1. Stefan Hild, Giovanni Losurdo and Andreas Freise Virgo week July 2008 Optimization of the Advanced Virgo Sensitivity for astrophysical Sources

2. Stefan HildVIRGO week July 2008Slide 2 Overview  Motivation for sensitivity optimization  Figures of merit for astrophysical sources  Parameters used for optimization  Optimized configurations

3. Stefan HildVIRGO week July 2008Slide 3 Motivation for sensitivity optimization  Reference is the AdV conceptual design.  In the meantime several things changed:  Thermo-optic noise  Ribbons (instead of cylindrical) Fibers  New code (Bench => GWINC)  So far optimization was done only for Inspiral sources Conceptual design: VIR-042A-07 Bench-GWINC note: VIR-055A-08 Conceptual design: SR-transmittance = 0.04, SR-tuning 0.07 BNS = 121 Mpc and BBH = 856 Mpc

4. Stefan HildVIRGO week July 2008Slide 4 What do we want to do ?? Optimisation for as many different astrophysical sources… … and we want to scan the full parameter space of the detector. http://hubblesite.org/ http://numrel.aei.mpg.de/Visualisations/

5. Stefan HildVIRGO week July 2008Slide 5 Figure of merit: Inspiral  Inspiral ranges for BHBH and NSNS coalesence:  Parameters usually used:  NS mass = 1.4 solar masses  BH mass = 10 solar masses  SNR = 8  Averaged sky location [1] Damour, Iyer and Sathyaprakash, Phys. Rev. D 62, 084036 (2000). [2] B. S. Sathyaprakash, “ Two PN Chirps for injection into GEO ”, GEO Internal Document Frequency of last stable orbit (BNS = 1570 Hz, BBH = 220 Hz) Spectral weighting = f -7/3 Total mass Symmetric mass ratio Detector sensitivity

6. Stefan HildVIRGO week July 2008Slide 6 Figure of merit: Pulsars  Tried to find a figure of merit for Pulsars. Used the summed SNR of all detectable known pulsars.  We start from ANTF Pulsar Catalogue (about 1500 radio Pulsars), http://www.atnf.csiro.au/research/pulsar/psrcat/ Observation time (we used 1yr) Integrated detector sensitivity Detector sensitivity Figure of merrit: Summed SNR Detectibility threshold Factor 10.4 = all-sky, all polarisation average 95% upper limit for the Bayesian known pulsar pipeline Spin down upper limit

7. Stefan HildVIRGO week July 2008Slide 7 Figure of merit: Bursts  Tried to find a figure of merit for Bursts.  Used the sum of the inverse sensitivity:  No spectral weighting.  in bad words: detector bandwidth / average sensitivity. Gives something like 10 25 sqrt(Hz) Detector sensitivity Width of the frequency bin No frequency cuts !

8. Stefan HildVIRGO week July 2008Slide 8 Starting point of the optimization  Analysis based on Gravitational Wave Interferometer Noise Calculator (GWINC)  Using noise levels as presented in VIR-055A-08 (S.Hild and G.Losurdo: “Advanced Virgo design: Comparison of the Advanced Virgo sensitivity from Bench 4 and GWINC (v1)”)  Figures of merrit:  Binary ranges (provided by gwinc)  Pulsar (our own Matlab function)  Burst (our own Matlab function)  Stochastic (provided by gwinc)  Multi-parameter analysis is done by Matlab-scripts querying GWINC.

9. Stefan HildVIRGO week July 2008Slide 9 Optimization Parameter 1: Signal-Recycling (de)tuning  Frequency of pure optical resonance goes down with SR-tuning.  Frequency of opto-mechanical resonanced goes up with SR-tuning Advanced Virgo, Power = 125W, SR-transmittance = 4% Pure optical resonance Opto-mechanical resonance Shot noise Radiation pressure noise

10. Stefan HildVIRGO week July 2008Slide 10 Optimization Parameter 2: Signal-Recycling mirror transmittance Advanced Virgo, Power = 125W, SR-tuning = 0.07  Resonances are less developed for larger SR transmittance.

11. Stefan HildVIRGO week July 2008Slide 11 Optimization Parameter 3: Laser-Input-Power Advanced Virgo, SR-tuning=0.07, SR-transmittance = 4%  High frequency sensitivity improves with higher power (Shotnoise)  Low frequency sensitivity decreases with higher power (Radiation pressure noise)

12. Stefan HildVIRGO week July 2008Slide 12 Limits of the Parameter optimization  Our optimisation is limited by Coating thermal noise and Gravity Gradient noise.

13. Stefan HildVIRGO week July 2008Slide 13 Example: Optimizing 2 Parameters  Inspiral ranges for free SR-tuning and free SRM- transmittance, but fixed Input power NSNS-range BHBH-range

14. Stefan HildVIRGO week July 2008Slide 14 Example: Optimizing 2 Parameters Maximum NSNS-range Maximum BHBH-range Parameters for maximum Parameters for maximum  Different source usually have their maxima at different operation points.  It is impossible to get the maximum for BNS AND BBH both at the same time !

15. Stefan HildVIRGO week July 2008Slide 15 Comparison of Baseline and Single source type optimum  The crossing point of the black lines indicate the SR- operation point from the concep- tual design.  The parameters of the conceptual de- sign were chosen to give a compro- mise between BNS and BBH range. Conceptual design: SR-transmittance = 0.04, SR-tuning 0.07 BNS = 121 Mpc and BBH = 856 Mpc

16. Stefan HildVIRGO week July 2008Slide 16 Example: Optimizing 3 Parameter for Inspiral range  Scanning 3 parameter at the same time:  SR-tuning  SR-trans  Input Power  Using a video to display 4th dimension.

17. Stefan HildVIRGO week July 2008Slide 17 Example: Inspiral ranges vs Power Already very low power (< initial Virgo) gives maximum range BBH sensitivity and 80% of the maximum BNS range. Please note: Each dot may represent Different SR-parameter.

18. Stefan HildVIRGO week July 2008Slide 18 2-Step approach for Advanced Virgo construction  The conceptual design proposes: 1.Start AdV without Signal- Recycling, but with high power. 2.At a later stage install Signal-Recycling.  We should investigate if it would not be easier to go for a different approach: 1.Start AdV with Signal- Recycling, but with only 5W (no TCS required) 2.At a later stage increase the power gradually Please note: Each dot may represent Different SR-parameter.

19. Stefan HildVIRGO week July 2008Slide 19 Maximum we can achieve  We performed identical 3-para- meter optimization for Burst, Stochastic and Pulsar sources.  Traces show the maximum value achievable vs input power. All traces are normalized to their maximum value.  BNS and Burst have their maximum at high power.  BBH, Pulsars and Stochastic have their maximum at low power (3-5W). Please note: Each dot may represent Different SR-parameter.

20. Stefan HildVIRGO week July 2008Slide 20 Optimal configurations Curves show the optimal sensitivity for a single source type.

21. Stefan HildVIRGO week July 2008Slide 21 Optimal configurations Optimal REFERNCE Scenarios:

22. Stefan HildVIRGO week July 2008Slide 22 Optimal configurations Using SR-parameters from Conceptual design with GWINC (new noise levels): Optimal REFERNCE Scenarios:

23. Stefan HildVIRGO week July 2008Slide 23 Heavy mirror option (63kg) PRELIMINARY: non comprehensive Analysis !!!  Used the reference BNS scenario (pink) and only changed the mirror weight to 63 kg (green dashed).  High frequency performance unchanged, but significant improvement below 50 Hz. 42kg: BNS= 152 Mpc BBH= 747 Mpc 63kg: BNS= 161 Mpc BBH= 961 Mpc

24. Stefan HildVIRGO week July 2008Slide 24 Summary  Recent results changed the noise boundaries of Advanced Virgo => new sensitivity optimization required.  Performed a detailed 3-parameter Optimization for 5 figures of merit (BNS, BBH, Burst, Pulsar, Stochastic).  Each source requires different parameters for optimal detection:  Can get optimal sensitivity only for a single source type.  Can get not optimal, but moderate sensitivity for more than one source type  We have now 5 optimized reference scenarios (optimal parameter sets + values of performance for each astrophysical source type).  Low power scenarios seem (at least for the beginning of Advanced Virgo) to be promising.

25. Stefan HildVIRGO week July 2008Slide 25 Outlook  Optimization is an continuous process … (until the design is fixed)  Include cavity finesse as 4th parameter into future optimization.  Proper modeling of suspension thermal noise (volunteers needed)  Investigate the effect of 62kg mirror option  Interaction with data analysis groups (Virgo and LSC):  How to make the trade-off between the different source-types? Can there be a sensible multi-source-type figure of merit?  Define the Science case (includes final figure of merit) Technical point of view: Scientific point of view:

26. Stefan HildVIRGO week July 2008Slide 26 More details … … will soon be available in:

27. Stefan HildVIRGO week July 2008Slide 27 E N D Acknowledgements  We wish to thank the LIGO Scientific Collaboration for making the GWINC code available for public use.  We are also very grateful to Matthew Pittkin (Glasgow) and Alberto Vecchio (Birmingham) for their help defining the figure of merit for pulsar sources.

28. Stefan HildVIRGO week July 2008Slide 28 Example: Optimising 3 Parameter for Burst figure of merit  Scanning 3 parameter at the same time:  SR-tuning  SR-trans  Input Power  Using a video to display 4th dimension.

29. Stefan HildVIRGO week July 2008Slide 29 Example: Optimising 3 Parameter for the Pulsar figure of merit  Scanning 3 parameter at the same time:  SR-tuning  SR-trans  Input Power  Using a video to display 4th dimension.

30. Stefan HildVIRGO week July 2008Slide 30 Figure of merrit: Stochastic  Omega is provided as standard output of GWinc

31. Stefan HildVIRGO week July 2008Slide 31 Example: Optimising 3 Parameter for 1/Omega(stochastic)  Scanning 3 parameter at the same time:  SR-tuning  SR-trans  Input Power  Using a video to display 4th dimension.

32. Stefan HildVIRGO week July 2008Slide 32 Limited Parameter space  After we fixed the design of Advanced Virgo, there will probably some restrictions about accessible range of SR-parameter.  For instance: for very small SR detunings the errorsignals “vanish”. S Hild et al 2007 Class. Quantum Grav. 24 1513-1523