1: \documentclass[12pt,preprint]{aastex}
2: %\documentclass[aps,prd, amssymb, 12pt]{revtex4}
3: %\usepackage{graphicx,amsmath,subfigure}
4:
5:
6: \newcommand{\Msun}{\ensuremath{M_{\odot}}}
7: \newcommand{\bel}[1]{\begin{equation}\label{#1}}
8: \newcommand{\be}{\begin{equation}}
9: \newcommand{\ee}{\end{equation}}
10: \newcommand{\ba}{\begin{eqnarray}}
11: \newcommand{\ea}{\end{eqnarray}}
12: \newcommand{\bal}[1]{\begin{eqnarray}\label{#1}}
13:
14: \shorttitle{Spins evolution via minor mergers}
15: \shortauthors{Mandel}
16:
17:
18: \begin{document}
19:
20: \title{Spin distribution following minor mergers and the effect of spin
21: on
22: the detection range for low-mass-ratio inspirals}
23: \author{Ilya Mandel} \affil{Theoretical Astrophysics, California
24: Institute of Technology, Pasadena, California
25: 91125}\email{ilya@caltech.edu}
26:
27: %\date{July 4, 2007}
28:
29: \begin{abstract}
30:
31: We compute the probability distribution for the spin of a black hole following
32: a series of minor mergers with isotropically distributed, non-spinning,
33: inspiraling compact objects. By solving the Fokker-Planck equation governing
34: this stochastic process, we obtain accurate analytical fits for the evolution
35: of the mean and standard deviation of the spin distribution in several
36: parameter regimes. We complement these analytical fits with numerical
37: Monte-Carlo simulations in situations when the Fokker-Planck analysis is not
38: applicable. We find that a $\sim 150\ M_\odot$ intermediate-mass black hole
39: that gained half of its mass through minor mergers with neutron stars will have
40: dimensionless spin parameter $\chi=a/M \sim 0.2 \pm 0.08$. We estimate the
41: effect of the spin of the central black hole on the detection range for
42: intermediate-mass-ratio inspiral (IMRI) detections by Advanced LIGO and
43: extreme-mass-ratio inspiral (EMRI) detections by LISA. We find that for
44: realistic black hole spins, the inclination-averaged Advanced-LIGO IMRI
45: detection range may be increased by up to $10\%$ relative to the range for
46: IMRIs into non-spinning intermediate-mass black holes. For LISA, we find that
47: the detection range for EMRIs into $10^5\ M_\odot$ massive black holes (MBHs)
48: is not significantly affected by MBH spin, the range for EMRIs into $10^6\
49: M_\odot$ MBHs is affected at the $\lesssim 10\%$ level, and EMRIs into
50: maximally spinning $10^7\ M_\odot$ MBHs are detectable to a distance $\sim 25$
51: times greater than EMRIs into non-spinning black holes. The resulting bias in
52: favor of detecting EMRIs into rapidly spinning MBHs will play a role when
53: extracting the MBH spin distribution from EMRI statistics.
54:
55: \end{abstract}
56:
57: \keywords{black hole physics --- gravitational waves}
58:
59: \maketitle
60:
61: \section{Introduction}
62:
63: A growing body of evidence from observations, numerical simulations, and
64: comparisons between the two, suggests the existence of a population of
65: intermediate-mass black holes with masses in the $M\sim 10^2-10^4\,M_\odot$
66: range (e.g., \citep{MC2004} and references therein). These
67: intermediate-mass black holes may capture compact objects (stellar-mass black
68: holes or neutrons stars) and merge with them
69: \citep{Tan00,MH02a,MH02b,MT02a,MT02b,GMH04,GMH06,OL06, Mandel2007}. In
70: addition
71: to adding to the black-hole mass, the merging compact objects will also
72: contribute their orbital angular momentum to the spin angular momentum of the
73: central black hole, leading to the evolution of the black-hole spin through a
74: sequence of such minor mergers.
75:
76: We might expect the typical spin of a black hole to be low if a significant
77: fraction of its mass has been added via minor mergers with compact objects
78: whose angular momentum at plunge is distributed isotropically. The angular
79: momentum imparted to the black hole of mass $M$ by a compact object of mass $m$
80: is $L_{\rm obj} \propto m M$. (We include only the orbital angular momentum,
81: not the spin angular momentum of the compact object, since the latter is lower
82: than the former by a factor of order $m/M$, which we assume to be small for
83: minor mergers.) This causes the dimensionless spin parameter of the hole
84: $\chi\equiv S_1/M^2 = a/M$ to change by $\sim L_{\rm obj}/M^2 \propto m/M$.
85: After $N \sim M/m$ such mergers, necessary for the hole to grow to mass $M$,
86: the typical dimensionless spin parameter of the hole will be $\chi \propto
87: (m/M)\sqrt{N} \sim \sqrt{m/M}$.
88:
89: As discussed by \citet{Miller2002} and \citet{HB}, the
90: angular momenta of black holes that grow through minor mergers undergo a damped
91: random walk. The damping comes about because retrograde orbits, which subtract
92: angular momentum from a black hole, plunge from a last stable orbit (LSO) at a
93: higher radius than prograde orbits, so more angular momentum is subtracted
94: following retrograde inspirals than is added following prograde ones.
