IPP Software Navigation Tools IPP Links Communication Pan-STARRS Links

Changeset 40612 for trunk


Ignore:
Timestamp:
Jan 26, 2019, 11:47:09 AM (7 years ago)
Author:
eugene
Message:

spell-check, add conclusions

File:
1 edited

Legend:

Unmodified
Added
Removed
  • trunk/doc/release.2015/ps1.datasystem/datasystem.tex

    r40599 r40612  
    8282\begin{abstract}
    8383
    84 The Pan-STARRS Image Processing Pipeline performs the processing
     84The Pan-STARRS Data Processing System is responsible for the steps
    8585needed to downloaded, archive, and process all images obtained by the
    86 Pan-STARRS telescopes.  This article describes the overall data
    87 analysis system.
     86Pan-STARRS telescopes, including real-time detection of transient
     87sources such as supernovae and moving objects including potentially
     88hazardous asteroids.  With a nightly data volume of up to 4 terabytes
     89and an archive of over 4 petabytes of raw imagery, Pan-STARRS is
     90solidly in the realm of Big Data astronomy.  The full data processing
     91system consists of several subsystems covering the wide range of
     92necessary capabilities.  This article describes the Image Processing
     93Pipeline and its connections to both the summit data systems and the
     94outward-facing systems downstream.  The latter include the Moving
     95Object Processing System (MOPS) \& the public database: the Published
     96Science Products Subsystem (PSPS).
    8897
    8998\end{abstract}
     
    130139Release 1 (DR1) on 16 December 2016.  DR1 contains the results of the
    131140third full reduction of the Pan-STARRS $3\pi$ Survey archival data,
    132 identified as PV3.  Previous reductions \citep[PV0, PV1, PV2;
    133   see][]{magnier2017.datasystem} were used internally for pipeline
    134 optimization and the development of the initial photometric and
    135 astrometric reference catalog \citep{magnier2017.calibration}.  The
    136 products from these reductions were not publicly released, but have
    137 been used to produce a wide range of scientific papers from the
    138 Pan-STARRS 1 Science Consortium members \citep{chambers2017}.  DR1
    139 contained only average information resulting from the many individual
    140 images obtained by the $3\pi$ Survey observations.  A second data
    141 release, DR2, was made available \note{20 January 2019}.  DR2 provides
    142 measurements from all of the individual exposures, and include an
    143 improved calibration of the PV3 processing of that dataset.
     141identified as PV3.  Previous reductions (PV0, PV1, PV2) were used
     142internally for pipeline optimization and the development of the
     143initial photometric and astrometric reference catalog
     144\citep{magnier2017.calibration}.  The products from these reductions
     145were not publicly released, but have been used to produce a wide range
     146of scientific papers from the Pan-STARRS 1 Science Consortium members
     147\citep{chambers2017}.  DR1 contained only average information
     148resulting from the many individual images obtained by the $3\pi$
     149Survey observations.  A second data release, DR2, was made available
     15028 January 2019.  DR2 provides measurements from all of the individual
     151exposures, and include an improved calibration of the PV3 processing
     152of that dataset.
    144153
    145154This is the second in a series of seven papers describing the
    146 Pan-STARRS1 Surveys, the data reduction techiques and the resulting
     155Pan-STARRS1 Surveys, the data reduction techniques and the resulting
    147156data products.  This paper (Paper II) presents a description of the
    148157Pan-STARRS data handling systems, with an emphasis on the Image
     
    198207%Pan-STARRS 1 Database and Data Products
    199208\citet[][Paper VI]{flewelling2017}
    200 describe the details of the resulting catalog data and its organization in the Pan-STARRS database.
     209describe the details of the resulting catalog data and its
     210organization in the Pan-STARRS database.
    201211
    202212%Huber et al. 2017 (Paper VII)
     
    218228used by the IPP for regular nightly operations and for processing the
    219229PV3 data release, with some details on the scale of computing needed
    220 to reduce this large number of exposures.  Finally,
    221 Section~\ref{sec:discussion} presents a discussion of some of the
    222 lessons learned in the creation of the IPP, and its utility in
    223 reducing data from other cameras and telescopes.
     230to reduce this large number of exposures. 
     231
     232% Finally,
     233% Section~\ref{sec:discussion} presents a discussion of some of the
     234% lessons learned in the creation of the IPP, and its utility in
     235% reducing data from other cameras and telescopes.
    224236
    225237%% {\color{red} {\em Note: These papers are being placed on arXiv.org to
     
