[daip] IMAGR bug
Eric Greisen
egreisen at nrao.edu
Wed Jun 11 12:11:54 EDT 2008
Crystal Brogan wrote:
> Hi,
>
> There appears to have been a bug introduced in IMAGR during the weekend
> update. I am using the computer shadowfax in CV.
>
> Attached is a single uv channel (channel 90) of a VLA 44 GHz dataset. The
> masers in this channel are very strong (~300 Jy).
>
> If I clean the image as follows below with AIPS version='TST', when the
> min allowed flux reaches
>
> Field 1 min allowed flux = -534.402 MilliJy
>
> It goes into an infinite loop where the min flux does not change but the
> total cleaned flux continues to grow and grow until the niter is reached.
> The exact min. flux where this happens varies but it happens for several
> very strong maser channels (the rms noise in a maser free channel is ~60
> mJy - the strong masers are very much dynamic range limited). If I do NOT
> use clean boxes, the flux threshold is reached without a problem.
>
> So the problem appears connected with the use of clean boxes.
>
> If I use version='old' WITH clean boxes (as specified below), the image
> cleans to 300 mJy with no abnormal behavior.
>
> AIPS 1: IMAGR: Wide field imaging/Clean task
> AIPS 1: Adverbs Values Comments
> AIPS 1: ----------------------------------------------------------------
> AIPS 1: INNAME '17209-35451' Input UV data (name)
> AIPS 1: INCLASS 'CH90' Input UV data (class)
> AIPS 1: INSEQ 1 Input UV data (seq. #)
> AIPS 1: INDISK 3 Input UV data disk drive #
> AIPS 1: SOURCES *all ' ' Source name
> AIPS 1: QUAL -1 Calibrator qualifier -1=>all
> AIPS 1: CALCODE ' ' Calibrator code ' '=>all
> AIPS 1: TIMERANG *all 0 Time range to use
> AIPS 1: SELBAND -1 Bandwidth to select (kHz)
> AIPS 1: SELFREQ -1 Frequency to select (MHz)
> AIPS 1: FREQID -1 Freq. ID to select.
> AIPS 1: SUBARRAY 0 Sub-array, 0=>all
> AIPS 1: ANTENNAS *all 0 Antennas to plot
> AIPS 1: BASELINE *all 0 Baselines with ANTENNAS
> AIPS 1: DOCALIB -1 > 0 calibrate data & weights
> AIPS 1: > 99 do NOT calibrate weights
> AIPS 1: GAINUSE 0 CL (or SN) table to apply
> AIPS 1: DOPOL 0 If >0.5 correct polarization.
> AIPS 1: BLVER 0 BL table to apply.
> AIPS 1: FLAGVER 0 Flag table version
> AIPS 1: DOBAND 0 If >0.5 apply bandpass cal.
> AIPS 1: Method used depends on value
> AIPS 1: of DOBAND (see HELP file).
> AIPS 1: BPVER 0 Bandpass table version
> AIPS 1: SMOOTH *all 0 Smoothing function. See
> AIPS 1: HELP SMOOTH for details.
> AIPS 1: STOKES ' ' Stokes parameters (see HELP)
> AIPS 1: BCHAN 90 Low freq. channel 0 for cont.
> AIPS 1: ECHAN 90 Highest freq channel
> AIPS 1: CHANNEL 0 Restart channel number
> AIPS 1: NCHAV 0 Number of chan. to average.
> AIPS 1: CHINC 1 Channel incr. between maps.
> AIPS 1: BIF 1 First IF in average.
> AIPS 1: EIF 0 Last IF in average.
> AIPS 1: OUTNAME ' ' Output image name (name)
> AIPS 1: OUTDISK 3 Output image disk drive #
> AIPS 1: OUTSEQ 0 Output seq. no.
