astronomy with gravitational-wave detectors vuk mandic university of minnesota 04/02/09
TRANSCRIPT
Astronomy with Gravitational-Wave Detectors
Vuk MandicUniversity of Minnesota
04/02/09
Outline
Gravitational Waves:» What are they?
LIGO - Laser Interferometer Gravitational-wave Observatory: » Overview and Status
Recent results highlights:» Crab pulsar» GRB search» Stochastic background of gravitational waves
Outlook for the future:» Enhanced and Advanced LIGO» Underground interferometers?
Einstein's General Relativity:» Mass/Energy and Space-Time are related.
Presence of mass distorts the fabric of space-time.» Straight lines not always shortest distances.
Gravity is an effect of curved space-time.
General Relativity
Gravitational Waves
Newtonian gravity: instantaneous action at a distance. General Relativity: the “signal” travels at the speed of light. Weak field limit: Einstein’s field equations reduce to the wave equation:
Two polarizations:a,b ~ f(t - k·x)
Gravitational Waves
Two Polarizations:
“+” Polarization “x” Polarization
Sources: Binary Coalescence
Compact binary objects:» Two neutron stars, ~1.4 Solar
Masses each.» Two black holes, large mass
ranges.» Neutron star and a black hole.
Inspiral toward each other.» Emit gravitational waves as they
inspiral. Amplitude and frequency of the
waves increases over time, until the merger.
Waveform relatively well understood.
Sources: Bursts
Various other transients exists:» Supernova explosions: e.g.
collapse of massive stars.» Gamma Ray Bursts:
collapse of rapidly rotating massive stars or neutron star mergers.
» Pulsar glitches: acretion. Physics not entirely
understood. Waveform not known – search
for power excess in the data.
Sources: Periodic
Pulsars with mass non-uniformity:» Small “mountain”.» Density non-uniformity.» Dynamic processes inside
neutron star. Produce gravitational-waves at
twice the rotational frequency. Waveform well understood:
» Sinusoidal, but Doppler-modulated.
Continuous source!
Sources: Stochastic Background
Incoherent superposition of many unresolved sources.
Cosmological:» Inflationary epoch» Cosmic strings» Alternative
cosmologies Astrophysical:
» Supernovae» Magnetars
Potentially could probe physics of the very-early Universe.
Interferometers as Gravitational Wave Detectors
Gravitational wave effectively stretches one arm while compressing the other.
Time
Gravitational wave effectively stretches one arm while compressing the other.
Interferometer measures the arm-length difference.» Suspended mirrors act as
“freely-falling”.» Dark fringe at the detector.
Time
Interferometers as Gravitational Wave Detectors
Gravitational wave effectively stretches one arm while compressing the other.
Interferometer measures the arm-length difference.» Suspended mirrors act as
“freely-falling”.» Dark fringe at the detector.
Fabry-Perot cavities in the arms» Effectively increase arm length
~100 times.
Time
Interferometers as Gravitational Wave Detectors
Gravitational wave effectively stretches one arm while compressing the other.
Interferometer measures the arm-length difference.» Suspended mirrors act as
“freely-falling”.» Dark fringe at the detector.
Fabry-Perot cavities in the arms» Effectively increase arm length
~100 times. Power-recycling mirror
» Another factor of ~40 in power.
