1. GridLab, Eger, 31 Mar-1 Apr 2003B.Sathyaprakash@astro.cf.ac.ukp1 Potential Gravitational Applications of Grid B.S. Sathyaprakash GridLab conference, 31 Mar-1 April, Eger, Hungary

2. GridLab, Eger, 31 Mar-1 Apr 2003B.Sathyaprakash@astro.cf.ac.ukp2 Modern Astronomy Cosmic micro-wave background and big bang Optical, radio, x- and gamma-ray telescopes have revealed a lot of new objects and phenomena Pulsars X-ray binaries; gamma-ray emitting sources Supermassive Black holes Quasars and Radio galaxies

3. GridLab, Eger, 31 Mar-1 Apr 2003B.Sathyaprakash@astro.cf.ac.ukp3 Astronomy has taught us that more than 90% of the Universe is dark Even this dark matter interacts gravitationally; we should be able to ‘see’ this matter via gravitational radiation it might emit But...

4. GridLab, Eger, 31 Mar-1 Apr 2003B.Sathyaprakash@astro.cf.ac.ukp4 Plan of the talk  Gravitational waves  brief overview of gravitational waves  astronomical sources  interferometric detector projects around the world  Gravitational wave data analysis and Grid  large data sets  big collaborations  huge data base records

5. GridLab, Eger, 31 Mar-1 Apr 2003B.Sathyaprakash@astro.cf.ac.ukp5 Gravitational Waves - A simple and brief overview of the theory

6. GridLab, Eger, 31 Mar-1 Apr 2003B.Sathyaprakash@astro.cf.ac.ukp6 Newton’s law of Gravity  The force of gravity between two masses m and M separated by a distance r is F = G m M / r 2  Newton’s law of gravity transmits force instantaneously - if body M changes its position it is felt by instantaneously by body m  If Newton’s gravity is right we will be able to build a ‘gravitational telegraph’ which can transmit signals instantaneously - a violation of Einstein’s special relativity

7. GridLab, Eger, 31 Mar-1 Apr 2003B.Sathyaprakash@astro.cf.ac.ukp7 Ripples in the Fabric of Spacetime  Gravitational disturbances too travel at a finite speed - indeed the same speed as light. This is what we call gravitational waves  According to Einstein gravity is nothing but warping of spacetime  Therefore, gravitational waves are ripples in space-time warping that propagates at the speed of light.

8. GridLab, Eger, 31 Mar-1 Apr 2003B.Sathyaprakash@astro.cf.ac.ukp8 Do Gravitational Waves Exist? Inspiral in Hulse-Taylor binary pulsar Two neutron stars in orbit  Each has mass 1.4 times the mass of the Sun; Orbital period 7.5 Hrs  stars are whirling around each other at a thousandth the speed of light Eventually the binary will coalesce emitting a burst of GW that will be observable using instruments that are currently being built But that will take another 100 million years  According to Einstein’s theory the binary should emit GW  GW carry rotational energy from the system which causes the two stars to spiral towards each other and a decrease in the period  Observed period change is about 10 micro seconds per year  This decrease in period is exactly as predicted by Einstein’s theory

9. GridLab, Eger, 31 Mar-1 Apr 2003B.Sathyaprakash@astro.cf.ac.ukp9 Stellar mass GW sources - observable from ground Supernovae and birth of black holes Spinning neutron stars in X-ray binaries Relativistic Instabilities in young NS Binaries of black holes & neutron stars

10. GridLab, Eger, 31 Mar-1 Apr 2003B.Sathyaprakash@astro.cf.ac.ukp10 GW Sources observable from space  Merging super-massive black holes in galactic centers  Signals from gravitational capture of small black holes by super-massive black holes

11. GridLab, Eger, 31 Mar-1 Apr 2003B.Sathyaprakash@astro.cf.ac.ukp11 Observing the origin of the Universe

12. GridLab, Eger, 31 Mar-1 Apr 2003B.Sathyaprakash@astro.cf.ac.ukp12 Gravitational Wave Detectors

13. GridLab, Eger, 31 Mar-1 Apr 2003B.Sathyaprakash@astro.cf.ac.ukp13 Interaction of Gravitational Waves Plus polarization Cross polarization

14. GridLab, Eger, 31 Mar-1 Apr 2003B.Sathyaprakash@astro.cf.ac.ukp14 Laser Interferometric Detectors Basic Principle of Operation Laser Beam Splitter Photo Diode Mirror Laser Beam

15. GridLab, Eger, 31 Mar-1 Apr 2003B.Sathyaprakash@astro.cf.ac.ukp15 LIGO VIRGO GEO TAMA ACIGA LISA

16. GridLab, Eger, 31 Mar-1 Apr 2003B.Sathyaprakash@astro.cf.ac.ukp16 Searching for Gravitational Waves How Grid Technology Can Help

17. GridLab, Eger, 31 Mar-1 Apr 2003B.Sathyaprakash@astro.cf.ac.ukp17 A list of the problems  Computationally limited searches - bigger computers means better science  Hundreds of collaborators requiring to access data from a network of detectors distributed round the world  Events are rare but data is poor with large false alarm rates - need to examine subsidiary channels of information  A large number of database records - making sense out of garbage

18. GridLab, Eger, 31 Mar-1 Apr 2003B.Sathyaprakash@astro.cf.ac.ukp18 What are we up against? Large Data Rates  environmental background  seismic disturbances  solar flares and magnetic storms, cosmic rays,...  instrumental noise  electronic noise in feedback systems, laser frequency and intensity noise, thermal fluctuations in mirrors, vibration of suspension systems,...  Important to understand detectors before any analysis begins  a large number of channels are collected to record detector state - any analysis should look at all this data  Interferometers collect data at rates of order 10 Mbytes per second, 24/7; 300 Tbytes per year  We want to be able analyse at least part of that data

19. GridLab, Eger, 31 Mar-1 Apr 2003B.Sathyaprakash@astro.cf.ac.ukp19 Distributed data  Interferometer projects work collaboratively - all data is accessible everyone in the collaboration wherever in the world they may be  How do we make all this data available to the community?  data replication to multiple sites - GriPhyN, Triana  guaranteeing data integrity  data discovery tools and P2P data access

20. GridLab, Eger, 31 Mar-1 Apr 2003B.Sathyaprakash@astro.cf.ac.ukp20 Types of gravitational wave signals  Transients - last for a short duration so that detector motion can be neglected  Transients with known shape, e.g. black hole binaries  Transients with unknown shape, e.g. supernovae  Stochastic backgrounds  population of astronomical sources  primordial stochastic gravitational wave signals  Continuous waves - last for a duration long enough so that detector motion cannot be neglected  Typically very weak amplitude, signal power a billion times smaller than noise power  long integration times needed  slowly changing frequency depending on several parameters

21. GridLab, Eger, 31 Mar-1 Apr 2003B.Sathyaprakash@astro.cf.ac.ukp21 Near all-sky sensitivity  All sky sensitivity  Quadrupolar antenna pattern  multiple detectors to determine direction to source  Wide band sensitivity  1 kHz around 100 Hz

22. GridLab, Eger, 31 Mar-1 Apr 2003B.Sathyaprakash@astro.cf.ac.ukp22 Why GW data analysis challenging?  Signals with known shapes but unknown parameters  large parameter space  for example, 10 parameters in black hole binary search  great number of wave cycles to integrate  for example, 10 10 wave cycles in a year from a neutron star  Signals of unknown shape  uncertain and inaccurate, physical models  for example waves from supernovae and black hole collisions  Very weak signal strengths  long integration times  for example up to a year for neutron star signals  a lot of pixels on the sky due to Doppler modulation Implies the need for large computational resources

23. GridLab, Eger, 31 Mar-1 Apr 2003B.Sathyaprakash@astro.cf.ac.ukp23 Compute-intensive searches - An example  Searching for black hole binaries that last for about a few seconds in the detector band  A pattern matching technique is employed since the signal shape is known, but...  signal parameters are not known before hand  must filter the data through a large number of templates corresponding to different parameters  a search in a 10-dimensional space  Triana is currently implementing this search on a compute cluster to be extended using Grids  issues - distributed data, on-line search, load balancing  data serial search is preferred due to astrophysical reasons

24. GridLab, Eger, 31 Mar-1 Apr 2003B.Sathyaprakash@astro.cf.ac.ukp24 Knowledge discovery  Not all problems are computational resource intensive - some can be handled computationally, for example short bursts of unknown shape as in supernovae, but produce huge data bases  millions of records inserted into the database each day  must go back to the original data set to veto out false alarms (that is, spurious non-GW events produced by instrumental and environmental background)  need an automatic bridge between analysis pipeline and database  Database query functionality built into Triana...

25. GridLab, Eger, 31 Mar-1 Apr 2003B.Sathyaprakash@astro.cf.ac.ukp25 Two searches that urgently require grid technology  Searching for black hole binaries  large parameter space  masses, spins, orientations,  need to go back to numerical simulations that produced the templates and to refine the search  need to analyse thousands of subsidiary channels to confirm or veto out events  All sky search for neutron stars  week signals warranting integration of large data sets  Doppler modulation in the signal caused by the motion of the detector means billions of pixels in the sky  currently the search is restricted to targeted known sources

26. GridLab, Eger, 31 Mar-1 Apr 2003B.Sathyaprakash@astro.cf.ac.ukp26 Scientific rewards from GW observations (Very) Early Universe Gravitational Wave Observations Solar, stellar interiorsCosmology Quantum theory Astrophysics Fundamental physics Extreme Gravity