IPP Software Navigation Tools IPP Links Communication Pan-STARRS Links

Changeset 41347


Ignore:
Timestamp:
Apr 24, 2020, 4:01:26 PM (6 years ago)
Author:
eugene
Message:

update the figures to fix a PDF bug

Location:
trunk/doc/release.2015/ps1.analysis
Files:
3 added
13 edited

Legend:

Unmodified
Added
Removed

  • trunk/doc/release.2015/ps1.analysis/analysis.tex

    r41333 r41347  
    4545\def\Princeton{2}
    4646\def\DUR{3}
    47 \def\MPIA{4}
     47\def\MPE{4}
    4848\def\CfA{5}
    4949
    … …  
    6161L. Denneau,\altaffilmark{\IfA}
    6262P.~W. Draper,\altaffilmark{\DUR}
    63 D. Farrow,\altaffilmark{\DUR,\MPIA}
     63D. Farrow,\altaffilmark{\DUR,\MPE}
    6464R. Jedicke,\altaffilmark{\IfA}
    6565K. W. Hodapp,\altaffilmark{\IfA}
    … …  
    8888% \altaffiltext{\USNO}{US Naval Observatory, Flagstaff Station, Flagstaff, AZ 86001, USA}
    8989% \altaffiltext{\JHU}{Department of Physics and Astronomy, Johns Hopkins University, 3400 North Charles Street, Baltimore, MD 21218, USA}
    90 \altaffiltext{\MPIA}{Max Planck Institute for Astronomy, K\"onigstuhl 17, D-69117 Heidelberg, Germany}
     90% \altaffiltext{\MPIA}{Max Planck Institute for Astronomy, K\"onigstuhl 17, D-69117 Heidelberg, Germany}
     91\altaffiltext{\MPE}{Max-Planck-Institut f\"ur extraterrestrische Physik, Giessenbachstrasse 1, 85748 Garching, Germany}
    9192\begin{abstract}
    9293
    … …  
    207208the analysis parameters to better suite the longer exposures.  This
    208209program as well as the rest of the Pan-STARRS Image Processing
    209 Pipeline (IPP) software suite is available for download from \url{http:ipp.ifa.hawaii.edu}}.
    210 
    211 \note{Generate a tarball of just the programs (skip certain directories)}
     210Pipeline (IPP) software suite is available for download from \url{http://ipp.ifa.hawaii.edu}}.
     211
     212% \note{Generate a tarball of just the programs (skip certain directories)}
    212213
    213214%Chambers et al. 2017 (Paper I)
    … …  
    538539  individual astrometric measurements is 16 milliarcseconds and the
    539540  Pan-STARRS Data Release 2 (DR2) astrometric system is tied to the
    540   Gaia DR1 coordinate frame with a systematic uncertainty of $\sim 5$
    541   milliarcseconds. }
     541  Gaia DR1 \citep{2016AA...595A...4L} coordinate frame with a
     542  systematic uncertainty of $\sim 5$ milliarcseconds. }
    542543
    543544\section{Basic Analysis}
    … …  
    14731474follow some of the observed PSF variations in the images.
    14741475
     1476Figure~\ref{fig:iq.exposure} illustrates the 2D variations in the PSF
     1477shapes seen in PS1 data.  This figure shows the FWHM, $e_1$, and $e_2$
     1478polarizations of the stars as a function of position in 4 exposures.
     1479For images with good image quality, variations of the PSF shape due to
     1480the optical aberrations can be see.  The optical aberrations vary as
     1481the active collimation and alignment are adjusted and as the focus
     1482changes.  These aberrations are coupled to the piston of the chips,
     1483which have been adjusted to crudely follow the focal surface
     1484\citep{chambers2017}.  During regular operations, image with large
     1485PSFs are usually caused by the atmosphere (seeing) or by telescope
     1486tracking errors, both of which result in common shapes across the
     1487field of the camera.  In the figure, the top panel shows the
     1488circularization of the PSF due to the atmosphere washes out the
     1489lower-level variations caused by the optics.
     1490
     1491Examples of 2D PSF variations.
     1492    Each row represents an exposure.  The left-most column shows the
     1493    distribution of FWHM across the camera; the median value in
     1494    arcseconds is given in the inset.  The middle column gives the
     1495    $e_1$ polarization measured from second moments (see
     1496    Section~\ref{sec:lensing.params} while the right column gives the
     1497    $e_2$ polarization.
     1498
    14751499% \note{write up the fitting process to define the grid?}
    14761500
    … …  
    14981522quality of the PSF fits.
    14991523
     1524% Figure 3: ** repaired PDF text **
     1525% pueo:psphot.iq.20200413/mana.sh : show.e12 (iq.exposures.pdf)
     1526\begin{figure}[htbp]
     1527  \begin{center}
     1528  \includegraphics[width=\hsize]{{\picdir/iq.exposures}.\plotext}
     1529  \caption{\label{fig:iq.exposure} Examples of 2D PSF variations.
     1530    Each row represents an exposure.  The left-most column shows the
     1531    distribution of FWHM across the camera; the median value in
     1532    arcseconds is given in the inset.  The middle column gives the
     1533    $e_1$ polarization measured from second moments (see
     1534    Section~\ref{sec:lensing.params}) while the right column gives the
     1535    $e_2$ polarization. }
     1536  \end{center}
     1537\end{figure}
     1538
    15001539For the PS1 GPC1 analysis, we used the \code{PS1_V1} model, which we
    15011540found by experimentation to match well to the observed profiles
    … …  
    15151554% buonanno : 1983A&AS...51...83B
    15161555
     1556% Figure 4:
    15171557% /data/kukui.3/eugene/psphot.20161214/mana.sh
    15181558\begin{figure}[htbp]
    … …  
    15751615\subsubsection{Candidate PSF Source Selection}
    15761616\label{sec:psf.source.selection}
     1617
     1618% Figure 5:
     1619% /data/kukui.3/eugene/psphot.20161214/mana.sh
     1620\begin{figure}[htbp]
     1621  \begin{center}
     1622  \includegraphics[width=\hsize]{{\picdir/moment.class}.\plotext}
     1623  \caption{\label{fig:moment.class} Illustration of PSF star selection
     1624    using the second moments. \textadd{Each point represents the
     1625      second moments in the $X_{\rm ccd}$ and $Y_{\rm ccd}$ directions
     1626      for sources measured in one chip (XY32) from a particular PS\,1
     1627      exposure (o6065g0428o)}.  The dominant clump is located in this
     1628    diagram \textadd{to identify the stars.}  Galaxies tend to have a range of
     1629    sizes and thus spread out above the stars.  Cosmic rays also have
     1630    a range of sizes, with one dimension smaller than the PSF.  The
     1631    red circle represents the PSF star candidates. }
     1632  \end{center}
     1633\end{figure}
    15771634
    15781635The first stage of determining the PSF model for an image is to
    … …  
    16181675most additional analyses and are marked with the flag bit
    16191676\code{PM_SOURCE_MODE_SATURATED}.
    1620 
    1621 % /data/kukui.3/eugene/psphot.20161214/mana.sh
    1622 \begin{figure}[htbp]
    1623   \begin{center}
    1624   \includegraphics[width=\hsize]{{\picdir/moment.class}.\plotext}
    1625   \caption{\label{fig:moment.class} Illustration of PSF star selection
    1626     using the second moments. \textadd{Each point represents the
    1627       second moments in the $X_{\rm ccd}$ and $Y_{\rm ccd}$ directions
    1628       for sources measured in one chip (XY32) from a particular PS\,1
    1629       exposure (o6065g0428o)}.  The dominant clump is located in this
    1630     diagram \textadd{to identify the stars.}  Galaxies tend to have a range of
    1631     sizes and thus spread out above the stars.  Cosmic rays also have
    1632     a range of sizes, with one dimension smaller than the PSF.  The
    1633     red circle represents the PSF star candidates. }
    1634   \end{center}
    1635 \end{figure}
    16361677
    16371678\subsubsection{Candidate PSF Source Model Fits}
    … …  
    20392080\code{PM_SOURCE_MODE_EXT_LIMIT} is set for the source.
    20402081
    2041 \textmod{We decided to use $\delta M_{\rm KP}$ metric for this
     2082\textmod{We decided to use the $\delta M_{\rm KP}$ metric for this
    20422083  assessment after we tested several possible star-galaxy separation
    20432084  statistics.  We found that the Kron-PSF comparison was more reliable
    … …  
    20842125% apScale = 4.5
    20852126
    2086 Once a PSF model has been selected for an image, \ippprog{psphot}
    2087 attempts to fit all of the detected sources, with signal-to-noise
    2088 ratio greater than a user-defined limit, with the PSF model.  In the
    2089 PV3 analysis of the $3\pi$ survey data, this limit was set to a
    2090 signal-to-noise ratio of 20.0 for all analysis stages.  In these fits,
    2091 the dependent parameters are fixed by the PSF model and only \textmod{the 3
    2092 independent source model parameters (position in $X$ and $Y$ and flux
    2093 normalization) are allowed to vary in the fit.  Note that we do {\em
    2094   not} allow the local sky to be fitted as a free parameters.  Since
    2095 we have subtracted a model for the background, allowing the sky to be
    2096 again at this stage is redundant.  In fact, in our testing, we found
    2097 that allowing the sky to float resulted in higher scatter for the flux
    2098 normalizations.}  \ippprog{psphot} again uses Levenberg-Marquardt
    2099 minimization for the non-linear fitting.  The sources are fitted in
    2100 their S/N order, starting with the brightest and working down to the
    2101 user-specified limit, with the other sources subtracted as discussed
    2102 above.  All sources for which a non-linear PSF model has been
    2103 attempted have the flag bit \code{PM_SOURCE_MODE_FITTED} set,
    2104 regardless of the quality of that fit.
