Release 12.0 (2016-06-15)¶
Source | Identifier |
---|---|
Git tag | 12.0 |
EUPS distrib | v12_0 |
Preamble¶
Major Feature Development¶
With a new release schedule, this saw prolonged development. This release features a complete backport of functionality from the Hyper-Suprime Cam pipeline, in addition to many new LSST-driven developments. Even Astropy views of LSST catalogs objects are available.
New: Install with Anaconda¶
In addition to the customary source installation, this release features two binary installation methods:
- Anaconda
- CernVM FS (contributed by Fabio Hernandez)
If you are unsure which one to use, we recommend you try the conda binaries. This is the first conda binary release though and it is always possible we did not anticipate all the problems “in the wild;” if you experience difficulties, please report them so that we can improve our testing (and help you!).
New Documentation at pipelines.lsst.io¶
Our release documentation and installation instructions are no longer on the corporate wiki; you can find them at http://pipelines.lsst.io/.
You are welcome and encouraged to contribute improvements to the repo on GitHub: https://github.com/lsst/pipelines_docs/.
New Release Schedule¶
The timing of the formal releases (which mark the end of each DM planning cycle) has precessed from Winter/Summer to Spring/Fall. This has been done to avoid conflict between DM cycle planning activities and the LSST All-Hands Meetings in August and February. In the future, formal releases will be in June and December. In order to precess the cycles to the new schedule, an Extra 3-month cycle (aka X16) was added, so 12.0 is a “bumper” release containing approximately 9 months of activity.
Join Us on the Community Forum¶
If you have questions or comments or comments about the LSST Science Pipelines, visit our new community forum for advice: https://community.lsst.org/.
Contents¶
- Major Functionality and Interface Changes
- Bug Fixes
- Build and Code Improvements
- Science User Interface
- Qserv and Data Access
- Documentation improvements
See also:
Major Functionality and Interface Changes¶
- Add the option of excluding mask planes before coaddition
- Dumping task configuration parameters now includes documentation
- Clean up interpolation tasks and implement useFallbackValueAtEdge
- HSC backport: Avoid I/O race conditions config write out
- HSC backport: Introduce SafeClipAssembleCoaddTask which extends AssembleCoaddTask to make clipping safer
- HSC backport: Allow for some fraction of PSF Candidates to be reserved from PSF fitting
- Updated pipeline configuration based on Hyper Suprime-Cam experience
- Define polygon bounds for CCDs based on vignetted regions
- Introduce --rerun option for command line tasks
- Introduce framework for injecting fake sources into data processing
- Add convenience routines for working with tracts
- Warp images based on an XYTransform
- Add getCoordSystem to Coord and add UNKNOWN CoordSystem enum
- Adapt joinMatchListWithCatalog to facilitate and simplify denormalizing a match list
- Add a script for visualizing skymaps and CCDs
- Add functions to generate “unpacked matches” to and from a catalog
- Add a measurement algorithm which records the focal plane coordinates of sources
- Add a measurement algorithm which records the Jacobian at the positions of sources
- Add a measurement algorithm which records the number of input images contributing to a coadd sources
- Add a measurement algorithm which records the variance at the positions of sources
- Add a measurement algorithm which records source flux in an aperture scaled to the PSF
- Add a measurement algorithm which quantifies the amount of “blendedness” of an object
- Add a “simple” shape measurement algorithm
- Add Hirata-Seljak-Mandelbaum shape measurement algorithms
- Add option to temporarily remove an interpolated background prior to detection
- Add function to average coordinates
- Integrate support for Hyper Suprime-Cam
- Reimplement PSF Shapelet approximations for CModel
- Propagate flags from individual visits to coadds
- Aperture correction on coadd measurements
- Return grown Footprints from detection
- Enable measurement of “sky objects” in coadd processing
- Specification of output directory is now mandatory
- Bright object masks
- CModel fitting improvements
- Astropy Table views into LSST Catalog objects
- Add an “afterburner” measurement facility
- Tasks can now be kept in registries
- New test utilities in afw: BoxGrid and makeRampImage
- Correcting non-linearity
- Amplifier information catalogs have changed
- Changes in how background subtraction is done
- Star selectors have changed
- Backward-incompatible changes to ProcessCcdTask and subtasks
Add the option of excluding mask planes before coaddition¶
The user-configurable parameter removeMaskPlanes
has been added to AssembleCoaddConfig
.
This is a list of mask planes which will not be propagated to the coadd; by default, the CROSSTALK
and NOT_DEBLENDED
mask planes are removed.
[DM-4866]
Dumping task configuration parameters now includes documentation¶
That is, running ${TASK} ... --show config
displays not only the names and values of the configuration but also associated documentation.
[DM-3811]
Clean up interpolation tasks and implement useFallbackValueAtEdge¶
The interpImageTask
in pipe_tasks has been refactored to implement a single run function for interpolation over a list of defects in an image.
This run function will accept an afw
image type of either MaskedImage
or Exposure
.
A defects list can be passed in directly or the name of mask plane can be passed from which a defects list will be created.
If a PSF is attached to the image, it will be used as the (required) argument for the interpolateoverDefects
function in meas_algorithms
used for the interpolation.
Otherwise a FWHM (in pixels) can be provided or the defaultFWHM
value in meas_algorithms
’ GaussianFactory
is used.
Note that while the PSF is a required argument for meas_algorithms
’ interpolateOverDefects
function it is currently not being used, so it is not necessary to pass in an accurate PSF.
The useFallbackValueAtEdge
option is now implemented.