95:
96: In this paper, we make an analytical approximation to the spin change induced
97: by a minor merger and solve the Fokker-Planck equation to obtain the evolution
98: of the spin probability distribution \citep{HB}. (We use a simpler
99: one-dimensional version of the Fokker-Planck equation than \citet{HB},
100: since we are interested only in the evolution of the
101: magnitude of the spin, not its direction.) We find that for black holes with
102: $\chi \gg \sqrt{m/M}$, the spin $\chi$ evolves proportionally to $M^{-2.63}$ as
103: the mass grows via minor mergers (rather than $M^{-2}$, which would be the case
104: without damping). We determine the asymptotic values of the expected mean of
105: the spin distribution and its standard deviation in the limit of infinitely
106: many minor mergers: $\bar{\chi}\to \sqrt{1.5 m /M}$ and $\sigma \to \sqrt{0.7
107: m /M}$. We also describe the evolution of the spin distribution in other
108: parameter regimes, e.g., when $\sqrt{m/M} \gg \chi \gg m/M$.
109:
110: Our Fokker-Planck analysis fails when the mass ratio $m/M$ is not
111: sufficiently low, so for those cases we resort to Monte-Carlo numerical
112: simulations. We find that if the mass of the central black hole grows
113: from $M=5 m$ to $M=10 m$ by capturing five objects of equal mass $m$, the
114: mean spin of the resulting black hole is $\bar{\chi} \approx 0.5$, nearly
115: independent of its initial spin (\citet{Miller2002} obtained similar
116: results). However, if the central black hole grows from $M=50 m$ to
117: $M=100 m$ (e.g., a $M=70\ M_\odot$ black hole growing to $M=140\ M_\odot$
118: by capturing fifty $m=1.4\ M_\odot$ neutron stars), its resulting spin is
119: rather low, $\chi \sim 0.2 \pm 0.08$.
120:
121: The combination of the spin of the central black hole and the inclination of
122: the inspiraling object's orbit can have a significant effect on the
123: gravitational-wave signal from a low-mass-ratio inspiral. We compute the
124: increase in the Advanced-LIGO detection range for intermediate-mass-ratio
125: inspirals (IMRIs) due to the spin of the central black hole. We find that the
126: detection range, averaged over orbital inclinations, may increase by $\sim
127: 3-10\%$ relative to the range for inspirals into non-spinning black holes for
128: the expected values of black hole mass and spin. We provide an approximate
129: expression for the dependence of the Advanced-LIGO IMRI detection range on spin
130: [see Eq.~(\ref{AdvLIGORatio})]. We also compute the change in the LISA
131: extreme-mass-ratio-inspiral (EMRI) detection range due to the spin of the
132: massive black hole. We find that the range for inspirals into $M=10^5\ M_\odot$
133: black holes is nearly independent of their spin, because the frequency at the
134: last stable orbit (LSO) is away from the minimum of the LISA noise curve. On
135: the other hand, the inclination-averaged detection range for IMRIs into rapidly
136: spinning $M=10^7\ M_\odot$ black holes is $\sim 25$ times greater than into
137: non-spinning ones. The detection volumes are proportional to the cube of the
138: range. This will create a bias in favor of detecting inspirals into
139: rapidly spinning black holes, which in turn will have consequences for the
140: extraction of massive-black-hole spin function from LISA EMRI statistics.
141:
142: The paper is organized as follows. In Sec.~\ref{SpinEvol}, we provide the
143: background for our calculation of the spin evolution via minor mergers. In
144: Sec.~\ref{FPSpinEvol}, we describe analytical solutions of the Fokker-Planck
145: equation for spin evolution. In Sec.~\ref{MCSpinEvol}, we describe
146: Markov-Chain numerical simulations of spin evolution. In Sec.~\ref{Boost}, we
147: evaluate the dependence of the detection ranges for low-mass-ratio inspirals
148: averaged over orbital inclination angles on the spin of the massive body, in
149: the context of both Advanced LIGO and LISA.
150:
151:
152:
153:
154: \section{Spin evolution}\label{SpinEvol}
155:
156: We assume that the distribution of the orbital inclination angle $\iota$
157: relative to the central black hole's spin is isotropic at
158: capture. Here $\iota$ is defined via
159: \bel{iota}
160: \cos{\iota}=\frac{L_z}{\sqrt{L_z^2+Q}},
161: \ee
162: $L_z$ is the object's orbital angular momentum in the direction of the black
163: hole's spin, and $Q$ is the Carter constant. We further assume that the
164: inclination angle $\iota$ remains approximately constant over the
165: inspiral~\citep{Hughes2000}, so the distribution of inclinations at the
166: LSO is also isotropic, ${\rm Pr}(\cos \iota)=1/2$.
167:
168: In the low-mass-ratio limit, the amount of angular momentum radiated in
169: gravitational waves during the plunge and ringdown is smaller by a factor of
170: $\sim m/M$ than the angular momentum at the LSO. Therefore, we assume that the
171: merging object contributes its orbital angular momentum at the LSO to the
172: angular momentum of the black hole. The spin of the black hole after a minor
173: merger, $\chi'$, is related to the original spin $\chi$ via
174: \bel{chievol}
175: \chi' \approx \frac{1}{(M+m)^2} \sqrt{(\chi M^2 + L_z)^2+Q},
176: \ee
177: where $m$ is the mass of the small object, $M$ is the mass of the hole, and we
178: assume $m \ll M$.
179:
180: The constants of motion $L_z$ and $Q$ at the LSO can be obtained as a function
181: of $\iota$ by demanding that
182: the potential $R$ and its first and second derivatives in $r$ are zero at the
183: LSO (see Chapter 33 of \citep{MTW}):
184: \bal{R}
185: \nonumber
186: R &=& \left[ E (r^2+\chi^2 M^2) - L_z \chi M \right]^2
187: - (r^2-2 M r + \chi^2 M^2)
188: \left[m^2 r^2 + (L_z- \chi M E)^2 + Q\right],
189: \\
190: R &=&0, \qquad \frac{dR}{dr} = 0, \qquad \frac{d^2 R}{dr^2}=0 \qquad
191: {\rm at\ LSO}.