    263275  ingests the calibrated measurements from the IPP, MOPS, and others
    264276  and generates a high-availability database with web-based
    265   interactions for public consumption \citet[][]{flewelling2017}.
     277  interactions for public consumption \citep[][]{flewelling2017}.
    266278
    267279\end{itemize}
    268280Management of the above set of analysis stages takes place at the IfA
    269281within the scope of responsibility of the Pan-STARRS Observatory.
    270 Across the wider Pan-STARRS colloboration(s), additional data analysis
     282Across the wider Pan-STARRS collaboration(s), additional data analysis
    271283operations are performed to support science results.  These
    272284collaboration-wide analysis operations range from those which are
     
    331343Pan-STARRS has performed several large-scale reprocessings of both the
    332344Medium Deep and $3\pi$ Survey data for internal consumption.  For the
    333 $3\pi$ Survey data, we identify these large-scale reprocessings as
    334 PV1, PV2, and PV3, with PV3 the analysis used for the first public
    335 data release, DR1.  We also refer to the nightly science analysis of
    336 the data as PV0.  For these reprocessing stages, the standard steps of
    337 \ippstage{chip} through \ippstage{warp}, plus \ippstage{stack} and
    338 \ippstage{diff} are performed, starting from raw data, usually using a
    339 single homogenous version of the data analysis procedures.  PV2 was a
    340 special case in which we started from the camera level products of PV1
    341 to speed up the turn-around to the community.  In addition to the
    342 analysis stages listed above which are shared with the nightly
    343 processing, these large-scale reprocessing analyses include additional
    344 processing steps.  A more detailed photometric analysis is performed
    345 on the stacks, including morphological analysis appropriate to
    346 galaxies.  The results of the stack photometry analysis are used to
    347 drive a forced-photometry analysis of the warp images.  These analysis
    348 steps are discussed in detail by
    349 \citet[][]{magnier2017.analysis}.  The data products from the
    350 camera, stack, and forced-warp photometry analysis stages
    351 are ingested into the internal calibration database (DVO, the Desktop
    352 Virtual Observatory) and used for photometric and astrometric
    353 calibrations \citet[see Section~\ref{sec:DVO} and][]{magnier2017.calibration}.
     345$3\pi$ Survey data, we identify these large-scale reprocessings as PV1
     346(Processing Version 1), PV2, and PV3, with PV3 the analysis used for
     347the first public data release, DR1.  We also refer to the nightly
     348science analysis of the data as PV0.  For these reprocessing stages,
     349the standard steps of \ippstage{chip} through \ippstage{warp}, plus
     350\ippstage{stack} and \ippstage{diff} are performed, starting from raw
     351data, usually using a single homogeneous version of the data analysis
     352procedures.  PV2 was a special case in which we started from the
     353camera level products of PV1 to speed up the turn-around to the
     354community.  In addition to the analysis stages listed above which are
     355shared with the nightly processing, these large-scale reprocessing
     356analyses include additional processing steps.  A more detailed
     357photometric analysis is performed on the stacks, including
     358morphological analysis appropriate to galaxies (model fits, Kron and
     359Petrosian aperture photometry, etc).  The results of the stack
     360photometry analysis are used to drive a forced-photometry analysis of
     361the warp images.  These analysis steps are discussed in detail by
     362\citet[][]{magnier2017.analysis}.  The data products from the camera,
     363stack, and forced-warp photometry analysis stages are ingested into
     364the internal calibration database (DVO, the Desktop Virtual
     365Observatory) and used for photometric and astrometric calibrations
     366\citet[see Section~\ref{sec:DVO} and][]{magnier2017.calibration}.
    354367
    355368\subsection{Data Access and Distribution}
     
    379392
    380393\begin{table*}
    381 \caption{GPC1 Database Schema Outline} %\vspace{-0.5cm}
     394\caption{GPC1 Database Schema Outline}
    382395\begin{center}
     396\footnotesize
    383397\begin{tabular}{llll}
    384398\hline
     