Time
Interferometers as Gravitational Wave Detectors
LIGO Sensitivity
Rough sensitivity estimate» Input laser power: ~5 Watt
Sensitivity (ΔL) ~ (~ 10-6 m)/ Number of Bounces in Arm (~100)/ Sqrt(Number of Photons (~1021))
~ 3 × 10-19 m Strain Sensitivity:
» h = ΔL / L ~ 10-22
» L = 4 km
Seismic Noise
LIGO Sensitivity
Seismic Noise» Active and passive isolation
LIGO Sensitivity
Seismic Noise» Active and passive isolation» Suspensions» Effective “Seismic Wall” at 40 Hz
LIGO Sensitivity
Seismic Noise (<40 Hz)» Active and passive isolation» Suspensions» Effective “Seismic Wall” at 40 Hz
Thermal Noise (40-150 Hz)» Suspension wires» Internal mirror modes
Shot noise (>150 Hz)
Substrates: SiO2
» 25 cm Diameter, 10 cm thick
» Internal mode Q’s > 2 x 106
Polishing
» Surface uniformity < 1 nm rms ( / 1000)
LIGO Sensitivity
LIGO Noise Budget
Fundamental noise sources:» Seismic noise, <40 Hz» Thermal noise, 40-150 Hz» Laser-power shot noise, >150 Hz
Many technical, noise sources:» Electronics, 60 Hz harmonics» Angle-to-length couplings» Auxiliary length degrees of freedom» Laser frequency noise» Laser intensity noise» Oscillator phase noise
LIGO Observatories
4 km + 2 km
4 km L1
H2
H1
Brief History of LIGO
Inauguration
1999 2000 2001 2002 20033 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4
E2Engineering
E3 E5 E9 E10E7 E8 E11
First Lock Full Lock all IFO
10-17 10-18 10-20 10-21
2004 20051 2 3 4 1 2 3 4 1 2 3 4
2006
First Science Data
S1 S4Science
S2 RunsS3 S5
10-22
4K strain noiseat 150 Hz Strain [Hz-1/2]3x10-23
20071 2 3 4
Sensitivity History
~4 orders of magnitude over ~4 years.
LIGO reached the design sensitivity in Nov. 2005.
Science run S5 completed in Oct. 2007.
~2-year long run at design sensitivity, 1-year of H1-H2-L1 coincident data.
Data analysis in progress.
S5 Results: Pulsar Searches
Crab pulsar:» One of the most promising known local
sources, one of the LIGO “targets”.» Spin down 3.710-10 Hz/s.» Maximum strain of 1.410-24.» Model: <40% of the spin-down is due to
gravitational waves. Two analyses performed, using data up to August
2006.» Time domain: using E-M timing.» Frequency domain: spans small range in
frequency and spin-down. Final results (time domain):
» 95% UL h0<3.410-25, ellipticity < 1.810-4
» Less than 6% of the lost energy goes into gravitational waves.
Major Milestone: Beating the Crab spin-down limit!Ap. J. Lett. 683, 45 (2008)
S5 Results: Transient Searches
Short Gamma Ray Bursts (GRBs): intense flashes of gamma rays, lasting <2s.» Nearby: soft gamma-ray repeaters (SGRs).» Distant: neutron star and/or black hole merger.
GRB 070201 was observed in the direction of Andromeda galaxy (M31) by several spacecraft (Konus-Wind, Integral, Messenger, Swift).
H1 and H2 operational at the time.» Search -120/+60 sec around the GRB time.» No gravitational-wave candidate was found
(Astrop. J. 681, 1419 (2008)). Inspiral search for compact binary merger
(M⊙<m1<3M⊙, M⊙<m
2<40M⊙):
» In M31 (770 kpc) excluded at 99% confidence.» Excluded at 90% confidence out to 3.5 Mpc.
Un-modeled burst: SGR in M31 not excluded.IPN3 error box overlaps with M31
Stochastic Background of Gravitational Waves
Energy density:
Characterized by log-frequency spectrum:
Related to the strain power spectrum:
Strain scale:
Detection Strategy Cross-correlation estimator
Theoretical variance
Optimal Filter
Overlap Reduction Function
For template:
Choose N such that:
Preliminary S5 Result
Use S5 data up to Jan 22, 2007.» 140 days of effective observing
time. Preliminary S5 H1L1 result:
» Ω σΩ = (1.0 5.2) × 10-6
» H = 72 km/s/Mpc The frequency band is selected to
include 99% of sensitivity, as measured by the integrand of σ-2.» 41.5-177.5 Hz
Bayesian 90% UL:» Prior on Ω: S4 Posterior» Marginalize over calibration
uncertainties
» 90% UL: 9.0 × 10-6
Signal Injections
Software injections:» Signal added to data in
software.» Successfully recovered
down to Ω ~ 4 × 10-5. Hardware injections:
» Physically moving the mirrors.
» Successfully performed and recovered three injections (within errors).
HardwareInjections
0.6 x 10-30.3 x 10-20.2Calibration Uncertainty
0.2 x 10-30.1 x 10-20.04Statistical Uncertainty
7.0 x 10-32.3 x 10-21.8Recovered Amplitude
6.3 x 10-31.7 x 10-21.9Injected amplitude
2152913Duration (min)
3 21Injection
Future Stochastic Searches Analysis of full S5 data in progress. H1H2 pair offers potentially 10x better sensitivity, but susceptible to instrumental
correlations.» Study coherence with auxiliary (environmental etc.) channels to identify
“bad” frequencies.» Time-shift one detector with respect to the other by ~1-sec. GW correlations
should disappear, but instrumental correlations may survive. Search at the higher resonance frequencies of the arm cavities (37.5, 75 kHz). Include non-LIGO interferometers in the search (VIRGO, GEO). Directional searches:
» Search for point-sources (S4 run)» Generate stochastic GW map (analogous to CMB maps).» Spherical-harmonic decomposition.
Landscape
- Surpassed BBN bound
- Beginning to probe models.
- Window around 1 Hz.
BBN Bound
Big-Bang Nucleosynthesis model and observations constrain the total energy at the time of BBN:
In our frequency band, 41.5-177.5 Hz, and assuming frequency independent spectrum, this becomes:
» Ω0 < 1.0 × 10-5
We have now surpassed this bound.» Another important LIGO milestone!
Landscape
- Surpassed BBN bound
- Beginning to probe models.
- Window around 1 Hz.
Cosmic (Super)Strings String cusps moving at the speed of light
produce bursts of gravitational waves. Integrating over the entire Universe gives a
stochastic background. Parameters (small-loop scenario):
» loop-size parametrized by: 10-13 < < 1
» String tension: 10-12 < G < 10-6
» Reconnection probability: 10-3 < p < 1
Expected Future Timeline
Year
LIGO
2007 2009 2010 2011 2012 2013 20142008
S5Enhanced LIGOCommissioning S6 AdvLIGO Commissioning
Enhanced LIGO» Relatively small upgrade over the Initial LIGO.» Improve strain sensitivity of 4-km interferometers 2-3 times.» Test some of the Advanced LIGO concepts.» NSF support: $150M over 2009-2013 for LIGO Lab operation!
Advanced LIGO» Major upgrade: essentially every aspect of the experiment will be
significantly improved.» Improve strain sensitivity of all interferometers >10 times.» Widen the sensitive band down to 10 Hz.» Substantial increase in the number of expected accessible sources.» NSF support: $205M over 2008-2015 for ALIGO construction!
Hardware installation nearly complete.
More powerful laser (35 W instead of 6 W), with upgraded input optics.
New, active seismic isolation platform in the output port.
Output mode cleaner, suspended, in vacuum.
New locking scheme. New earthquake stops (SiO
2) to
suppress charging of the mirrors. New magnets (SmCo instead of
NdFeB) to minimize non-linear, up-conversion effects.
Enhanced LIGO
Enhanced LIGO
Modifications only planned for 4-km interferometers: H1 and L1.
H2 will not be upgraded, currently running in Astrowatch mode.
Astrowatch:» Stay “ON”, just in case...» Run by students, when there is no
interference from H1 commissioning.» Excellent training for students, while
experts can focus on H1. H2 maintains the S5 strain sensitivity,
with about 50% duty cycle.
Advanced LIGO Keep the same facilities, but redesign
all subsystems.» Improve sensitivity over the whole
frequency range.
Advanced LIGO Keep the same facilities, but redesign
all subsystems.» Improve sensitivity over the whole
frequency range. Increase laser power in arms.
Advanced LIGO Keep the same facilities, but redesign
all subsystems.» Improve sensitivity over the whole
frequency range. Increase laser power in arms. Better seismic isolation.
» Quadruple pendula for each mass
Advanced LIGO Keep the same facilities, but redesign
all subsystems.» Improve sensitivity over the whole
frequency range. Increase laser power in arms. Better seismic isolation.
» Quadruple pendula for each mass Larger mirrors to suppress thermal
noise. Silica wires to suppress suspension
thermal noise.
Advanced LIGO Keep the same facilities, but redesign
all subsystems.» Improve sensitivity over the whole
frequency range. Increase laser power in arms. Better seismic isolation.
» Quadruple pendula for each mass Larger mirrors to suppress thermal
noise. Silica wires to suppress suspension
thermal noise. “New” noise source due to increased
laser power: radiation pressure noise. Signal recycling mirror
» Allows tuning sensitivity for a particular frequency range.
Advanced LIGO
Three 4km interferometers.» Two at Hanford, one at Livingston.
Significantly larger expected number of GW sources.
Project started in April 2008. Presently finalizing designs, placing
orders.» Preliminary tests of various
subsystems.» Some of the hardware and
concepts to be tested by Enhanced LIGO.
Installation expected to start in 2011.» First results: ~2014.
Beyond Advanced LIGO
Third generation GW interferometers will have to confront (and beat) the Standard Quantum Limit» Shot noise ~ P-1/2 » Radiation pressure noise ~ P1/2
» Together define an optimal power and a maximum sensitivity for a “conventional” interferometer
» Require non-classical states of light, special interferometer configurations, …
Underground interferometers?» Significant reduction in seismic noise and gravity gradient could
allow exploring low frequencies (~1 Hz)» Further suppression of thermal noise
– Cryogenic?– Larger mirror masses?
» Lower power?
Probing ~1 Hz
Since
probing lower frequencies quickly improves sensitivity to ΩGW.
Astrophysical stochastic foregrounds appear to be minimized in this band.» Could pursue the most interesting
cosmological sources, such as inflationary models etc.
Other scientific motivation:» Most pulsars are below 10 Hz
(periodic sources).» Inspirals: larger total mass, longer
observations, possibly a dark energy probe.
ATNF Pulsar Catalogue,The Astronomical Journal 129, 1993 (2005)
Pulsar Period Distribution
Landscape
- Surpassed BBN bound
- Beginning to probe models.
- Window around 1 Hz.
Probing ~1 Hz
Since
probing lower frequencies quickly improves sensitivity to ΩGW.
Astrophysical stochastic foregrounds appear to be minimized in this band.» Could pursue the most interesting
cosmological sources, such as inflationary models etc.
Other scientific motivation:» Most pulsars are below 10 Hz
(periodic sources).» Inspirals: larger total mass, longer
observations, possibly a dark energy probe.
ATNF Pulsar Catalogue,The Astronomical Journal 129, 1993 (2005)
Pulsar Period Distribution
47
Gravity Gradient Noise
Saulson (PRD30, 732, (1984)) made first estimates.» At 1 Hz, predicts equivalent
strain of ~10-20 Hz-1/2.» Also predicts seismic and
atmospheric contributions to be similar.
Later, more detailed calculations agree with Saulson:
– Hughes and Thorne, PRD58, 122002 (1998).
– Beccaria et al, CQG15, 3339 (1998).
We need ~103x suppression at 1 Hz.
48
Why Underground?
Atmospheric fluctuations are irrelevant. Local disturbances much more
controlled:» Limited access, no passers-by... » Controlled use of machinery.» Ventilation well understood and
controllable. Much more stable environment.
» Temperature, pressure, humidity... Seismic noise is reduced exponentially.
» At 1 Hz expect ~10x suppression» Depends on depth and rock
structure.Depth (m)
Seismic suppression due to depth (Cella)
49
Why Underground?
Correlation lengths are significantly longer (several km at 1 Hz):» Speed of sound is ~5 km/s, compared to ~500 m/s on surface.» Uniform medium, not very dissipative.
– Depends on location, rock composition, geological history... Two consequences:
» Gravity gradients correlated between mirrors, depending on frequency (at 0.1 Hz, expect ~50 km wavelength...)
» Track the seismic noise with a relatively few instruments spanning the entire detector.
– Active suppression may be more plausible.– And the requirement is a bit more relaxed: need ~100x.
Generation of gravity gradient noise is different:» Not surface waves!» Bulk waves, within probably several wavelength away... » Developing finite-element models to understand this better.
50
What Do We Know About Underground?
Not much! How does seismic noise amplitude depend on depth and frequency?
» How deep should we go?» How much does rock content matter?» How much does rock non-uniformity matter?
How does correlation length change as a function of depth and frequency? » How deep should we go?» How much does rock content matter?» How much does rock non-uniformity matter?
How much can we gain by building large cavities?» Less sensitive to tilts?
What does this mean for the gravity gradient noise? How does this constrain the design of an underground GW detector? Many questions!
» Need detailed underground measurements and finite-element models to answer these questions!
DUSEL at Homestake Mine
Homestake mine, SD, chosen to host Deep Underground Science and Engineering Laboratory (DUSEL).» Expected to be operational in ~2013.
Driven by low radioactive background requirements (dark matter, neutrino experiments etc).
DUGL at DUSEL
Deep Underground Gravity Lab at DUSEL:» Benefit from the low gravitational and vibrational background.» Developing a collaboration to characterize the gravitational field noise
underground.» Minnesota, Caltech, U of Florida, Carleton, Louisiana State.
First trip in February 2008.» Preliminary measurements, established contacts.
Initiated a more detailed campaign in the summer 2008.» Build an array of seismic stations.» Monitor environmental conditions (magnetometers, thermometers...)» Combine with theoretical models to estimate gravity gradient noise.» Goal: inform the design of a multi-km gravitational-wave interferometer.
Seismic Stations
*
*
*
*N
- 300 ft Station- Ellison Formation- 250 ft from shaft
Ross Shaft
- 800 ft station- Tip of Homestake formation ledge-600 ft from shaft
- 2000 ft station- Homestake formation ledge- 2600 ft from shaft
Stations:- Concrete slabs on bedrock.- Leveled granite plates supportseismometers.- Nested rigid-foam huts to suppress ambient noise.- DAQ in a separate hut.
300ft Station
2000ft Station
First Results
In the summer 2008, operated STS-2 and two T240's next to each other at 2000ft level.
Nanometrics provided the T240's and the data acquisition for them.
Simultaneous data:» Instrument self-noise.
Good measurement of the seismic noise level at 2000 ft.» ~10x quieter than LLO at 1 Hz.
Residual microseismic peaks in the self-noise spectra probably indicate imperfections in the set-up.» Made improvements since.
LLO
2000 ft
Homestake Low Noise Model
Minimize over time for each frequency bin.
Similar (or even quieter) than the USGS New Low Noise Model (Peterson 1993)!
300 ft level also has quiet times, although not quite as good as 2000 ft.
Conclusion
LIGO achieved design sensitivity and completed a year-long science run.
Searches beginning to have astrophysical implications:» New upper limit on stochastic
background: ΩGW < 9.0×10-6
Significant improvements expected in the coming years:» Enhanced LIGO nearly
completed.» Advanced LIGO construction
under way.» Already thinking about third
generation. Stay tuned...
Environment Stability
Analyzed about 12 days of 2000ft data in Jan/Feb 2009 (Jan Harms).
Environment very stable!
Relative Humidity
Pressure
Temperature
Magnetic Field at 2000 ft
Ground Velocity
300 ft: Guralp not reliable below 50 mHz
2000 ft: above 0.7 Hz dominated by digitizer noise.
Note: 300 ft is NOT surface!
Horizontal
Vertical
300 ft
2000 ft
300 ft
2000 ft
Homestake Horizontal Motion
300ft appears to have significantly more local disturbances.Possibly due to proximity of the shaft and the hoist?
Homestake Vertical Motion
300ft appears to have significantly more local disturbances.Possibly due to proximity of the shaft and the hoist?
Coherence Between Levels
Microseismic peak appears coherent, especially in horizontal DOF.
Elsewhere the coherence is strongly reduced:» Reliable range is: 50 mHz –
0.7 Hz.» Effect of local disturbances at
300 ft?» Effect of the rock structure
between 300ft and 2000ft? Of course, we really care about
coherence at a given level.» Hopefully soon...
Horizontal
Vertical
Analysis Details Data divided into segments:
» Yi and i calculated for each interval i.
» Weighed average performed. Sliding Point Estimate:
» Avoid bias in point estimate» Allows stationarity (Δσ) cut
Data manipulation:» Down-sample to 1024 Hz» High-pass filter (32 Hz cutoff)
50% overlapping Hann windows:» Overlap in order to recover the
SNR loss due to windowing.
ii
iii Y
Y2
2
opt
i
i22
opt
i-1 i i+1 i+2
Yi
i
Science Run S5
LIGO science run S5 started on November 4 (14), 2005 at Hanford (Livingston) and ended on September 30, 2007.
One year of coincidence running at LIGO design sensitivity.