192: \ea
193:
194: It is possible to make analytic approximations to the values of $L_z$ and $Q$
195: at the LSO based on appropriately averaging the
196: analytically known constants of motion at the LSO for prograde and retrograde
197: equatorial orbits (cf.~Eq.~(9) of \citep{HB}). In particular, for $\chi
198: \ll
199: 1$, the plunging object's dimensionless ``total angular momentum'' is given by
200: \bel{Lapprox}
201: \hat{L}=\frac{\sqrt{L_z^2+Q}}{Mm} \approx M m \sqrt{12}
202: \left[ 1- \frac{1}{2}\left(\frac{2}{3}\right)^{3/2} \chi \cos \iota
203: \right],
204: \ee
205: where we correct a mistake in Eq.~(4) of \citep{Miller2002}. Then $L_z$
206: and $Q$ follow from Eq.~(\ref{iota}):
207: \bel{LzQ}
208: L_z = \cos{\iota} \sqrt{L_z^2+Q}; \qquad Q=\sin{\iota} \sqrt{L_z^2+Q} .
209: \ee
210:
211:
212:
213: \section{Fokker-Planck equation for spin evolution}\label{FPSpinEvol}
214:
215: The black-hole spin evolution is a stochastic process. The probability
216: distribution function of a stochastic process, however, can be described the
217: deterministic Fokker-Planck equation:
218: \bel{FP}
219: \frac{\partial}{\partial t} f(x,t)=
220: -\frac{\partial}{\partial x} \left[\mu(x,t) f(x,t)\right]+
221: \frac{1}{2}\frac{\partial^2}{\partial x^2}
222: \left[\sigma^2(x,t) f(x,t)\right],
223: \ee
224: where $\mu=\langle dx \rangle/dt$ is the mean drift and
225: $\sigma^2=\langle(dx)^2\rangle/dt$ is the stochastic variance. In this Section,
226: we derive approximate analytical solutions to the Fokker-Planck equation in
227: several interesting parameter regimes.
228:
229: For simplicity, assume that all merging objects have the same mass $m$. We
230: parametrize the mass of the black hole by a dimensionless ``time''
231: parameter $t=M/m$. The change in the spin $\chi$ after a merger follows from
232: Eq.~(\ref{chievol}):
233: \bel{dchi}
234: d\chi=\frac{1}{(t+1)^2}\sqrt{\chi^2 t^4 + \hat{L}^2 t^2 + 2\chi \hat{L} t^3
235: \cos{\iota}} -\chi.
236: \ee
237:
238: We can compute $\hat{L}$ at plunge as a function of $\chi$ and $\cos{\iota}$ by
239: solving Eqs.~(\ref{R}), then substituting the result into Eq.~(\ref{dchi}) to
240: obtain $d\chi$ as a function of $t$, $\chi$, and $\cos{\iota}$. Although this
241: process is simple in principle, such a numerical computation makes it
242: impossible to obtain analytic expressions for $\langle d\chi \rangle$ and
243: $\langle (d\chi)^2 \rangle$, which are necessary if we wish to solve the
244: Fokker-Planck equation. (Here, brackets denote averaging over $\cos{\iota}$.)
245:
246: We could, of course, try to obtain empirical analytic fits to the numerical
247: solutions for $\langle d\chi \rangle$ and $\langle (d\chi)^2 \rangle$, but it
248: turns out that there is a simpler approach. The approximate formula for
249: $\hat{L}$ given in Eq.~(\ref{Lapprox}) is valid only when $\chi \ll 1$; when
250: $\chi \sim 1$, Eq.~(\ref{Lapprox}) overestimates $\hat{L}$ by as much as
251: $40\%$. Remarkably, however, using this incorrect approximation for $\hat{L}$
252: in Eq.~(\ref{dchi}) generally yields very accurate expressions for $\langle
253: d\chi \rangle$ for a wide range of $\chi$. So long as $\chi t \gg 1$ (i.e.,
254: $\chi \gg m/M$), an expansion of Eq.~(\ref{dchi}) to the first order in
255: $1/(\chi t)$ yields the following simple analytic expression for the mean drift
256: in $\chi$:
257: \bel{mu}
258: \mu(\chi,t)=\frac{\langle d\chi \rangle}{dt} =
259: \frac{\chi}{t} \left(-2-\frac{4\sqrt{2}}{9}\right)+\frac{4}{\chi t^2}.
260: \ee
261: This expression is accurate to about $1\%$ for all values of $\chi$ so long as
262: $\chi t \gtrsim 10$. Similarly, the analytic expression for the stochastic
263: variance of the spin is
264: \bel{sigma} \sigma^2(\chi,t)=\frac{\langle (d\chi)^2 \rangle}{dt}
265: =\frac{4}{t^2} \left(1+\frac{4 \sqrt{2} \chi^2}{9}-\chi^2\right).
266: \ee
267: This expression underestimates the variance by $\gtrsim 10\%$ for very high
268: spins, but is generally accurate to a few percent for lower spins which are
269: expected as a consequence of minor mergers in the Advanced LIGO setting.