This tapers the interpolation to a fallbackValue
towards the image edge.
The fallbackValue
can be set via config parameters to be either computed as a statistical representation of the image data (MEAN
, MEDIAN
, or MEANCLIP
) or set by providing a specific value.
Allowance for a negative fallbackValue
is also controlled though a config parameter.
[DM-3677]
HSC backport: Avoid I/O race conditions config write out¶
This adds functionality to avoid potential I/O race conditions when running multiple simultaneous process.
This is accomplished by writing to temporary files and then renaming them to the correct destination filename in a single operation.
Also, to avoid similar race conditions in the backup file creation (e.g. config.py~1
, config.py~2
, …), a --no-backup-config
option (to be used with --clobber-config
) is added here to prevent the backup copies being made.
The outcome for this option is that the config that are still recorded are for the most recent run.
[DM-3911]
HSC backport: Introduce SafeClipAssembleCoaddTask which extends AssembleCoaddTask to make clipping safer¶
SafeClipAssembleCoaddTask
does the following,
- build both clipped and unclipped coadds and difference them first
- detect on the difference
- identify difference footprints that overlap appreciably with only one input image
- use identified difference footprints as a mask into a final coadd
- set a “clipped” bit on the final coadd for any pixel that did not include all input frames within the valid polygons.
To support the new coadd task, the pixel flags measurement plugin has been modified to accept two new configuration parameters.
These new parameters, named masksFpCenter
and masksFpAnywhere
, each accept a list of mask planes.
When the pixel flags measurement plugin searches for mask planes to set corresponding flags, it will now additionally search the user supplied mask plane lists.
The masksFpCenter
parameter specifies mask planes that, if found within the center of a footprint, will have a corresponding pixel flag set.
The masksFpCenter
parameter specifies mask planes that, if found anywhere in a footprint, will have a corresponding pixel flag set.
The masksFpAnywhere
parameter now has it’s defaults set within the stack which specify the clipped mask plane created by SafeClipAssembleCoaddTask
.
SafeClipAssebleCoaddTask
is now the default method for building a coadd within the LSST Stack.
The assembleCoadd.py CommandLineTask
now supports the --legacy
command line flag which will allow the original AssembleCoaddTask
to be run.
If the legacy task is run, the clipped mask plane must be removed from the masksFpAnywhere
configuration parameter.
[DM-2915]
HSC backport: Allow for some fraction of PSF Candidates to be reserved from PSF fitting¶
This adds functionality that allows for the reservation of a fraction of the PSF Candidates from PSF fitting.
The reserved fraction can then be used to test for over-fitting, do cross-validation, etc..
To support this functionality, the run methods of CalibrateTask
and MeasurePsfTask
now accept the additional keyword argument expId
.
Dummy versions of this keyword were added to the SdssCalibrateTask
and CfhtCalibrateTask
to maintain a consistent API.
[DM-3692]
Updated pipeline configuration based on Hyper Suprime-Cam experience¶
Configuration defaults and metadata through the LSST single-processing pipeline have been updated to match the current best practice established on HSC. Major changes include:
- Support for narrow band filters;
- Updated CCD defect lists;
- Optimized (in terms of CPU time) deblender settings;
- Avoiding failure in certain corner cases (e.g. operating on zero-length arrays, taking log10 of zero).
[DM-3942]
Define polygon bounds for CCDs based on vignetted regions¶
This adds a function in ip_isr’s isrTask.py to set a “valid polygon” for a given CCD exposure as the intersection of a polygon defined in focal plane coordinates and the CCD corners.
It is currently being used in obs_subaru
’s isr.py to set the polygon bounds (added in DM-2981) for a CCD exposure to include the non-vignetted regions.
The settings for the vignetted region is in a separate config file so that it can be used in different places in the code.
[DM-3259]
Introduce --rerun
option for command line tasks¶
This new command line option streamlines the process of specifying input and output repositories for command line tasks.
In its simplest form, the --rerun
option is a shorthand for specifying an output repository: output is written to a location relative to the input.
Thus
CmdLineTask /path/to/inputroot --rerun useroutput
is equivalent to
CmdLineTask /path/to/inputroot --output /path/to/inputroot/rerun/useroutput
Often, one task will process the output of a previous rerun.
For this situation, --rerun
provides a two-valued form which specifies relative locations of both input and output.
In this mode,
CmdLineTask /path/to/inputroot --rerun process1:useroutput
is equivalent to
CmdLineTask /path/to/inputroot --input /path/to/inputroot/rerun/process1 --output /path/to/inputroot/rerun/useroutput
Introduce framework for injecting fake sources into data processing¶
A new boolean option (doFakes
) and a retargetable task (fakes) have been introduced into the ProcessCcdTask
.
This (along with a new class called BaseFakeSourcesTask
) sets up a frame work that others may use to introduce known fake sources into the data processing stream.
However, this framework itself does not actually insert any fake data itself, but provides an interface others may use to define their own fake source injection task.
To implement a fake injection task one must create a child class of BaseFakeSourcesTask
(located in lsst.pipe.tasks.fakes
) and overload the run method to do the work of injecting the sources.
Each source that is injected should have a corresponding bit set in the FAKE
maskplane which can be accessed with the bitmask
attribute of BaseFakeSourcesTask
.
Once a task has been created, the config field fakes in processCcd must be retargeted to point to the user created task.
Additionally, the task will not be run unless the doFakes
configuration option in processCcd is set to True
.
However, if this variable is set to True
, and the task is not retargeted processCcd will fail.