New result uses data up to Jan 22, 2007.» About 140 days of effective
observing time. Currently analyzing the rest of the
run:» Duty cycle and sensitivity
improved during the run.» Expect 2x more data and ~2x
better final sensitivity.
Typical strain sensitivitiesduring S5.
S5: H1L1 Coherence
Calculated using S5 data up to 04/07.» Using the same data segments as in
stochastic analysis. Very clean:
» 48.0 Hz (3rd harmonic of 16 Hz).» 108.85 Hz – simulated pulsar.» 60 Hz (only 0.1 Hz resolution).» Additional very weak lines at 46.45
Hz and 47.58 Hz (10 mHz resolution).– Likely instrumental.– Notching them has negligible
effect on the final result.– Currently still included.
No sign of the 1 Hz harmonics, which were present in the S4 search.
1 mHz resolution
0.1 Hz resolution
Coh = CSD2 / PSD1 / PSD2
S5: H1L1 Coherence
Calculated using S5 data up to 04/07.» Using the same data segments as in
stochastic analysis. Very clean:
» 48.0 Hz (3rd harmonic of 16 Hz).» 108.85 Hz – simulated pulsar.» 60 Hz (only 0.1 Hz resolution).» Additional very weak lines at 46.45
Hz and 47.58 Hz (10 mHz resolution).– Likely instrumental.– Notching them has negligible
effect on the final result.– Currently still included.
No sign of the 1 Hz harmonics, which were present in the S4 search.
Coh = CSD2 / PSD1 / PSD2
Data Quality
Frequency notches:
» 48.0 Hz
» 60 Hz harmonics
» Simulated pulsar lines Data Quality Cuts:
» Require |σi±1 – σi| / σi < 20% (removes 4-5% of the segments).
» Reject a handful of segments identified to contain glitches in coherence studies.
All cuts defined blindly (with unphysical time-shift).
Residual distribution consistent with gaussian.
» Passes the Kolmogorov-Smirnov test.
Residual distributionis Gaussian.
S5 H1L1 Results (1)
Point estimate integrand Y(f) reveals no signal.
S5 H1L1 Results (2)
Inverse FFT of Y(f): - Shows no signal. - Gaussian behavior.
S5 H1L1 Results (3)
Preliminary S5 H1L1 result: » Ω σΩ = (1.0 5.2) × 10-6
» H = 72 km/s/Mpc The frequency band is selected to
include 99% of sensitivity, as measured by the integrand of σ-2.» 41.5-177.5 Hz
Bayesian 90% UL:» Prior on Ω: S4 Posterior» Marginalize over calibration
uncertainties– Gaussian priors with standard
deviation 8% for L1, 13% for H1.
» 90% UL: 9.0 × 10-6
Reach as a Function of Spectral Slope
- S3 H1L1: Bayesian 90% UL.- S4 H1L1+H2L1: Bayesian 90% UL.- S5 H1L1 preliminary: Bayesian 90% UL.
- Expected S5: design strain sensitivity and 1 year exposure.
- AdvLIGO: sensitivity optimized for binary neutron star search, and 1 year exposure.
Cosmic Strings: Model Topological defects formed during phase
transitions in the early Universe. Or, fundamental or Dirichlet strings (in
string theory). Cosmic string cusps, with large Lorentz
boosts, can create large GW signals. Look for the stochastic background
created by superposing cusp signals throughout the Universe.
Calculation done by Siemens, Mandic & Creighton, PRL98, 111101 (2007).» Update on Damour & Vilenkin,
PRD71, 063510 (2005).» There are uncertainties in the
calculation.» Some of them can be resolved by
improving simulations of cosmic strings networks.
Small-loop Casep = 5× 10-3
Gµ = 10-7
Small Loop Case If loop-size at formation is determined by
gravitational back-reaction, the loops are small and of the same size.
Parameters:
» loop-size parametrized by: 10-13 < < 1
» String tension: 10-12 < G < 10-6
– Upper bound from CMB observations.
» Reconnection probability: 10-3 < p < 1
– Determines the density of strings. Spectrum has a low-frequency cutoff.
» Determined by the string length and the angle at which we observe the cusp.
Small or G push the cutoff to higher frequencies.
Spectrum amplitude increases with G and with 1/p.