    386400{\bf Stage} & {\bf Primary Table} & {\bf Secondary Table(s)} & {\bf Key} \\% & {\bf Notes} \\
    387401%%D \begin{deluxetable}{llll}
    388 \footnotesize
    389402%%D   \tablecolumns{5}
    390403%%D   \tablewidth{0pc}
     
    741754The guess astrometry is used to match the reference catalog to the
    742755observed stellar positions in the focal plane coordinate system
    743 \citep[see][]{magnier2017.calibration}. 
     756\citep[see][]{magnier2017.calibration}.  Early on in the PS1SC
     757mission, the nightly processing (PV0) used a reference catalog based
     758on a combination of external catalogs (2MASS, Tycho, USNO).  Later,
     759reference catalogs based on Pan-STARRS data was used.  For the $3\pi$ PV3 analysis,
     760the reference catalog was based on Pan-STARRS data from the PV2
     761analysis \citep[see][for more details]{magnier2017.calibration}.
    744762
    745763Once an acceptable match is found, the astrometric calibration of the
    746764individual chips is performed, including a fit to a single model for
    747 the distortion introduced by the camera optics.  After the astrometic
    748 analysis is completed, the photometric calibration is determined using
    749 the final match to the reference catalog.  At this stage,
    750 pre-determined color terms may be included to convert the reference
    751 photometry to an appropriate photometric system.  For PS1, this is
    752 used to generate synthetic w-band photometry for areas where no
    753 PS1-based calibrated w-band photometry is available.  For more
    754 details, see \cite{magnier2017.calibration}.  The result of these
    755 calibrations is stored as a single multi-extension FITS table
    756 containing the results from each OTA as a separate extension.
     765the distortion introduced by the camera optics.  The astrometric model
     766includes a set of 3rd order polynomials for the transformations from the chip
     767coordinate system to the camera focal plane coordinate system and a
     768single additional 3rd order polynomial transformation from the camera focal
     769plane coordinate system to the tangent plane of a tangent projection.
     770For the $3\pi$ PV3 analysis, the typical astrometric residuals are in
     771the range of 20 - 30 milliarcseconds, sufficient to match observations
     772of the same objects between different exposures.  There are, however,
     773inevitable outliers.  Certain chips occasionally have systematically worse
     774astrometry, with OTA XY17 notably poor in this respect.
     775
     776After the astrometic analysis is completed, the photometric
     777calibration is determined using the final match to the reference
     778catalog.  A single photometric zero point is determined for each
     779exposure, with the airmass term fixed to the nominal linear slope for
     780each filter.  No color terms are measured between the observed
     781photometry and the reference photometry.  However, at this stage,
     782pre-determined color terms may be used to transform the reference
     783photometry to an appropriate photometric system.  For the PS1 nightly
     784processing, the reference catalog does not include \wps\ photometry,
     785so a fixed color transformation is used to generate synthetic w-band
     786photometry from the \rps\ \& \ips\ photometry.  For more details, see
     787\cite{magnier2017.calibration}.  The result of these calibrations is
     788stored as a single multi-extension FITS table containing the results
     789from each OTA as a separate extension.
    757790
    758791In addition to the astrometric and photometric calibrations, the
     
    842875\ippdbtable{warpSkyCellMap} table in the database, which contains a
    843876row for each skycell and OTA that overlap.  Each skycell may contain
    844 contributions from multiple OTAs.
     877contributions from multiple OTAs; since they are similar in size, in a
     878typical situation the warp is constructed from 4-6 neighboring OTAs.
    845879
    846880Once this mapping has been defined, jobs to warp the pixels onto each
     
