMES': np.array(self.
rawExpTimes[ampName]).tolist()
527 'RAW_MEANS': np.array(self.
rawMeans[ampName]).tolist()
528 if len(self.
rawMeans[ampName])
else np.nan,
529 'RAW_VARS': np.array(self.
rawVars[ampName]).tolist()
530 if len(self.
rawVars[ampName])
else np.nan,
531 'GAIN': self.
gain[ampName],
532 'GAIN_ERR': self.
gainErr[ampName],
533 'NOISE': self.
noise[ampName],
534 'NOISE_ERR': self.
noiseErr[ampName],
535 'PTC_FIT_PARS': np.array(self.
ptcFitPars[ampName]).tolist(),
536 'PTC_FIT_PARS_ERROR': np.array(self.
ptcFitParsError[ampName]).tolist(),
539 'COVARIANCES': np.pad(np.array(self.
covariances[ampName]),
540 ((0, nPadPoints[ampName]), (0, 0), (0, 0)),
541 'constant', constant_values=np.nan).reshape(
542 nSignalPoints*covMatrixSide**2).tolist(),
544 ((0, nPadPoints[ampName]), (0, 0), (0, 0)),
545 'constant', constant_values=np.nan).reshape(
546 nSignalPoints*covMatrixSide**2).tolist(),
548 ((0, nPadPoints[ampName]), (0, 0), (0, 0)),
549 'constant', constant_values=0.0).reshape(
550 nSignalPoints*covMatrixSide**2).tolist(),
551 'A_MATRIX': np.array(self.
aMatrix[ampName]).reshape(covMatrixSide**2).tolist(),
552 'B_MATRIX': np.array(self.
bMatrix[ampName]).reshape(covMatrixSide**2).tolist(),
553 'COVARIANCES_MODEL_NO_B':
555 ((0, nPadPoints[ampName]), (0, 0), (0, 0)),
556 'constant', constant_values=np.nan).reshape(
557 nSignalPoints*covMatrixSide**2).tolist(),
558 'A_MATRIX_NO_B': np.array(self.
aMatrixNoB[ampName]).reshape(
559 covMatrixSide**2).tolist(),
560 'FINAL_VARS': np.pad(np.array(self.
finalVars[ampName]), (0, nPadPoints[ampName]),
561 'constant', constant_values=np.nan).tolist(),
562 'FINAL_MODEL_VARS': np.pad(np.array(self.
finalModelVars[ampName]),
563 (0, nPadPoints[ampName]),
564 'constant', constant_values=np.nan).tolist(),
565 'FINAL_MEANS': np.pad(np.array(self.
finalMeans[ampName]),
566 (0, nPadPoints[ampName]),
567 'constant', constant_values=np.nan).tolist(),
568 'BAD_AMPS': np.array(self.
badAmps).tolist()
if len(self.
badAmps)
else np.nan,
569 'PHOTO_CHARGE': np.array(self.
photoCharge[ampName]).tolist(),
572 outMeta = {k: v
for k, v
in inMeta.items()
if v
is not None}
573 outMeta.update({k:
"" for k, v
in inMeta.items()
if v
is None})
574 catalog.meta = outMeta
575 tableList.append(catalog)
580 """Get the exposures used, i.e. not discarded, for a given amp.
581 If no mask has been created yet, all exposures are returned.
592 return [(exp1, exp2)
for ((exp1, exp2), m)
in zip(pairs, mask)
if bool(m)
is True]
def requiredAttributes(self, value)
def updateMetadata(self, camera=None, detector=None, filterName=None, setCalibId=False, setCalibInfo=False, setDate=False, **kwargs)
def requiredAttributes(self)
def __init__(self, ampNames=[], ptcFitType=None, covMatrixSide=1, **kwargs)
def getExpIdsUsed(self, ampName)
def fromDict(cls, dictionary)
def fromTable(cls, tableList)
def updateMetadata(self, setDate=False, **kwargs)
def setAmpValues(self, ampName, inputExpIdPair=[(np.nan, np.nan)], expIdMask=[np.nan], rawExpTime=[np.nan], rawMean=[np.nan], rawVar=[np.nan], photoCharge=[np.nan], gain=np.nan, gainErr=np.nan, noise=np.nan, noiseErr=np.nan, ptcFitPars=[np.nan], ptcFitParsError=[np.nan], ptcFitChiSq=np.nan, ptcTurnoff=np.nan, covArray=[], covArrayModel=[], covSqrtWeights=[], aMatrix=[], bMatrix=[], covArrayModelNoB=[], aMatrixNoB=[], finalVar=[np.nan], finalModelVar=[np.nan], finalMean=[np.nan])