     2127\textadd{At this point, we have a PSF model for the image, we have an
     2128assessment of the size (PSF-like, extended, or cosmic-ray) for each
     2129object, and we have fitted the PSF model for the normalization to each
     2130source (Section~\ref{sec:ensemble.fitting}).  However, the positions
     2131for the sources have been fixed to the position determined from the
     2132peak detection stage (Section~\ref{sec:peaks}) or the centroid from
     2133the analysis of the moments (Section~\ref{sec:moments}).  A better
     2134position, and thus a better normalization, can be determined by
     2135simultaneously fitting for all three parameters.  We therefore go
     2136through the image and re-fit the PSF model to each source
     2137one-at-a-time with all other sources subtracted based on the earlier
     2138fit.}
     2139
     2140\textmod{This re-fitting analysis is performed for all of the sources
     2141  with signal-to-noise ratio greater than a user-defined limit.  In
     2142  the PV3 analysis of the $3\pi$ survey data, this limit was set to a
     2143  signal-to-noise ratio of 20.0 for the \ippstage{chip} and
     2144  \ippstage{stack} analysis stages.  In these fits, the dependent
     2145  parameters are fixed by the PSF model and only the 3 independent
     2146  source model parameters (position in $X$ and $Y$ and flux
     2147  normalization) are allowed to vary in the fit.  Note that we do {\em
     2148    not} allow the local sky to be fitted as a free parameters.  Since
     2149  we have subtracted a model for the background, allowing the sky to
     2150  be again at this stage is redundant.  In fact, in our testing, we
     2151  found that allowing the sky to float resulted in higher scatter for
     2152  the flux normalizations.  For the non-linear fitting,
     2153\ippprog{psphot} again uses the Levenberg-Marquardt technique.}  The
     2154sources are fitted in their S/N order, starting with the brightest and
     2155working down to the user-specified limit, with the other sources
     2156subtracted as discussed above.  All sources for which a non-linear PSF
     2157model has been attempted have the flag bit
     2158\code{PM_SOURCE_MODE_FITTED} set, regardless of the quality of that
     2159fit.
    21052160
    21062161Since the PSF model describes the variation of the PSF across the
    … …  
    21212176Section~\ref{sec:moments}).  For the PV3 $3\pi$ analysis, the PSF fit
    21222177window radius is $7 \times \sigma_w$.
    2123 
    2124 Sources which are blended with other sources may be fitted together as a
    2125 set of PSFs.  Blended objects are identified by first searching for
    2126 objects for which the PSF fit windows overlap.  For a group of such
    2127 neighboring objects, a contour is determined in the flux image at
    2128 $25\%$ of the peak of the brightest source in the group.  All objects
    2129 lying within this contour are treated as blends of this brightest
    2130 source.  If other objects in this group exist, the brightest object
    2131 not already assigned to a blend is used to define the contour for
    2132 blends of this next object.  All objects in the image are tested as
    2133 possible blends.  A single multi-source fit is performed on each group
    2134 of blended peaks.  Sources which are identified as members of a
    2135 blended group have the flag bit \code{PM_SOURCE_MODE_BLEND} set, while
    2136 those for which a blended PSF fit succeeds have the flag bit
    2137 \code{PM_SOURCE_MODE_BLEND_FIT} set.  {\em Note that for DR1 \& DR2,
    2138   this option was not used because it tended to prevent galaxies from
    2139   being fitted as extended objects; the rules for identifying blended
    2140   stars and galaxies will be revisited in future re-analyses.}
    21412178
    21422179%% Once a solution has been achieved for a source, \ippprog{psphot} attempts to
    … …  
    22262263represented and may have larger residual significance.
    22272264
    2228 For sources in groups of blended stars, the resulting fits are
    2229 evaluated independently.  Any which are determined to be valid PSF
    2230 fits are subtracted from the image and kept for future analysis.
    2231 
    2232 \subsubsection{Double and Extended Sources}
    2233 
    2234 Sources which are judged to be non-PSF-like are confronted with two
    2235 possible alternative choices.  First, the source is fitted with a
    2236 double-source model.  In this pass, the assumption is made that there
    2237 are two neighboring sources, but the peaks are not resolved.  The
    2238 initial guess for the two peaks is made by splitting the flux of the
    2239 single source in half and locating the two starting peaks at +/- 2
    2240 pixels from the original peak along the direction of the semi-major
    2241 axis of the sources, as measured from the second moments.  In order
    2242 for the two-source model to be accepted, both sources must be judged
    2243 as a valid PSF source.  Otherwise, the double-PSF model is rejected
    2244 and the source is fitted with the available non-PSF model or models.
    2245 Sources for which a double-PSF model is fitted have the flag bit
    2246 \code{PM_SOURCE_MODE_PAIR} set.
     2265\subsubsection{Double and Blended Sources}
     2266
     2267\textmod{In fields with high stellar density, the non-linear source fitting can
     2268be adversely affected by close neighbors.  We implemented two
     2269modifications of the non-linear fitting code to address this issue for
     2270different scales to the nearby neighbors.  One version addresses the
     2271case of nearby sources which are separately detected in the
     2272peak-detection stage; the other version of the analysis attempts to
     2273fit a pair of PSFs for sources which are apparently extended.  {\em
     2274  Note that for DR1 \& DR2, neither of these options were used because
     2275  they tended to prevent galaxies from being fitted as extended
     2276  objects; these rules for distinguishing blended stars and galaxies will
     2277  be revisited in future re-analyses.}  We outline the strategy below
     2278although it was not used for these data releases.}
     2279
     2280{\TEXTADD Sources which are blended with other sources may be fitted together as
     2281a set of PSFs.  Blended objects are identified by first searching for
     2282objects for which the PSF fit windows overlap.  For a group of such
     2283neighboring objects, a contour is determined in the flux image at
     2284$25\%$ of the peak of the brightest source in the group.  All objects
     2285lying within this contour are treated as blends of this brightest
     2286source.  If other objects in this group exist, the brightest object
     2287not already assigned to a blend is used to define the contour for
     2288blends of this next object.  All objects in the image are tested as
     2289possible blends.  A single multi-source fit is performed on each group
     2290of blended peaks.  Sources which are identified as members of a
     2291blended group have the flag bit \code{PM_SOURCE_MODE_BLEND} set, while
     2292those for which a blended PSF fit succeeds have the flag bit
     2293\code{PM_SOURCE_MODE_BLEND_FIT} set.  For sources in groups of blended
     2294stars, the resulting fits are evaluated independently.  Any which are
     2295determined to be valid PSF fits are subtracted from the image and kept
     2296for future analysis.}
     2297
     2298{\TEXTADD Sources which are judged to be non-PSF-like are confronted with two
     2299possible alternative choices: double-star or extended source model
     2300(see next section).  For the double-star model, the assumption is made
     2301that there are two neighboring PSF-like sources, but the peaks are not
     2302resolved.  The initial guess for the two peaks is made by splitting
     2303the flux of the single source in half and locating the two starting
     2304peaks at +/- 2 pixels from the original peak along the direction of
     2305the semi-major axis of the sources, as measured from the second
     2306moments.  In order for the two-source model to be accepted, both
     2307sources must be judged as a valid PSF source.  Otherwise, the
     2308double-PSF model is rejected and the source is fitted with the
     2309available non-PSF model or models.  Sources for which a double-PSF
     2310model is fitted have the flag bit \code{PM_SOURCE_MODE_PAIR} set. }
    22472311
    22482312\subsubsection{Non-PSF Sources}
    22492313\label{sec:nonlinear.galaxy.model}
    22502314
    2251 Once every source (above the S/N cutoff) has been confronted with the
     2315\textmod{Once every source (above the S/N cutoff) has been confronted with the
    22522316PSF model, the sources which are thought to be extended (resolved) can
    22532317now be fit with an appropriate model (e.g., galaxy profile or other
    2254 likely extended shapes).  Again, the fitting stage starts with the
     2318likely extended shapes).}  Again, the fitting stage starts with the
    22552319brightest sources (as judged by the rough S/N measured from the
    22562320moments aperture) and working to a user defined S/N limit.
    22572321
    2258 \ippprog{psphot} will use the user-selected extended source model to
    2259 attempt these fits.  In the configuration system, the keyword
    2260 \code{EXT_MODEL} is set to the model of interest.  All suspected
    2261 extended sources are fitted with the model, allowing all of the
    2262 parameters to float.  The initial parameter guesses are critical here
    2263 to achieving convergence on the model fits in a reasonable time.  The
    2264 moments and the pixel flux distribution are used to make the initial
    2265 parameter guess.  Many of the source parameters can be accurately
    2266 guessed from the first and second moments.  The power-law slope can be
    2267 guessed by measuring the isophotal level at two elliptical radii and
    2268 comparing the ratio to that expected.
     2322{\TEXTADD The choice of extended source model or models is set by the user for a given
     2323analysis.  In the configuration system, the keyword \code{EXT_MODEL}
     2324is set to the model of interest.}  All suspected extended sources are
     2325fitted with the model, allowing all of the parameters to float.  The
     2326initial parameter guesses are critical here to achieving convergence
     2327on the model fits in a reasonable time.  The moments and the pixel
     2328flux distribution are used to make the initial parameter guess.  Many
     2329of the source parameters can be accurately guessed from the first and
     2330second moments.  The power-law slope can be guessed by measuring the
     2331isophotal level at two elliptical radii and comparing the ratio to
     2332that expected.