[DM-3380]
Add convenience routines for working with tracts¶
Added two new data id containers:
PerTractCcdDataIdContainer
: determines the set of tracts each visit touches and adds a data reference with those tracts for each of the visit components. This allows for the user to run a command line task forcedPhotCcd.py for a given visit without having to know which tracts overlap the visit. Note this will also be used by meas_mosaic if/when it gets incorporated into the LSST Stack.TractDataIdContainer
: generates a list of data references for patches within a given tract (effectively a “data reference list” that points to the entire tract). Note that, at the time of writing, this is only being used by a QA analysis script currently under development.
[DM-4373]
Warp images based on an XYTransform¶
Added the ability to warp images using a transformation defined by an lsst::afw::geom::XYTransform
.
[DM-4162]
Add getCoordSystem to Coord and add UNKNOWN CoordSystem enum¶
Added method getCoordSystem
to lsst::afw::coord::Coord
.
Also added UNKNOWN=-1
as a new CoordSystem
enum (the existing enums retain their existing value).
DM-4606
Adapt joinMatchListWithCatalog to facilitate and simplify denormalizing a match list¶
The match lists created when performing image calibration (astrometry and photometry) are normalized (i.e. stripped down to a list of the matched reference and source ids and their distance) prior to being persisted.
The ability to denormalize a match catalog is very useful (for post QA analysis, for example).
This can now be done using the joinMatchListWithCatalog
function in meas_algorithms
’s LoadReferenceObjectsTask
.
It has been moved from meas_astrom
’s ANetBasicAstrometryTask
so that it can be easily accessed (requiring only that a reference object loader be initiated) and to allow it to work with any kind of reference catalog (i.e. other than a_net
).
[DM-3633]
Add a script for visualizing skymaps and CCDs¶
The skymap
package now contains the script showVisitSkyMap.py which provides a convenient way of visualizing the tracts, patches and CCDs contained in a set of visits.
[DM-4095]
Add functions to generate “unpacked matches” to and from a catalog¶
Functions have been added to lsst::afw::catalogMatches
to provide the ability to convert a match list into a catalog and vice versa (this can be useful for post-processing analyses; QA analysis, for example).
[DM-4729]
Add a measurement algorithm which records the focal plane coordinates of sources¶
The SingleFrameFPPositionPlugin
measurement plugin, available in meas_base
, records the positions of source centroids in focal plane coordinates (which may be convenient for plotting).
This plugin is not enabled by default, but may be switched on by requesting base_FPPosition
in measurement configuration.
[DM-4234]
Add a measurement algorithm which records the Jacobian at the positions of sources¶
The SingleFrameJacobianPlugin
calculates the ratio between the nominal Jacobian determinant at the source centroid (as determined by a user-specified pixel scale) and the actual Jacobian determinant as derived from the astrometric solution.
This plugin is not enabled by default, but may be switched on by requesting base_Jacobian
in measurement configuration.
[DM-4234]
Add a measurement algorithm which records the number of input images contributing to a coadd sources¶
When run on a source detected on a coadd, SingleFrameInputCountPlugin
records the number of input images which were stacked to create the coadd at the position corresponding to the source centroid.
The plugin is referred to as base_CountInputs
, and is enabled by default when performing measurement on coadded images.
It is not appropriate to enable this plugin when processing single visit (i.e., not coadded) images.
[DM-4235]
Add a measurement algorithm which records the variance at the positions of sources¶
The SingleFrameVariancePlugin
records the median variance in the background around the position of the source being measured.
The plugin is referred to as base_Variance
and is enabled by default when performing single frame measurement.
[DM-4235, DM-5427]
Add a measurement algorithm which records source flux in an aperture scaled to the PSF¶
The ScaledApertureFluxAlgorithm
measures the flux in a circular aperture with radius scaled to some user-specified multiple of the PSF FWHM.
This plugin is not enabled by default, but may be switched on by requesting the base_ScaledApertureFlux
in measurement configuration.
[DM-3257]
Add a measurement algorithm which quantifies the amount of “blendedness” of an object¶
The BlendednessAlgorithm
measures the amount to which an object is blended.
Both the flux and shape of each child object are compared to measurements at the same point on the full image.
The size of the weight function used on both images is determined from the child object.
The blendedness metric implemented is defined as 1-childFlux/parentFlux
.
The plugin is referred to as base_Blendedness
and is not enabled by default.
[DM-4847]
Add a “simple” shape measurement algorithm¶
The SimpleShape
algorithm, provided in the meas_extensions_simpleShape
package, computes the non-adaptive elliptical Gaussian-weighted moments of an image.
The plugin is referred to as ext_simpleShape_SimpleShape
and is not enabled by default.
[DM-5284]
Add Hirata-Seljak-Mandelbaum shape measurement algorithms¶
The meas_extensions_shapeHSM
package has been added to the distribution.
This provides a series of measurement algorithms based on the work by Hirata and Seljak (2003) and Mandelbaum et al (2005).
Please cite those works if publishing results based on this code.
These algorithms are disabled by default; they can be enabled by requesting ext_shapeHSM_HsmShapeBj
, ext_shapeHSM_HsmShapeLinear
, ext_shapeHSM_HsmShapeKsb
, ext_shapeHSM_HsmShapeRegauss
, ext_shapeHSM_HsmSourceMoments
and/or ext_shapeHSM_HsmPsfMoments
in the measurement configuration.
[DM-2141, DM-3384, DM-4780]
Add option to temporarily remove an interpolated background prior to detection¶
This has the potential for removing a large number of junk detections around bright objects due to noise fluctuations in the elevated local background.
