1: %\documentclass[12pt,preprint]{aastex}
2: \documentclass[12pt,preprint]{emulateapj}
3:
4: \usepackage{amsmath, amssymb, latexsym}
5:
6: \newcommand{\Msun}{\textrm{M}_{\odot}}
7: \newcommand{\etal}{et al.}
8: \newcommand{\Mbh}{M_{\bullet}}
9:
10: \begin{document}
11:
12: \title{Event Rate for Extreme Mass Ratio Burst Signals in the Laser
13: Interferometer Space Antenna Band}
14:
15: \author{Louis J. Rubbo, Kelly Holley-Bockelmann, and Lee Samuel Finn}
16: \affil{Center for Gravitational Wave Physics, Pennsylvania State
17: University, University Park, PA 16802}
18:
19:
20:
21: \begin{abstract}
22: Stellar mass compact objects in short period ($P\lesssim10^3$~s)
23: orbits about a $10^{4.5}$--$10^{7.5}\,\rm{M}_{\odot}$ massive black
24: hole (MBH) are thought to be a significant continuous-wave source of
25: gravitational radiation for the ESA/NASA Laser Interferometer Space
26: Antenna (LISA) gravitational wave detector. These extreme
27: mass-ratio inspiral sources began in long-period, nearly parabolic
28: orbits that have multiple close encounters with the MBH. The
29: gravitational radiation emitted during the close encounters may be
30: detectable by LISA as a gravitational wave burst if the
31: characteristic passage timescale is less than $10^5$~s. Scaling a
32: static, spherical model to the size and mass of the Milky Way bulge
33: we estimate an event rate of $\sim\!15~\textrm{yr}^{-1}$ for such
34: burst signals, detectable by LISA with signal-to-noise ratio greater
35: than five, originating in our Galaxy. When extended to include
36: Virgo Cluster galaxies our estimate increases to a gravitational
37: wave burst rate of $\sim\!18~\textrm{yr}^{-1}$. We conclude that
38: these extreme mass-ratio burst sources may be a steady and
39: significant source of gravitational radiation in the LISA data
40: streams.
41: \end{abstract}
42:
43:
44: \keywords{black hole physics --- Galaxy: nucleus --- gravitational
45: waves --- stellar dynamics}
46:
47:
48: \section{Introduction} \label{sec:intro}
49:
50: The inspiral of compact objects onto massive black holes (MBHs) in
51: galactic nuclei is an anticipated important gravitational radiation
52: source for the ESA/NASA Laser Interferometer Space Antenna (LISA)
53: \citep{hils:1995:gac, sigurdsson:1997:csm, glampedakis:2002:zwe,
54: glampedakis:2002:ait, ivanov:2002:frc, freitag:2003:gws,
55: gair:2004:ere, danzmann:2003:ltc, sumner:2004:ogw, jennrich:2005:pil,
56: seto:2001:pdm}. These extreme mass-ratio inspiral (EMRI) sources will
57: complete $>10^4$ orbits during the LISA mission lifetime
58: ($\sim\!5$~yrs) and it is the accumulation of the signal power,
59: emitted continuously with frequency $\gtrsim10^{-3}$~Hz, that makes
60: them visible to LISA.
61:
62: However, before emerging as EMRIs these objects were on long-period,
63: nearly radial orbits that were ``captured'' by the MBH after the
64: orbital energy was reduced by a series of short, intense bursts of
65: gravitational radiation emitted during each close encounter with the
66: MBH. If the encounter timescale is less than $\sim\!10^5$~s the
67: gravitational wave burst will be in the LISA band and, if strong
68: enough, detectable by LISA. Here we describe a preliminary analysis
69: of this extreme mass ratio burst (EMRB) phenomenon, characterizing the
70: detectability and estimating the rate of EMRB events in LISA. Our
71: estimates show that the Milky Way may be responsible for $\sim\!15$
72: observable EMRB events per year, and that the Virgo Cluster galaxies
73: may be responsible for an additional $\sim\!3$ events per year, making
74: EMRBs a significant source of gravitational waves for LISA.
