Simulations in the context of SPHERE Exoplanet Imaging Workshop David Mouillet Lecture 27 Feb 2012 Numerous contributors in the simulation work for SPHERE: Kjetil Dohlen, Anthony Boccaletti, Arthur Vigan, Dino Mesa, Raffaele Gratton, Marcel Carbillet, …

SPHERE overall presentation Many similarities wrt GPI (see Lisa’s presentation) in terms of high level purposes, main components and challenges : Many similarities wrt GPI (see Lisa’s presentation) in terms of high level purposes, main components and challenges : High order AO, coronagraphs, multi-wavelength differential imaging, angular differential imaging, budget analysis up to 8-10 target magnitude, … High order AO, coronagraphs, multi-wavelength differential imaging, angular differential imaging, budget analysis up to 8-10 target magnitude, … With some significant differences: With some significant differences: Focus: Nasmyth vs Cassegrain (flexures, telescope vs instrument WFE) Focus: Nasmyth vs Cassegrain (flexures, telescope vs instrument WFE) Wavelength coverage: Wavelength coverage: A VISIBLE imager with differential polarimetry: ZIMPOL nm A VISIBLE imager with differential polarimetry: ZIMPOL nm Wider NIR simultaneous coverage: 0.95 – 1.65 mic (IFS + diff imager IRDIS, or IFS only) Wider NIR simultaneous coverage: 0.95 – 1.65 mic (IFS + diff imager IRDIS, or IFS only) Variable / Differential WFE: Variable / Differential WFE: pupil de-rotator, 2 ADCs: some rotating surfaces (eventhough in a predictible manner) pupil de-rotator, 2 ADCs: some rotating surfaces (eventhough in a predictible manner) No « CAL » system (NIR tilt/defocus sensor only) No « CAL » system (NIR tilt/defocus sensor only)

High frequency AO correction (41x41 act.) High stability : image / pupil control Visible – NIR Refraction correction FoV = 12.5’’ 40x40 SH-WFS in visible 1.2 KHz, RON < 1e- Pupil apodisation, Focal masks: Lyot, A4Q, ALC. IR-TT sensor for fine entering Coronagraphic imaging: Dual polarimetry, direct BB + NB. λ = 0.5 – 0.9 µm, λ 0.6 µm, FoV = 3.5” 0.95 – 1.35/1.65 µm λ 0.95 µm, Spectral resolution: R = 54 / 33 FoV = 1.77” 0.95 – 2.32 µm; λ 0.95 µm Differential imaging: 2 wavelengths, R~30, FoV = 12.5’’ Long Slit spectro: R~50 & 400 Differential polarization Nasmyth platform, static bench, Temperature control, cleanliness control Active vibration control Beam control (DM, TT, PTT, derotation) Pola control Calibration Concept overview

Implementation CPI IRDIS IFS ZIMPOL ITTM PTTM DM DTTP DTTS WFS De-rotator VIS ADC NIR ADC Focus 1 Focus 2 Focus 3 Focus 4 NIR corono VIS corono HWP2 HWP1 Polar Cal

IRDIS IFS ZIMPOL CPI (structure, damping system..) Calibration sources SPHERE in test environment

SPHERE small house…

Simulations: risks and difficulties Variety of effects to be taken into account, various timescales Variety of effects to be taken into account, various timescales Needed accuracy Needed accuracy Metric to consider ? Metric to consider ? Impact on final contrast performance after data reduction Impact on final contrast performance after data reduction Inter-dependency of various contributors to the budget Inter-dependency of various contributors to the budget Risk: forget an important limitation source, system imperfection Risk: forget an important limitation source, system imperfection Risk: in the limit of very bright targets, including only well understood defects (also considered in data reduction): syndroma of infinite performance Risk: in the limit of very bright targets, including only well understood defects (also considered in data reduction): syndroma of infinite performance

separate limitation types: separate limitation types: « noise » contributors derived from mean image properties and system properties: stellar halo, detector, background, FF « noise » contributors derived from mean image properties and system properties: stellar halo, detector, background, FF speckle calibration residuals speckle calibration residuals Simulation approach

