Gravitational Wave Detection with Pulsar Timing Arrays

with pulsar timing arrays

Michael Kramer

Max-Planck-Institut für Radioastronomie, Bonn Jodrell Bank Centre for Astrophysics, Univeristy of Manchester Outline

• Introduction Pulsar timing Pulsar timing arrays (PTAs) • Science with PTAs Theories of gravity Gravitational Wave Astronomy On-going and future experiments Status & prospects Recent progress • Conclusions Pulsars… • …cosmic lighthouses • …almost Black Holes: mass of ~1.4 Solar Mass within 20km • …objects of extreme matter – 10x nuclear density 13 – B ~ Bcr = 4.4 x 10 Gauss – Voltage drops ~ 1012 volts 10-12 – FEM = 10 Fgravity – High-temperature & superfluid superconductor • …very stable clocks:

The best pulsars keep time with an accuracy of 1 millionth of a second over 30 years (comparable to the best atomic clocks) Pulsar Timing “Simple” and clean experiment: - compare arrival times of pulses at telescope with model including… • Astrometric position, proper motion and parallax • Spin-down behaviour • Interstellar dispersion delays • Movement about possible binary companion & rel. effects - Measurement & determination is result of iterative process trying to minimize “post-fit residuals”, e.g

Fold Fold

Model

TOA Residual Coherent timing solution about 1,000,000 more precise than Doppler method! Pulsar Timing “Simple” and clean experiment: - compare arrival times of pulses at telescope with model including… • Astrometric position, proper motion and parallax • Spin-down behaviour • Interstellar dispersion delays • Movement about possible binary companion & rel. effects - Measurement & determination is result of iterative process trying to minimize “post-fit residuals”

- extremely precise measurements:

PSR J1012+5307: 15 years of observations with EPTA

P = 0.005255749014115410 ± 0.000000000000000015 s

[ Lazaridis et al. 2009 ]

! 100 billion rotations since discovery & not lost a single count! High precision measurements indeed! – “Best of” examples… Masses:

• Masses of neutron stars: m1 = 1.4398(2) M and m2 = 1.3886(2) M (Weisberg et al. 2010) • Mass of WD companion: 0.207(2) M (Hotan et al. 2006)

• Mass of millisecond pulsar: 1.67(2) M (Freire et al. 2010)

• Main sequence star companion: 1.029(8) M (Freire et al. 2010) -4 • Mass of Jupiter and moons: 9.547921(2) x 10 M (Champion et a. 2010) Spin parameters: • Period: 5.757451924362137(2) ms Note: 2 atto seconds uncertainty! (Verbiest et al. 2008) Orbital parameters: • Period: 0.102251562479(8) day (Kramer et al. in prep.) • Eccentricity: 3.5 (1.1) × 10−7 (Freire et al. in 2012) Astrometry: • Distance: 157(1) pc (Verbiest et al. 2008) • Proper motion: 140.915(1) mas/yr (Verbiest e t al. 2008) Tests of general relativity: • Periastron advance: 4.226598(4) deg/yr (Weisberg et al. 2010) • Shrinkage due to GW emission: 7.152(8) mm/day (Kramer et al. in prep) • GR validity (obs/exp): 1.0000(5) (Kramer et al. in prep.) • Constancy of grav. Constant, dG/dt/G: (9±12) x 10-13 yr-1 (Zhu et al. in prep) Gravitational wave detection: • Change in relative distance: 100m / 1 lightyear (PTAs) But with the SKA…we can do so much more, for instance: • Measure SGR A* properties: mass to 10-6, spin to 10-4 to 10-3, no hair to 10-3 to 10-2: No hair! (Liu et al. 2012) Pulsar Timing - Ability to measure small effects scales with various parameters, mostly sensitivity: W T 1 P 3 / 2 sys δ σTOA ≅ ∝ × × S /N Aeff tobsΔν SPSR

Telescope Obs. Pulsar

- Indicator:€ RMS timing residuals σTOA, current state of art ~ 50 ns

σTOA Amin,GWB ∝ 5 / 3 T NPSR × NTOA / PSR

- Important factors: - Instrumental calibration € - Interstellar weather - Source specific: Strength, period, pulse shape, “timing noise” - Need to know things accurately: time, solar system ephemerides