270:
271: We can now substitute Eqs.~(\ref{mu}) and (\ref{sigma}) into the Fokker-Planck
272: equation for the probability evolution (\ref{FP}) to obtain
273: \bal{FPspin}
274: \frac{\partial}{\partial t} f(\chi,t)&=&
275: -\frac{\partial}{\partial \chi} \left[
276: \frac{\chi}{t} \left(-2-\frac{4\sqrt{2}}{9}
277: +\frac{4}{\chi^2 t}\right) f(\chi,t)\right]\\
278: \nonumber
279: &+&
280: \frac{1}{2}\frac{\partial^2}{\partial \chi^2}
281: \left[\frac{4}{t^2}
282: \left(1+\frac{4 \sqrt{2} \chi^2}{9}-\chi^2\right) f(\chi,t)\right].
283: \ea
284: This is a one-dimensional equation unlike the three-dimensional equation
285: derived in \citep{HB}, since we choose to focus on the evolution of the
286: magnitude of the spin, not its direction. Still, this is a rather complicated
287: equation that does not easily separate. Fortunately, for many applications
288: it is not necessary to solve the complete equation.
289:
290: Equation (\ref{FPspin}) was derived under the assumption $\chi t \gg 1$. If we
291: further assume that $\chi^2 t \gg 1$ (i.e., $\chi \gg \sqrt{m/M}$,
292: then the mean spin evolution is dominated by
293: \bel{DriftSpinApprox}
294: \frac{d \bar{\chi}}{dt} \approx a \frac{\bar{\chi}}{t},
295: \ee
296: where $a \equiv -2-4\sqrt{2}/9\approx-2.63$. (This result can also be obtained
297: directly from Eq.~(\ref{mu}).) Thus, the mean spin evolves according to
298: \bel{SpinApprox}
299: \bar{\chi} \approx \bar{\chi}_0 \left(\frac{t}{t_0}\right)^a
300: \approx \bar{\chi}_0 \left(\frac{M_0}{M}\right)^{2.63}
301: \ee
302: (compare with Eq.~(26) of \citep{HB}, where the exponent is approximated
303: by
304: $2.4$).
305:
306: If the assumption $\chi^2 t \gg 1$ is not satisfied, and instead $\chi^2 t \ll
307: 1$, but $\chi t \gg 1$ so that Eq.~(\ref{FPspin}) still holds, the evolution
308: of the probability function may be approximated as
309: \be
310: \frac{\partial f(t,\chi)}{\partial t}= -\frac{\partial}{\partial \chi}
311: \left(\frac{4 f(t,\chi)}{\chi t^2}\right)
312: +\frac{1}{2}\frac{\partial^2}{\partial \chi^2}
313: \left(\frac{4 f(t,\chi)}{t^2}\right).
314: \ee
315: This equation can be solved by separation of variables: $f(t, \chi)=T(t) X(\chi)$,
316: where the solution for $T$ is $T(t)=\exp(-k/t)$, $X$ is the solution to
317: \be
318: 2\chi^2 X'' - 4 \chi X + 4 X - k \chi^2 X = 0,
319: \ee
320: and $k$ is a constant. The mean spin grows roughly as
321: \bel{SpinApprox2}
322: \bar{\chi} \sim \sqrt{\frac{2}{t_0}-\frac{2}{t}},
323: \ee
324: so after $t \gtrsim 2 t_0$ (i.e., after the black hole captures half its mass
325: via minor mergers), $\chi^2 t \gtrsim 1$.
326:
327: The spin growth and spin decay terms in
328: Eq.~(\ref{FPspin}) cancel when the spin is approximately equal to
329: \bel{SpinAsympt}
330: \bar{\chi} \to \sqrt{\frac{4}{-a t}} \approx \sqrt{\frac{1.5}{t}}.
331: \ee
332: (Compare with \citet{Miller2002}, who estimated the mean spin to be
333: $\sqrt{2} \sqrt{(m/M)} = \sqrt{2/t}$ based on numerical simulations.)
334:
335: We can estimate the second moment of the probability distribution by
336: approximating the solution to Eq.~(\ref{FPspin}) by a Gaussian
337: (as suggested by \citet{Miller2002}):
338: \bel{GaussianGuess}
339: f(\chi, t)=\frac{1}{\sqrt{2\pi}\sigma} \exp\left[
340: -\frac{\left(\chi-\bar{\chi} (t)\right)^2}{2\sigma^2(t)}
341: \right].
342: \ee
343: (A Gaussian turns out to be a good approximation except at small
344: $\bar{\chi}$, when the tails at $\chi>\bar{\chi}$ are larger than those
345: at $\chi<\bar{\chi}$.) Substituting this Gaussian into Eq.~(\ref{FPspin}),
346: keeping only the lowest-order terms in $t\chi$, and setting $\chi =\bar{\chi}$,
347: we obtain
348: \bel{DriftSigmaApprox}
349: -\frac{1}{\sigma}\frac{d\sigma}{dt} =
350: -\frac{a}{t}-\frac{2}{t^2\sigma^2}(1+b\bar{\chi}^2),
351: \ee
352: where $b \equiv 4\sqrt{2}/9-1$.
353: If $\sigma^2 t \gg 1$, then $\sigma$ evolves in the same way as
354: $\bar{\chi}$ when $\chi^2 t \gg 1$:
355: \bel{SigmaApprox}
356: \sigma \approx \sigma_0 \left(\frac{t}{t_0}\right)^a
357: \approx \sigma_0 \left(\frac{M_0}{M}\right)^{2.63}.
