MUSE: instrumentMUSE: instrument and DRL overviewand DRL overview

Peter WeilbacherJoris Gerssen

Arlette Pecontal-RoussetMartin Roth

and the MUSE team

Garching, 18thApr07

p.2

Instrument propertiesInstrument properties

3 operating modes– No AO WFM– AO WFM– AO NFM

4096 pixels

90,000 spaxels

370 106

pixels

AO

Laser guide stars

Garching, 18thApr07

p.3

Field splittingField splitting1 arcmin

2.4 arcsec + x0.017 arcsec

12

43

5

24

6……………………………………………….

X arcsec

X arcsec

Garching, 18thApr07

p.4

SubSub--field slicingfield slicing12

43

12

…………

15 arcsec0.2 arcsec

1

5

9

2

6

10

3

7

11

4

8

12

400

pix

984 pix

7 pix

75 pix

13

17

21

X 4

Garching, 18thApr07

p.5

Pattern on CCD Pattern on CCD

70 pix

2 pix

3867 pix3870 pix

Garching, 18thApr07

p.6

Mock Mock science science

datadata

1/24th of a full exposure!

Garching, 18thApr07

p.7

Mock Mock science science

datadata

1/24th of a full exposure!

Garching, 18thApr07

p.8

Data volumeData volumeOne exposure (raw) 40962px x 24 CCDs x 16bit 805 MB + headersOne exposure (reduced) ~4.8 GB + headers! (Euro3D format with 3 extensions) Typical raw science data ~30 GB/night (min. ~10 GB, max. 145 GB)Calibration raw data between 10 and 100 GB per night, typical ~25 GB

Garching, 18thApr07

p.9

Pipeline DesignPipeline DesignTwo levels: CCD based and whole instrumentAll (?) calibrations and first part of science data reduction CCD basedSecond half of science data reduction depends on data from all IFUsFew recipes that combine most steps into one external callParallelization perhaps trivial for first part, more difficult for second part (CPL + OpenMP?)

Garching, 18thApr07

p.10

RecipesRecipes(muse_quick_image) muse_masterbias and -darkmuse_masterflatmuse_distortionmuse_arcmuse_scibasicmuse_scipost

Garching, 18thApr07

p.11

Science recipesScience recipesmuse_scibasic: apply calibrations and transform input data into 3D lookup table (“apply” trace and wavelength calibration) muse_scipost: most complex recipe:– Some common steps for all inputs, but output depends

on type of input data (std vs. science) – Merge fields– Subtract sky / create sky spectrum– Apply flux calibration / compute response– Correct atmospheric refraction– Combine multiple exposures

Garching, 18thApr07

p.12

Reduction approachReduction approach

Output data Euro3D: data spectrum, bad pixel spectrum, and statistics (error) spectrum“Direct” measure of the S/N of an exposureDifficult to estimate correct errors through multiple resampling steps only use one resampling?Create lookup tables (matrices?) in 3D, similar to wavelength maps in 2DApply steps that need resampling to the “coordinates” in these lookup tables instead of data itselfIn the end one routine does the resampling using three dimensional interpolation in the lookup table

Garching, 18thApr07

p.13

Visualization toolVisualization tool

E3D tool too slow, difficult to maintain, Tcl-basedLicense problem with PGPLOTCreate new visualization tool for 3D dataSimilar basic feature set of E3D but more user friendlyBoth command line and GUI (like IRAF/MIDAS/IDL) Able to access plots as variables from the shellPython + Gtk or Qt, C+CPL for fast routines?“Plugins” for extended functionality?

Garching, 18thApr07

p.14

StatusStatusPDR July 2007, FDR mid 2008, first light 2012Worked with CPL quite a bit so farCPL based coding: prototype code can (almost) determine and apply trace and wavelength calibration (still to do: write Euro3D output) Further prototype code in IDL, Python, cl/IRAF e.g. for sky subtractionExperimented with OpenMP: works (only) using direct array access not when using CPL APICreated first automated test scripts (IDL + shell scripts) for recipes

MUSE: instrument�and DRL overviewInstrument propertiesField splittingSub-field slicingPattern on CCD Mock science dataMock science dataData volumePipeline DesignRecipesScience recipesReduction approachVisualization toolStatus