We even measure the masses of our planets…

Champion et al. (2010) Putting the planets on the pulsar scale…

• incorrect planet masses have severe impact on our timing residuals = measured – expected pulse times of arrival • in the future, a PTA should measure planet masses very precisely

Champion et al. (2010)

Effect of a Jupiter mass modified PTA with biweekly observations

-10 by 5x10 M! of 20 pulsars A pulsar time scale

• There is not absolutely accurate clock on Earth • Only, some 2clocksG. are Hobbs better et al. than others (or corrected later…)

the motion of the Earth about the solar system barycentre. Timing residuals are the difference between the arrival times converted to the solar system barycentre and predic- tions of those times based upon the timing model (see e.g., Edwards, Hobbs & Manchester 2006 for details). Non-zero residuals can result from an incorrect conversion from the measured ToAs to barycentric arrival times. Our ability to convert to barycentric arrival times relies, for instance, upon Hobbs et al. (2012) the accuracy of the solar system ephemeris. Many pulsars also display irregularities in rotation and changes in pulse shape that make timing difficult (e.g., Lyne et al. 2011 and references therein). A subset of pulsars, the “millisecond pulsars”, have shorter pulse periods and much more stable rotation than the “normal pulsars”. However, precise observations of millisecond pulsars show some unexplained timing irregularities which we refer to as “timing noise”. Figure 1. The top panel shows the difference between Some of the variations in the timing residuals are caused TT(BIPM11) and TT(TAI) since the year 1994. The bottom If the clock is wrong, all pulsars will be affected in the sameby way! processes that are correlated between different pulsars. panel shows the same, but after a quadratic polynomial has been These can be identified by observing an ensemble of pul- fitted and removed. sars, a so-called “Pulsar Timing Array” (PTA, e.g., Foster &Backer1990).Errorsintheterrestrialtimestandardwill (which does conform to the SI second) from EAL, various introduce exactly the same signal in the residuals for each frequency adjustments are necessary. These are determined pulsar. In contrast, errors in the planetary ephemeris used in using primary frequency standards. Frequency adjustments the timing analysis will induce timing residuals which have a are generally made slowly, a process referred to as “steer- dipolar signature on the sky and gravitational waves prop- ing”. In 1996, a decision was made to change the realization agating past the pulsar and the Earth will induce timing of the SI second that resulted in a frequency shift of about residuals with a quadrupolar signature. As shown later in 2 × 10?14.ThatshiftwasprogressivelyintroducedintoTAI this paper, it is not possible to obtain an unbiased estimate over a period of two years. As TAI itself is never retroac- of the time standard errors simply by forming a weighted tively corrected, only the post-corrected versions of TT, e.g., average of the timing residuals for different pulsars. This is TT(BIPM11), have the earlier data corrected. This leads to because of the coupling between the timing model for each the “bump” that we observe in Figure 1 around the year pulsar and the measurement of the correlated signal as well 1998. as the differing data spans for each pulsar. Although numerous clocks are used in forming TAI and In this paper we analyse data from the Parkes Pulsar there is continuous development of atomic clocks, stability Timing Array (PPTA) project (Manchester et al. 2012) to over decades is difficult to measure and maintain. It is there- develop a pulsar-based timescale which we label an Ensem- fore desirable to have an independent precise timescale valid ble Pulsar Scale (EPS). This scale has similarities to the on such long intervals. In this paper, we describe the devel- free atomic timescale EAL. The frequency of EAL needs to opment of such a timescale based on the rotation of pulsars. be steered using primary frequency standards to realise a Radio pulsars are rotating, magnetised neutron stars timescale based on the SI second. Similarly, since the intrin- that radiate beams of electromagnetic waves. For a fortu- sic pulsar pulse periods and their time derivatives are un- itous line of sight to the pulsar, these can be observed at the known for the pulsars in a PTA, the EPS is not an absolute Earth as pulses. The pulse times of arrival (ToAs) from the timescale and it must be “steered” to a reference timescale brightest and fastest-spinning pulsars can be measured with which conforms to the SI. This is achieved by first forming aprecisionof∼ 100 ns in an observation time of ∼ 1hour. timing residuals for each pulsar with respect to the reference This precision is significantly worse than that obtainable timescale, TT(TAI) in our case, and subsequently fitting a from atomic clocks, but, in contrast to individual clocks, quadratic polynomial to the residuals. Fluctuations in the can be maintained for a very long time. We note that a reference timescale with respect to the EPS can be identified pulsar-based timescale provides: and used to provide a set of corrections to that realisation of TT, thereby realising a new pulsar-based timescale. We • an independent check on terrestrial timescales using a refer to the timescale derived in this paper as TT(PPTA11). system that is not terrestrial in origin. The bottom panel of Figure 1 shows the difference between • atimescalebasedonmacroscopicobjectsofstellarmass TT(BIPM11) and TT(TAI) after a quadratic polynomial instead of being based on atomic clocks that are based on has been fitted and removed. It is this signal that we expect quantum processes. to see in comparing TT(PPTA11) with TT(TAI). • atimescalethatiscontinuousandwillremainvalidfar Earlier attempts to develop a pulsar timescale have been longer than any clock we can construct. made by Guinot & Petit (1991), Petit & Tavella (1996), 2 In order to develop a pulsar-based timescale, all phe- Rodin (2008) and Rodin & Chen (2011) .Wewillshowbe- nomena affecting the pulse ToAs must be taken into ac- low that, in contrast to our method, these earlier attempts count. These are incorporated into a “pulsar timing model” that contains the pulsar’s astrometric, rotational and orbital parameters, the effects of the interstellar medium and 2 Note that some authors (e.g., Petit & Tavella 1996; Rodin, A pulsar time scale