Requirements for simulation Tools To estimate mean image profile in various conditions (AO correction conditions, coronagraphs, filters)  photon noise, FF noise… To estimate sensitivity to WFE dependencies, in particular as a function of wavelength (various sources of chromatism), and time  system specifications To estimate sensitivity to WFE dependencies, in particular as a function of wavelength (various sources of chromatism), and time  system specifications to estimate long-term speckle pattern evolution to estimate long-term speckle pattern evolution Detailed AO loop simulation  system specification on stability, noise propagation, control laws, vibration filtering, impact of calib errors IFS internal behaviour  system specifications on cross-talk, samplings, detector calibration accuracy… IFS internal behaviour  system specifications on cross-talk, samplings, detector calibration accuracy…

Baseline diffraction code Baseline including: Baseline including: Fraunhofer propagation through pupil and focal planes Fraunhofer propagation through pupil and focal planes AO-filtered turbulence residuals, based on analytical model: simulation of n independent realization of residuals AO-filtered turbulence residuals, based on analytical model: simulation of n independent realization of residuals realistic optics wfe in terms of rms and DSP realistic optics wfe in terms of rms and DSP various coronagraph types various coronagraph types WFE: clear distinction between before/after coronagraph, time variable, chromaticity WFE: clear distinction between before/after coronagraph, time variable, chromaticity Limitations or effects treated on a case-by-case basis no out-of-pupil WFE in baseline. Fresnel effects only roughly estimated for some of the most sensitive surfaces (PROPER code, see J. Krist’s presentation) no out-of-pupil WFE in baseline. Fresnel effects only roughly estimated for some of the most sensitive surfaces (PROPER code, see J. Krist’s presentation) Amplitude defects: few test cases but not treated exhaustively Chromatism not treated exhaustively: mainly residual tilt, defocus, diff WFE for IRDIS dual band imaging Difficulty to handle very different timescales from ms to hr

Fulfilling the Requirements To estimate mean image profile in various conditions (AO correction conditions, coronagraphs, filters)  photon noise, FF noise… To estimate sensitivity to WFE dependencies, in particular as a function of wavelength (various sources of chromatism), and time  system specifications To estimate sensitivity to WFE dependencies, in particular as a function of wavelength (various sources of chromatism), and time  system specifications to estimate long-term speckle pattern evolution to estimate long-term speckle pattern evolution Detailed AO loop simulation  system specification on stability, noise propagation, control laws, vibration filtering, impact of calib errors IFS internal behaviour  system specifications on cross-talk, samplings, detector calibration accuracy… IFS internal behaviour  system specifications on cross-talk, samplings, detector calibration accuracy…

Fulfilling the Requirements From x,y,  cubes to simulated detector images: Including fraction on lenslets and downstream optics Possible detector and calibration defects Time consuming: can often be restricted to a small part of the image On operational point of view: definite support to data reduction testing On Signal/performance point of view: Some tests about sensitivity to cross-talk Analysis not so easy ; some coupling with image spatial structure, data extraction algo, and final impact depends also on the temporal structure (in case of artifact, how wold they smooth down ?) A point to be precised in very good performance regime (other aspects well handled) and/or if problem in terms of IFS internal calibration and signal extraction IFS internal behaviour

Fulfilling the Requirements End-2-end simulation from incoming turbulent WFE, through AO signal at each iteration and AO residual estimates Time consuming: Used for AO loop system analysis and specifications Used to validate simpler analytical tools (based on DSP filtering) NOT used for numerous overall instrument image estimates Good feedback from tests: good agreement between end-2-end simus and actual loop behaviour. good agreement between end-2-end simus and actual loop behaviour. Validation of analytical tool Pending confirmation concerning lowest flux AO loop detailed study