358: \ee
359:
360: What if $\sigma^2 t \ll 1$? This might be the case of interest if, say, the
361: initial spin of a black hole created during some process is known precisely,
362: and we wish to estimate future spin evolution through minor mergers. In this
363: case, the second term on the right-hand side of Eq.~(\ref{DriftSigmaApprox})
364: dominates, and if $\bar{\chi}$ is small or does not change
365: significantly, $\sigma$ grows according to
366: \bel{SigmaApprox2}
367: \sigma \approx \sqrt{4 \left(1+b \bar{\chi}^2\right)
368: \left(\frac{1}{t_0}-\frac{1}{t}\right)+\sigma_0^2}.
369: \ee
370:
371: In either case, $\sigma$ asymptotes to the solution
372: \bel{SigmaAsympt}
373: \sigma \to \sqrt{\frac{2(1+b \bar{\chi}^2)}{-at}}.
374: \ee
375: For large $t$, $\sigma \sim \sqrt{2/(-at)} \approx \sqrt{0.7/t}$;
376: \citet{Miller2002} estimated $\sigma$ to be
377: $\sqrt{(m/M)}/\sqrt{2} = \sqrt{1/(2t)}$ based on numerical simulations.
378:
379: Lastly, consider the case when $\chi t \lesssim 1$. In this case the orbital
380: angular momentum of the plunging object is comparable to the spin angular
381: momentum of the black hole, and Eq.~(\ref{FPspin}) is incorrect, since it was
382: derived under the assumption $\chi t \gg 1$. If the black hole is initially
383: non-spinning or has spin $\chi \lesssim 1/t$, however, a single minor merger
384: will bring its spin to $\chi \sim \sqrt{12}/t$ according to Eq.~(\ref{dchi}).
385: This case can be treated with a Monte-Carlo numerical simulation as described
386: in the next section.
387:
388:
389: \section{Spin evolution via Monte Carlo simulations}\label{MCSpinEvol}
390:
391: We have carried out Monte Carlo simulations of spin evolution through minor
392: mergers in order to confirm the analytical estimates presented above, based on
393: the approximate Fokker-Planck equation. Our simulations also allow us to
394: access the small-$t$ regime where the Fokker-Planck approach is not valid, but
395: where our physical approximations for low-mass-ratio inspirals still
396: hold. Since these simulations were performed numerically, there was no need
397: to make analytical approximations to $d\chi$ following a merger; instead, we
398: solved Eqs.~(\ref{R}) directly and obtained $d\chi$ via Eq.~(\ref{dchi}).
399:
400: In Figure \ref{Hist510} we plot the spin distribution of a black hole of mass
401: $t=M/m=10$ that started out with either spin $\chi=0.1$ or $\chi=0.9$ at
402: $t=M/m=5$ before growing via minor mergers. This corresponds, for example, to
403: an intermediate-mass black hole that grows from $M=50\ M_\odot$ to $M=100\
404: M_\odot$ by capturing $m=10\ M_\odot$ black holes. The distributions for both
405: values of initial spin are roughly Gaussian, although with
406: shorter-than-Gaussian tails (we plot the actual Monte-Carlo histogram for the
407: $\chi=0.9$ case for comparison with a fitted Gaussian). We see that for these
408: small values of $t$, the initial value of the spin is largely forgotten after
409: the black hole captures half of its mass through minor mergers. The means of
410: the spin at $t=10$ are $\bar{\chi}=0.49$ for the initially
411: slowly-spinning hole and $\bar{\chi}=0.51$ for the initially
412: rapidly-spinning hole. The standard deviations at $t=10$ are $\sigma=0.17$
413: for initial spin $\chi=0.1$ and $\sigma=0.18$ for initial spin $\chi=0.9$ (the
414: initial standard deviations are zero in both cases, i.e., the initial spins are
415: presumed to be precisely determined). These results agree with Fig.~1 of
416: \citep{Miller2002}. Because the values of $t$ involved are so small, the
417: Fokker-Planck equation (\ref{FPspin}) does not apply: at $t=5$, the angular
418: momentum of the inspiraling object at the LSO is comparable to or larger than
419: the spin angular momentum of the black hole even for large initial black hole
420: spins.
421:
422: \begin{figure}
423: \includegraphics[keepaspectratio=true,width=6in]{f1.eps}
424: \caption{Monte-Carlo predictions for the black-hole spin distribution
425: following black-hole growth via minor mergers from $t=M/m=5$ to $t=M/m=10$.
426: The histogram shows the spin distribution at $t=10$ for a black hole with
427: initial spin $\chi=0.9$, and the solid curve is a Gaussian fit to that
428: distribution. The dashed curve is a Gaussian fit to the spin distribution at
429: $t=10$ for a black hole that has initial spin $\chi=0.1$ at $t=5$.