    22692333
    22702334For each type of extended source model (in fact for all source
    … …  
    23032367\subsection{Faint Source Analysis}
    23042368\label{sec:faint.psf.model}
    2305 
    2306 % pueo:/home/real/eugene/ppsim.20200407
    2307 \begin{figure}[htbp]
    2308   \begin{center}
    2309  \includegraphics[width=\hsize,clip]{\picdir/{completion.ppsim}.pdf}
    2310   \caption{\label{fig:complete.ppsim} Completeness as a function of
    2311     magnitude (blue curves) for different stellar densities in
    2312     simulated data.  The curves are labeled with the logarithm of the
    2313     stellar density at the detection threshold of the low-density
    2314     image.  The dotted red line shows the detection limit expected for
    2315     the sky level and seeing.  The solid red curve shows the
    2316     completeness estimated for the low-density image based on
    2317     injection and recovery.}
    2318   \end{center}
    2319 \end{figure}
    2320 
    2321 % pueo:/home/real/eugene/ppsim.20200407
    2322 \begin{figure}[htbp]
    2323   \begin{center}
    2324  \includegraphics[width=\hsize,clip]{\picdir/{psphot.complete.pv3}.pdf}
    2325   \caption{\label{fig:complete.pv3} Completeness and bogus fraction
    2326     as a function of magnitude for different stellar densities in real
    2327     PS1 exposures.  Each panel represents an exposure at different
    2328     Galactic latitudes towards anti-center, labeled by the density of
    2329     stars at the detection limit of the low-density exposure.  In each
    2330     panel, the completeness (compared to deep stack data) and fraction
    2331     of false detections (bogus fraction) is shown for a series of
    2332     cuts.  The gold curves show all detections in the exposures.  The
    2333     dotted black curve shows the impact of cutting detections
    2334     identified by {\tt psphot} as cosmic rays.  The blue curve
    2335     excludes cosmic rays and detections with {\tt PSF\_QF} $< 0.95$
    2336     while the red curve excludes cosmic rays and detections with {\tt
    2337       PSF\_QF\_PERFECT} $< 0.95$.}
    2338   \end{center}
    2339 \end{figure}
    23402369
    23412370After a first pass through the image, in which the brighter sources
    … …  
    24372466actual source flux.
    24382467
    2439 % /data/kukui.1/eugene/psphot.examples.20190423/compare.sh
     2468Aperture photometry attempts to avoid these problems, but introduces
     2469other difficulties.  In aperture photometry, if a large enough
     2470aperture is chosen, the amount of flux which is lost will be a small
     2471fraction of the total source flux.  Even more importantly, as the
     2472image conditions change, the amount lost will change by an even
     2473smaller fraction, at least for a large aperture. 
     2474%
     2475% This can be seen by
     2476% the fact that the dominant variations in the image quality are in the
     2477% focus, tracking and seeing.  All of these errors initially affect the
     2478% cores of the stellar images, rather than the wide wings.  The wide
     2479% wings are largely dominated by scattering in the optics and scattering
     2480% in the atmosphere.  The amplitude and distribution of these two
     2481% scattering functions do not change significantly or quickly for a
     2482% single telescope and site. 
     2483%
     2484Aperture photometry can then be used to
     2485correct the PSF photometry.
     2486
     2487The difficulty for aperture photometry is the need to make an accurate
     2488measurement of the local background for each source.  As the aperture
     2489grows, errors in the measurement of the sky flux start to become
     2490dominant.  If the aperture is too small, then variations in the image
     2491quality are dominant.  The brighter is the source, the smaller is the
     2492error introduced by the large size of the aperture.  However, the
     2493number of very bright stars is limited in any image, and of course the
     2494brighter stars are more likely to suffer from non-linearity or
     2495saturation. 
     2496
     2497In order to thread the needle between these effects, \ippprog{psphot}
     2498measures the aperture photometry on a modest-sized aperture, and then
     2499uses the PSF model to extrapolate to a large aperture.  When the PSF
     2500fluxes are calculated, the aperture flux for the modest-sized aperture
     2501is also determined.  The aperture is a circular aperture with radius
     2502set to a fixed multiple (\code{PSF_APERTURE_SCALE}) of $\sigma_w$, the
     2503width of the Gaussian window function determined based on the analysis
     2504of the second moments (see Section~\ref{sec:moments}).  For the PV3
     2505$3\pi$ analysis, the aperture window radius is $4.5 \times \sigma_w$,
     2506while the large reference aperture radius is set to 25 pixels
     2507(\code{PSF_REF_RADIUS} = 6\farcs4).  These corrected aperture
     2508magnitudes are saved in the output catalogs as \code{AP_MAG}, the
     2509uncorrected aperture magnitudes are saved as \code{AP_MAG_RAW}, and
     2510the radius used to measure the raw aperture flux is saved as
     2511\code{AP_MAG_RADIUS}.  The corresponding flux and the flux error are
     2512saved as \code{AP_FLUX} and \code{AP_FLUX_SIG}.
     2513
     2514% Figure 6:  ** repaired PDF text **
     2515% pueo:/home/real/eugene/ppsim.20200407/tap_psphot_deteff.pro : all.complete
     2516\begin{figure}[htbp]
     2517  \begin{center}
     2518 \includegraphics[width=\hsize,clip]{\picdir/{completion.ppsim}.pdf}
     2519  \caption{\label{fig:complete.ppsim} Completeness as a function of
     2520    magnitude (blue curves) for different stellar densities in
     2521    simulated data.  The curves are labeled with the logarithm of the
     2522    stellar density at the detection threshold of the low-density
     2523    image.  The dotted red line shows the detection limit expected for
     2524    the sky level and seeing.  The solid red curve shows the
     2525    completeness estimated for the low-density image based on
     2526    injection and recovery.}
     2527  \end{center}
     2528\end{figure}
     2529
     2530With these aperture magnitudes in hand, it is now possible to make an
     2531average correction to the PSF magnitudes to bring the PSF and aperture
     2532magnitudes to the same system.  This correction is measured using the
     2533same stars from which the PSF model is measured, as long as the PSF
     2534magnitude error for the star is less than 0.03 mag.  The correction is
     2535calculated using the weighted average of the values $m_{\rm AP} -
     2536m_{\rm PSF}$.  Since the PSF may vary across the image, the correction
     2537is determined as a function of position in the image.  Like the PSF
     2538model, the 2D variations of the aperture correction may be modeled as
     2539a polynomial or via interpolation in a grid.  For the $3\pi$ PV3
     2540analysis, a grid with a maximum of $6\times 6$ samples per GPC1 chip
     2541image was used.  The reported PSF magnitudes for all objects have this
     2542aperture correction applied.  \textadd{Note that an initial aperture correction was
     2543measured during the initial steps of the analysis before the PSF model
     2544was chosen.  However, since the sources in the image were not yet
     2545measured and subtracted, that aperture could be contaminated by
     2546neighbors.  The analysis here is performed one fairly bright star at a
     2547time with all other sources subtracted in order to minimize such contamination.}
     2548
     2549% growth curve analysis in psphot:
     2550% in psphotChoosePSF : call psphotMakeGrowthCurve
     2551% in psphotMakeGrowthCurve : boolean GROWTH_FROM_SOURCES, use
     2552%% pmGrowthCurveGenerateFromSources or
     2553%% pmGrowthCurveGenerate (uses PSF model only)
     2554%% GROWTH_FROM_SOURCES is set to TRUE for default recipe
     2555
     2556%% ApTrend:
     2557%% in psphotApResid, called by psphotReadout near the end of the
     2558%% analysis
     2559%% ApTrend = f(x,y) for (apMag - psfMag) for psfMagErr <= 0.03
     2560%% apMag is growth curve corrected
     2561%% psfMag is raw
     2562
     2563%% raw psfMag and raw apMag are measured
     2564%% apMag = apMagRaw + growth curve correction (from apRadius to 25 pix
     2565%% = PSF_REF_RADIUS)
     2566%% psfMag = psfMagRaw + aperture trend (<ap - psf> + growth curve)
     2567
     2568% How important is this effect?  Consider a typical bright source with a
     2569% flux of (say) 40,000 counts in an image of background 1000 counts per
     2570% pixel, with FWHM of 4 pixels.  In principle, the flux of this source
     2571% should be measurable with an accuracy of roughly 0.57\%
     2572% ($\frac{\sqrt{40000 + 1000 \times 12}}{40000}$).  However, the
     2573% measurement of the sky is limited at some finite level by Poisson
     2574% statistics.  If we are required to use an aperture of (say) 25 pixels
     2575% in radius (eg, 5 arcseconds for an 0.2 arcsec / pixel detector), and
     2576% we have an annulus of twice this radius to measure the local sky, then
     2577% we will have an error of XXX.
     2578%
     2579% \note{outline the variation of {\em ApResid} as a function of
     2580% magnitude}.
     2581
     2582%%% \ippprog{psphot} measures the aperture correction ({\em ApResid}) for every PSF
     2583%%% candidate source, then calculates the trend of this correction as a
     2584%%% function of the magnitude.  This trend is fitted with a line.  The
     2585%%% resulting function can be used to determine the effective aperture
     2586%%% correction for an infinite flux source and the average bias inherent
     2587%%% in the sky measurement for the image.  The scatter of the
     2588%%% PSF-candidate source measurements about this trend is a measure of how
     2589%%% well we can measure photometry from the image by applying the specific
     2590%%% PSF model.  The slope of this trend is a measure of the bias in the
     2591%%% local sky measurment for each source.  In principal, the measured sky
     2592%%% levels could be modified by this bias.  More generally, the measured
     2593%%% bias in a collection of images could be used to improve the model
     2594%%% fitting or sky fitting portion of the software the remove the bias
     2595%%% term.