The extra subtracted interpolated background is added back in after detection.
Currently, the default setting for the config parameter doTempLocalBackround
is set to False
.
[DM-4821]
Add function to average coordinates¶
Added function lsst.afw.coord.averageCoord
, which will return an average coordinate (accounting for spherical geometry) given a list of input coordinates.
[DM-4933]
Integrate support for Hyper Suprime-Cam¶
The obs_subaru
camera package, which enables the stack to operate on data taken with the Suprime-Cam and Hyper Suprime-Cam instruments on Subaru, has been modernized, resolving build and test issues and integrating it with LSST’s continuous integration system.
It will now be included as part of the lsst_distrib release.
Note, though, that usage of Suprime-Cam with the stack is unsupported and unmaintained at present.
[DM-3518, DONE DM-4358, DM-5007]
Reimplement PSF Shapelet approximations for CModel¶
A new algorithm for computing multi-shapelet approximations (DoubleShapeletPsfApprox
) has been added to meas_modelfit
.
This is simpler and more robust than the old algorithm, which has been renamed to GeneralShapeletPsfApprox
.
The new algorithm is recommended for production use, and is now the default.
[DM-5197]
Propagate flags from individual visits to coadds¶
A task has been added which can propagate flags from individual visit catalogs to coadd catalogs. This is useful, for example, to track which stars in the coadd were used for measuring PSFs on the individual visits. [DM-4878, DM-5084]
Prototype Brighter-Fatter correction¶
Code for correcting for the Brighter-Fatter effect on CCDs is now available in the stack.
It is enabled using the doBrighterFatter
configuration option to IsrTask
.
It requires a pre-generated correction kernel.
Calculation of this kernel is not currently performed within the stack: a prototype exists, and will be merged to the Calibration Products Pipeline in a future cycle.
[DM-4837, DM-5082, DM-5130]
Aperture correction on coadd measurements¶
Aperture corrections are now enabled for measurements performed on coadds. [DM-5086]
Return grown Footprints from detection¶
By default, Footprints
returned by SourceDetectionTask
are now grown by a multiple of the PSF size.
[DM-4410]
Enable measurement of “sky objects” in coadd processing¶
Sky objects correspond to source properties measured at positions when no objects have been detected.
This enables us to better characterize the depth of the survey.
This functionality is enabled by default; it can be disabled by setting nSkySourcesPerPatch
to zero in the configuration of MergeDetectionsTask
.
[DM-4840, DM-5288]
Specification of output directory is now mandatory¶
When running a command line task which produces output it is now mandatory to specify an output directory (previously, if no output location was specified, data products were written back to the input repository).
Output locations may be specified with the --rerun
or --output
command line options.
More information is available on community.lsst.org.
[DM-4236]
Bright object masks¶
Given an input catalog listing the known positions and sizes of bright objects, a bit is set in the mask plane for every pixel lying within the object’s footprint. [DM-4831]
CModel fitting improvements¶
CModel is a model fitting approach in which a pure exponential and pure de Vaucouleur are each fit separately, and then their linear combination is fit while the ellipse parameters are held fixed. Improvements in this release make CModel fitting faster and improves results on objects which are detected with an unphysically large likelihood radius. This has been achieved in three ways:
- The initial approximate exponential fit that is used to determine the starting parameters and pixel region to use for the exp and dev fit now uses per-pixel variances;
- The method used to determine the pixel region used in fitting has been adjusted to make smarter choices, using fewer pixels on average for all objects and many fewer pixels for unphysically large objects;
- A new semi-empirical Bayesian prior on radius and ellipticity based on COSMOS distributions has been introduced.
[DM-4768]
Astropy Table views into LSST Catalog objects¶
Astropy Table views into LSST catalog objects can now be created. These views share underlying data buffers (aside from flag fields), making them read-write, but rows and columns added on either side will not be visible in the other. Two equivalent interfaces are available:
astropy_table = lsst_catalog.asAstropy()
and (in Astropy >= v1.2):
astropy_table = astropy.table.Table(lsst_catalog)
QTable objects can also be used, resulting in columns that use Astropy’s units package to enforce unit correctness.
These interfaces have multiple options to control the details of the view, including how to handle columns that require copies; see the Python on-line help for asAstropy
for more information.
While LSST’s catalog objects have features that make them particularly useful in building pipelines, Astropy’s are much more appropriate for most analysis tasks, and we strongly recommend using them for any analysis tasks that need to add columns to tables or combine columns from multiple tables.
Add an “afterburner” measurement facility¶
This new functionality makes it possible to register plugins to calculate quantities based on the results of pixel measurement algorithms. This might include, for example, star-galaxy separation or applying aperture corrections. Afterburners of this type are run after measurement plugins, and do not have access to pixel data. [DM-4887]
Tasks can now be kept in registries¶
Related sets of tasks should now be kept in a registry as per RFC-183, with a common abstract base class.
Tasks can now use an lsst.pex.config.RegistryField
config field to specify a subtask if that subtask is in a registry (DM-6074).
The task is built and used the same way as if it was specified in an lsst.pex.config.ConfigurableField
, but retargeting and overriding config parameters is different.
See task documentation for more information.
See also How to Write a Task for guidelines for choosing between using lsst.pex.config.RegistryField
and lsst.pex.config.ConfigurableField
to hold a subtask.
PSF determiners are now tasks that inherit from an abstract base class lsst.meas.algorithms.PsfDeterminerTask
(DM-6077).
However, the effect on existing code was negligible because they were already configurables used from a registry.