Small Loop Case If loop-size at formation is determined by
gravitational back-reaction, the loops are small and of the same size.
Parameters:
» loop-size parametrized by: 10-13 < < 1
» String tension: 10-12 < G < 10-6
– Upper bound from CMB observations.
» Reconnection probability: 10-3 < p < 1
– Determines the density of strings. Spectrum has a low-frequency cutoff.
» Determined by the string length and the angle at which we observe the cusp.
Small or G push the cutoff to higher frequencies.
Spectrum amplitude increases with G and with 1/p.
Small Loop Case If loop-size at formation is determined by
gravitational back-reaction, the loops are small and of the same size.
Parameters:
» loop-size parametrized by: 10-13 < < 1
» String tension: 10-12 < G < 10-6
– Upper bound from CMB observations.
» Reconnection probability: 10-3 < p < 1
– Determines the density of strings. Spectrum has a low-frequency cutoff.
» Determined by the string length and the angle at which we observe the cusp.
Small or G push the cutoff to higher frequencies.
Spectrum amplitude increases with G and with 1/p.
Small Loop Case If loop-size at formation is determined by
gravitational back-reaction, the loops are small and of the same size.
Parameters:
» loop-size parametrized by: 10-13 < < 1
» String tension: 10-12 < G < 10-6
– Upper bound from CMB observations.
» Reconnection probability: 10-3 < p < 1
– Determines the density of strings. Spectrum has a low-frequency cutoff.
» Determined by the string length and the angle at which we observe the cusp.
Small or G push the cutoff to higher frequencies.
Spectrum amplitude increases with G and with 1/p.
Large Loop Case Recent simulations indicate that loops
could be large at formation, and therefore long-lived.
Loop distribution more complex.» Larger amplitudes of gravitational-
wave spectra. Free parameters:
» String tension: 10-12 < G < 10-6
» Reconnection probability: 10-4 < p < 1 Assuming that loop-size is 10% of the
horizon at the formation time. » Some simulations indicate that a
more complicated distribution would be more accurate, involving both small and large loops.
Interferometer Controls
6 mirrors: many degrees of freedom» Length» Angular
Input field is phase modulated:» Ein = E0 ei**cos(t)
» Carrier + sidebands at ±25 MHz.
» Carrier resonates in arms, sidebands do not!
Sample the beam at several locations, measure with photodiodes.» Output voltage is demodulated.» Pound-Drever-Hall locking.
Reflected Port
Pick-off Port
Antisymmetric Port
Example: White Noise
Define: si(f) = ni(f) + h(f) ni(f) – white gaussian noise h(f) – gaussian, colored, signal
Q(f) = 1 Coherence – “normalized” cross spectrum, averaged over N samples: Coh = CSD2 / PSD1 / PSD2
Running Point Estimate
Pre-Big-Bang: Model
Mechanism similar to inflation:» Amplification of vacuum fluctuations» Super-horizon modes are amplified
during transitions between phases. Universe goes through several phases:
» Dilaton-dominated phase» Stringy phase» Radiation, followed by matter phase.
2 free parameters:» 1 < < 1.5
» fs – essentially unconstrained But: High-frequency amplitude goes as f1
4.
» f1 depends on string related parameters, which are not well known.
» We vary it by factor of 10 around the most “natural” value.
~f3
~f3-2
fs
Mandic & Buonanno, PRD73, 063008, (2006).
Pre-Big-Bang: Results
Scan f1 - plane for fs=30 Hz. For each model, calculate ΩGW(f) and
check if it is within reach of current or future expected LIGO results.
Beginning to probe the allowed parameter space.
Currently sensitive only to large values of f1.
Sensitive only to spectra close to flat at high-frequency.
But, not yet as sensitive as the BBN bound:
S3
S4
S5 H1L1
S5 H1H2
AdvLIGO H1H2
BBN
Pre-Big-Bang: Results
Can also define:
» zs = f1/fs is the total redshift in the stringy phase.
» gs/g1 = (fs/f1), where 2 = |2 - 3|
– gs (g1) are string couplings at the beginning (end) of the stringy phase
Probe fundamental, string-related parameters, in the framework of PBB models.
Assumed f1 = 4.3 × 1011 Hz (relatively large).