• It is possible to recover the same clock variation with pulsars Development of a pulsar-based timescale 7

Hobbs et al. (2012)

Figure 6. The top panel shows the sampling for the 19 pul- Figure 7. The correlated signal caused by (a) errors in the Solar sars in our sample. The lower panel shows the diﬀerence between System ephemeris and (b) from one realisation of a gravitational Hence, weTT(BIPM11) can use pulsars and TT(TAI) to also as define the solid a line.new Thetime data standard points – inwave progress! background with dimensionless amplitude of 10−15.The indicate the diﬀerence between TT(PPTA11) and TT(TAI). solid line indicates the expected correlated signal caused by the steering of TAI.

Acarefulstatisticalanalysisisnon-trivialas1)theerrorbars by coalescing supermassive black-hole binary systems (e.g., on the data points are correlated and 2) the clock errors are Sesana, Vecchio & Colacino 2008). For each simulation we constrained so that they do not include a quadratic poly- use the real sampling and ToA uncertainties as in the actual nomial. The reduced-χ2 value obtained by comparing the observations. The results from our algorithm for one realisa- expected clock signal with our data is 2.7. This simple station are shown in the bottom panel of Figure 7. This shows tistical test assumes that each data point is independent, but that, for current data spans, it is unlikely that such a signal the value does indicate that there are no large discrepancies will significantly affect the stability of the pulsar timescale. between the expected clock errors and the measurements. However, with increasing data lengths and with improve- The most obvious discrepancies between our values and ments in the ToA precision achievable, the gravitational- the expectation occur between the years 1995 and 2003. wave background could become a significant factor. In this However, 1) our pulsar data set has sparse sampling around case the clock estimation algorithm would need to be mod- this time and has not been corrected for dispersion mea- ified to make it orthogonal to the gravitational wave back- sure variations and 2) the observed discrepancies would re- ground. − quire an error in the frequency of TT(BIPM11) of ∼ 10 14 From our data, we therefore conclude that whereas the uncertainty on this frequency is thought to be − • the difference between TT(TAI) and TT(BIPM11) can ∼ 1 × 10 15 around the year 2003 (Petit 2003). This sug- be detected using pulsar data and that this difference, as gests that the discrepancies result from the determination expected, results from the deliberate steering of TAI. of ∆c(t). There may be sufficient archival observations from • there are no large unexpected errors in TT(BIPM11) other observatories to improve the clock error estimates dur- over our data span. ing this period and thus to confirm or deny these possible • the variations in TT(TAI) are at a significant level com- errors in TT(BIPM11) and its estimated uncertainty. pared with the precision of current pulsar timing array ob- It is possible that errors in the solar system ephemeris servations. We note that Guinot (1988) and subsequent pa- could lead to correlated signals in the timing residuals. To pers from the clock community have already pointed out see the maximum size of any such signal in our data, we that TT(TAI) is not suitable for high time precision pulsar have simulated observations using the