The way you retarget PSF determiners and override their config parameters remains unchanged.
Reimplemented the registry for star selectors that was lost in DM-5532: lsst.meas.algorithms.starSelectorRegistry
(DM-6474)
One backwards incompatible change: in [DM-6474] MeasurePsfTask
and MeasureApCorrTask
both now specify their star selectors using an lsst.pex.config.RegistryField
.
This means the format for retargeting star selectors and overriding their config parameters has changed.
The config override files in the various obs_
packages are updated accordingly.
New test utilities in afw: BoxGrid and makeRampImage¶
lsst.afw.geom.testUtils.BoxGrid
divides a bounding box into nCol x nRow
equal sized sub-boxes (as equal sized as possible for integer boxes that do not divide evenly) that tile the larger box and have the same type.
lsst.afw.image.testUtils.makeRampImage
makes an image (ImageX
where X
is any available type) with values that increase linearly between specified limits (linearly to the extent possible, for integer images).
[DM-5462]
Correcting non-linearity¶
Introduced a standard way to correct non-linearity (linearize data) as part of Instrument Signature Removal (ISR).
Linearization is performed by new functors in ip_isr
:
LinearizeBase
is an abstract base class. It is called with an image and the detector information and the correction is performed in place (like all other ISR corrections inIsrTask
).LinearizeSquared
performs a simple square correction:corrImage = uncorrImage + c0*uncorrImage^2
wherec0
is the first coefficient in in the linearity coefficients of the amp into catalog. This is the model used byobs_subaru
for SuprimeCam and HSC.LinearizeLookupTable
uses a lookup table to determine an offset (read the code doc string for details). The lookup table is saved with the linearizer, but the linearizer also performs a sanity check against the provided detector when called.- You can easily add other linearizers as desired.
- Each linearizer has a class variable
LinearizationType
, a string whose value should be used as the linearization type in the amplifier info catalog. The linearizer checks this value when performing linearization.
All detector in a camera must use the same type of linearizer.
However linearization can easily be disabled on a detector-by-detector basis by setting linearity type to lsst.afw.cameraGeom.NullLinearityType
.
For a camera that does not need linearization, do this for all detectors.
Linearizers are obtained from the butler, like any other calibration product.
- For
LinearizeSquared
and other linearizers that get coefficients from the amplifier info catalog, only one instance is needed for all detectors. In that case the simplest technique is to definemap_linearize
andbypass_linearize
methods on the camera mapper to return an instance. See theobs_subaru
package for an example. - For
LinearizeLookupTable
and other linearizers that store detector-specific data, theobs_
package developer must pickle one linearizer for each detector and make them available as dataset type “linearizer”. - If the camera does not want linearization then no “linearizer” dataset type is required because
IsrTask
realizes linearization is not wanted before it tries to unpersist the linearizer. You may leaveIsrConfig.doLinearize
set to its default value ofTrue
without significant performance penalty.
Amplifier information catalogs have changed¶
The format of amplifier information catalogs has changed. Your versions of afw and associated obs_ packages must be compatible or else you will get errors when building a camera mapper (thus when running any nearly any command-line task).
Amplifier information catalogs have a new field as of DM-6147: suspect level.
If the value is not nan
then pixels whose values are above this level are masked as SUSPECT
.
The only cameras that specify a suspect level, so far, are HSC and SuprimeCam.
However, a value can be set for any camera, if desired.
SUSPECT
is intended to indicate pixels with doubtful values due to errors that are difficult to correct accurately, e.g. a regime where linearity correction is not very stable.
In addition, saturation level in the amplifier information catalog is now a floating point value (instead of an integer) and a value of nan
means ‘do not mask pixels as SAT.’
[DM-6147]
Changes in how background subtraction is done¶
Background estimation in Python is now done using different routines in meas_algorithms
.
There is a new task SubtractBackgroundTask
, with full documentation and a working example.
The existing function getBackground
(which fits a background) is replaced by SubtractBackgroundTask.fitBackground
.
Changes from getBackground
are:
getBackground
could returnNone
if the fit failed; in that situationfitBackround
will raiseRuntimeError
instead of returningNone
.- The argument
image
was renamed tomaskedImage
, for clarity. - The config is not passed as an argument.
- The debug display code uses different keys and is updated to use
afw.display
.
The existing function estimateBackground
(which subtract a background from an exposure) is replaced by SubtractBackgroundTask.run
.
Changes from estimateBackground
are:
- You may pass in a background model (an
lsst.afw.math.BackgroundList
). - It returns a struct containing the updated background model.
- The config is not passed as an argument.
- The debug display code displays the unsubtracted image and uses different keys and is updated to use
afw.display
.
The task’s config SubtractBackgroundConfig
replaces the old BackgroundConfig
.
The field algorithm
may no longer be None
; you must use the string "NONE"
, instead.
See discussion on Community for details.
Star selectors have changed¶
Star selectors are now tasks. They were already configurable and many added logs; now they are standard tasks.
The star selector registry starSelectorRegistry
was gone for awhile.
Now that it is back, using a registry field from that registry is the preferred way to specify a star selector as a subtask of another task.
Added BaseStarSelectorTask
(but for awhile it was called StarSelectorTask
) an abstract base class for star selectors with the following methods:
selectStars
an abstract method that takes a catalog of sources and returns a catalog of stars.makePsfCandidates
a concrete method that takes a catalog of stars (as returned byselectStars
and produces PSF candidates; it also returns a sub-catalog of those stars that were successfully turned into PSF candidates (which is usually all of them).run
a concrete method that selects stars, makes them into PSF candidates and optionally flags the stars.