75:
76:
77: \section{Inspirals vs. Bursts} \label{sec:process}
78:
79: To illustrate the evolution from EMRB to EMRI consider a $0.6~\Msun$
80: white dwarf in a nearly radial Keplerian orbit about a MBH of mass
81: $\Mbh\sim10^{6}~\Msun$. If the initial orbit has an apocenter of
82: 100~pc and a pericenter of 0.2~AU (corresponding to 10 Schwarzschild
83: radii for the MBH), then during the first pericenter pass the orbit
84: loses enough energy via gravitational radiation that the next
85: apocenter is $\sim\!60$~pc, while the change in pericenter is
86: negligible. With each subsequent pericenter pass, a strong burst of
87: gravitational radiation reduces the subsequent apocenter and the orbit
88: becomes more circular. LISA is sensitive to gravitational waves in the
89: approximately $10^{-5}$ to $10^{-1}$~Hz band, so we classify as EMRBs
90: those systems with orbital periods greater than $10^5$~s and with
91: pericenter passage timescales of less than $10^5$~s. Conversely, EMRIs
92: have orbital periods less than or on order $10^3$~s and radiate
93: continuously in the LISA band.
94:
95: EMRIs and EMRBs sample different components of a galactic nucleus.
96: Every EMRI source was, in its past, a \emph{possible} EMRB source:
97: i.e., it was on a highly eccentric orbit that, owing to gravitational
98: wave emission, circularized and decayed until it entered the LISA band
99: as a continuous source. However, if the MBH encounter timescale is too
100: long, the orbit may become an EMRI without any significant radiation
101: burst in the LISA band.
102:
103: Not all EMRBs evolve to become EMRIs, either. An object on a highly
104: eccentric orbit may be scattered by other stars after encountering the
105: MBH, or may plunge directly into the MBH. In addition, while massive
106: main sequence stars will be disrupted long before appearing as EMRIs,
107: the periapsis of an EMRB orbit can be much greater than the tidal
108: disruption radius for low-mass main sequence stars. For example, an
109: M2V star could pass within 0.65~AU of the Milky Way's MBH without
110: disruption and, if on a nearly radial orbit so that its pericenter
111: velocity was highly relativistic, may radiate a gravitational wave
112: burst with a characteristic frequency of several mHz.
113:
114:
115: \section{Stellar Model} \label{sec:model}
116:
117: To evaluate the EMRB event rate we begin with a model of our Galaxy
118: center. The Milky Way bulge consists of a $\Mbh =
119: 3.7\times10^6~\Msun$~MBH \citep{ghez:2005:soa} embedded in a stellar
120: ellipsoid \citep[e.g.][]{binney:1997:psi, stanek:1997:mgb,
121: lopez-corredoira:2000:iss} of mass $(1.3$--$2)\times10^{10}\,\Msun$
122: \citep[e.g.][]{gerhard:2002:mdo}, with axis lengths of
123: $\sim\!1.8:0.7:0.5$~kpc \citep{bissantz:2002:sab}, and with a cuspy
124: density profile that tends toward $\rho \sim r^{-1.8}$ as $r
125: \rightarrow 0$ \citep{matsumoto:1982:boc}. Although the Milky Way is a
126: barred galaxy, we treat the bulge as spherical with a density profile
127: described by an $\eta$ model \citep{tremaine:1994:fms}. Since $\eta$
128: models are static, spherical, and isotropic their distribution
129: function depends only on energy, making the model an ideal first
130: choice to analytically estimate the number of bursts in the bulge. We
131: choose $\eta=1.25$ to match the observed slope of the Milky Way inner
132: cusp and set the effective radius equal to 2~kpc, total stellar mass
133: to $2 \times 10^{10}~\Msun$, and MBH mass to $4 \times 10^6~\Msun$.
134:
135: Embedding an $\eta$ model with a pre-existing central MBH leaves the
136: stellar density profile unchanged, but changes the cluster potential
137: so that $\Phi_{\bullet}(r) \equiv \Phi_{\star}(r) - \Mbh/r$, where
138: $\Phi_{\bullet}$ is the potential of the MBH-embedded model and
139: $\Phi_{\star}$ is the stellar potential of an $\eta$ model. The new
140: potential naturally adjusts the distribution function, which can be
141: obtained from a spherical, isotropic density profile via Eddington's
142: formula \citep{binney:1987:gd}, which for a MBH-embedded model can be
143: solved analytically near the MBH:
144: \begin{equation}
145: f(\epsilon) = \frac{\eta \Gamma(4-\eta)}{2^{7/2} \pi^{5/2}
146: \Mbh^{(3-\eta)} \Gamma(5/2-\eta)} \; \epsilon^{3/2-\eta} \,,
147: \end{equation}
148: where $\epsilon$ is the absolute value of the energy per unit mass.