430: \label{Hist510}}
431: \end{figure}
432:
433:
434: In Figure \ref{Hist50100} we plot the spin distribution for a black hole of
435: mass $t=M/m=100$ that started out at $t=M/m=50$ at either spin $\chi=0.1$ or
436: $\chi=0.9$ before growing via minor mergers. This corresponds, for example, to
437: an intermediate-mass black hole that grows from $70\ M_\odot$ to $140\ M_\odot$
438: by capturing $M=1.4\ M_\odot$ neutron stars. The means of the spin at $t=100$
439: are $\bar{\chi}=0.162$ for the initially slowly-spinning hole and
440: $\bar{\chi}=0.233$ for the initially rapidly-spinning hole. The final spin in
441: the initially rapidly-spinning case decreases as $\bar{\chi} \sim \chi_0
442: (t/t_0)^{-2}$, rather than $\bar{\chi} \sim \chi_0 (t/t_0)^{-2.63}$ as
443: predicted by Eq.~(\ref{SpinApprox}). That is because the spin begins to
444: approach the asymptotic value of $\bar{\chi} \approx \sqrt{1.5/t} \approx
445: 0.12$ as predicted by Eq.~(\ref{SpinAsympt}), and the rate of spin evolution
446: decreases because $\chi^2 t$ is no longer much greater than one. The
447: initially slowly-spinning case does not quite satisfy $\chi t \gg 1$, so the
448: Fokker-Planck analysis is suspect; however, Eq.~(\ref{SpinApprox2}), relevant
449: since $\chi^2 t < 1$ in this case, provides a roughly accurate estimate of spin
450: growth. The standard deviations at $t=100$ are $\sigma=0.066$ for initial spin
451: $\chi=0.1$ and $\sigma=0.084$ for initial spin $\chi=0.9$; the predicted
452: asymptotic value of the standard deviation according to Eq.~(\ref{SigmaAsympt})
453: is $\sigma=0.087$. The mass ratios considered in this paragraph may be
454: plausible for intermediate-mass-ratio inspirals into intermediate-mass black
455: holes that would be detectable with Advanced LIGO \citep{Mandel2007}.
456:
457:
458: \begin{figure}
459: \includegraphics[keepaspectratio=true,width=6in]{f2.eps}
460: \caption{Monte-Carlo black-hole spin distribution
461: following black hole growth via minor mergers from $t=M/m=50$ to $t=M/m=100$.
462: The spin distribution for a black hole with initial spin
463: $\chi=0.9$ is shown with a solid curve, and one
464: for initial spin $\chi=0.1$ is shown with a dashed curve.\label{Hist50100}}
465: \end{figure}
466:
467: Finally, we perform a Monte-Carlo simulation of the evolution of a spin
468: distribution from $t=1100$ to $t=1200$ where the starting mean spin is
469: $\bar{\chi}=0.72$ and the starting standard deviation is
470: $\sigma=0.016$. In this case, $\chi^2 t \gg 1$ holds throughout the evolution,
471: so this example can be viewed as a test of our Fokker-Planck analysis. Based
472: on Eq.~(\ref{SpinApprox}), we expect the spin at $t=1200$ to decrease to
473: $\bar{\chi}=0.57$; in fact, we find $\bar{\chi}
474: (t=1200)=0.58$. Since $\sigma^2 t \ll 1$, we expect the standard deviation to
475: grow via Eq.~(\ref{SigmaApprox2}) to $\sigma=0.022$ at $t=1200$; in fact,
476: $\sigma(t=1200)=0.021$.
477:
478: The Fokker-Planck analysis should give excellent results in the regime of very
479: large $t$, such as those corresponding to minor mergers of stellar-mass compact
480: objects with $\sim 10^6\ M_\odot$ massive black holes in galactic centers. (The
481: extreme-mass-ratio inspirals preceding such minor mergers are an interesting
482: class of potential LISA sources~\citep{Pau}.) On the other hand, if a
483: large
484: range of $t$ must be covered, Monte-Carlo simulations become expensive. Thus,
485: the Monte-Carlo numerical methods and Fokker-Planck analysis can be viewed as
486: complementary techniques.
487:
488:
489: \section{Effect of black-hole spin on detection ranges for low-mass-ratio
490: inspirals} \label{Boost}
491:
492: The frequency of the last stable orbit before plunge is strongly influenced by
493: the black-hole spin and the orbital inclination. Prograde inspirals into
494: rapidly spinning black holes will have much higher LSO frequencies than
495: inspirals into non-spinning black holes or polar inspirals into spinning black
496: holes of the same mass, while retrograde inspirals into rapidly spinning black
497: holes will have lower LSO frequencies. For example, for a maximally spinning
498: Kerr black hole, the frequency of the LSO of a retrograde equatorial inspiral
499: is twice lower than for a polar orbit, while the LSO frequency of a prograde
500: equatorial inspiral is six times higher than for a polar orbit. Even for a
501: more moderately spinning black hole with $\chi=0.4$, there is almost a factor
502: of two difference between LSO frequencies for prograde and retrograde
503: inspirals.
504:
505: The signal-to-noise ratio (SNR) for the detection of gravitational waves from
506: inspirals depends on where the LSO frequency falls on the noise power spectral
507: density curve of the detector. Although some inclination angles will increase
508: SNR and others will decrease it, we might generally expect that average
509: detection range for inspirals into spinning black holes will be higher than
510: into non-spinning ones. (``Average'' refers to averaging over the
511: isotropically distributed orbital inclination angles of the inspiraling
512: object.) This is because of the cubic dependence of the detection volume on
513: detection range, which is proportional to SNR: if, say, $10\%$ of all inspirals
514: have their SNR boosted by a factor of three, these will be seen three times
515: further and the detection volume for these kinds of inspirals will go up by a
516: factor of $27$, so the average volume in which detections can be made will
517: increase by a factor of $\sim 3$, and the average detection range will grow by
518: the cube root of $3$.
519:
520: Conversely, this average detection range increase can manifest itself as a bias
521: in favor of detecting inspirals into rapidly spinning black holes rather than
522: slowly spinning ones. Thus, a numerical estimate of the detection range
523: increase due to black hole spin is useful for determining whether a high
524: fraction of rapidly spinning black holes among detected inspirals is an
525: indication of the prevalence of such black holes in the universe, or whether
526: this is merely a selection effect.