Backward-incompatible changes to ProcessCcdTask and subtasks¶
Code changes¶
ProcessCoaddTask
is gone, along with all bin scripts that run it. Use the new Multi-Band code, instead.ProcessCcdTask
has been rewritten, so its config has changed. Config override files will need to be updated. This will be done for theobs_
packages as part of the merge, but if you have personal config override files then you will probably need to update them.Camera-specific variants of
ProcessCcdTask
are gone. You will run processCcdTask.py to process images for all cameras.For DECam processCcdTask.py will use the LSST Stack’s ISR by default. To read
instcal
files from the DECam Community Pipeline, replace the ISR task withDecamNullIsrTask
by using a config override file containing the following:from lsst.obs.decam.decamNullIsr import DecamNullIsrTask config.isr.retarget(DecamNullIsrTask)
A new dynamic dataset type is available for adding data ID arguments to the argument parser for command-line tasks:
ConfigDatasetType
obtains the dataset type from a config parameter.Various subtasks have changed, including:
- New camera-specific ISR task variants for SDSS and DECam:
SdssNullIsrTask
andDecamNullIsrTask
. - New task
DetectAndMeasureTask
detects and deblends sources and performs single-frame measurement. - New task
CharacterizeImageTask
measures PSF and aperture correction, among other things. CalibrateTask
has been rewritten. It now performs deep detection and measurement, astrometry and photometry.- Camera-specific variants of
CalibrateTask
are gone. ProcessImageTask
(formerly a base class forProcessCcdTask
andProcessCoaddTask
) is gone.
- New camera-specific ISR task variants for SDSS and DECam:
Data product changes¶
icSrc
no longer includes RA/Dec coordinate data, because the fit WCS is not available when the catalog is constructed. You will have to set the coord field yourself if you need it.icExp
andicExpBackground
can optionally be written byCharacterizeImageTask
. They are so close tocalexp
andcalexpBackground
that they are not written by default.icMatch
is no longer being written.
Algorithm changes¶
- PSF is fit somewhat differently. The new task fits the PSF in using a configurable number of iterations. By default each iteration starts with a simple Gaussian PSF whose sigma matches the PSF of the previous fit, but you can use the actual PSF each time. Using a Gaussian causes convergence in 2 iterations. Using the fit PSF slows convergence.
- Sources in the
icSrc
catalog should have a more consistent minimum SNR for varying seeing. The old code detected once, using a Gaussian PSF with FWHM set by a config parameter. The new code performs detection using the PSF in the final PSF iteration. - The default star selector for
MeasurePsfTask
isobjectSize
rather thansizeMagnitude
. TheobjectSize
star selector is preferred and was already being specified as an override by HSC. - The icSrc catalog is not matched to an astrometric reference catalog unless the star selector used to measure PSF can use the matches (which is unusual).
- The astrometric and photometric solution now use the deeper
src
catalog instead of the shallowericSrc
catalog, though with a new SNR cutoff whose default provides a depth similar to theicSrc
catalog. - Fake source handling is temporarily missing; it will be re-added in DM-5310.
Bug Fixes¶
- Persist LTVn headers as floating point numbers
- Fix bug when identifying existing peaks in a merge
- Fix situation in which the getChildren method of SourceCatalog may return the wrong information
- Fix warping when the WCS have different coordinate systems
- Correct bad default minInitialRadius for CModel
- Correct algebraic error in CModel uncertainty calculation
- NaiveDipoleCentroid and NaiveDipoleFlux algorithms no longer require a centroid slot
- Update (some) example code to run with recent stack versions
- Fix a failure to appropriately log failed task execution
- Updates to Skymap packages
- Fix variance in coadded images
- Fix variance in deblended sources
- Fix logic for applying aperture corrections
- More uniform support for assigning to catalog columns
- Upgraded WCSLIB to version 5.13
- Fix rotation for instrument signature removal in obs_subaru
- Fix a silent failure to apply config overrides
- Correctly set flags for bad SdssShape measurements.
- Fix CONSTANT background interpolation of bad data
- Correct misleading filter fallback error message
- Accommodate pixel padding when unpersisting reference catalog matches
Persist LTVn headers as floating point numbers¶
When persisting to a FITS file, these header cards were previously, incorrectly, stored as integers. [DM-4133]
Fix bug when identifying existing peaks in a merge¶
If two separate footprints from the same catalog are merged due to an existing merged object which overlaps both of them the flags of which peaks are being detected were not being propagated. This issue caused apparent dropouts of sources and has now been fixed. [DM-2978]
Fix situation in which the getChildren method of SourceCatalog may return the wrong information¶
The getChildren
method requires that the result must be sorted by parent.
This is naturally the case when the catalog is produced by detection or deblending tasks.
However, if multiple catalogs are concatenated together this condition may no longer be true.
The getChildren
method was updated to raise an exception if the precondition of sorting is not met.
[DM-2976]
Fix warping when the WCS have different coordinate systems¶
Warping assumed that the sky representation of both WCS was identical. [DM-4162]
Correct bad default minInitialRadius for CModel¶
The minInitialRadius
configuration parameter had a default that is too small, causing many galaxies to be fit with point source models, leading to bad star/galaxy classifications.
