MIPS EnhancerMIPS Enhancer
MIPS_Enhancer OverviewMIPS Data Processing and
Calibration WorkshopApril 12, 2004
How to Run How to Run MIPS_ENHANCERMIPS_ENHANCER
• Command line options– mips_enhancer –f filename.fits (40+ options)
• Namelist options– mips_enhancer –n namelist.in
What have we learned What have we learned since launch ?since launch ?
• Need to make MIPS_enhancer more efficient in terms of:– Memory – Speed
• Improve method of combing surface bright nesses.
Memory and SpeedMemory and Speed• Tiling
– -x #, -y #– Only operating on tile sub-pixels
• Do not hold mosaic in memory – -m
• Memory limited to current tile• Write tiles out • Stitch mosaic from tiles
– Necessary to change format of mosaic to limit memory needed.
• Primary header (history, information from first DCE)• Extensions: surface brightness, uncertainty, pixel flag,
overlap coverage)
Conserving Surface Conserving Surface BrightnessBrightness
• Improvements in:– Outlier rejection algorithm– Weighting overlapping pixels in
mosaic
Outlier Rejection Outlier Rejection AlgorithmAlgorithm
• For each sub-pixel there is a stack of overlapping BCD pixels.
Sub-Pixel Outlier RejectionSub-Pixel Outlier Rejection
N BCD overlapping pixel valuesStart Iteration i = 1.• Find the median value: Medianorg
• Find Standard Deviation: STDEVorg
• Find N differences: |(Medianorg – fluxk)|• Sort differences • Withhold 10% largest differences (minimum 1)• Find new median and standard deviation of remaining points:
Mediani, STDEVi
• Test Most deviant point:– (Fluxmost deviant point – Mediani)/STDEVi> Sigma Clipping (SC)– If reject – iterate
Outlier Rejection cont.Outlier Rejection cont.
• Continue iterating until:– Remain points < Minimum Points (SP)– Iteration # > Maximum # iterations
(SM)
– ((STDEVorg – STDEVnew)/STDEVorg )*100 > Sigma Tolerance (ST)
Variation on Outlier Variation on Outlier RejectionRejection
• Option: SKTest Most deviant point:
(Fluxmost deviant point – Mediani)/STDEVi> Sigma Clipping (SC)
• Option: SUInstead of using the standard deviation of stack: STDEVi
Combine in quadrature: uncertainty of BCD pixel and standard deviation of Median
Obsolete Outlier Rejection Obsolete Outlier Rejection optionsoptions
• Quartiles• May retire thresholding (t)
– (Fluxi/Median Flux) > t continue rejecting
– (Fluxi/Median Flux) < t stop rejection
Conserving Surface Conserving Surface BrightnessBrightness
• For each sub-pixel we have N BCD pixels overlapping– Surface Brightness of BCD– Uncertainty associated with surface
brightness– Overlap Area
• How do we combine information to conserve surface brightness in mosaic ?
DistortionsDistortions
Weighting MethodsWeighting Methods
Most probable value is a weighted average of points: wi xi/ wi
• W1= overlap area, uncertainties• W2=overlap area• W3=overlap area, S/N (SB/uncertainty)• W4=overlap area, SQRT(S/N)• W5=overlap area, SQRT(S)
Useful Options for Useful Options for MIPS_enhancerMIPS_enhancer
• -f filename (-i list of files)• -o output_name• -RR (remove rotation)• -s # (scale factor, recommended = 0.25)• dy (correct for distortion, default)• -x #, -y #, -m (memory & speed options)• -r # (70 microns, # BCDs to reject after stim,
recommened 1)• -IB (ignore bias boost, 24 microns, default)• -W2 (W1,W3,W4,W5) Weight options• -SC #, ST #, SP #, SM # SK, SU (outlier rejection options
Recommened:SC 4, ST 10, SP 3, SM 30, SK)• -FM (FD) sub-pixel flux weighted mean (median)• -D, -C, -FL (sets other appropriate options)
Near Term ImprovementsNear Term Improvements
• SED mode• Distortion (verify or tweak distortion
coefficients produced from CODE-V) • Study on surface brightness conservation
– Add weighting which only uses the number of reads that went into the slope determination.
– Develop testing routine to see how well we haveUse simulating routine (Emeric) to add sources to cal file.
• Move on to 160 micron data• Start enhancing framework
Suggestions ?Suggestions ?