527:
528: We use the simple scaling
529: \be
530: |\tilde{h}(f)^2| \propto f^{-7/3}
531: \ee
532: for the frequency-domain gravitational wave. The square of the signal-to-noise
533: ratio $\rho^2$ is proportional to
534: \bel{SNR}
535: \rho^2 \propto
536: \int_{f_{\rm min}}^{f_{\rm max}} \frac{|\tilde{h}(f)^2|}{S_n(f)} df \propto
537: \int_{f_{\rm min}}^{f_{\rm max}} \frac{f^{-7/3}}{S_n(f)} df.
538: \ee
539: Here, $S_n(f)$ is the noise power spectral density of the detector,
540: $f_{\rm max}$ is the frequency of gravitational waves from the
541: last stable orbit, and $f_{m\rm in}$ is the low-frequency cutoff for the
542: detector for Advanced LIGO, where $f_{\rm min}= 10$ Hz, or the frequency of
543: gravitational waves one year before plunge for LISA. We
544: set $f_{\rm max}$ equal to twice the orbital frequency at the LSO, which we
545: obtain numerically as a function of the black-hole mass $M$ and spin $\chi$ and
546: of the orbital inclination angle $\cos{\iota}$ by solving Eq.~(\ref{R}).
547:
548: The distance to which an event can be seen is proportional to SNR, $\rho$, so
549: the detection volume is proportional to $\rho^3$. Therefore, we average
550: $\rho^3$, computed via Eq.~(\ref{SNR}), over the different inclinations
551: $\cos{\iota}$ (uniformly distributed through the range $[-1,\ 1]$) in order to
552: compute the expected increase in the detection volume for a given values of
553: $\chi$, and then take the cube root to compute the increase in the average
554: detection range.
555:
556: \begin{figure}
557: \includegraphics[keepaspectratio=true,width=6in]{f3.eps}
558: \caption{The ratio between the inclination-averaged Advanced-LIGO detection
559: range for intermediate-mass-ratio inspirals into Kerr black holes of a given
560: spin and the detection range for IMRIs into non-spinning black holes. The
561: solid curve represents black holes with mass $M=100\ M_\odot$; the dashed
562: curve, mass $M=200\ M_\odot$.
563: \label{AdvLIGOBoost}}
564: \end{figure}
565:
566: We have computed detection ranges for Advanced LIGO using this method with the
567: noise power spectral density $S_n(|f|)$ taken from~\citep{Fritschel}.
568: Fig.~\ref{AdvLIGOBoost} shows our computed ratio between (i) the average
569: Advanced-LIGO detection range for intermediate-mass-ratio inspirals into black
570: holes of a given mass and spin and (ii) the detection range for IMRIs into
571: Schwarzschild black holes with the same mass. For low spins $\chi \lesssim
572: 0.4$, which are typical for intermediate-mass black holes of $\sim
573: 100-200$ solar masses that gained a significant fraction of their mass via
574: minor mergers, we can approximate the detection range increase due to the
575: inclusion of central black hole spin as
576: \bel{AdvLIGORatio}
577: \frac{\rm Range_{\rm spin}}{\rm Range_{\rm no-spin}}
578: \sim 1+0.6\chi^2 \left( \frac{M}{100\ M_\odot} \right).
579: \ee
580: This is the ratio of detection ranges; the ratio of detection volumes is a
581: cube of this ratio.
582:
583: The effects of cosmological redshift are not significant for Advanced-LIGO
584: IMRIs when the black-hole spin is small. Even prograde equatorial inspirals of
585: neutron stars into $M=100\ M_\odot$ black holes spinning at $\chi=0.9$ are only
586: detectable to $z \approx 0.2$ at an SNR threshold of $8$. The cosmological
587: redshift has the same effect as increasing the black-hole mass, so including
588: redshift increases the ratio of detection volumes at higher spins. For the
589: purposes of including redshift in Fig.~\ref{AdvLIGOBoost}, the inspiraling
590: object mass was set to $m=1.4\ M_\odot$ and a detection threshold of ${\rm
591: SNR}=8$ was assumed.
592:
593: The results described here do not include higher-order ($m \neq 2$) harmonics
594: of the orbital frequency. Higher harmonics are not significant when black-hole
595: spins are small, since in that case they affect both the spinning and the
596: non-spinning rates roughly equally, and so the ratio does not change. However,
597: for high values of spin, the ratios would probably drop somewhat relative to
598: those given in Fig.~\ref{AdvLIGOBoost}, since including higher-frequency
599: harmonics would contribute more to increasing the detection range for inspirals
600: into non-spinning holes than into rapidly holes with prograde orbits
601: (cf.~Fig.~6 of \citep{Mandel2007}).
602:
603: We also compute the dependence of the LISA EMRI detection range on the massive
604: black hole spin. We consider EMRIs of $m=10\ M_\odot$ objects into $M=10^5\
605: M_\odot$, $M=10^6\ M_\odot$, and $M=10^7\ M_\odot$ massive black holes. We
606: assume that a detection is possible at an SNR threshold of 30. (Setting the
607: threshold to 15 changes the results at the $10-20\%$ level.) Cosmological
608: redshift must be included for LISA EMRIs since they can be seen to $z \sim
609: 1-2$. This means we must specify the inspiraling object mass and the SNR
610: detection threshold, since these are necessary to determine the cosmological
611: redshift of the most distant detectable source.
612:
613: LISA EMRIs only sweep through a fraction of the frequency band during the
614: observation time. Therefore, $f_{m\rm in}$ for LISA is set not by the detector
615: threshold, but by the frequency of the gravitational waves emitted one year
616: before plunge. We compute $f_{m\rm in}$ by evolving the gravitational-wave
617: frequency back in time from plunge for one year using the prescription of
618: \citet{BarackCutler2004} (Eqs.~(28) and (29)).