same sampling and experiments and that TT(BIPM) should always be used. We ToA uncertainties as the real data using the JPL DE421 confirm that TT(BIPM11) is adequate for current millisec- solar system ephemeris, but without any clock errors. We ond pulsar timing experiments. then processed the data using the earlier JPL DE414 solar system ephemeris. The resulting estimate of the “clock er- Our results do not show the steering of TAI as clearly rors” are shown in the top panel of Figure 7. The maximum as in Figure 5a. This is mainly because the timing resid- deviation for recent data is < 100 ns. As we use the most re- uals for PSR J0437−4715 dominate the data set: it is the cent ephemeris, DE421, for our analysis it is likely that the only pulsar that was observed around the year 2000 and actual correlated signal caused by the planetary ephemeris has a large number of observations and very small ToA un- is significantly smaller than this. certainties. However, the statistical properties of the tim- In order to test whether a gravitational-wave back- ing residuals for this pulsar suddenly change around the ground signal could be mis-identified as an error in the ter- year 2006. Prior to this date, observations were made in restrial time standard, we have simulated multiple realisa- the 20 cm band and have not been corrected for dispersion tions of a gravitational-wave background (Hobbs et al. 2009) measure variations. After this date, the observations were − with a dimensionless strain amplitude of 10 15.Thisampli- made with new instrumentation, in the 10 cm band and the tude is typical of that expected for a background created dispersion measure variations have been measured and re- Pulsars as Gravitational Wave (GW) detectors

Pulse arrival times will be affected by low-frequency gravitational waves – correlated across sky!

In a “Pulsar Timing Array” (PTA) pulsars act as the arms of a cosmic gravitational wave detector:

[ Hellings & Downs 1983 ] Pulsars as Gravitational Wave Detectors

(Kramer'et'al.'2004)'

• PTA is sensitive to -6 Current PTA 0.9K PULSARS Blackbody nHz gravitational waves LIGO I Spectrum LISA Local Advanced strings LIGO • Complementary to LISA, -8 LEAP LIGO and CMB-pol band 1st Order Extended EW Phase Inflation MBH−MBH Transition • Expected sources: COBE Binaries -10 LISA LIGO II - binary super-massive 1st Order/VIRGO COBE EW Phase black holes in early Global Transition -12 strings Galaxy evolution SKA−PTA - Cosmic strings -14 - Cosmological sources Slow-roll inflation - Upper Bound CMB−POL • Types of signals: Inflation - stochastic (multiple) -16 Inflation - periodic (single) -15 -10 -5 0 5 10 - burst (single) Stochastic background

• Earliest signal expected from binary super-massive black holes in early 7 galaxy evolution (PTA only way to detect M>10 M! Porb~10-20yr) • Amplitude depends on merger rate, galaxy evolution and cosmology but could be “soon” detectable (e.g. Sesana et al. 2008)

1km in 1 lt-yr

Sesana et al. (2008), Sesana & Vecchio (2009)

1m in 1 lt-yr

10 yr 1 yr

Amplitude,strain characteristic h 1 10 100 GW frequency (nHz) Stochastic background

1km in 1 lt-yr 2010 Note: • Current best limits from European, North-American, Australian timing array are all very similar: EPTA: van Haasteren et al. (2011) NanoGrav: Demorest et al. (2013) 1m in 1 lt-yr PPTA: Manchester et al. (2013) • All are tantalizingly close to 10 yr 1 yr expected detection limit!