149: When normalized to the total mass of the model,
150: $f(\epsilon)d{\boldsymbol r} d{\boldsymbol v}$ is the mass contained
151: in the local volume element centered on ${\boldsymbol r}, {\boldsymbol
152: v}$.
153:
154:
155: \section{Burst Waveforms} \label{sec:waveforms}
156:
157: Orbits that may generate observable EMRBs are characterized by large
158: orbital periods ($\sim\!1 - 10^{9}$~yrs) and high
159: eccentricity. Objects on these orbits may be highly relativistic
160: during periapsis passage, with velocities an appreciable fraction of
161: the speed of light ($v_{p} \sim 0.3c$). The gravitational wave
162: emission from such a ``gravitational bremsstrahlung'' event emerges as
163: a burst, lasting as long as the MBH encounter and beamed in the
164: direction of the secondary's velocity at periapsis. For this
165: preliminary exploration we treat trajectories near periapsis as
166: Keplerian,\footnote{Although we only considered elliptical systems
167: within the $\eta$ model, it is worth noting that the Keplerian
168: equivalent orbit near the MBH is not necessarily elliptical.} with the
169: same $v_{p}$, $r_{p}$, and mass enclosed within $r_{p}$ as dictated by
170: the $\eta$ model, and we calculate the gravitational radiation as if
171: it were well described by the quadrupole formula. A more detailed
172: treatment would include the effect of beaming, the radiative
173: contributions from terms in $v/c$ higher than quadrupole, and take
174: into account the actual relativistic trajectory, all of which may be
175: important for trajectories with small $r_{p}/\Mbh$ or large $v/c$
176: \citep[cf.][]{gair:2005:sagr}.
177:
178: From the (quadrupole) waveforms we calculate the signal-to-noise ratio
179: (SNR) in the LISA detector assuming the source is at a distance of
180: 8~kpc. In general, the square of the SNR is given by,
181: \begin{equation} \label{eq:snr_defined}
182: \rho^{2} = 4 \int_{0}^{\infty}
183: \frac{|\widetilde{h}(f)|^{2}}{S_{n}(f)} \, df \,,
184: \end{equation}
185: where $\widetilde{h}(f)$ is the Fourier transform of the gravitational
186: wave signal projected onto the LISA detector and $S_{n}(f)$ is the
187: one-sided instrument noise power spectral density. For our purposes,
188: the values of $S_{n}(f)$ are taken from the \textit{Online Sensitivity
189: Curve Generator} \citep{larson:2000:scs} with the standard LISA
190: settings and the inclusion of a white dwarf background ``noise''
191: contribution \citep{bender:1997:cnl}.
192:
193: To gain insight into the magnitude and scaling of the SNR,
194: equation~\ref{eq:snr_defined} can be approximated by taking advantage
195: of the simple burst structure in the waveforms, which allows us to
196: quickly approximate the integral and Fourier transform,
197: \begin{eqnarray}
198: \rho &\approx& 100 \left( \frac{R}{8~\textrm{kpc}} \right)^{-1} \left(
199: \frac{M_{\star}}{\Msun} \right) \bigg( \frac{v_{p}}{0.3 c}
200: \bigg)^{2} \left( \frac{f_{\star}}{1~\textrm{mHz}} \right)^{-1/2}
201: \nonumber\\
202: && \times \left( \frac{S_{n}(f_{\star}/1~\textrm{mHz})}{4 \times
203: 10^{-37}~\textrm{Hz}^{-1}} \right)^{-1/2} \;,
204: \end{eqnarray}
205: where $R$ is the distance to the EMRB, $M_{\star}$ is the mass of the
206: secondary, and $f_{\star}= v_{p}/r_{p}$ is the inverse of the burst
207: duration. The value of $f_{\star}$ can be viewed as a characteristic
208: frequency for an EMRB event, but these events are extremely broadband
209: in nature. Note that the noise spectral density scales as $f^{-4}$
210: for frequencies below 1~mHz and the pre-factor accounts for the white
211: dwarf binary background.