[DM-3821]
Correct algebraic error in CModel uncertainty calculation¶
There was a simple but important algebra error in the uncertainty calculation, making the uncertainty a strong function of magnitude. [DM-3821]
NaiveDipoleCentroid and NaiveDipoleFlux algorithms no longer require a centroid slot¶
Previously, initializing these algorithms was only possible if a centroid was already defined. That was not only unnecessary, but also made them more complicated to use, particularly in testing. [DM-3940]
Update (some) example code to run with recent stack versions¶
Changes in afw::table
had broken examples/calibrateTask.py
in pipe_tasks
.
It has now been updated to comply with the latest afw::table
API.
[DM-4125]
Fix a failure to appropriately log failed task execution¶
When task execution fails, we add a message to the log (with level FATAL
).
In some cases, the very act of attempting to log this message could throw an exception, and information about the original error was lost.
This has now been resolved.
[DM-4218]
Updates to Skymap packages¶
Add functions to return patches and tracts which contain given coordinates, i.e. conversions between celestial coordinates and tract,patch
indices.
Functions include findClosestTractPatchList
, findAllTract
, and findTractPatchList
which finds the closets tract and patch that overlaps coordinates, finds all tracts which include the specified coordinate, and finds tracts and patches that overlap a region respectively.
[DM-3775]
Fix variance in coadded images¶
Warping images in order to coadd them loses variance into covariance. This is mitigated by scaling the variance plane of the coadd. The scaling was being applied incorrectly in some cases. This has now been fixed. [DM-4798]
Fix variance in deblended sources¶
The deblender incorrectly scaled the variance plane in deblended sources with the fraction of the total flux assigned to the source. This has been corrected. [DM-4845]
Fix logic for applying aperture corrections¶
This fixes a bug whereby the aperture corrections were being applied only after all the measurement plugins had run through, independent of their execution order.
This resulted in plugins whose measurements rely on aperture corrected fluxes (i.e. with execution order > APCORR_ORDER
) being applied prior to the aperture correction, leading to erroneous results.
The only plugin that was affected by this at this time was base_ClassificationExtendedness
.
[DM-4836]
More uniform support for assigning to catalog columns¶
Assignment of scalars or NumPy arrays to columns of afw.table.Catalog objects (e.g. catalog["column"] = value
) is now more uniformly supported across types (support was inconsistent before, and never allowed scalar or augmented assignment).
Flag columns still do not support column assignment, and Flag column access still returns a copy, not a view, because Flag values are stored internally as individual bits within a larger integer.
[DM-4856]
Upgraded WCSLIB to version 5.13¶
Version 5.13 of WCSLIB resolves memory corruption errors that could crash the stack in some circumstances. [DM-4904, RFC-89, DM-4946, DM-3793]
Fix rotation for instrument signature removal in obs_subaru¶
Approximately half of the HSC CCDs are rotated 180 deg with respect to the others.
Two others have 90 deg rotations and another two have 270 deg rotations (see HSC CCD layout).
The raw images for the rotated CCDs thus need to be rotated to match the rotation of their associated calibration frames (in the context of how they have currently been ingested) prior to applying the corrections.
This is accomplished by rotating the exposure using the rotated context manager function in obs_subaru
’s isr.py and the nQuarter
specification in the policy file for each CCD.
Currently, rotated uses afw
’s rotateImageBy90
(which apparently rotates in a counter-clockwise direction) to rotated the exposure by 4 - nQuarter
turns.
This turns out to be the wrong rotation for the odd nQuarter
CCDs.
This issue fixes this bug, leading to much improved processing of HSC CCD’s 100, 101, 102, and 103.
Note that, in the future, the ingestion of the calibration data will be updated such that no rotations are necessary (so they will then be removed from obs_subaru
accordingly).
[DM-4998]
Fix a silent failure to apply config overrides¶
When applying a config override, using a variable which hadn’t been defined should throw a NameError
, which ultimately propagates to the end user to notify them that something has gone awry.
This warning was being incorrectly suppressed.
[DM-5729]
Correctly set flags for bad SdssShape measurements.¶
The SdssShape
algorithm provides both shape and flux measurements.
In some cases, a failed shape measurement could go un-noticed, resulting in an incorrect and unflagged flux measurement being associated with that source.
This is now checked for, and bad fluxes are appropriately flagged.
[DM-3935]
Fix CONSTANT background interpolation of bad data¶
When performing interpolation over bad data (e.g. every pixel masked), all interpolation types other than CONSTANT
would return NaN
s; CONSTANT
would throw.
This has now been changed so that CONSTANT
also returns NaN
s.
[DM-5797]
Accommodate pixel padding when unpersisting reference catalog matches¶
The reference object loader in meas_algorithm
’s loadReferenceObjects.py grows the bbox
by the config parameter pixelMargin
(padding to add to 4 all edges of the bounding box [pixels]) when setting the radius of the sky circle to be searched in the reference catalog.
This is set to 50 by default but was not reflected by the radius parameter set in the metadata, which left open the possibility that some matches could reside outside the circle searched within the unpersisted radius.
Additionally, the match metadata was being set after the exposure’s WCS had been updated, also leading to an inconsistency with the sky circle that was actually searched.
We now ensure that the actual sky circle that was searched for reference objects is the one set and persisted in the match metadata.
[DM-5686]
Correct misleading filter fallback error message¶
When failing to load a calib
, if fallbackFilterName
was not set, a confusing and apparently unrelated error message would be generated (Unknown value type for filter: <type 'NoneType'>
).
This has been corrected to properly inform the user about the issue.
[DM-6151]
Build and Code Improvements¶
- Work-around incompatibilities with NumPy 1.10
- When building boost warn user if user-config.jam or site-config.jam exists
- Remove deprecated Task.display() method
- Efficiency improvement in converting Masks to DefectLists
- Add a new task parallelization framework
- Add parallel-processing top level tasks
- Adjust test tolerances to be compatible with MKL-based NumPy
- Now possible to directly get a Filter’s canonical name and aliases
- Fix build issues with recent clang
- Fix incorrect linking against Anaconda-provided libraries when using CMake
- Execute afw test suite when afwdata is not available
- Disable implicit threading
- Migrate to standard smart pointers
Work-around incompatibilities with NumPy 1.10¶
NumPy 1.10 introduced API changes which were incompatible with existing usage in the stack. The latter has been updated to match. [DM-4063, DM-4071, DM-4238].
When building boost warn user if user-config.jam or site-config.jam exists¶
Building boost can fail if a user-config.jam
or site-config.jam
exist and have options which conflict with the LSST build configuration process.
Introduce a warning message if either of these files are found to notify the user.
[DM-4198]
Remove deprecated Task.display() method¶
This method has been deprecated since release 9.2 (S14).
It has been removed from the codebase, and all stack code updated to directly interface with afw.display
or to use helper functions defined in meas_astrom
.
[DM-4428]
Efficiency improvement in converting Masks to DefectLists¶
The previous version of routine was extremely memory intensive when large numbers of pixels were masked. [DM-4800]
Add a new task parallelization framework¶
The ctrl_pool
package has been added to the LSST stack.
This is a high-level parallelization framework used for distributing Task execution across a cluster, based on an MPI process pool.
It is based on work carried out on Hyper Suprime-Cam.
It is not intended to be the long-term solution to parallelized processing in the LSST stack, but meets our data processing needs until the fully-fledged parallelization middleware is available.
[DM-2983, DM-4835, DM-5409]
Add parallel-processing top level tasks¶
The new pipe_drivers
package builds upon ctrl_pool
, above, to provide command-line scripts which coordinate distributed execution of the single-frame, coaddition and multiband processing steps using either the Python multiprocessing module or with a SLURM batch scheduler on a cluster.
[DM-3368, DM-3369, DM-3370]
Adjust test tolerances to be compatible with MKL-based NumPy¶
Anaconda 2.5 ships, by default, with a version of NumPy built against Intel MKL rather than OpenBLAS. This can change some numerical results slightly, necessitating a change to test tolerances. [DM-5108]
Now possible to directly get a Filter’s canonical name and aliases¶
Added the convenience methods getCanonicalName
and getAliases
to lsst.afw.image.Filter
, accessible from both C++ and Python.
These return the canonical name and the aliases, respectively, of the filter. This information was previously only available through an awkward sequence of method calls.
[DM-4816]
Fix build issues with recent clang¶
Recent releases of the clang C/C++ compiler, as shipped with Apple XCode, caused build failures in the stack. Although we believe this may be a problem with clang, we have worked around it within the stack code. We hope to track down the source of the error and, if appropriate, report it to the clang developers in future. [DM-5590, DM-5609]
Fix incorrect linking against Anaconda-provided libraries when using CMake¶
Some external packages—mariadb and mariadbclient—use a CMake based build system. This can incorrectly link against some libraries bundled with the Anaconda Python distribution, rather than the system-provided equivalents, resulting in a build failure. We have adjusted the build process of the affected packages to work around this error. [DM-5595]
Execute afw test suite when afwdata is not available¶
Some tests in the afw
package rely on data from the afwdata
package.
The test suite would search for afwdata
, and skip all of the afw tests if afwdata
is not available.
This check has been made smarter, so that only tests which actually require afwdata
are now skipped.
[DM-609]
Disable implicit threading¶
Low-level threading packages (such as OpenBLAS or MKL) can implicitly use many threads.
Since the LSST stack also parallelizes at a higher level (e.g. using Python’s multiprocessing module), this can cause undesirable contention.
We now disable implicit threading when explicitly parallelizing at a higher level to protect the user from this.
Implicit threading can be re-enabled by setting the LSST_ALLOW_IMPLICIT_THREADS
environment variable.
For more details, see this Community post.
[DM-4719]
Science User Interface¶
During cycles winter and extra 2016, SUIT group has embarked on a major rewrite of the Firefly client side code, converting the Java/GWT based code to pure JavaScript (ES6) code, adopting React/Redux framework. The goal is to finish more than 90% of the client code conversion by the end of X16. We will make a release version of the code by end of September 2016.
Documentation improvements¶
A number of documentation improvements were made, three of which may be of general interest:
DM Developer Guide¶
New life has been breathed into the DM Developer Guide both on the surface and behind the scenes. While the primary audience for this are LSST developers, it can be a useful guide to anyone who wishes to participate in our open source development—all the fun without the paperwork! You can find it at developer.lsst.io
LSST Technotes¶
We have a lightweight bootstrap that allows contributors to produce ReStructuredText documents that are then published via our web services using Sphinx and some of our won sauce. You can see a rich example in Colin Slater’s technote, “DMTN-006: False Positive Rates in the LSST Image Differencing Pipeline,” at dmtn-006.lsst.io. A list of available technotes is currently curated on the Community forum.
If you have been thinking hard about an LSST-related technical or scientific issue that you feel is below your “worth a paper” threshold, why not contribute it as a technote? We have a lsst-technote-bootstrap project to get you get started.
Release Notes and Installation¶
You’re reading them now on pipelines.lsst.io! We have moved our release note and installation instructions to our Spinx-based platform as well. This means that if you find a deficiency with our notes that needs clarification, or discover a new issue, you are quite welcome to fork-and-PR them; the GitHub repository is pipelines_docs.