home

PCTS Workshop Day 2 Recordings:

PCTS Workshop Day 2 Talks: Neil Cornish Richard O'Shaughnessy

PCTS Workshop Day 2 Discussion Panel:

References and supporting information for talks: - Richard (source populations): > Globular clusters >>
 * Comparing predictions to LIGO:
 * Range vs horizon distance
 * Real network sensitivity: complicated; sensitivity fluctuates (1203.2674); pipeline
 * + Empirical event rates (directly relevant, testable):
 * - pulsars : see[| LSC], and [|ROS and Kim]for updates
 * - short GRBs :[| Wen-Fai and Edo's paper]
 * + Empirical mass distributions: NS (Lattimer reviews; Ozel et al), BH (Farr et al)
 * + syntheticuniverse.org : binary evolution example, based on :[| Belczynsky et al.]
 * Example mass distributions for different sets of assumptions.
 * Go to double compact objects, then comparisons, compare the standard model with 'variation 3'.
 * The blue arrows next to each plot will let you cycle through different options that are largely similar, except for two key changes (metal content and CE evolution).
 * O'Leary et al 2006 astro-ph/0508224, 2007 0701887 (start with segregation); Downing, Benacquista et al 2011 (a,b), improved code with real evolution; see also Banerjee et al [|2010 MNRAS];
 * + Supernova physics
 * - kicks and binary evolution: Kalogera et al 2008 [|AIPC]; kicks (Hobbs, Arzoumanian, ...). Kick misaligment of spin of NS (Farr et al 2011). Kick tilt of orbital plane
 * Electron capture SN channel. Implications for tilts
 * - remnant mass distributions (Ott; Fryer et al)
 * Miscellaneous short GRB
 * - Predicted offset distributions: Kelley et al 2010 [|ApJL]; Belczynski et al...
 * LIGO publications: https://www.lsc-group.phys.uwm.edu/ppcomm/Papers.html


 * Questions:**


 * 1) "What are our expectations for event rates of different sources (binaries with BHs and/or NSs, supernovae) in the Advanced detector era?"**

>
 * 2) "How uncertain are these? Is a null result a real possibility?"**
 * Worst case scenario (Richard): Extrapolate from the double neutron star merger rate, assuming constant star formation like the MW, and only look for double neutron stars. Chunglee Kim has [|a paper with updated event rate predictions] and another submitted.
 * Are these the most updated papers to read/discuss? [|LSC][|Belczynsky et al.][|Coward et al.] (Alessandra). Note also [|Wen-Fai and Edo's paper] on the opening angle of a recent short GRB (Richard)
 * How probable is to observe eccentric binary systems with Advanced LIGO? (Alessandra)
 * What is the effect of eccentricity on searches? (Duncan)
 * For Initial LIGO the effect on matched filter searches is negligible if e < 0.1 at 40 Hz. Brown and Zimmerman (2010)
 * For aLIGO it can be a concern for smaller values, as small secular effects have a long time to accumulate. The above studies needs to be performed carefully for aLIGO, as the above paper primarily looked at initial LIGO. However, this is not likely to be a concern for the very low eccentrics in field binaries, only those formed by the O'Leary and Wen mechanisms below. (Duncan)
 * However, the Burst group are also looking for eccentric binaries, although with less efficiency (Duncan)
 * And what values of eccentricity are most likely? 0.1? 0.01? 0? 0.9? (//Nico//)
 * There are distinct populations with different eccentricity distributions in the LIGO band: e<10 -4 [|O'Leary et al. (2006)], 0.1-0.5 [|Wen (2003)], 0.98 [|O'Leary, Kocsis, Loeb (2009)]. The rates are uncertain for each type, and even more uncertain is the relative fraction of different types, see[| LSC]. //(Bence)//
 * What spins and what orientations relative to the orbital angular momentum should we expect in compact binary inspirals? (Nico) I know only these papers by [|Kalogera] and by [|Postnov and Kuranov] (Alessandra)
 * 3) "How well do we think we understand the signals? through what portion of parameter space?"**