212:
213:
214: \section{Event Rates} \label{sec:results}
215:
216: To arrive at an EMRB event rate estimate, we divided our $\eta$ model
217: into five million elements in $(r,v)$ phase space. We chose the
218: element centers uniformly spaced in $\log(r)$ and $\log(v)$ with
219: $r\in[7.6\times10^{-7},2\times10^4]$~pc and
220: $v\in(0,2\times10^5)~\textrm{km s}^{-1}$. For each phase-space
221: element we calculated $f(\epsilon)$, the average orbital period
222: $P_{orb}$, periapsis $r_{p}$, and velocity at periapsis $v_{p}$.
223:
224: To arrive at the number of compact objects and low mass main sequence
225: stars in our model we do the following. Assuming all of the mass
226: density in our model results from a single burst of star formation
227: 10~Gyr ago, and that the number of stars per unit mass follows a
228: Kroupa IMF\footnote{We varied the IMF choice to include Salpeter and
229: Scalo IMFs; this did not change our rate significantly.}
230: \citep{kroupa:2001:vim}, we converted $f(\epsilon)$ into the total
231: number of stars per mass per phase space element. This number can be
232: less than one per phase space element. To determine the number of
233: main sequence stars per phase space element we calculated the maximum
234: stellar mass that would survive tidal disruption by the MBH
235: \citep{sridhar:1992:ic} and integrate the IMF up to this mass, setting
236: the main sequence stellar mass equal to the average mass that survives
237: tidal disruption. We determined the white dwarf population by
238: assuming white dwarfs formed from all stars between $1-8~\Msun$ and
239: the final white dwarf mass is $0.6~\Msun$. Similarly, neutron stars
240: were assigned masses of $1.4~\Msun$ and form from stars $8-20~\Msun$,
241: while black holes are $10~\Msun$ objects formed from $20-120~\Msun$
242: stars. Our calculation neglects the effects of mass segregation,
243: multiple episodes of star formation, and forces all stars between
244: $1-120~\Msun$ to be a stellar remnant.
245:
246: Only a small fraction of the initial phase space contain EMRBs. To
247: isolate the EMRB orbits we made a series of cuts. We immediately
248: discarded those orbits that were either unbound or had periapsis
249: within four Schwarzschild radii. We excluded orbits with $P_{orb}$
250: greater than the relaxation time, $t_{relax} \equiv v^2/D(\Delta
251: v_\parallel^2)$, where $D(\Delta v_\parallel^2)$ is the Fokker-Planck
252: diffusion coefficient \citep[eq.~8-68]{binney:1987:gd}. We also
253: excluded orbits with $P_{orb}$ less than $10^{5}$~s, which would yield
254: a continuous signal in the LISA band and are more accurately
255: classified as EMRIs. We also excluded orbits that radiate so strongly
256: that they plunge into the MBH faster than a dynamical time, where the
257: plunge timescale $T_{plunge}$ is
258: \begin{eqnarray}
259: T_{plunge} &\approx& 3.2\times 10^{6}~\textrm{yrs} \left(
260: \frac{\Mbh}{10^{6} \Msun} \right)^{2} \left( \frac{M_{\star}}{1
261: \Msun} \right)^{-1} \nonumber\\
262: & & \times \left( \frac{r_{p, i}}{10 R_{s}} \right)^{4}
263: \left( \frac{1 - e_{i}}{10^{-5}} \right)^{-1/2} \;.
264: \end{eqnarray}
265: Here $r_{p,i}$ is the initial pericenter distance and $e_{i}$ is the
266: initial eccentricity. Finally, we excluded all orbits with an
267: encounter timescale $\Delta t = r_{p}/v_{p} > 10^5$~s, as these passes
268: emit radiation outside the LISA band. The remaining stars constituted
269: our potential EMRB population.
270:
271: Figure~\ref{fig:parmspace} shows the phase space remaining after these
272: cuts; $\sim\!1,800$ phase space elements remained, which does not
273: imply 1800 distinct objects; in fact there are typically $10^{-7}$
274: white dwarfs per $d{\boldsymbol r} d{\boldsymbol v}$ in this region.
275: Overlaid in the same figure is the subset of the white dwarf phase
276: space with SNR~$>5$.