    914948The \ippstage{stack} jobs pass the information about the input images
    915949and catalogs to the \ippprog{ppStack} program, which performs the
    916 image combinations.  See~\cite{waters2017} for details on the stack
    917 combination algorithm.  In addition to the standard image, mask, and
    918 variance produced at other stages, additional images are constructed
    919 with information about the contributions to each pixel.  A number
    920 image contains the number of input exposures used for each pixel,
    921 along with an exposure time map, and a weighted exposure time map that
    922 scales the exposure time based on the relative variance of each input.
    923 These images for the $3\pi$ analysis are currently available from the
    924 MAST image extraction tools at STSci.
     950image combinations.  Input warps are combined based on a weighting
     951defined by the median variance for each image; see~\cite{waters2017}
     952for details on the stack combination algorithm.  In addition to the
     953standard image, mask, and variance produced at other stages,
     954additional images are constructed with information about the
     955contributions to each pixel.  A number image contains the number of
     956input exposures used for each pixel, along with an exposure time map,
     957and a weighted exposure time map that scales the exposure time based
     958on the relative variance of each input.  These images for the $3\pi$
     959analysis are currently available from the MAST image extraction tools
     960at STScI.
    925961
    926962Upon completing the generation of these images, a row is added to the
     
    961997that faint, high-redshift quasars may be detected in \yps{} band only.
    962998Sources detected only in \yps{} band are therefore more likely to have
    963 a higher false-positive rate than the other stack sources.
     999a higher false-positive rate than the other stack sources.  The
     1000parameters of the PSF model are allowed to vary with position in the
     1001skycell.  The PSF model is also used to convolve the analytical galaxy
     1002models, which are the fitted to the observed flux distributions.
     1003Galaxy models include S\'ersic, DeVaucouleur, and Exponential
     1004profiles.
    9641005
    9651006The stack photometry output files consist of a set of FITS table
     
    10101051particularly severe for the Pan-STARRS $3\pi$ survey stacks due to the
    10111052combination of the substantial mask fraction of the individual input
    1012 exposures, the large instrinsic image quality variations within a
     1053exposures, the large intrinsic image quality variations within a
    10131054single exposure, and the wide range of image quality conditions under
    10141055which data were obtained and used to generate the $3\pi$ PV3 stacks.
     
    10521093skycell and filter as a single unit within the processing database,
    10531094while individual warps are processed individually in parallel as
    1054 separate processing jobs.
     1095separate processing jobs.  A separate PSF model is determine for each
     1096of the warp images so that the combined measurement is reliable.
    10551097
    10561098When processing is queued for this stage, an entry is added to the
     
    11071149analysis measurements into a single value.  The output catalogs listed
    11081150in the \ippdbtable{fullForceResult} table are passed to the
    1109 \ippprog{psphotFullForceSummary} to do this averaging.  When this
    1110 completes, an entry is added to the \ippdbtable{fullForceSummary}, and
    1111 the \ippdbtable{fullForceRun} entry is marked as completed.
     1151\ippprog{psphotFullForceSummary} to calculate the averages of the
     1152individual warp measurements, weighted by the signal-to-noise of the
     1153flux measurements.  When this analysis completes, an entry is added to
     1154the \ippdbtable{fullForceSummary}, and the \ippdbtable{fullForceRun}
     1155entry is marked as completed.
     1156
     1157% flux averaging takes place in psphotFullForceSummaryReadout.c:409
    11121158
    11131159\subsection{Difference Images}
     
    11151161
    11161162Two of the primary science drivers for the Pan-STARRS system are the
    1117 search hazardous asteroids and the search for Type Ia supernovae to
     1163search for hazardous asteroids and the search for Type Ia supernovae to
    11181164measure the history of the expansion of the universe.  Both of these
    11191165projects require the discovery of faint, transient source in the
     
    12141260% intro
    12151261The Pan-STARRS IPP uses an internal database system, distinct from the
    1216 publically visible database system, to determine the association
     1262publicly visible database system, to determine the association
    12171263between multiple detections of the same astronomical object and as
    12181264part of the astrometric and photometric calibration process.  This
     
    12281274astronomical objects; 2) measurements of those objects (from which the
    12291275average properties are derived); 3) properties of the images which
    1230 provided some or all of the measuements.  In addition, certain
     1276provided some or all of the measurements.  In addition, certain
    12311277metadata tables define general features of the database.
    12321278Table~\ref{tab:DVO_schema} lists the full collection of database
     