 * To detect and localize NS-NS, I would say that we know the waveforms (PN waveforms). (Alessandra)
 * We do need to worry about spin for BNS in aLIGO, though. It is not negligible like it is for iLIGO. However, methods to do this are in hand. (Duncan)
 * To detect and localize NS-BH, if BH carries spin, we need to model accurately the precession effects. (Alessandra)
 * To detect BH-BH, if total mass is below 25 Msun, mass ratio is between 1 and 10, and BH doesn't carry spin, I think we have templates, including higher harmonics, for detection. (Alessandra)
 * I agree, although we are losing some SNR by ignoring higher order modes in the searches (Duncan).
 * However, for large mass ratio (~4-10) and if BHs carry spin we don't have waveforms covering the entire parameter space and we have not yet develop detection strategies for highly precessing systems. For large total masses (>25 Msun) for which the merger contributes significantly to the SNR, and if BHs carry spin, we don't have yet templates for the precessing case, but we have templates for the aligned/antialigned case [|Ajith et al.], [|Santamaria et al.], [|Taracchini et al.]. If BHs don't carry spin we have templates, including the first five dominant modes [|Pan et al.]. A general concern for templates calibrated to NR waveforms is that they have been calibrated using NR waveforms that do not cover the entire aLIGO bandwidth [|Damour et al.] [|Ohme et al.] (Alessandra)
 * For parameter estimation of BH-BH, up to total mass around 12 Msun, all PN models are close enough [|Buonanno et al.]. For larger total masses the PN models differ, but I didn't see any thorough study in the literature that has concluded that systematics are larger than statistical and for which region of the parameter space (mass ratio and spins) this happens. (Alessandra)


 * But if we see a BH-BH, do we have sufficient models today to detect ? (Nico)
 * Detection of non-spinning stellar mass black holes is fairly robust. The problem is spinning NSBH, or asymmetric BBH with spin as Alessandra mentions above (Duncan).
 * What about parameter estimation? Do we have sufficiently accurate waveforms today for this? (Nico)
 * People have been investigating the systematic errors induced by using different templates families versus the statistical error and concluded that the latter can be much larger than the former for certain systems and certain template family models. Does this imply we should be discarding certain families? Or would it be better to determine where the PN approximation needs to be stopped to avoid introducing large systematic errors? Of course, there are certain template families that are built to be valid during the entire coalescence, so does this mean these are good everywhere in the frequency band for all masses and system parameters? (Nico)


 * 4) "What accuracies can we hope to achieve for measuring position and distance in near-real time for the expected populations?"**


 * I think we should try negative latency for NS-NS or NS-BH systems, because of Tsang et al. (Yanbei)
 * Estimates of intrinsic limits (due to SNR) and computational limits (based on the "SVD" approach) are given by Cannon et al., for NS-NS binaries. An alternative computational approach (also for NS-NS) given by Luan et al. Although these existing algorithms have not demonstrated sky localization in nearly real time, I suspect that if spin effects are negligible, then better algorithms (plus more efficient ways of moving data) will eventually reach the limit imposed by SNR. (Yanbei)


 * 5) "What accuracies can ultimately be achieved for all parameters, and with what latency?"**


 * Is this a good paper to read/discuss? [|Fairhurst et al.](Alessandra)
 * Yes, this is a good paper which covers the basics (Duncan).
 * Or this one? [|Nissanke, Sievers, Dalal, Holz (2011)] //(Bence) and this one [|Veitch] (Alessandra)//
 * Are the spins detectable? //(Bence) There are those papers [|Raymond] and [|van der Sluys] (Alessandra).//


 * 6) "What are the implications of these accuracies for finding electromagnetic counterparts? What additional science can we reasonably expect with GW+EM?"**


 * I know that what I'm going to say is not strictly astrophysics now, but if we had a coincident GW+EM event, say from a SGRB, then we have a smoking gun to constrain certain modifications of GR, such as possible Lorentz-violating terms in the gravitational interaction. That is, the coincident detection can help us learn a lot beyond astrophysics. (Nico)

>
 * //Can you be more specific? (Alessandra)//
 * This question is probably related to the properties of the relevant follow-up instruments (which I think are discussed on day 5). If we are going to trigger off of GWs, it's only useful if there's something that's reasonably matched to the GW error box; the LIGO capabilities should set the follow-up strategy. The most straightforward additional science will probably be in the GRB domain. LIGO will be able to confirm/deny binary origins, as well as constrain beaming, timescales, event rates/population synthesis, etc. (Daniel)