277: \begin{figure}
278: \epsscale{0.9}
279: \plotone{f1.eps}
280: \caption{Orbital parameters of potential EMRB sources. Here $e$ is
281: the eccentricity of the orbit within the total bulge potential,
282: and $R_{p}$ is in units of Schwarzschild radii. Typical sources
283: are more eccentric and have larger pericenter distances than the
284: average EMRI. Overlaid is the subset of white dwarf orbits that
285: are observable with LISA at a SNR~$> 5$.}
286: \label{fig:parmspace}
287: \end{figure}
288:
289: The total event rate $\nu$ is the sum over all orbits with LISA SNR
290: greater than 5:
291: \begin{equation}
292: \nu = \sum_{\rho > 5} \left(
293: \frac{N_{\textrm{LMMS,survive}}}{P_{\textrm{orb}}} +
294: \frac{N_{\textrm{WD}}}{P_{orb}}+
295: \frac{N_{\textrm{NS}}}{P_{orb}} +
296: \frac{N_{\textrm{BH}}}{P_{orb}} \right) \,.
297: \end{equation}
298: For our Milky Way bulge model, we find $\nu_{\textrm{WD}} =
299: 3~\textrm{yr}^{-1}$, $\nu_{\textrm{NS}} = 0.1~\textrm{yr}^{-1}$,
300: $\nu_{\textrm{BH}} = 0.06~\textrm{yr}^{-1}$, and
301: $\nu_{\textrm{LMMS,survive}} = 12~\textrm{yr}^{-1}$. These event
302: rates imply that EMRBs may be an important, heretofore unrecognized
303: source of gravitational waves in the LISA band. The left panel of
304: figure~\ref{fig:TauVsP} shows the accumulative event rate as a
305: function of orbital period. Most EMRBs originate from orbits that
306: would produce multiple bursts of radiation per year in the LISA band.
307: The right panel of figure~\ref{fig:TauVsP} shows the number of black
308: hole EMRBs as a function of SNR; note that some have SNRs high enough
309: to be seen out to Virgo distances. If we repeat our analysis with the
310: EMRB distances set to 16~Mpc and multiply by the number of galaxies in
311: the cluster with suitable mass MBHs, we find a Virgo event rate of
312: $\sim\!3~\textrm{yr}^{-1}$, all due to encounters of low mass black
313: holes with central MBHs.
314:
315: \begin{figure}
316: \plottwo{f2a.eps}{f2b.eps}
317: \caption{Left: Accumulative event rate as a function of the orbital
318: period. MS = main sequence; WD = white dwarf; NS = neutron star;
319: BH = black hole. Right: Number of black hole sources versus
320: signal to noise ratio. All sources above the light grey section
321: have SNR $>5$. Note the small fraction of EMRBs with SNR$>10^{4}$
322: (dark grey), which would be visible to Virgo distances. }
323: \label{fig:TauVsP}
324: \end{figure}
325:
326:
327:
328: \section{Discussion} \label{sec:discussion}
329:
330: EMRBs may comprise a significant new low frequency gravitational wave
331: source, rivaling the total number of EMRI events visible in any year
332: of LISA observation. For example, \citet{freitag:2001:mcc} estimates
333: that $\mathcal{O}(0.1)$ white dwarfs per year are in the EMRI phase in
334: the Milky Way. We find $\mathcal{O}(1)$ EMRBs per year from close
335: white dwarf encounters. The relaxation requirements on EMRBs are much
336: less strict: EMRIs require thousands of undisturbed orbits to
337: accumulate a SNR large enough to be detected ($T_{plunge} > (1-e)
338: T_{relax}$), while EMRBs merely need to pass by the MBH before
339: scattering via 2-body relaxation.