    12361282%illustrates the schematic relationship between these types of
    12371283%measurements.
    1238 
    1239 \begin{figure*}[htbp]
    1240   \begin{center}
    1241  \includegraphics[width=\hsize,clip]{skypartition.png}
    1242   \caption{\label{fig:sky.partition} Level 3 sky paritioning.  The
    1243     blue grid shows the outlines of the different regions assigned to
    1244     separate tables in the sky partitioning scheme.  The Galactic
    1245     plane is shown as a solid red line while the ecliptic is shown in
    1246     green.  This organization of the sky duplicates that used by the
    1247     HST Guide Star Catalog \citep{1988IAUS..133..239J}. 
    1248  }
    1249 \end{center}
    1250 \end{figure*}
    12511284
    12521285In the most basic implementation, a collection of measurements for
     
    13761409The \ippdbtable{Galphot} table stores the results of the forced galaxy
    13771410fitting analysis for each object that has been measured.  This table
    1378 contains one row per filter and model type (Sersic, Exponential, or
     1411contains one row per filter and model type (S\'ersic, Exponential, or
    13791412DeVaucouleur) if forced galaxy models have been calculate for the
    13801413object.
     
    15141547\label{sec:SkyPartition}
    15151548
     1549\begin{figure*}[htbp]
     1550  \begin{center}
     1551 \includegraphics[width=\hsize,clip]{skypartition.png}
     1552  \caption{\label{fig:sky.partition} Level 3 sky paritioning.  The
     1553    blue grid shows the outlines of the different regions assigned to
     1554    separate tables in the sky partitioning scheme.  The Galactic
     1555    plane is shown as a solid red line while the ecliptic is shown in
     1556    green.  This organization of the sky duplicates that used by the
     1557    HST Guide Star Catalog \citep{1988IAUS..133..239J}. 
     1558 }
     1559\end{center}
     1560\end{figure*}
     1561
    15161562Tables within DVO containing information about astronomical objects
    15171563are partitioned on the basis of position in the sky: objects within a
     
    15351581subdivides these declination bands in the RA direction, with spacing
    15361582related to the stellar density.  Level 3 divides these RA chunks into
    1537 4 - 8 smaller partitions.  This level exactly matches the HST GSC
    1538 layout, and uses the same naming convention to identify the
    1539 partitions: \code{n0000/0000}, etc. Level 4 further divides these
    1540 regions by a factor of 16.  In the \ippdbtable{SkyTable}, a region at
    1541 one level has a pointer to its parent region (the one which contains
    1542 it) and a sequence pointing to its children (regions it contains).
    1543 The \ippdbtable{SkyTable} enables fast lookups of the on-disk
    1544 partitions which map to a specific coordinate on the sky.  In general,
    1545 a single DVO will have the full sky represented with tables at a
    1546 single level, although it is possible for mixed levels to be used.
    1547 For the PV3 master database, the partitioning is at Level 4, resulting
    1548 in \approx 150,000 regions to cover the full sky, of which \approx
    1549 110,000 are used for the PV3 $3\pi$ data.  The densest portions of the
    1550 bulge contain at most \approx 300,000 astronomical objects in the
    1551 database files, with an associated maximum of \approx 30 million
    1552 measurements in these files.  With the compression scheme described
    1553 below, the largest database files are \approx 3GB, which can be loaded
    1554 into memory in 30 seconds on the processing machines that contain
    1555 partition data.
     15834 - 8 smaller partitions (see Figure~\ref{fig:sky.partition}).  This
     1584level exactly matches the HST GSC layout, and uses the same naming
     1585convention to identify the partitions: \code{n0000/0000}, etc. Level 4
     1586further divides these regions by a factor of 16.  In the
     1587\ippdbtable{SkyTable}, a region at one level has a pointer to its
     1588parent region (the one which contains it) and a sequence pointing to
     1589its children (regions it contains).  The \ippdbtable{SkyTable} enables
     1590fast lookups of the on-disk partitions which map to a specific
     1591coordinate on the sky.  In general, a single DVO will have the full
     1592sky represented with tables at a single level, although it is possible
     1593for mixed levels to be used.  For the PV3 master database, the
     1594partitioning is at Level 4, resulting in \approx 150,000 regions to
     1595cover the full sky, of which \approx 110,000 are used for the PV3
     1596$3\pi$ data.  The densest portions of the bulge contain at most
     1597\approx 300,000 astronomical objects in the database files, with an
     1598associated maximum of \approx 30 million measurements in these files.
     1599With the compression scheme described below, the largest database
     1600files are \approx 3GB, which can be loaded into memory in 30 seconds
     1601on the processing machines that contain partition data.
    15561602
    15571603% parallel partitions
     