Amplitude,strain characteristic h 1 10 100 GW frequency (nHz) Stochastic background

• Earliest signal expected from binary super-massive black holes in early 7 galaxy evolution (PTA only way to detect M>10 M! Porb~10-20yr)

• Amplitude depends on merger rate, galaxy evolution and cosmology but could be “soon” detectable (e.g. Sesana et al. 2008)

1km in 1 lt-yr Note: • Current best limits from European, North-American, Australian timing

array are all very similar: 2015 EPTA: van Haasteren et al. (2011) NanoGrav: Demorest et al. (2013) 1m in 1 lt-yr PPTA: Manchester et al. (2013) • All are tantalizingly close to 10 yr 1 yr expected detection limit!

Amplitude,strain characteristic h 1 10 100 GW frequency (nHz) Stochastic background

1km in 1 lt-yr Note: • Current best limits from European, North-American, Australian timing array are all very similar: 2025 EPTA: van Haasteren et al. (2011) NanoGrav: Demorest et al. (2013) 1m in 1 lt-yr PPTA: Manchester et al. (2013) • All are tantalizingly close to 10 yr 1 yr expected detection limit!

Amplitude,strain characteristic h 1 10 100 See: Sesana (2013), GW frequency (nHz) & McWilliams et al. (2013) Properties of the graviton

• We can do more than “only” detect gravitational waves

• With SKA sensitivity we can study GW properties: polarisation & graviton mass

GW polarisation (Lee et al. 2009):

• Shape of curve depends on GW polarisation • Precise measurement gives test for theories of gravity! Properties of the graviton

• We can do more than “only” detect gravitational waves

• With SKA sensitivity we can study GW properties: polarisation & graviton mass

Lee et al. (2010)

Graviton mass:

present solar system limit [eV] ) [eV] g ( m log Single source detection

• Single binary super-massive black hole produces periodic signal

• Perhaps rare but complementary in mass range to LISA (Sesana et al. 2009, Sesana & Vecchio 2010) • If SNR is high (or source and orbital period known!) we can search for signature

• Expect periodic signal but also dc-term due to memory effect (van Haasteren & Levin 2010) • Signal contains information from two distinct epochs: t and t-d/c Example: if binary super-massive BH exists in 3C66B, we expected signature in

timing data of PSR B1855+09 (Jenet et al 2003): d

Observed Expected Individual sources

[ Kocsis & Sesana 2010 ]

In reach - already?

Keep looking! GW'amplitude'

…and try to match in EM window! Individual sources

[ Kocsis & Sesana 2010 ]

In reach - already?

Pobs = 5 yr Keep looking! GW'amplitude'

…and try to match in EM window! Gravitational Wave Astronomy

• We can pinpoint a single GW source:

Lee et al. (2011)

1 Possible by amazing Astrometry abilities of SKA! The European Pulsar Timing Array (EPTA)

An array of 100-m class telescopes to form a pulsar timing array

SRT,'Sardinia,'Italy'

Lovell,'Jodrell'Bank,' Eﬀelsberg'100Bm,'Germany' UK' Plus theory:

NRT,'Nancay,'France' WSRT,'Westerbork,'NL'

and ultimately forming the Large European Array for Pulsars (LEAP) The EPTA partners

Mission: “Perform high precision timing to study pulsar binaries, theories of gravity and to detect gravitational waves” Observational efforts: • MPI for Radioastronomy, Bonn, Germany • Jodrell Bank Centre for Astrophysics, Uni. Manchester, UK • ASTRON, The Netherlands • CNRS, France • INAF, Italy Complemented by strong theoretical efforts by these members: • Albert Einstein Institut, Germany: limits, detection methods • MPIfR, Germany: sources, detection & observing strategies, tests of theories of gravity • Uni. of Birmingham, UK: sources, detection • Uni. of Manchester, UK: cosmic strings • University of Cambridge (just joined!) Introducing LEAP

A Large European Array for Pulsars = LEAP! Large European Array for Pulsars Coherently add pulsar observations from 5 of the Coherently add pulsar observations from the 5 largest telescopes in Europe (and the world!) to large radio telescopes in Europe to obtain most obtain most precise TOA’s for GW detection. precise TOA’s for GW detection. Combine telescopes to form a phased array,a telescope with equivalent size of a 200 m dish.

Combine telescopes to form a phased array, a A LEAP in collecting area. telescope with equivalent size of a 200 m dish. Funding

ERC grant to Michael Kramer (2.5Me) for 2 A LEAP in collecting area. senior PDRA’s, 5 junior PDRA’s

Cees Bassa (JBCA) The LEAP Project JBCA Internal Symposium: June 16th, 2010 6 / 9 Funded by ERC Advanced Grant (PI Kramer) Almost there: 160-m dish! Only 10-sec of data: Other complementing ways to improve sensitivity

Increase sensitivity by much wider bandwidth:

The “Ultra-Broad-Band Receiver” on 100-m telescope

Project BEACON funded by ERC Grant to Paulo Freire: Competitive with Arecibo but full sky

Less than two years from conception to first light!

Removing the interstellar weather!! Other complementing ways to improve sensitivity

Increase sensitivity by much wider bandwidth:

The “Ultra-Broad-Band Receiver” on 100-m telescope

Project BEACON funded by ERC Grant to Paulo Freire: Competitive with Arecibo but full sky

Removing the interstellar weather! A Phased-Array-Feed to Find more pulsars!

On 100-m in 2013/Q4

Survey speed dramatically increased.

Note: in both cases, we need to fight RFI effectively! Detection significance

• EPTA and LEAP are well placed, with current sample • Ideally, we want much more – also more pulsars!

Van Haasteren et al. (2008)

Ideally…

• We are joining forces and have formed the IPTA (see e.g. Hobbs et al. 2009) • We may not far away, in particular with recent progress, but ultimate goal is the SKA – then it will be easy! Until then… “High time resolution Universe” all-sky survey - The sensitivity to periodic signals depends on the ability to resolve pulses:

! 2 2 2 2 2 #1/2 τ obs = τ intrinsic +τ samp +τ DM +τ acc +τ scatt " $ - Recent improvements in hardware that utilize field programmable gate arrays to produce high resolution digital filterbanks. - Dispersion smearing across individual channels reduced. - New all-sky survey to find highly dispersed MSPs! Collaboration: JBCA, MPIfR, CASS, Swinburne, Cagliari Parkes Parkes BPSR analogue digital filterbank filterbank Number of 96 1024 channels Channel 3 MHz 0.39 MHz bandwidth Dispersion 10 µs per DM 1.3 µs per DM smearing unit unit central channel

HTRU north - Effelsberg - All sky survey divided into three zones: low lat, mid lat, high lat. Wavelength: 21cm Integration times: 25min, 3min, 1.5min Sampling interval: 54 us Number of channels: 512 Total data volume: PB - Lots of ground work before the survey could begin: Receiver orientation &sensitivity, Parallactic angle feed rotation, Scheduler, Observations database, Processing database, Remote observing, Processing pipeline, Tape archives..... - PhD work of Ewan Barr. HTRU north - Effelsberg - All sky survey divided into three zones: low lat, mid lat, high lat. Wavelength: 21cm Integration times: 25min, 3min, 1.5min Sampling interval: 54 us Number of channels: 512 Total data volume: PB - Lots of ground work before the survey could begin: Receiver orientation &sensitivity, Parallactic angle feed rotation, Scheduler, Observations database, Processing database, Remote observing, Processing pipeline, Tape archives..... - PhD work of Ewan Barr. HTRU south – Parkes - Same three zones: low lat, mid lat, high lat (med-lat already complete) Wavelength: 21cm Integration time: 70 min Sampling interval: 64 us Number of channels: 1024 Total data volume: 250 TB HYDRA@JBCA' - Deep survey PhD work of Cherry Ng. - Significant processing requirements: Manchester HYDRA 2000 CPU cluster.

Current status: Detected all known PSRs, more than 30 pulsars already (>100 overall)

- All HTRU will produce about 2 Petabyte

- Processing requirements comparable to GW detection.

A short-cut to find PTA pulsars…

FERMI does not only detects known radio pulsars (about 50 already!)… …. But a lot of unidentified FERMI point sources are pulsars, in particular MSPs!

Ge#ng&a&li*le&from&friends…&

FZ'Jülich'

• Processing, analysis, archiving • Applications today: HTRU, P-ALFA, LOFAR, FERMI • GPU extremely successful • AEI' Future: MeerKAT, ASKAP, SKA The Square Kilometre Array (SKA)

Kramer et al. 2004, Cordes et al. 2004, Smits et al. 2008)

We gain twice: - find many more pulsars - time them to much higher precision

- ~30,000 normal pulsars - ~2,000 millisecond psrs - ~100 relativistic binaries - first pulsars in Galactic Centre - first extragalactic pulsars

" timing precision is expected to increase by factor ~100 " also rare and exotic pulsars and binary systems: PSR-BH systems! " GW detection should be “easy” Important implications: Significant demand also on signal transport, processing & storage

Today’s best: European Very Long Baseline Interferometry Network (1 Gbit/s) (see EC-FP6 EXPReS)

PAF/UBB: ~100 GB/s " On-line processing!

Extreme - SKA data rates: 80 Gbit/s/beam/dish (<200km) ~100'Tbits/sec!' 40 Gbit/s/station (20 dishes) (>200km)

About 1 Exabyte per day!!!! Synergies and Complementarities

LISA & PTA: Tests of strong-field gravity Gravitational wave astronomy & astrophysics In both cases, PTAs and LISA observe:

Similar sources Identical sources Complementary sources (“counterparts”) e.g. different e.g. combining different e.g. different evolutionary stages accessible information mass ranges Synergies and Complementarities

LISA & PTA: Tests of strong-field gravity Gravitational wave astronomy & astrophysics In both cases, PTAs and LISA observe:

Similar sources Identical sources Complementary sources (“counterparts”) e.g. different e.g. combining different e.g. different evolutionary stages accessible information mass ranges

Population of (S)MBH and galaxy evolution: • For systems with similar mass, LISA population will be at later evolutionary stage than those observed with PTA • First BHs in Universe seen with SKA compared to LISA population in normal galaxies & AGNs

Synergies and Complementarities

LISA & PTA: Tests of strong-field gravity Gravitational wave astronomy & astrophysics In both cases, PTAs and LISA observe:

Similar sources Identical sources Complementary sources (“counterparts”) e.g. different e.g. combining different e.g. different evolutionary stages accessible information mass ranges

Population of (S)MBH and galaxy evolution: • Triggered search in PTA data for pre-ring-down counterparts of

massive LISA sources (Pitkin et al. 2008) Compact binary systems (SKA in general):

• Simultaneous observations provide FULL description (distance!) (Gopukumar et al. in prep.) Distance scale (SKA in general) • HI redshifts for counterparts to GW sources in 1-billion galaxy survey

Synergies and Complementarities

LISA & PTA: Tests of strong-field gravity Gravitational wave astronomy & astrophysics In both cases, PTAs and LISA observe:

Similar sources Identical sources Complementary sources (“counterparts”) e.g. different e.g. combining different e.g. different evolutionary stages accessible information mass ranges

Properties of differently sized BHs: • LISAs (S)MBH and intermediate mass BH and LIGOs stellar BHs vs. radio

pulsar binaries around stellar BH, in clusters and around SGR A*

(S)MBH binaries: • low-mass observations with LISA, high-mass detection with PTA, ie. 7 only means to detect binary BHs with ≥10 M! (e.g. Sesana & Vecchio 2010) • combined will provide superb constraints on galaxy evolution

Conclusions

• Pulsar Timing Arrays offers the means to detect nHz-gravitational waves • The frequency coverage and science is complementary to ground-based detectors and LISA • We can see combine results to address and understand galaxy and massive binary black hole formation across the history of the Universe • Stochastic, single and burst sources can be detected • With the SKA it should be easy to do real GW astronomy and to test GW properties in GR and alternative theories of gravity • The current experiments try hard and may soon detect something

Stay tuned!