Fulfilling the Requirements Simple use of diffraction code, low sensitivity to exact WFE values. Main dependencies: AO correction quality Coronagraph Mean wavelength. (+ also, spectral width if speckle constrast is concerned) combined use with analytical formulae for extrapolation to numerous astrophysical cases (star type, distance, magnitude, total integraton time, individual detector integration time…) and comparison of various noise sources Mean Images

Fulfilling the Requirements Strong support to system specifications, with numerous and coupled system parameters, also depending on signal extraction: need for an underlying very schematic approach for differential imaging: whatever the exact signal extraction algo, we want to minimize The chromaticity of speckle pattern obtained simultaneously The image variability over time, with specific attention to two timescales: Minutes : time for companion differential imaging with field rotation (or polarimetric/filter switches) Hour: typical type for deep integration, potential re-calibration of the system Numerous tests made with just simple differences between two distinct images: Not representative to final perf But good estimate to pvarious parameter sensitivity Use of baseline diffraction code, in various conditions; Note: AO residuals not critical here. Sensitivity to WFE dependencies

Fulfilling the Requirements Support and typical example for various ADI data reduction: 4 hr continuous observation, crossing meridian, sampled with 144 mean images (1 every 100 s). Use of diffraction code including various timescales and variability effects: Approximative representation of turbulence residuals (not sampling AO loop iterations): average of 100 independent realization, statistics not fully converged but correlated on successive images. Predictible WFE evolution: rotating surfaces, ADC residuals associated to airmass Slow random effects: Turbulence non-stationarity: seeing, windspeed Thermal drifts effects on centering, defocus Production of x,y,l,t cubes (limited to small number of wavelengths up to now…) Long-term speckle pattern evolution

Fulfilling the Requirements Long-term speckle pattern evolution eg: DEC = -45° from Paranal, HA = -2 -> +2 hr eg: DEC = -45° from Paranal, HA = -2 -> +2 hr Airmass and Field rotation rate Evolution of pre-coro WFE var Evolution of pre-coro variability between successive images PSD of difference between WFE: 0 -1 PSD of difference between WFE: 0 -71

Fulfilling the Requirements Long-term speckle pattern evolution eg: DEC = -45° from Paranal, HA = -2 -> +2 hr eg: DEC = -45° from Paranal, HA = -2 -> +2 hr turbulence r0 Wind speed Variable de-centering

Fulfilling the Requirements Long-term speckle pattern evolution

Lessons learnt We have the ability to simulate many effects eventhough some care/effort may be involved, and possibly some simplifying tricks (for computation time saving) We have the ability to simulate many effects eventhough some care/effort may be involved, and possibly some simplifying tricks (for computation time saving) Eg from very short ot long timescales Fresnel effects on numerous surfaces Detailed effects internal to IFS… A major challenge for system/performance analysis = the number of parameters involved and combination to signal extraction approach Strong relation to system/signal: an a priori analysis of defects is needed first qualitatively => simulations provide the quantitative answer Single parameter sensitivity analysis in context of a given system paradigm = more robust than multi-parameter investigation Conclusions for a given system may not be extrapolated to another system Major risk for performance prediction = missing significant contributor ??

Risk items risk probably reasonable up to contrast risk probably reasonable up to contrast timescales around minutes at very fine levels (eventhough the system is specified for this) timescales around minutes at very fine levels (eventhough the system is specified for this) vibrations, turbulence non-stationarity vibrations, turbulence non-stationarity rotating surfaces, + ADC correction rotating surfaces, + ADC correction thermal drifts, slow loops residuals thermal drifts, slow loops residuals ultimate performance for WFE calibration ? (current assumptions for simus=quite conservative): margin for improvement and upgrades ultimate performance for WFE calibration ? (current assumptions for simus=quite conservative): margin for improvement and upgrades Amplitude/Fresnel/chromatic effects, crosstalk effects for contrasts >~10 7 Amplitude/Fresnel/chromatic effects, crosstalk effects for contrasts >~10 7 Fine detector calibration residuals ? + the unexpected !