    15611607and the location of the database partition on the disks of that
    15621608machine.  The \ippdbtable{SkyTable} contains elements to define by ID
    1563 the parition host to which a set of tables has been assigned.
     1609the partition host to which a set of tables has been assigned.
    15641610Operations which query the database, or perform other operations on
    15651611the database, are aware of the partitioning scheme and will launch
     
    15741620When the parallel partitioning for a DVO instance is defined, the
    15751621tables are randomly assigned to the partition hosts.  As a result,
    1576 queries which span more than a single parition are likely to spread
     1622queries which span more than a single partition are likely to spread
    15771623the I/O load across a large number of machines.  This parallelization
    15781624is critical to querying and manipulating the enormous database on a
     
    16051651The \ippdbtable{Measure} table, containing the detections of objects
    16061652from individual exposures or stack, or the (potentially
    1607 non-signficant) measurements from a warp, uses the 32-bit integer
     1653non-significant) measurements from a warp, uses the 32-bit integer
    16081654fields \ippdbcolumn{detID} and \ippdbcolumn{imageID} to uniquely
    16091655identify each entry.  The \ippdbcolumn{imageID} is the running
     
    16531699sequence within the image to form a single unique 64-bit integer value.
    16541700For detections from the stack images, the MJD is not unique, so a
    1655 different rubrick is used to define IDs for those detections.  The
     1701different rubric is used to define IDs for those detections.  The
    16561702field \ippdbcolumn{XstackDetectID} (where '\ippdbcolumn{X}' is one of
    16571703g,r,i,z,y) is constructed from the GPC1 stack ID
     
    16841730describe the location and size of the compressed data in the HEAP
    16851731section; the information about the uncompressed data is moved to a
    1686 keyword with ``Z'' prepended (e.g., ZFORM1) and an additional field is
     1732keyword with ``Z'' pre-pended (e.g., ZFORM1) and an additional field is
    16871733added to define the compression algorithm (e.g., ZCTYP1).  The column
    16881734names (e.g., TTYPE1) and units (e.g., TUNIT1) are retained in their
     
    16941740compressed in turn (this must be the same for all columns).
    16951741Additional header information is added to describe the block sizes and
    1696 infomation needed to describe the HEAP data section.  The compression
     1742information needed to describe the HEAP data section.  The compression
    16971743algorithms currently defined consist of the GZIP, RICE, PLIO, and
    16981744HCOMPRESS (REFS).  For GZIP, the compression algorithm may transpose
    16991745the byte order before compression: for floating point data of a
    1700 similiar dynamic range, this choice may allow for better compression
     1746similar dynamic range, this choice may allow for better compression
    17011747as each byte in the 4 or 8 byte floating point value is more likely to
    17021748be similar to the same byte in other rows than to the other bytes of
     
    17581804added to the \ippdbtable{addProcessedExp} table.
    17591805
    1760 After the master DVO is contructed containing the PS1 data, data from
     1806After the master DVO is constructed containing the PS1 data, data from
    17611807other sources are also added to the database.  For the PV3 DVO
    17621808database, data was added from 2MASS, WISE, Gaia DR1, and Tycho.  These
     
    18431889\label{sec:ipp2psps}
    18441890
    1845 The publically-visible Pan-STARRS database is hosted by the Space
     1891The publicly-visible Pan-STARRS database is hosted by the Space
    18461892Telescope Sciences Institute through their Mikulski Archive for Space
    1847 Telescopes (MAST).  The underying database at MAST is a copy of a
     1893Telescopes (MAST).  The underlying database at MAST is a copy of a
    18481894database generated at the IfA by the Published Science Products
    18491895Subsystem (PSPS).  The construction of the PSPS version of the PS1
     