> AIPS 1: OUTVER 0 CC ver. no (Continuum only)
> AIPS 1: *** SET OUTVER ON RESTARTS
> AIPS 1: IN2NAME ' ' UV work file name
> AIPS 1: IN2CLASS ' ' UV work file class
> AIPS 1: IN2SEQ 0 UV work file seq
> AIPS 1: *** SET TO KEEP WORK FILE
> AIPS 1: IN2DISK 0 UV work file disk
> AIPS 1: CELLSIZE 0.08 0.08 (X,Y) size of grid in asec
> AIPS 1: IMSIZE 2048 2048 Minimum image size
> AIPS 1: NFIELD 1 Number of fields (max 4096)
> AIPS 1: DO3DIMAG 1 > 0 => use different tangent
> AIPS 1: points for each field
> AIPS 1: FLDSIZE *all 0 Clean size of each field.
> AIPS 1: RASHIFT *all 0 RA shift per field (asec)
> AIPS 1: DECSHIFT *all 0 DEC shift per field (asec)
> AIPS 1: UVTAPER 0 0 (U,V) Gaussian taper
> AIPS 1: units are kilo-lambda
> AIPS 1: UVRANGE 0 0 Min & max baseline (klambda)
> AIPS 1: GUARD 0 0 x,y guard band fractional
> AIPS 1: radius
> AIPS 1: ROTATE 0 Rotate image CCW from N by
> AIPS 1: ROTATE degrees
> AIPS 1: ZEROSP *all 0 0-spacing fluxes and weights
> AIPS 1: SEE HELP!!
> AIPS 1: UVWTFN ' ' UV dist. weight function
> AIPS 1: UVSIZE 0 0 Array size for doing uniform
> AIPS 1: weights. 0 -> actual field
> AIPS 1: size.
> AIPS 1: ROBUST 5 Robustness power: -5 -> pure
> AIPS 1: uniform weights, 5 => natural
> AIPS 1: UVBOX 0 Additional rows and columns
> AIPS 1: used in weighting.
> AIPS 1: UVBXFN 1 Box function type when UVBOX
> AIPS 1: > 0. 0 -> 1 round pill box.
> AIPS 1: XTYPE 5 Conv. function type in x
> AIPS 1: default spheroidal
> AIPS 1: YTYPE 5 Conv. function type in y
> AIPS 1: default spheroidal
> AIPS 1: XPARM *all 0 Conv. function parms for x
> AIPS 1: YPARM *all 0 Conv. function parms for y
> AIPS 1: NITER 1000000 Maximum # of Clean components
> AIPS 1: BCOMP *all 0 Begin at BCOMP component
> AIPS 1: Specify for each field.
> AIPS 1: ALLOKAY 0 For restart: > 0 => beams
> AIPS 1: okay, > 1 => work file too
> AIPS 1: NBOXES 23 Number of boxes for Clean
> AIPS 1: NB: field 1 only.
> AIPS 1: CLBOX -1 12 Four coordinates for each box
> AIPS 1: 958 1150 -1 12
> AIPS 1: 946 1138 -1 12
> AIPS 1: 988 1030 -1 9
> AIPS 1: 982 1138 -1 12
> AIPS 1: 1240 982 -1 12
> AIPS 1: 1285 964 -1 9
> AIPS 1: 1306 910 -1 18
> AIPS 1: 1129 865 -1 15
> AIPS 1: 1153 877 -1 12
> AIPS 1: 1015 775 -1 12
> AIPS 1: 943 808 -1 15
> AIPS 1: 931 799 -1 15
> AIPS 1: 1069 982 -1 15
> AIPS 1: 1054 970 -1 12
> AIPS 1: 1045 952 -1 12
> AIPS 1: 1096 973 -1 12
> AIPS 1: 793 742 -1 21
> AIPS 1: 832 757 -1 21
> AIPS 1: 853 778 -1 18
> AIPS 1: 877 859 -1 12
> AIPS 1: 955 1018 -1 12
> AIPS 1: 985 1069 -1 18
> AIPS 1: 751 388 *rest 0
> AIPS 1: BOXFILE *all ' ' Input file of field params
> AIPS 1: and Clean boxes; ' ' => use
> AIPS 1: FLDSIZE, RASHIFT, DECSHIFT,
> AIPS 1: NBOXES, CLBOX only.