     2596
     2597\subsection{Completeness \& Contamination}
     2598
     2599{\TEXTADD At the end of the \ippprog{psphot} analysis of the sources in the
     2600image, an analysis is performed to test the detection efficiency.  A
     2601number of fake PSF sources are injected into the image and the peak
     2602detection analysis (Section~\ref{sec:peaks}) is use to determine if
     2603these sources would have been recovered.  The PSF model fluxes are
     2604measured for the source which are detected.  For a given image, the
     2605detection threshold is predicted based on the median image variance
     2606and the seeing.  A series of brightness bins straddling the threshold
     2607are defined and a number of sources are injected with magnitudes
     2608corresponding to each of these bin values.  The \ippprog{psphot}
     2609recipe value \code{EFF.NUM} specifies the number of sources in each
     2610brightness bin (500 the PV3), and the value \code{@EFF.MAG} specifies
     2611the bins as magnitudes above and below the threshold.  For PV3, the 13
     2612magnitude offsets were (-2.0, -1.0, -0.5, -0.25, -0.1, -0.05, 0.0,
     26130.05, 0.1, 0.25, 0.5, 1.0, 2.0), providing fine sampling near the
     2614limit, but more coarse coverage further away.  Poisson noise
     2615appropriate to the photon counts of the injected sources are included
     2616in the image.  Injected sources are uniformly distributed across the
     2617image in $X$ and $Y$ pixel coordinates {\em without any consideration
     2618  of the masked regions}.  This last point means the recovered
     2619fraction in the bright bins can be used to test the masking fraction.}
     2620
     2621{\TEXTADD As the stellar density increases, the completeness suffers due to
     2622crowding and confusion.  Since the injection and recovery analysis of
     2623the fake sources operates on the source-subtracted image and does not
     2624attempt to fully discovery the sources, this analysis over-estimates
     2625the completeness in crowded fields.  To explore the completeness in
     2626crowded field images, we generate a series of simulated images using a
     2627Gaussian PSF with FWHM = 1\arcsec\ for a range of stellar densities.
     2628We generate fake stars with fluxes as faint as $\frac{1}{5}$ of the
     2629flux as the low-density detection limit, with densities ranging from
     2630\approx 14,000 stars per square degree at low-density detection limit
     2631to \approx 4.8 million stars per square degree at the low-density
     2632detection limit.  The latter is comparable to observed densities in
     2633the Galactic plane.  We run the \ippprog{psphot} analysis on these
     2634simulated images and compare the detected stars to those injected to
     2635calculate the completeness for each image as a function of the true
     2636magnitude of the stars.  Figure~\ref{fig:complete.ppsim} shows the measured
     2637completeness for each of the six simulated images, labeled by the
     2638logarithm of their faint-end stellar density. The red dashed line
     2639shows the expected detection limit based on the background and seeing,
     2640while the red curve shows the completeness curve calculated
     2641automatically by \ippprog{psphot} using the injection and recovery
     2642analysis.}
     2643
     2644{\TEXTADD For low-density fields, the completeness function determined by
     2645injection and recovery is similar to that measured by the simulation,
     2646with the 50\% completeness threshold roughly 0.3 magnitudes too faint.
     2647As the stellar density increases, the true 50\% completeness magnitude
     2648rises relative to the value estimated by injection and recovery.}
     2649
     2650{\TEXTADD Ideally, all sources detected by \ippprog{psphot} would correspond to
     2651real astrophysical objects.  In reality, many sources are detected in
     2652the images which do not correspond to real sources in the sky.  In the
     2653very simplified simulations discussed above, which do not include
     2654realistic detector artifacts, we find that the fraction of bogus
     2655detections is extremely low, even at the very faint end.  In real
     2656data, bogus detections are due to a variety of typical instrumental
     2657features including cosmic rays, diffraction spikes, satelite tracks,
     2658glows, non-Gaussian noise, variance mis-estimation, etc.  See paper III
     2659for extensive discussion of instrumental artifacts in the Pan-STARRS images.}
     2660
     2661% Figure 7: ** repaired PDF text **
     2662% pueo:/home/real/eugene/psphot.complete.20200407/complete.sh : full.figure.all
     2663\begin{figure}[htbp]
     2664  \begin{center}
     2665 \includegraphics[width=\hsize,clip]{\picdir/{psphot.complete.pv3}.pdf}
     2666  \caption{\label{fig:complete.pv3} Completeness and bogus fraction
     2667    as a function of magnitude for different stellar densities in real
     2668    PS1 exposures.  Each panel represents an exposure at different
     2669    Galactic latitudes towards anti-center, labeled by the density of
     2670    stars at the detection limit of the low-density exposure.  In each
     2671    panel, the completeness (compared to deep stack data) and fraction
     2672    of false detections (bogus fraction) is shown for a series of
     2673    cuts.  The gold curves show all detections in the exposures.  The
     2674    dotted black curve shows the impact of cutting detections
     2675    identified by {\tt psphot} as cosmic rays.  The blue curve
     2676    excludes cosmic rays and detections with {\tt PSF\_QF} $< 0.95$
     2677    while the red curve excludes cosmic rays and detections with {\tt
     2678      PSF\_QF\_PERFECT} $< 0.95$.}
     2679  \end{center}
     2680\end{figure}
     2681
     2682% Figure 8:  ** repaired PDF text **
     2683% /data/kukui.1/eugene/psphot.examples.20190423/compare.sh : mkfigure
    24402684\begin{figure*}[htbp]
    24412685  \begin{center}
    … …  
    24602704\end{figure*}
    24612705
    2462 % /data/kukui.1/eugene/psphot.examples.20190423/compare.sh
    2463 \begin{figure*}[htbp]
    2464   \begin{center}
    2465  \includegraphics[width=\hsize,clip]{\picdir/{mag.resid.aper.v1}.\plotext}
    2466   \caption{\label{fig:mag.resid.aper} Aperture Photometry
    2467     demonstration.  The plots show identical measurements to those in
    2468     Figure~\ref{fig:mag.resid.psf}, but for aperture photometry, as discussed in
    2469     Section~\ref{sec:aperture.correction}, rather than PSF photometry.}
    2470   \end{center}
    2471 \end{figure*}
    2472 
    2473 % on pueo ~eugene
    2474 % /data/kukui.1/eugene/psphot.examples.20190423/compare.sh
    2475 \begin{figure}[htbp]
    2476   \begin{center}
    2477  \includegraphics[width=\hsize,clip]{\picdir/{bright.mag.resid}.\plotext}
    2478   \caption{\label{fig:mag.resid.stdevs} Demonstration of photometric
    2479     accuracy using the image sequence from
    2480     Figure~\ref{fig:mag.resid.psf}. Using only bright stars (7 - 8
    2481     magnitudes above the detection threshold), we calculate the
    2482     difference between the magnitudes in the first image and the other
    2483     17 images.  The plotted dots show for each pair the mean
    2484     difference vs the standard deviation of the difference.  Red dots
    2485     show the PSF magnitudes and blue dots show aperture
    2486     magnitudes. Despite real transparency variations of 0.4 over the
    2487     50 minutes of this sequence, magnitudes are consistent at the few
    2488     millimagnitude level.  Aperture magnitudes have scatter in
    2489     the 2 - 7 millimagnitude range, while the PSF magnitudes have
    2490     scatter of 7 - 14 millimagntiudes. 
    2491 }
    2492 \end{center}
    2493 \end{figure}
    2494 
    2495 % on pueo ~eugene/zpts.20200406/mana.sh
    2496 \begin{figure*}[htbp]
    2497   \begin{center}
    2498  \includegraphics[width=\hsize,clip]{\picdir/{zpt.mjd.v0.i}.\plotext}
    2499   \caption{\label{fig:zpt.iband} Historical \ips-band zero points.
    2500     Blue dots are the individual exposure zero points, corrected to
    2501     airmass at the zenith.  Red dots are the median of zero points
    2502     from images groups in bins of 10 nights.  The grey line is a
    2503     spline fit to these median values.  }
    2504 \end{center}
    2505 \end{figure*}
    2506 
    2507 % on pueo ~eugene/zpts.20200406/mana.sh
    2508 \begin{figure}[htbp]
    2509   \begin{center}
    2510  \includegraphics[width=\hsize,clip]{\picdir/{zptres.hist.v0.i}.\plotext}
    2511   \caption{\label{fig:zpt.resid.hist} Historical \ips-band zero-point
    2512     residual variations.  Log-histogram (black line) of the
    2513     per-exposure zero points, corrected to the zenith, after
    2514     subtracting a spline fit to the median of image groups in bins of
    2515     10 nights.  The inset shows the core of the distribution.  In
    2516     both, the red line is a Gaussian fit to the distribution.  The
    2517     large negative tails are due to clouds and haze.  }
    2518 \end{center}
    2519 \end{figure}
    2520 
    2521 Aperture photometry attempts to avoid these problems, but introduces
    2522 other difficulties.  In aperture photometry, if a large enough
    2523 aperture is chosen, the amount of flux which is lost will be a small
    2524 fraction of the total source flux.  Even more importantly, as the
    2525 image conditions change, the amount lost will change by an even
    2526 smaller fraction, at least for a large aperture. 
    2527 %
    2528 % This can be seen by
    2529 % the fact that the dominant variations in the image quality are in the
    2530 % focus, tracking and seeing.  All of these errors initially affect the
    2531 % cores of the stellar images, rather than the wide wings.  The wide
    2532 % wings are largely dominated by scattering in the optics and scattering
    2533 % in the atmosphere.  The amplitude and distribution of these two
    2534 % scattering functions do not change significantly or quickly for a
    2535 % single telescope and site. 
    2536 %
    2537 Aperture photometry can then be used to
    2538 correct the PSF photometry.
    2539 
    2540 The difficulty for aperture photometry is the need to make an accurate
    2541 measurement of the local background for each source.  As the aperture
    2542 grows, errors in the measurement of the sky flux start to become
    2543 dominant.  If the aperture is too small, then variations in the image
    2544 quality are dominant.  The brighter is the source, the smaller is the
    2545 error introduced by the large size of the aperture.  However, the
    2546 number of very bright stars is limited in any image, and of course the
    2547 brighter stars are more likely to suffer from non-linearity or
    2548 saturation. 
    2549 
    2550 In order to thread the needle between these effects, \ippprog{psphot}
    2551 measures the aperture photometry on a modest-sized aperture, and then
    2552 uses the PSF model to extrapolate to a large aperture.  When the PSF
    2553 fluxes are calculated, the aperture flux for the modest-sized aperture
    2554 is also determined.  The aperture is a circular aperture with radius
    2555 set to a fixed multiple (\code{PSF_APERTURE_SCALE}) of $\sigma_w$, the
    2556 width of the Gaussian window function determined based on the analysis
    2557 of the second moments (see Section~\ref{sec:moments}).  For the PV3
    2558 $3\pi$ analysis, the aperture window radius is $4.5 \times \sigma_w$,
    2559 while the large reference aperture radius is set to 25 pixels
    2560 (\code{PSF_REF_RADIUS} = 6\farcs4).  These corrected aperture
    2561 magnitudes are saved in the output catalogs as \code{AP_MAG}, the
    2562 uncorrected aperture magnitudes are saved as \code{AP_MAG_RAW}, and
    2563 the radius used to measure the raw aperture flux is saved as
    2564 \code{AP_MAG_RADIUS}.  The corresponding flux and the flux error are
    2565 saved as \code{AP_FLUX} and \code{AP_FLUX_SIG}.
    2566 
    2567 With these aperture magnitudes in hand, it is now possible to make an
    2568 average correction to the PSF magnitudes to bring the PSF and aperture
    2569 magnitudes to the same system.  This correction is measured using the
    2570 same stars from which the PSF model is measured, as long as the PSF
    2571 magnitude error for the star is less than 0.03 mag.  The correction is
    2572 calculated using the weighted average of the values $m_{\rm AP} -
    2573 m_{\rm PSF}$.  Since the PSF may vary across the image, the correction
    2574 is determined as a function of position in the image.  Like the PSF
    2575 model, the 2D variations of the aperture correction may be modeled as
    2576 a polynomial or via interpolation in a grid.  For the $3\pi$ PV3
    2577 analysis, a grid with a maximum of $6\times 6$ samples per GPC1 chip
    2578 image was used.  The reported PSF magnitudes for all objects have this
    2579 aperture correction applied.  \textadd{Note that an initial aperture correction was
    2580 measured during the initial steps of the analysis before the PSF model
    2581 was chosen.  However, since the sources in the image were not yet
    2582 measured and subtracted, that aperture could be contaminated by
    2583 neighbors.  The analysis here is performed one fairly bright star at a
    2584 time with all other sources subtracted in order to minimize such contamination.}
    2585 
    2586 % growth curve analysis in psphot:
    2587 % in psphotChoosePSF : call psphotMakeGrowthCurve
    2588 % in psphotMakeGrowthCurve : boolean GROWTH_FROM_SOURCES, use
    2589 %% pmGrowthCurveGenerateFromSources or
    2590 %% pmGrowthCurveGenerate (uses PSF model only)
    2591 %% GROWTH_FROM_SOURCES is set to TRUE for default recipe
    2592 
    2593 %% ApTrend:
    2594 %% in psphotApResid, called by psphotReadout near the end of the
    2595 %% analysis
    2596 %% ApTrend = f(x,y) for (apMag - psfMag) for psfMagErr <= 0.03
    2597 %% apMag is growth curve corrected
    2598 %% psfMag is raw
    2599 
    2600 %% raw psfMag and raw apMag are measured
    2601 %% apMag = apMagRaw + growth curve correction (from apRadius to 25 pix
    2602 %% = PSF_REF_RADIUS)
    2603 %% psfMag = psfMagRaw + aperture trend (<ap - psf> + growth curve)
    2604 
    2605 % How important is this effect?  Consider a typical bright source with a
    2606 % flux of (say) 40,000 counts in an image of background 1000 counts per
    2607 % pixel, with FWHM of 4 pixels.  In principle, the flux of this source
    2608 % should be measurable with an accuracy of roughly 0.57\%
    2609 % ($\frac{\sqrt{40000 + 1000 \times 12}}{40000}$).  However, the
    2610 % measurement of the sky is limited at some finite level by Poisson
    2611 % statistics.  If we are required to use an aperture of (say) 25 pixels
    2612 % in radius (eg, 5 arcseconds for an 0.2 arcsec / pixel detector), and
    2613 % we have an annulus of twice this radius to measure the local sky, then
    2614 % we will have an error of XXX.
    2615 %
    2616 % \note{outline the variation of {\em ApResid} as a function of
    2617 % magnitude}.
    2618 
    2619 %%% \ippprog{psphot} measures the aperture correction ({\em ApResid}) for every PSF
    2620 %%% candidate source, then calculates the trend of this correction as a
    2621 %%% function of the magnitude.  This trend is fitted with a line.  The
    2622 %%% resulting function can be used to determine the effective aperture
    2623 %%% correction for an infinite flux source and the average bias inherent
    2624 %%% in the sky measurement for the image.  The scatter of the
    2625 %%% PSF-candidate source measurements about this trend is a measure of how
    2626 %%% well we can measure photometry from the image by applying the specific
    2627 %%% PSF model.  The slope of this trend is a measure of the bias in the
    2628 %%% local sky measurment for each source.  In principal, the measured sky
    2629 %%% levels could be modified by this bias.  More generally, the measured
    2630 %%% bias in a collection of images could be used to improve the model
    2631 %%% fitting or sky fitting portion of the software the remove the bias
    2632 %%% term.
    2633 
    2634 \subsection{Completeness \& Contamination}
    2635 
    2636 At the end of the \ippprog{psphot} analysis of the sources in the
    2637 image, an analysis is performed to test the detection efficiency.  A
    2638 number of fake PSF sources are injected into the image and the peak
    2639 detection analysis (Section~\ref{sec:peaks}) is use to determine if
    2640 these sources would have been recovered.  The PSF model fluxes are
    2641 measured for the source which are detected.  For a given image, the
    2642 detection threshold is predicted based on the median image variance
    2643 and the seeing.  A series of brightness bins straddling the threshold
    2644 are defined and a number of sources are injected with magnitudes
    2645 corresponding to each of these bin values.  The \ippprog{psphot}
    2646 recipe value \code{EFF.NUM} specifies the number of sources in each
    2647 brightness bin (500 the PV3), and the value \code{@EFF.MAG} specifies
    2648 the bins as magnitudes above and below the threshold.  For PV3, the 13
    2649 magnitude offsets were (-2.0, -1.0, -0.5, -0.25, -0.1, -0.05, 0.0,
    2650 0.05, 0.1, 0.25, 0.5, 1.0, 2.0), providing fine sampling near the
    2651 limit, but more coarse coverage further away.  Poisson noise
    2652 appropriate to the photon counts of the injected sources are included
    2653 in the image.  Injected sources are uniformly distributed across the
    2654 image in $X$ and $Y$ pixel coordinates {\em without any consideration
    2655   of the masked regions}.  This last point means the recovered
    2656 fraction in the bright bins can be used to test the masking fraction.
    2657 
    2658 As the stellar density increases, the completeness suffers due to
    2659 crowding and confusion.  Since the injection and recovery analysis of
    2660 the fake sources operates on the source-subtracted image and does not
    2661 attempt to fully discovery the sources, this analysis over-estimates
    2662 the completeness in crowded fields.  To explore the completeness in
    2663 crowded field images, we generate a series of simulated images using a
    2664 Gaussian PSF with FWHM = 1\arcsec for a range of stellar densities.
    2665 We generate fake stars with fluxes as faint as $\frac{1}{5}$ of the
    2666 flux as the low-density detection limit, with densities ranging from
    2667 \approx 14,000 stars per square degree at low-density detection limit
    2668 to \approx 4.8 million stars per square degree at the low-density
    2669 detection limit.  The latter is comparable to observed densities in
    2670 the Galactic plane.  We run the \ippprog{psphot} analysis on these
    2671 simulated images and compare the detected stars to those injected to
    2672 calculate the completeness for each image as a function of the true
    2673 magnitude of the stars.  Figure~\ref{fig:complete.ppsim} shows the measured
    2674 completeness for each of the six simulated images, labeled by the
    2675 logarithm of their faint-end stellar density. The red dashed line
    2676 shows the expected detection limit based on the background and seeing,
    2677 while the red curve shows the completeness curve calculated
    2678 automatically by \ippprog{psphot} using the injection and recovery
    2679 analysis.
    2680 
    2681 For low-density fields, the completeness function determined by
    2682 injection and recovery is similar to that measured by the simulation,
    2683 with the 50\% completeness threshold roughly 0.3 magnitudes too faint.
    2684 As the stellar density increases, the true 50\% completeness magnitude
    2685 rises relative to the value estimated by injection and recovery.
    2686 
    2687 Ideally, all sources detected by \ippprog{psphot} would correspond to
    2688 real astrophysical objects.  In reality, many sources are detected in
    2689 the images which do not correspond to real sources in the sky.  In the
    2690 very simplified simulations discussed above, which do not include
    2691 realistic detector artifacts, we find that the fraction of bogus
    2692 detections is extremely low, even at the very faint end.  In real
    2693 data, bogus detections are due to a variety of typical instrumental
    2694 features including cosmic rays, diffraction spikes, satelite tracks,
    2695 glows, non-Gaussian noise, variance mis-estimation, etc.  See paper III
    2696 for extensive discussion of instrumental artifacts in the Pan-STARRS images.
    2697 
    2698 Figure~\ref{fig:complete.pv3} illustrates the completeness and bogus
     2706{\TEXTADD Figure~\ref{fig:complete.pv3} illustrates the completeness and bogus
    26992707detection fraction for a set of 4 real PS1 exposures from the $3\pi$
    27002708Survey.  This figure uses \ips-band exposures with Galactic longitude
    … …  
    27172725also exclude  detections with \ippmisc{PSF_QF_PERFECT} less than
    271827260.95.  This cut removes detections on residual persistent glows and
    2719 diffraction spikes.
    2720 
    2721 For the exposures at high-Galactic latitude, with a relatively low
     2727diffraction spikes.}
     2728
     2729% Figure 9: ** repaired PDF text **
     2730% /data/kukui.1/eugene/psphot.examples.20190423/compare.sh
     2731\begin{figure*}[htbp]
     2732  \begin{center}
     2733 \includegraphics[width=\hsize,clip]{\picdir/{mag.resid.aper.v1}.\plotext}
     2734  \caption{\label{fig:mag.resid.aper} Aperture Photometry
     2735    demonstration.  The plots show identical measurements to those in
     2736    Figure~\ref{fig:mag.resid.psf}, but for aperture photometry, as discussed in
     2737    Section~\ref{sec:aperture.correction}, rather than PSF photometry.}
     2738  \end{center}
     2739\end{figure*}
     2740
     2741{\TEXTADD For the exposures at high-Galactic latitude, with a relatively low
    27222742density of sources, the cosmic rays represent a significant
    27232743contamination, as seen in the excess of bogus sources with \ips-band
    … …  
    27302750because the chance of having a source lie on the diffraction spikes or
    27312751persistence glows is greatly increased at higher stellar densities.
    2732 The impact of the crowding on the completeness is also clear in this dataset.
     2752The impact of the crowding on the completeness is also clear in this dataset.}
    27332753
    27342754\subsection{Stellar Photometry Example}
    … …  
    27732793the reported photometry for both PSF and aperture magnitudes.
    27742794
     2795% Figure 10: ** repaired PDF text **
     2796% on pueo ~eugene
     2797% /data/kukui.1/eugene/psphot.examples.20190423/compare.sh : figure.resids
     2798\begin{figure}[t]
     2799  \begin{center}
     2800 \includegraphics[width=\hsize,clip]{\picdir/{bright.mag.resid}.\plotext}
     2801  \caption{\label{fig:mag.resid.stdevs} Demonstration of photometric
     2802    accuracy using the image sequence from
     2803    Figure~\ref{fig:mag.resid.psf}. Using only bright stars (7 - 8
     2804    magnitudes above the detection threshold), we calculate the
     2805    difference between the magnitudes in the first image and the other
     2806    17 images.  The plotted dots show for each pair the mean
     2807    difference vs the standard deviation of the difference.  Red dots
     2808    show the PSF magnitudes and blue dots show aperture
     2809    magnitudes. Despite real transparency variations of 0.4 over the
     2810    50 minutes of this sequence, magnitudes are consistent at the few
     2811    millimagnitude level.  Aperture magnitudes have scatter in
     2812    the 2 - 7 millimagnitude range, while the PSF magnitudes have
     2813    scatter of 7 - 14 millimagntiudes. 
     2814}
     2815\end{center}
     2816\end{figure}
     2817
    27752818We believe the observed behavior at the faint end is primarily a
    27762819result of the increased crowding.  Aperture photometry is more
    … …  
    27792822with the aperture photometry degrading rapidly as the flux of the star
    27802823decreases. 
     2824
     2825% Figure 11: ** repaired PDF text **
     2826% on pueo ~eugene/zpts.20200406/mana.sh : go.zpt.stats i
     2827\begin{figure*}[tb]
     2828  \begin{center}
     2829 \includegraphics[width=\hsize,clip]{\picdir/{zpt.mjd.v0.i}.\plotext}
     2830  \caption{\label{fig:zpt.iband} Historical \ips-band zero points.
     2831    Blue dots are the individual exposure zero points, corrected to
     2832    airmass at the zenith.  Red dots are the median of zero points
     2833    from images groups in bins of 10 nights.  The grey line is a
     2834    spline fit to these median values.  }
     2835\end{center}
     2836\end{figure*}
    27812837
    27822838{\TEXTADD The figures above show the relative photometric accuracy for
    … …  
    28322888  of the \ips-band zero points after subtracting a smoothly varying
    28332889  spline fit to the median of groups of 10 nights.  A Gaussian fit to
    2834   this distribution has $\sigma = 28.4$ millimags.  If we
     2890  this distribution has $\sigma = 26.6$ millimags.  If we
    28352891  alternatively subtract a median zero point for each night, the
    2836   standard deviation is reduced to 18.9 millimags.  These values can be
     2892  standard deviation is reduced to 17.6 millimags.  These values can be
    28372893  compared to the results of \cite{2012ApJ...756..158S} in which only
    28382894  photometric nights were included, yielding a standard deviation of
    … …  
    28422898  which are not expected from the normal effects of weather.  We
    28432899  believe these are largely due to aperture correction errors.}
     2900
     2901% Figure 12: ** repaired PDF text **
     2902% on pueo ~eugene/zpts.20200406/mana.sh
     2903\begin{figure}[b]
     2904  \begin{center}
     2905 \includegraphics[width=\hsize,clip]{\picdir/{zptres.hist.v0.i}.\plotext}
     2906  \caption{\label{fig:zpt.resid.hist} Historical \ips-band zero-point
     2907    residual variations.  Log-histogram (black line) of the
     2908    per-exposure zero points, corrected to the zenith, after
     2909    subtracting a spline fit to the median of image groups in bins of
     2910    10 nights.  The inset shows the core of the distribution.  In
     2911    both, the red line is a Gaussian fit to the distribution.  The
     2912    large negative tails are due to clouds and haze.  }
     2913\end{center}
     2914\end{figure}
    28442915
    28452916\subsection{Basic Analysis Summary}
    … …  
    29072978cut was defined by $|b| > b_{\rm min}$ where $b_{\rm min} = b_0 + r_b
    29082979e^{\frac{-l^2}{2 \sigma_b^2}}$.  For the PV3 analysis, $b_0 =
    2909 $20\degree, $r_b = $15\degree, $\sigma_b = $50\degree.  \textadd{The Galactic plane cut is made on an object-by-object basis.}  This contour
    2910 avoids the denser portions of the Galactic plane and bulge, limiting
    2911 the total time spent on the galaxy modeling analysis at the expense of
    2912 galaxy photometry in the plane (though Kron photometry is available
    2913 for those sources). 
    2914 
     2980$20\degree, $r_b = $15\degree, $\sigma_b = $50\degree.  See
     2981Figure~\ref{fig:galplanecut} for an illustration of the cut used for PV3.  \textadd{The
     2982  Galactic plane cut is made on an object-by-object basis.}  This
     2983contour avoids the denser portions of the Galactic plane and bulge,
     2984limiting the total time spent on the galaxy modeling analysis at the
     2985expense of galaxy photometry in the plane (though Kron photometry is
     2986available for those sources).
     2987
     2988% galaxy model fits performed based on limits set in psphotChooseAnalysisOptions.c
     2989
     2990% petrosian analysis performed on same objects as galaxy model fits
     2991% if EXTENDED_SOURCE_PETROSIAN == TRUE (TRUE for PV3 stack - STACKPHOT)
     2992
     2993% galaxy model fits are performed on :
     2994% all if (PSPHOT.EXT.FIT.ALL.SOURCES == TRUE) (FALSE for PV3 stack)
     2995%   (even so, if density is higher than PSPHOT.EXT.FIT.ALL.THRESH, skip)
     2996
     2997% only extended sources (based on EXT.NSIGMA) if EXT.NSIGMA.LIMIT.USE
     2998% == TRUE (FALSE for PV3 stacks)
     2999
     3000% fit sources / measure petrosian to fixed flux limit if limits are
     3001% defined (they are for PV3)
     3002
     3003% mag limits by filter, e.g., : petro 25, extfit 21.5
     3004% are translated to flux in counts and compared to Kron flux
     3005
     3006% SN limit is used only if fixed flux limits are not set
     3007% SN limit set to EXTENDED_SOURCE_SN_LIM (10.0 for PV3)
     3008% S/N limit for Kron flux,
     3009
     3010% S/N lim values set to 0.0 for all models in PV3
     3011
     3012% galaxy coordinate limits:
     3013% if |b| > 20.0 + 15.0 exp(-long^2 / (2 * 50^2))
     3014
     3015\subsection{Radial Profiles}
     3016\label{sec:radial.profile.v2}
     3017
     3018Galaxies with regular profiles, such as elliptical galaxies and
     3019regular spiral galaxies, may be described as primarily a radial
     3020surface brightness profile, with additional structure acting as a
     3021perturbation on that profile.  For many galaxies, the azimuthal shape
     3022at a given isophotal level may be described as an elliptical contour.
     3023To first order, a galaxy may be well described with a single elliptical
     3024contour and radial profile. 
     3025
     3026% Figure 13
    29153027% uses plots.sh in this directory
    2916 \begin{figure}[htbp]
     3028\begin{figure}[b]
    29173029 \begin{center}
    29183030 \includegraphics[width=\hsize,clip]{\picdir/galplanecut.pdf}
    … …  
    29233035\end{figure}
    29243036
    2925 % galaxy model fits performed based on limits set in psphotChooseAnalysisOptions.c
    2926 
    2927 % petrosian analysis performed on same objects as galaxy model fits
    2928 % if EXTENDED_SOURCE_PETROSIAN == TRUE (TRUE for PV3 stack - STACKPHOT)
    2929 
    2930 % galaxy model fits are performed on :
    2931 % all if (PSPHOT.EXT.FIT.ALL.SOURCES == TRUE) (FALSE for PV3 stack)
    2932 %   (even so, if density is higher than PSPHOT.EXT.FIT.ALL.THRESH, skip)
    2933 
    2934 % only extended sources (based on EXT.NSIGMA) if EXT.NSIGMA.LIMIT.USE
    2935 % == TRUE (FALSE for PV3 stacks)
    2936 
    2937 % fit sources / measure petrosian to fixed flux limit if limits are
    2938 % defined (they are for PV3)
    2939 
    2940 % mag limits by filter, e.g., : petro 25, extfit 21.5
    2941 % are translated to flux in counts and compared to Kron flux
    2942 
    2943 % SN limit is used only if fixed flux limits are not set
    2944 % SN limit set to EXTENDED_SOURCE_SN_LIM (10.0 for PV3)
    2945 % S/N limit for Kron flux,
    2946 
    2947 % S/N lim values set to 0.0 for all models in PV3
    2948 
    2949 % galaxy coordinate limits:
    2950 % if |b| > 20.0 + 15.0 exp(-long^2 / (2 * 50^2))
    2951 
    2952 \subsection{Radial Profiles}
    2953 \label{sec:radial.profile.v2}
    2954 
    2955 Galaxies with regular profiles, such as elliptical galaxies and
    2956 regular spiral galaxies, may be described as primarily a radial
    2957 surface brightness profile, with additional structure acting as a
    2958 perturbation on that profile.  For many galaxies, the azimuthal shape
    2959 at a given isophotal level may be described as an elliptical contour.
    2960 To first order, a galaxy may be well described with a single elliptical
    2961 contour and radial profile. 
     3037% Figure 14  ** repaired PDF text **
     3038% on pueo ~eugene/sdss.psphot.2020414/mana.sh : go.figure
     3039\begin{figure*}[htbp]
     3040 \begin{center}
     3041 \includegraphics[width=\hsize,clip]{\picdir/{petrosians.mags}.pdf}
     3042  \caption{\label{fig:petrosians} Comparison of PS1 ({\tt psphot}) and
     3043    SDSS Petrosian parameters for objects identified as galaxies by
     3044    SDSS.  Panel (a) shows the difference in the measured Petrosian
     3045    magnitudes as a function of the Petrosian magnitude.  Panel (b)
     3046    shows the magnitude difference as a function of the measured
     3047    difference in the Petrosian radius.  }
     3048  \end{center}
     3049\end{figure*}
    29623050
    29633051In order to facilitate the Petrosian photometry analysis below, \ippprog{psphot}
    … …  
    30793167available from the PSPS \ippdbtable{StackPetrosian} table.}
    30803168
     3169Our implementation of the Petrosian apertures and fluxes is designed
     3170to match the SDSS implementation \citep{2002AJ....123..485S} and
     3171therefore the measured parameters should be quite comparable between
     3172the two surveys.  Figure~\ref{fig:petrosians} compare the Petrosian
     3173magnitudes and radii as measured by \ippprog{psphot} on the $3\pi$
     3174Survey observations and the values measured by SDSS for the same
     3175objects.  Objects identified by SDSS as galaxies ({\tt probPSF\_r} $<
     31760.5$) near the Galactic north pole ($\alpha$ = 180\degrees\ to
     3177190\degrees, $\delta$ = 25\degrees\ to 35\degrees) are selected from
     3178the PS1 $3\pi$ Survey dataset base on positional coincidence.  The
     3179figure shows the difference in the $r$-band Petrosian magnitudes as a
     3180function of the Petrosian magnitude and as a function of the
     3181difference in the measured Petrosian radii.  Differences in the
     3182measured magnitudes are driven by differences in the size estimates
     3183from the two datasets and analysis methods.  The PS1 analysis tends to
     3184find larger radii for the same objects than the SDSS analysis, with
     3185a mean difference of 0.3 arcseconds.  The larger aperture results in
     3186more flux captured in the aperture and thus brighter magnitudes for
     3187the same object: the mean difference is -0.23 magnitude in the sense
     3188of larger fluxes for the PS1 measurements.
    30813189
    30823190\subsection{Convolved Galaxy Model Fits}
    … …  
    32703378%% about the center of the pixel.  do this?
    32713379
    3272 In order to accurately compare the observed galaxy flux distribution
     3380\textmod{In order to accurately compare the observed galaxy flux distribution
    32733381to a model, it is necessary to integrate the pixel flux for a given
    3274 set of model parameter values.  This could be done in a numerical
    3275 fashion, but in practice brute-force evaluation of the numerical
    3276 integral is computationally very expensive, requiring many evaluations
    3277 of the model function.  Within \ippprog{psphot}, we bypass this
    3278 problem by defining a set of pre-calculated images for the central 9
    3279 pixels (the $3 \times 3$ grid of pixels centered on the peak).  These
    3280 pixel images are defined at higher resolution, with 11 subpixels per
    3281 real CCD pixel.  The pre-calculated images are generated for a series
    3282 of values for the following parameters: S\'ersic index, effective
    3283 radius, axial ratio.  We then select the closest image to our specific
    3284 case, and integrate over the true sub-pixels relevant for our position
    3285 and model.  We have thus turned the problem from thousands of
    3286 evaluations of the full galaxy model to \approx 100 straight
    3287 additions, or up to $6 \times$ that number if we interpolate between
    3288 any of the parameters.
    3289 
    3290 \note{how much error does this approximation introduce?}
     3382set of model parameter values.  In the \ippprog{psphot}
     3383implementation, we currently use a brute-force numerical evaluation of
     3384the integral, dividing the central pixel into a grid of subpixels,
     3385with the sampling set by the S\'ersic index of the model being
     3386evaluated as $N_{\rm sub} = 2 Integer(6n / R_{\rm min})$ where $N_{\rm sub}$
     3387is subpixel scale $n$ is the S\'ersic index and $R_{\rm min}$ is the
     3388size of the minor axis in pixel units.  The value of $N_{\rm sub}$ is
     3389constrained to be in the range 11 to 121, so the number of subpixels
     3390evaluations ranges from 121 to $121^2 = 14,641$.  Faster
     3391approximations to this analysis were explored but they resulted in
     3392unsatisfactory results.  This is definitely an area where
     3393\ippprog{psphot} could benefit from some of the lessons in the
     3394literature \citep[e.g.][]{2013PASP..125..719H}.}
     3395
     3396%% This could be done in a numerical
     3397%% fashion, but in practice brute-force evaluation of the numerical
     3398%% integral is computationally very expensive, requiring many evaluations
     3399%% of the model function.  Within \ippprog{psphot}, we bypass this
     3400%% problem by defining a set of pre-calculated images for the central 9
     3401%% pixels (the $3 \times 3$ grid of pixels centered on the peak).  These
     3402%% pixel images are defined at higher resolution, with 11 subpixels per
     3403%% real CCD pixel.  The pre-calculated images are generated for a series
     3404%% of values for the following parameters: S\'ersic index, effective
     3405%% radius, axial ratio.  We then select the closest image to our specific
     3406%% case, and integrate over the true sub-pixels relevant for our position
     3407%% and model.  We have thus turned the problem from thousands of
     3408%% evaluations of the full galaxy model to \approx 100 straight
     3409%% additions, or up to $6 \times$ that number if we interpolate between
     3410%% any of the parameters.
    32913411
    32923412The convolved galaxy model fit results are available in one of three
    … …  
    32943414\ippdbtable{StackModelFitDeV}, \ippdbtable{StackModelFitSer} for the
    32953415Exponential, DeVaucouleur, and S\'ersic models, respectively.
    3296 
    32973416
    32983417\subsection{Fixed Aperture Photometry}
    … …  
    33773496 sets of measurements joined together for ease of access.}
    33783497
    3379 \note{test SDSS radial apertures?}
     3498% \note{test SDSS radial apertures?}
    33803499
    33813500% at least out to aperture # RADIAL_AP_MIN (= 4), but no further than
    … …  
    34513570earlier work were generally compact.
    34523571
    3453 % /data/kukui.1/eugene/galaxies.20190425/tap_psphot_galaxies.pro : go.bigtest.ckgalaxy
     3572% Figure 15: ** repaired PDF text **
     3573% was /data/kukui.1/eugene/galaxies.20190425/tap_psphot_galaxies.pro : go.bigtest.ckgalaxy
     3574% is pueo:galaxies.20190425/tap_psphot_galaxies.pro : go.bigtest.ckgalaxy
    34543575\begin{figure}[htbp]
    34553576  \begin{center}
    … …  
    34973618accurate for the larger galaxies.
    34983619
    3499 % /data/kukui.1/eugene/galaxies.20190425/tap_psphot_galaxies.pro : go.bigtest.ckgalaxy
     3620% Figure 16 ** repaired PDF text **
     3621% was /data/kukui.1/eugene/galaxies.20190425/tap_psphot_galaxies.pro : go.bigtest.ckgalaxy
     3622% is pueo:galaxies.20190425/tap_psphot_galaxies.pro : go.bigtest.ckgalaxy
    35003623\begin{figure*}[htbp]
    35013624  \begin{center}
    35023625 \includegraphics[width=\hsize,clip]{\picdir/{galaxy.exp.params}.\plotext}
    3503 
    35043626  \caption{\label{fig:exp.params} Parameter recovery for simulated
    35053627    galaxies with Exponential profiles.  In each panel, we show
    … …  
    35193641\end{figure*}
    35203642
    3521 % /data/kukui.1/eugene/galaxies.20190425/tap_psphot_galaxies.pro : go.bigtest.ckgalaxy
     3643% Figure 17 ** repaired PDF text **
     3644% was /data/kukui.1/eugene/galaxies.20190425/tap_psphot_galaxies.pro : go.bigtest.ckgalaxy
     3645% is pueo:galaxies.20190425/tap_psphot_galaxies.pro : go.bigtest.ckgalaxy
    35223646\begin{figure*}[htbp]
    35233647  \begin{center}
    … …  
    35323656\label{sec:psf.forced.fit}
    35333657
    3534 \note{reference to multifit / cfht lens?}
    3535 
    35363658Traditionally, projects which use multiple exposures to increase the
    35373659depth and sensitivity of the observations have generated something
    35383660equivalent to the stack images produced by the IPP analysis,
    3539 \textadd{as done for example by the CFHT Legacy Survey
     3661\textadd{as done for example by the Canada-France-Hawaii Telescope (CFHT) Legacy Survey
    35403662  \citep{2006ApJ...647..116H} or the Cosmic Evolution Survey
    35413663  \citep[COSMOS][]{2007ApJS..172...99C}}.  In theory, the photometry
    … …  
    36413763\ippdbtable{ForcedMeanObject} tables.}
    36423764
    3643 \note{discuss the relative quality of average exposure, forced warp
    3644   average, and stack photometry. reference to Best et al}
     3765% Figure 18 ** repaired PDF text **
     3766% on pueo ~eugene/sdss.psphot.2020414/photcompare.sh : go.figure
     3767\begin{figure}[htbp]
     3768 \begin{center}
     3769 \includegraphics[width=\hsize,clip]{\picdir/{compare.mags}.pdf}
     3770  \caption{\label{fig:compare.mags} Comparison of {\tt psphot} average
     3771    chip photometry, average forced-warp photometry, and stack
     3772    photometry from $3\pi$ Survey data to average forced-warp
     3773    photometry from the Pan-STARRS\,1 Medium-Deep Survey field MD06
     3774    At bright magnitudes, average chip photometry is the most
     3775    accurate while the stack photometry is degraded by the
     3776    highly-texturd PSF.  At faint magnitudes, average chip magnitudes
     3777    are biased to artifically bright values.}
     3778  \end{center}
     3779\end{figure}
     3780
     3781{\TEXTADD With the inclusion of the forced-warp photometry, we have three
     3782distinct methods for measuring the PSF photometry of stars in the
     3783Pan-STARRS survey data: the average of the \ippstage{chip}-stage
     3784photometry from the individual exposures; the measurement from the
     3785stacks, and the average of the forced-warp photometry described here.
     3786It is worth considering which of these should be used in which
     3787circumstance.  Figure~\ref{fig:compare.mags} shows a comparison of
     3788these three different methods to deeper data from the Medium Deep
     3789Survey observations (MD06 field).  Our conclusion from this and other
     3790analysis is that the average \ippstage{chip}-stage photometry is the
     3791best (most accurate) measurement for brighter objects, where the
     3792signal-to-noise is roughly 10 or more.  This is the photometry source
     3793which was used for the global photometry solution discussed by
     3794\cite{2012ApJ...756..158S} and used in the overall calibration (see
     3795Paper V).}
     3796
     3797{\TEXTADD As can be clearly seen in the figure, the average from the forced-warp
     3798photometry is slightly worse than the chip photometry, while the stack
     3799PSF photometry is significantly degraded.  We attribute the latter
     3800effect to the highly-textured PSF observed in the stack images due to
     3801the combination of variable PSFs in each exposure and significant
     3802masking fraction in the PS1 camera.  At the faint end, the chip
     3803photometry is significantly worse that both average warp and stack
     3804photometry.  First, in order to have a measurement, a source must be
     3805detected above the detection threshold in at least one of the
     3806exposures, limiting the depth possible of the average chip
     3807photometry. Second, at the faint end, only bright fluctuations will be
     3808detected, resulting in a bright bias. This latter effect is clearly
     3809seen in Figure~\ref{fig:compare.mags} as the average chip magnitudes
     3810diverge from the deeper Medium Deep photometry measurements.  As has
     3811been noted elsewhere \citep{2018ApJS..234....1B}, the warp and stack
     3812photometry is also degraded for objects which have significant proper
     3813motion over the course of the $3\pi$ Survey since the position is held
     3814constant for all epochs, while the average chip photometry is
     3815calculated on detections which are cross-matched in the database.
     3816Thus, warp and stack photometry should be avoided for sources with
     3817proper motion greater than roughly 100 milliarcseconds per year.}
    36453818
    36463819\subsection{Forced Galaxy Models}
    … …  
    36633836the same time the best normalization corresponding to the best
    36643837elliptical shape, and thus the best galaxy magnitude value.
     3838\textadd{This technique is similar to the joint fitting of multiple
     3839  exposures performed by the CFHT Lensing Survey team \citep{2013MNRAS.429.2858M}.}
    36653840
    36663841For each warp image, the stack values for the major and minor axis are
    … …  
    38944069from the PSPS database \ippdbtable{ForcedWarpLensing} table while the
    38954070average values calculated over the warps is found in the
    3896 \ippdbtable{ForcedMeanLensing} tables.
     4071\ippdbtable{ForcedMeanLensing} tables.  \textadd{Although the software used
     4072here was not involved in any of the GRavitational lEnsing Accuracy
     4073Testing (GREAT) challenges, it is similar to the code of the EPFL\_KSB
     4074team \citep{2015MNRAS.450.2963M} and likely to perform similarly.}
    38974075
    38984076% \note{example of using the lensing elements for binaries?}
    … …  
    40424220\section{Conclusions}
    40434221
    4044 \note{add lessons learned here}
    4045 
    4046 \begin{verbatim}
    4047 Suggestions for improvements / changes
    4048 * use more external knowledge:
    4049   ** Gaia or PS1 to select stars as PSF sources
    4050   ** pre-seed information about the very bright or very crowded
    4051                 regions
    4052 * background model
    4053   ** allow the superpixel scale to change as a function of environment
    4054   ** do not use the lower-end model unless region is known to be dense
    4055 * use galactic latitude or local stellar density to smoothly
    4056   transition from double / multi-PSF to galaxy model fitting
    4057 \end{verbatim}
    4058 
    40594222The Pan-STARRS Image Processing Pipeline has used the \ippprog{psphot}
    40604223software to detect and characterize astronomical sources in images
    … …  
    40694232million PS\,1 exposures have been characterized (some representing
    40704233repeated measurements of the same exposures). 
     4234
     4235There is always room for improvement, however.  A number of
     4236possible improvements to \ippprog{psphot} have been identified which
     4237could result in more reliable measurements for either stars or
     4238galaxies.  Here we discuss improvements beyond simply tuning
     4239parameters for a specific dataset.
     4240
     4241In general, the improvements we identify share the characteristic of
     4242making use of external information in the analysis.  As described
     4243above, essentially all operations of \ippprog{psphot}, except in the
     4244context of forced photometry, approach each image with no prior
     4245knowledge.  This was necessary in the early stages of the Pan-STARRS
     4246project when we had not yet observed the sky with our instrument and
     4247comparable observations were only available in the SDSS Galactic cap
     4248regions.  However, the sky is now much better known, not only from
     4249PS1, but also for example due to Gaia.
     4250
     4251Several improvements to the \ippprog{psphot} analysis could be made by
     4252including as much information from external catalogs about the
     4253positions and characteristics of sources in the images as possible.
     4254For example, known stars (e.g., based on proper motions from Gaia or
     4255colors and morphology from PS1) could be used for PSF sources.  In
     4256areas of high density, especially in known globular or even open
     4257clusters, existing high-resolution imagery could be used to provide a
     4258constraint on location of stars.  External information could also be
     4259used to control the scale on which the background is modelled: a finer
     4260sampling is helpful in regions of known nebulosity and large galaxies
     4261such as M31.  Finally, the galactic latitude or the externally-defined
     4262stellar density could be used to control the choice of fitting double
     4263stars or galaxy models.  This would be a step beyond the current
     4264capability of choosing to fit galaxy models as a function of galactic
     4265latitude.
    40714266
    40724267% PS2 reference:

  • trunk/doc/release.2015/ps1.analysis/response.txt

    r41333 r41347  
    11
    2 ---------------------------------------------------------------------
    32Referee Report
    43Reviewer's Comments:
    … …  
    9897that the photometric goals are achieved
    9998
    100 **** TBD : discuss relative quality of chip, forced, stack photometry
     99** added comparion discussion of chip, warp, stack photometry at the end of Sec 6.1
    101100
    102101- Sec 7, where the image differencing detections and photometry is used
    … …  
    126125in one place would be a useful service.
    127126
    128 **** TBD : summarize the lessons learned
     127** added suggested improvements in conclusion
    129128
    130129Abstract:
    … …  
    331330for a typical exposure.
    332331
    333 **** TBD: SHOW SOME EXAMPLES of PSF variations
    334      choose 3 exposures: 1 with good IQ, one with bad IQ, but round, one with bad IQ but not round,
    335      plot some IQ stats (Mxx - Myy) / (Mxx + Myy)
     332** we have added a figure to show examples of the image quality
     333   variations observed in PS1 in both good and bad seeing data.
    336334
    337335- Please state whether the PSF model is this set of formulae
    … …  
    435433and presented as a future development effort.
    436434
    437 **** TBD : wording of full PSF model section 4.6.6
     435** reworded to explain that this step, unlike 4.6.2, does a
     436   simultaneous fit to the position and normalization for sources
     437   one-at-a-time.
    438438
    439439- Remind the reader that the 4 independent parameters includes a local sky
    … …  
    455455range.
    456456
    457 **** TBD: double-star mode: was this turned on for PV3? ppSim to show recovery
     457** In reviewing the code, we discovered that this approach to close
     458   neighbors was turned off for PV3, similar to the blend fits
     459   discussed above.  We have moved both of these crowded field
     460   analysis concepts to a single section, identified as deactivated
     461   for PV3.
    458462
    459463Sec 4.7:
    … …  
    558562compare well to those in the PS1 catalog?
    559563
    560 **** TBD: compare Petrosian mags to SDSS for some example
     564** These agree to first order, but there is a tendency for the PS1
     565   measurements to have larger radii and smaller (brighter)
     566   magnitudes.  Added text and a figure to illustrate
    561567
    562568Sec 5.3:
    … …  
    585591error of this approximation should be stated.
    586592
    587 **** TBD: model central pixel errors for Sersic models
     593** In trying to answer this question, we realized that, while we
     594   experimented with this technique, the as implemented psphot in fact
     595   used brute-force numerical evaluation.  These implementation
     596   experiments did not pan out so we went ahead with something that
     597   worked, even if it was slower.  We have updated the text to
     598   describe the actual implementation.
    588599
    589600Sec 5.4:
    … …  
    688699and if not, which code would it be most similar to?
    689700
    690 **** TBD : check on GREAT challenge to compare code
     701** psphot was not used in any of the GREAT challenges, but is similar
     702   to the EPFL_KSB team's code.  added this to the text
    691703
    692704- Define "KSB" and "HFK" references in-line
Note: See TracChangeset for help on using the changeset viewer.