340:
341: Our estimated Milky Way EMRB event rate of $\sim\!15~\textrm{yr}^{-1}$
342: may appear high in light of other processes, such as the white dwarf
343: capture rate, which estimates place $\mathcal{O}(10^{-7})$ per year
344: \citep{freitag:2001:mcc}; however, the greatest contribution to the
345: event rate we calculate arises from sources that burst multiple times
346: a year, yet individually have a small phase space density. In other
347: words, our rate does not imply that $15~\Msun$ of material interact with
348: the black hole per year; rather, given these EMRB orbits we calculate
349: that Milky Way black hole grows by less than $10^{4}~\Msun$ over a
350: Hubble time via EMRB decay, with many EMRBs bursting hundreds of times
351: before falling into the MBH. Our simple stellar model passes other
352: sanity-checks as well: given the capture definition of
353: \citet{freitag:2001:mcc}, we expect a white dwarf capture rate of
354: $\mathcal{O}(10^{-6})$ per year and find, in our calculations, a tidal
355: disruption rate of $\mathcal{O}(10^{-5})$ per year, which is
356: consistent with observational estimates \citep{donley:2002:lxray}.
357:
358: We have introduced EMRBs as the evolutionary precursor to EMRIs; many
359: EMRBs will inspiral and eventually evolve into EMRIs as gravitational
360: radiation carries away energy and angular momentum. This implies that
361: the lifespan of the typical EMRB phase may only be $\mathcal{O}(10^7)
362: - \mathcal{O}(10^9)$ years after galaxy formation or a star formation
363: episode. We have assumed, however, that EMRB phase space is always
364: occupied. For there to be a significant population of EMRBs in the
365: present-day galaxy sample, there must be a refilling mechanism. There
366: are several mechanisms that may refill the burst reservoir. One such
367: mechanism, the dynamical migration of a stellar cluster, may have
368: recently left its mark in the inner 0.04 pc of our galaxy in the form
369: of the young S stars \citep{kim:2004:dfg}. Other mechanisms, such as
370: triaxiality \citep{merritt:2004:apj, khb:2006:astroph} or resonant
371: relaxation \citep{hopman:2006:astroph}, can act to refill the loss
372: cone on timescales much shorter than the two-body relaxation time.
373:
374: Since we were primarily concerned with determining whether EMRBs were
375: an overlooked class of LISA sources, we considered the simplest
376: possible model for the Milky Way bulge, the radiation properties, and
377: the ability of gravitational wave analysis techniques to detect the
378: signal. Under these approximations the predicted event rate implies
379: that bright EMRBs are so numerous that the local universe will
380: generate a background of gravitational wave bursts with a mean rate of
381: once per three weeks. Changing the assumptions, however, can alter
382: this rate by several orders of magnitude; modeling the bulge as a bar,
383: for example, can increase the capture rate by two orders of magnitude
384: \citep{khb:2006:astroph}, while mass-segregation can {\em decrease}
385: the EMRB rate from low mass main sequence stars by two orders of
386: magnitude \citep{freitag:2003:gws}.
387:
388: The relatively large event rate calculated here suggests that EMRBs
389: may in fact be a significant source for the LISA detector and
390: indicates that further, more accurate modeling needs to be undertaken.
391: To better characterize EMRBs and study how they can constrain models
392: of galactic nuclei, future work needs to include a more accurate model
393: for the gravitational wave emission, a better treatment of the orbit
394: in the relativistic regime, more realistic time-dependent galaxy and
395: star formation models, and a proper treatment of stellar dynamics,
396: including mass segregation. In addition, the potential presence of
397: these burst gravitational wave sources introduces a new set of data
398: analysis challenges, associated with the detection and
399: characterization of the EMRB signal from LISA data. As daunting as
400: the prospect of these challenges are their potential payoff: the
401: signal from EMRBs probe the central region of our Galaxy, and galaxies
402: in the Virgo Cluster, at scales less than $10^{-5}$ pc, far greater
403: resolution than we can ever hope to achieve electromagnetically.
404:
405:
406: \acknowledgments
407:
408: The authors are grateful to T. Bogdanovic, M. Freitag, P. Laguna,
409: S. Larson, C. Miller, S. Sigurdsson, and D. Shoemaker for numerous
410: helpful discussions on all aspects of this project. This work was
411: supported by NASA NNG04GU99G, NASA NNG05GF71G, NSF PHY 00-99559 and
412: the Center for Gravitational Wave Physics, which is funded by the
413: National Science Foundation under cooperative agreement PHY 01-14375.