 * 7) "Which information can be extracted if we could observe with Advanced LIGO the late inspiral and/or merger of NS-NS or NS-BH binaries?"**
 * According to [|Lackley et al. (2012)], apart from their mass, you can only extract a single physical parameter Λ, describing its deformability. This can be used to constrain the equation of state. //(Bence)//
 * //There is also this recent paper by [|Bauswein et al.] (Alessandra)//


 * 8) "How accurately do we know the gravitational signal from the late inspiral and merger of NS-NS and NS-BH binaries?"**


 * Again this paper can be useful [|Lackley et al. (2012)] (Alessandra)


 * 9) "Can we learn anything about the astrophysical environment in which compact binaries exist from an observation of compact binary inspiral ?"**
 * That is, you have another compact object nearby, could this affect the GW signal? The answer I think is that it would have to be really close, and the likelihood of this is astrophysically very small...is this true? (Nico)
 * If the triples can drive the binary to merger, than the LIGO rate from triples may be large regardless of the fact that singles or wider binaries are even more numerous that LIGO does not detect[| Wen (2003)]. Although the effects of the third object may be negligible in the LIGO band, the eccentricity is a kind of fossil imprint of the previous evolution of the binary, telling about the possible presence of the third object. //(Bence)//
 * Another possibility is that the galaxy in which these binaries are might affect the GW signal? Again, I think the answer to this is no, because the galaxy potential is rather weak relative to the gravitational interaction of compact binaries...do we agree? (Nico)
 * Again, similar to the previous point, I think it doesn't affect the GW signal, only through the initial eccentricity. //(Bence)//
 * For EMRIs, I think these answers would be different, but of course, LIGO can't see those. Do we want to talk about 3rd generation detectors? Probably not. (Nico)


 * 10) "How many templates would be needed for detection? How about for parameter estimation?"**


 * I know the answers to these questions depend on the sources, but recent studies (see Farr and Mandel, eg), suggest that for detection of quasi-circular inspiral precessing systems, one might need on the order of 10^7 templates. Do we really need that many? If so, we'll need a better method to populate the template manifold in the late regime, because 10^7 numerical simulations of the merger would be out of the question. One might argue that EOB type templates calibrated to a "few" numerical simulations could be used to interpolate from template to template in the template manifold, but is this feasible if you still need 10^7 EOB evolutions? (these are cheap, but not free as orbital evolutions are still numerical!). How many numerical simulations do we need to calibrate the EOB for such complicated spinning precessing systems? (Nico)
 * People are suggesting alternative ways to detect waveforms from compact binaries, see e.g., [|Field et al.] and [|Kannon et al.] I am not sure I understand where that 10^(7) comes from. In some regions of the parameter space, a non-spinning template can detect spinning waveforms. It is only in some cases (large mass ratio and large spins) that you would need to increase the dimensionality of the template bank. In fact, even in the case of precession, several studies have shown that for detection you might need to add only one or two spin parameters.[|Pan et al.][|Buonanno et al.][|Ajith][|Harry et al.](Alessandra)
 * I was referring here to parameter estimation, not detection. The studies here are carried out via a Bayesian approach with MCMC methods. The posteriors are determined by the number of times the chains visit a given point in the template manifold. The problem is that, according to Ilya and Will, you need on the order of 10^7 templates to carry out such an analysis. This was discussed extensively at the April APS Meeting in Atlanta by Kalogera's group and it is currently unpublished, but results are mature enough to present at conferences with converged posteriors and chains. The systems they were modeling were unequal mass, quasi circular BH BH binaries (order 10ish Msun, I think), with spins of about 0.9 that are misaligned with the orbital angular momentum. This is not a special configuration; they could have changed the masses, mass ratios, spin magnitudes and initial alignments, and they still find they need about 10^7. Parameter estimation with non-spinning templates leads to absurd errors in the recovered parameters that are way bigger than the statistical ones.

eccentricity distributions and repeated bursts: [|eccentricity.pdf]