    19401986Each task must at a minimum define a command to generate.  Commands
    19411987may be static or dynamic.  For a task with a static command, the
    1942 command is explicity defined in the task block (see code example in
     1988command is explicitly defined in the task block (see code example in
    19431989Figure~\ref{fig:task_example}) and is identical each time the task is
    19441990executed.  A dynamic command is defined within a special block of the
    1945 task, called \code{task.exec}.  This block is a snipet of code (in the
     1991task, called \code{task.exec}.  This block is a snippet of code (in the
    19461992\ippprog{opihi} language) which is run each time the task is executed.  The
    19471993\code{task.exec} code may refer to variables or other data structures
     
    19872033These options may be dynamically reset by the \code{task.exec} macro.
    19882034Some options control the number of jobs, such as limiting the number
    1989 of currently-outsanding jobs for a given task, or limiting the total
     2035of currently-outstanding jobs for a given task, or limiting the total
    19902036number generated.  Other options can be used to control the time when
    19912037jobs of a certain task are allowed to run.  It is also possible to
     
    20302076
    20312077When \ippprog{pcontrol} is provided with the name of a computer, it will attempt
    2032 to make an connection to that machine via ssh (or rsh?).  When a
     2078to make an connection to that machine via ssh.  When a
    20332079connection is made, the remote shell is used to run a special
    20342080interface program call \ippprog{pclient}.  This program accepts
     
    20362082individual commands in the local shell environment.  A single ssh
    20372083connection to a remote host keeps a single \ippprog{pclient} shell running for a
    2038 somewhat arbirarly long time, excuting many shell commands as needed.
     2084somewhat arbitrarly long time, executing many shell commands as needed.
    20392085This architecture avoids wasting overhead making the ssh connection to
    20402086the remote machine each time a command is executed, allowing for rapid
    2041 excution of many commands.  As a result, a single job within the IPP
     2087execution of many commands.  As a result, a single job within the IPP
    20422088architecture is allowed to be very light and short running if needed.
    20432089
     
    22552301ippstage{diff} analysis stage.
    22562302
    2257 Once observations have been completed for the night (signalled by the
     2303Once observations have been completed for the night (signaled by the
    22582304end-of-night dark exposures that are taken each morning after the
    22592305telescope closes), and the script has generated all \ippstage{diff}
     
    23212367challenge of storing and managing the large volume of data that is
    23222368generated by the GPC1 camera.  The \ippprog{Nebulous} system was
    2323 designed to aid in thie process.  \ippprog{Nebulous} is not a file
     2369designed to aid in this process.  \ippprog{Nebulous} is not a file
    23242370system per-se, but only a method of tracking the locations of files
    23252371within the file system, and of tracking duplicate copies of the same
     
    26222668unification of configuration options between config files on disk and
    26232669the options specified on the command line is handled by
    2624 \ippmisc{psModules} functions, as is the contruction of data
     2670\ippmisc{psModules} functions, as is the construction of data
    26252671structures in memory to represent the astronomical camera based on the
    26262672header information in the input file.  The functions to generate and
     
    27922838to hang until the job time limit is reached.  These stacks were
    27932839instead processed on the regular IPP cluster, where hosts with
    2794 sufficent memory were available.
     2840sufficient memory were available.
    27952841
    27962842\subsection{UH Cray Cluster}
     
    28202866994,890 runs processed there.
    28212867
    2822 \section{Discussion}
    2823 \label{sec:discussion}
     2868\section{Conclusion}
     2869
     2870Since the Pan-STARRS\,1 telescope first came online in 2007, this
     2871telescope has obtained 1.43 million exposures with GPC1, amounting to
     2872a raw data volume of 4.32 petabytes.  The Pan-STARRS Image Processing
     2873Pipeline has archived and processed these images on-the-fly to produce
     2874discoveries of transient events and hazardous asteroids in real-time.
     2875The IPP has been used to perform several re-processings of large
     2876fractions of the science exposures to produce a well-calibrated data
     2877release of the $3\pi$ Survey data.  To date, and including repeated
     2878analysis, the IPP has processed 2.1 million exposures, detecting 900 billion
     2879sources in those exposures (real and otherwise!).  The Pan-STARRS data
     2880processing system represents a real example of astronomy data
     2881processing on the very large scale.
    28242882
    28252883\acknowledgments
Note: See TracChangeset for help on using the changeset viewer.