414:
415:
416: \begin{thebibliography}{32}
417: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
418:
419: \bibitem[{Bender \& Hils(1997)}]{bender:1997:cnl}
420: Bender, P.~L. \& Hils, D. 1997, Class. Quantum Grav., 14, 1439
421:
422: \bibitem[{{Binney} {et~al.}(1997){Binney}, {Gerhard}, \&
423: {Spergel}}]{binney:1997:psi}
424: {Binney}, J., {Gerhard}, O., \& {Spergel}, D. 1997, \mnras, 288, 365
425:
426: \bibitem[{{Binney} \& {Tremaine}(1987)}]{binney:1987:gd}
427: {Binney}, J. \& {Tremaine}, S. 1987, {Galactic dynamics} (Princeton, NJ,
428: Princeton University Press, 1987, 747 p.)
429:
430: \bibitem[{{Bissantz} \& {Gerhard}(2002)}]{bissantz:2002:sab}
431: {Bissantz}, N. \& {Gerhard}, O. 2002, \mnras, 330, 591
432:
433: \bibitem[{{Danzmann} \& {R{\"u}diger}(2003)}]{danzmann:2003:ltc}
434: {Danzmann}, K. \& {R{\"u}diger}, A. 2003, Class. Quantum Grav., 20, 1
435:
436: \bibitem[{{Donley} {et~al.}(2002){Donley}, {Brandt}, {Eracleous}, \&
437: {Boller}}]{donley:2002:lxray}
438: {Donley}, J.~L., {Brandt}, W.~N., {Eracleous}, M., \& {Boller}, T. 2002, \aj,
439: 124, 1308
440:
441: \bibitem[{{Freitag}(2001)}]{freitag:2001:mcc}
442: {Freitag}, M. 2001, Class. Quantum Grav., 18, 4033
443:
444: \bibitem[{{Freitag}(2003)}]{freitag:2003:gws}
445: ---. 2003, \apjl, 583, L21
446:
447: \bibitem[{{Gair} {et~al.}(2004){Gair}, {Barack}, {Creighton}, {Cutler},
448: {Larson}, {Phinney}, \& {Vallisneri}}]{gair:2004:ere}
449: {Gair}, J.~R. \etal 2004, Class. Quantum Grav.,
450: 21, 1595
451:
452: \bibitem[{{Gair} {et~al.}(2005){Gair}, {Kennefick}, \&
453: {Larson}}]{gair:2005:sagr}
454: {Gair}, J.~R., {Kennefick}, D.~J., \& {Larson}, S.~L. 2005, \prd, 72, 084009
455:
456: \bibitem[{{Gerhard}(2002)}]{gerhard:2002:mdo}
457: {Gerhard}, O. 2002, \ssr, 100, 129
458:
459: \bibitem[{{Ghez} {et~al.}(2005){Ghez}, {Salim}, {Hornstein}, {Tanner}, {Lu},
460: {Morris}, {Becklin}, \& {Duch{\^e}ne}}]{ghez:2005:soa}
461: {Ghez}, A.~M. \etal 2005, \apj, 620, 744
462:
463: \bibitem[{{Glampedakis} {et~al.}(2002){Glampedakis}, {Hughes}, \&
464: {Kennefick}}]{glampedakis:2002:ait}
465: {Glampedakis}, K., {Hughes}, S.~A., \& {Kennefick}, D. 2002, \prd, 66, 064005
466:
467: \bibitem[{{Glampedakis} \& {Kennefick}(2002)}]{glampedakis:2002:zwe}
468: {Glampedakis}, K. \& {Kennefick}, D. 2002, \prd, 66, 044002
469:
470: \bibitem[{Hils \& Bender(1995)}]{hils:1995:gac}
471: Hils, D. \& Bender, P. 1995, \apjl, 445, L7
472:
473: \bibitem[{{Holley-Bockelmann} \& {Sigurdsson}(2006)}]{khb:2006:astroph}
474: {Holley-Bockelmann}, K. \& {Sigurdsson}, S. 2006, preprint
475: (astro-ph/0601520)
476:
477: \bibitem[{{Hopman} \& {Alexander}(2006)}]{hopman:2006:astroph}
478: {Hopman}, C. \& {Alexander}, T. 2006, ApJ, 645, 1152
479:
480: \bibitem[{{Ivanov}(2002)}]{ivanov:2002:frc}
481: {Ivanov}, P.~B. 2002, \mnras, 336, 373
482:
483: %\bibitem[{{Jennrich}(2004)}]{jennrich:2004:lmd}
484: %{Jennrich}, O. 2004, in Optical Fabrication, Metrology, and Material
485: % Advancements for Telescopes. Edited by Atad-Ettedgui, Eli; Dierickx,
486: % Philippe. Proceedings of the SPIE, Volume 5500, pp. 113-119 (2004)., 113--119
487:
488: \bibitem[{{Jennrich}(2005)}]{jennrich:2005:pil}
489: {Jennrich}, O. 2005, Class. Quantum Grav., 22
490:
491: \bibitem[{{Kim} {et~al.}(2004){Kim}, {Figer}, \& {Morris}}]{kim:2004:dfg}
492: {Kim}, S.~S., {Figer}, D.~F., \& {Morris}, M. 2004, \apjl, 607, L123
493:
494: \bibitem[{{Kroupa}(2001)}]{kroupa:2001:vim}
495: {Kroupa}, P. 2001, \mnras, 322, 231
496:
497: \bibitem[{{Larson} {et~al.}(2000){Larson}, {Hiscock}, \&
498: {Hellings}}]{larson:2000:scs}
499: {Larson}, S.~L., {Hiscock}, W.~A., \& {Hellings}, R.~W. 2000, \prd, 62, 062001
500:
501: \bibitem[{{L{\'o}pez-Corredoira} {et~al.}(2000){L{\'o}pez-Corredoira},
502: {Hammersley}, {Garz{\'o}n}, {Simonneau}, \&
503: {Mahoney}}]{lopez-corredoira:2000:iss}
504: {L{\'o}pez-Corredoira}, M. \etal 2000, \mnras, 313, 392
505:
506: \bibitem[{{Matsumoto} {et~al.}(1982){Matsumoto}, {Hayakawa}, {Koizumi},
507: {Murakami}, {Uyama}, {Yamagami}, \& {Thomas}}]{matsumoto:1982:boc}
508: {Matsumoto}, T. \etal 1982, in AIP Conf. Proc. 83: The Galactic
509: Center, 48--52
510:
511: \bibitem[{{Merritt} \& {Poon}(2004)}]{merritt:2004:apj}
512: {Merritt}, D. \& {Poon}, M.~Y. 2004, \apj, 606, 788
513:
514: \bibitem[{{Seto} {et~al.}(2001){Seto}, {Kawamura}, \&
515: {Nakamura}}]{seto:2001:pdm}
516: {Seto}, N., {Kawamura}, S., \& {Nakamura}, T. 2001, \prl,
517: 87, 221103
518:
519: \bibitem[{Sigurdsson \& Rees(1997)}]{sigurdsson:1997:csm}
520: Sigurdsson, S. \& Rees, M.~J. 1997, \mnras, 284, 318
521:
522: \bibitem[{{Sridhar} \& {Tremaine}(1992)}]{sridhar:1992:ic}
523: {Sridhar}, S. \& {Tremaine}, S. 1992, Icarus, 95, 86
524:
525: \bibitem[{{Stanek} {et~al.}(1997){Stanek}, {Udalski}, {Szymanski}, {Kaluzny},
526: {Kubiak}, {Mateo}, \& {Krzeminski}}]{stanek:1997:mgb}
527: {Stanek}, K.~Z. \etal 1997, \apj, 477, 163
528:
529: \bibitem[{{Sumner} \& {Shaul}(2004)}]{sumner:2004:ogw} {Sumner},
530: T.~J. \& {Shaul}, D.~N.~A. 2004, Mod. Phys. Lett. A, 19, 785
531:
532: \bibitem[{{Tremaine} {et~al.}(1994){Tremaine}, {Richstone}, {Byun}, {Dressler},
533: {Faber}, {Grillmair}, {Kormendy}, \& {Lauer}}]{tremaine:1994:fms}
534: {Tremaine}, S. \etal 1994, \aj, 107, 634
535:
536: \end{thebibliography}
537:
538:
539: %\clearpage
540:
541: %\begin{figure}
542: %\plotone{f1.eps}
543: %\caption{Quadrupole EMRB waveforms. The large $h_{+,\times}(t)$
544: % values are attributed to the relativistic velocities associated
545: % with the object's pericenter pass.}
546: %\label{fig:waveforms}
547: %\end{figure}
548:
549: %\clearpage
550:
551: %\clearpage
552:
553:
554: \end{document}
555: