=== Amaldi GW detection ===

Yoichi Aso - IceCube/LIGO coincident observations

- IceCube neutrino detector
	- uses 1 Gton of antarctic ice, instrumented by "strings"; to be completed 2011
- expect ultralow coincidence rate between IceCube and LIGO
	- sources: GRBs, AGNs, largely unknown
	- some overlap during LIGO S5

- position determination
	- with LIGO only, a ring on the sky
	- with LIGO/VIRGO, a spot on the sky
	- with IceCube, a spot on the (northern) sky
- background determination
	- time shifts for GW detectors
	- Monte Carlo for IceCube
- GW-neutrino time coincidence
	- unknown, but use time windows 0.1-10 s
- simulation: 13 events/day LIGO, 2 events/day IceCube
	- false alarm rate (1/435) (p/1%) (Tw/1 sec)

+ coherent method, include VIRGO, do injections, more neutrino detectors

---

Nick Fotopolous - stochastic-background search with the LIGO Hanford detectors

- stochastic search is signal correlation
- overlap reduction function is 10x better than H1-L1
- same arguments apply to LCGT

- how to identify non-GW cross-correlation contributions?
	- IFO-PEM coherence: linear environmental couplings
	- IFO-IFO timeshift: all narrow band signals
- then what?
	- veto bad frequencies: IFO-PEM and timeshift agree
		- but what about nonstationary features?
	- run search on remaining frequencies
	- subtract estimate Omega_PEM
		- but PEM coverage can never be complete
	- estimate uncertainty
		- negative residual Omega_env can cancel Omega_GW and give false null; what's the probability of two large numbers canceling?
		- positive residual Omega_GW may give false positive

---

Sukanta Bose - Coherent inspiral search

- Inspiral coherent statistic
- Follow up coincidence search with coherent search
	- In non-Gaussian noise, modify statistic to include goodness of fit
	- One interferometer can dominate
- Null-stream statistic
	- Fitting data to signal as opposed to maximizing SNR
- Coherent search run on S5

---

Ruben Marinho - Astrophysics from spherical GW detectors

- Mario Schenberg started operation in Sep 2006
- All detector modes can be used to solve for both polarizations
- Can get source position
- 5 TT parameters can be gotten even with 5 transducers only
- Independent bars method

---

Antony Searle - Robust Bayesian detection of unmodeled bursts

- Bayesian statistics requires priors, criticized as subjective
	- But frequentist also has implicit/unexamined priors
- Detection problem for bursts
	- Know some things (observatory characteristics)
	- Want to know others (GW features): need priors for this
- Toy model: direction
	- Source presence prior: how plausible is it that a GW is present? p(H_signal)/p(H_noise)
	- Waveform prior: on the space of possible strain waveforms
		- Example: white-noise bursts with power-law energies
	- Can marginalize over strain, sky position, and get Bayesian odds ratio
		p(H_signal|x) / p(H_noise|x)
	- Before marginalizing, can get Bayesian sky map
		- Comparable to other frequentist methods proposed for this purpose
		- Can turn Bayesian into Gursel-Tinto with odd choice of priors: very large signal energy, and source directions distributed according to network sensitivity
		- Klimenko's soft constraint seems to have priors of infinitely small signals and nonuniform direction priors
		- Tikhonov regularization implies nonuniform direction prior
	- All these techniques work when evidence can overwhelm prior
	- But Bayesian is the most effective analysis
- Robust noise model needed to avoid being fooled by incoherent glitches
	- Can do a glitch model similar to signal model
		- Should be easily and cheaply integrated into Bayesian analysis
- Bayesian approach to bursts
	- Supersedes previously proposed methods
	- Computationally tractable