> AIPS 1: OBOXFILE *all ' ' Output file for final Clean
> AIPS 1: boxes
> AIPS 1: GAIN 0.1 Clean loop gain
> AIPS 1: FLUX 0.3 Minimum Clean component (Jy)
> AIPS 1: MINPATCH 121 Min. BEAM half-width in AP.
> AIPS 1: BMAJ 0 FWHM(asec) major axis Clean
> AIPS 1: restoring beam.
> AIPS 1: BMIN 0 FWHM(asec) minor axis Clean
> AIPS 1: restoring beam.
> AIPS 1: BPA 0 Clean beam position angle
> AIPS 1: OVERLAP 2 1 => restore components to
> AIPS 1: overlapped fields, >=2=>
> AIPS 1: expect overlaps in Cleaning
> AIPS 1: PHAT 0 Prussian hat height.
> AIPS 1: FACTOR 0 Speedup factor see HELP
> AIPS 1: CMETHOD ' ' Modeling method:
> AIPS 1: 'DFT','GRID',' '
> AIPS 1: IMAGRPRM *all 0 Task enrichment parameters
> AIPS 1: (1) Antenna diameter (m)
> AIPS 1: (2) Source Spectral index
> AIPS 1: (3) Frequency scaling factor
> AIPS 1: (4) > 0 -> SDI Clean factor
> AIPS 1: (5) >0 => scale residuals
> AIPS 1: (6) Half-width in x of box
> AIPS 1: (7) Half-width in y of box
> AIPS 1: (8) Filter components whose
> AIPS 1: neighborhood is weaker than
> AIPS 1: IMAGRPRM(8) Jy. 0 -> don't
> AIPS 1: (9) Radius in pixels for the
> AIPS 1: IMAGRPRM(8) test.
> AIPS 1: (10) multiplier of image size
> AIPS 1: to get beam size: 0 => 2;
> AIPS 1: 2, 1, 0.5 0.25 supported
> AIPS 1: (11-16) Multi-resolution
> AIPS 1: added controls
> AIPS 1: (17) spectral index radius
> AIPS 1: 0 -> no correction
> AIPS 1: (19) Dynamic range limit
> AIPS 1: (20) Retry factor (see help)
> AIPS 1: NGAUSS 0 Number of resolutions to use
> AIPS 1: WGAUSS *all 0 Resolutions in arc sec >= 0
> AIPS 1: FGAUSS *all 0 Minimum flux for each resol.
> AIPS 1: MAXPIXEL 60000 Maximum pixels searched in
> AIPS 1: each major cycle.
> AIPS 1: IN3NAME ' ' Spectral index image name
> AIPS 1: IN3CLASS ' ' Spectral index image class
> AIPS 1: IN3SEQ 0 Spectral index image sequence
> AIPS 1: number
> AIPS 1: IN3DISK 0 Spectral index image disk
> AIPS 1: IN4NAME ' ' Spectral curvature name
> AIPS 1: IN4CLASS ' ' Spectral curvature class
> AIPS 1: IN4SEQ 0 Spectral curvature sequence
> AIPS 1: number
> AIPS 1: IN4DISK 0 Spectral curvature disk
> AIPS 1: FQTOL -1 Frequency tolerance in kHz
> AIPS 1: (primary beam & spec index)
> AIPS 1: DOTV 1 Display residuals on TV ?
> AIPS 1: Start with field = DOTV
> AIPS 1: GRCHAN 0 Graphics channel of boundary
> AIPS 1: BADDISK *all 0 Disks to avoid for scratch.
>
>
>
> _______________________________________________
> Daip mailing list
> Daip at listmgr.cv.nrao.edu
> http://listmgr.cv.nrao.edu/mailman/listinfo/daip
I have found the bug - odd that with no windows you did not hit it.
When IMAGR switched to gridded subtraction it did not ask for a big
enough AP and so subtracted no components and so found the same ones
over and over. I have added to the request more things to account for
overhead. I suspect that we added enough to manage smaller images but
2048 x 2048 have more overhead of one sort that was not adequately
provided.
Thanks - tomorrow's MNJ will fix
Eric Greisen
More information about the Daip
mailing list