619:
620: For $M=10^5\ M_\odot$, the spin of the black hole is almost irrelevant: once we
621: average over orbital inclinations, the spin affects the detection range at a
622: level of at most a few percent. This is because at these low masses, most of
623: the SNR comes from the portion of the inspiral at much higher radii than the
624: LSO, so the exact frequency of the LSO does not play a very significant role
625: (cf.~Fig.~8 and associated discussion in \citep{Pau}).
626:
627:
628: \begin{figure}
629: \includegraphics[keepaspectratio=true,width=6in]{f4.eps}
630: \caption{The ratio between LISA detection ranges (at SNR$=30$)
631: for extreme-mass-ratio inspirals of $m=10\ M_\odot$ compact objects
632: into Kerr black holes of mass $M=10^6\ M_\odot$ and
633: a given spin vs.~non-spinning black holes.
634: \label{LISABoost6}}
635: \end{figure}
636:
637: \begin{figure}
638: \includegraphics[keepaspectratio=true,width=6in]{f5.eps}
639: \caption{The ratio between LISA detection ranges (at SNR$=30$)
640: for extreme-mass-ratio inspirals of $m=10\ M_\odot$ compact objects
641: into Kerr black holes of mass $M=10^7\ M_\odot$ and
642: a given spin vs.~non-spinning black holes.
643: \label{LISABoost7}}
644: \end{figure}
645:
646: Figure~\ref{LISABoost6} shows the dependence of the average EMRI detection
647: range on the massive-black-hole spin for $M=10^6\ M_\odot$. The average
648: detection range for EMRIs into rapidly spinning black holes of mass $M=10^6\
649: M_\odot$ is $\sim 13\%$ larger than for EMRIs into non-spinning black holes.
650: For $M=10^7\ M_\odot$, the detection range for EMRIs into rapidly spinning
651: black holes is increased by a factor of $\sim 25$ over those into non-spinning
652: black holes, as shown in Fig.~\ref{LISABoost7}. This greater sensitivity to
653: black hole spin is expected, since for these massive black holes most of the
654: SNR comes from the cycles near the LSO. However, this should not be taken to
655: mean that inspirals into rapidly spinning $M=10^7\ M_\odot$ black holes are
656: likely to dominate LISA EMRI observations. Figures \ref{LISABoost6} and
657: \ref{LISABoost7} show detection range ratios only; the inclination-averaged
658: detection range for an EMRI into a maximally spinning $M=10^7\ M_\odot$ black
659: hole is actually less than the detection range for an EMRI into a non-spinning
660: $M=10^6\ M_\odot$ black hole. On the other hand, this large ratio does mean
661: that there is a strong detection bias in favor of rapidly spinning black holes,
662: which must be taken into account when statistics of EMRI observations are
663: inverted to gather information about the massive-black-hole spin distribution.
664:
665:
666:
667: \begin{acknowledgments}
668: I thank Kip Thorne and Curt Cutler for suggesting this problem, and Luc
669: Bouten, Jonathan Gair and Cole Miller for helpful discussions, and Kip Thorne
670: for thorough comments on the manuscript. I was partially
671: supported by NSF Grant PHY-0099568, NASA Grant NAG5-12834, and the Brinson
672: Foundation.
673: \end{acknowledgments}
674:
675:
676:
677: \begin{thebibliography}{}
678:
679: \bibitem[Barack \& Cutler(2004)]{BarackCutler2004}
680: {Barack}, L.~\& {Cutler}, C. 2004, \prd {\bf 69}, 082005
681:
682: \bibitem[Fritschel(2003)]{Fritschel}
683: Fritschel, P. 2003, arXiv:gr-qc/0308090
684:
685: \bibitem[G\"ultekin, Miller, \& Hamilton(2004)]{GMH04}
686: {G\"ultekin}, K., {Miller}, M.~C., \& {Hamilton}, D.~P. 2004,
687: ApJ, 616, 221
688:
689: \bibitem[G\"ultekin, Miller, \& Hamilton(2006)]{GMH06}
690: {G\"ultekin}, K., {Miller}, M.~C., \& {Hamilton}, D.~P. 2006,
691: ApJ, 640, 156
692:
693: \bibitem[Hughes(2000)]{Hughes2000}
694: Hughes, S.~A. 2000, \prd {\bf 61}, 084004
695:
696: \bibitem[Hughes \& Blandford(2004)]{HB}
697: Hughes, S.~A. \& Blandford, R.~D. 2004, ApJ, 585, L101
698:
699: \bibitem[Mandel et al.(2007)]{Mandel2007}
700: Mandel, I., Brown, D.~A., Gair, J.~R., Miller, M.~C.
701: 2007, submitted to ApJ, arXiv:0705.0285
702:
703: \bibitem[Miller(2002)]{Miller2002}
704: Miller, M.~C. 2002, ApJ, 581, 438
705:
706: \bibitem[Miller \& Colbert(2004)]{MC2004}
707: {Miller}, M.~C., \& {Colbert}, E.~J.~M. 2004, IJMPD, 13, 1
708:
709: \bibitem[Miller \& Hamilton(2002a)]{MH02a}
710: {Miller}, M.~C., \& {Hamilton}, D.~P. 2002a, MNRAS, 330, 232
711:
712: \bibitem[Miller \& Hamilton(2002b)]{MH02b}
713: {Miller}, M.~C., \& {Hamilton}, D.~P. 2002b, ApJ, 576, 894
714:
715: \bibitem[Misner, Thorne, \& Wheeler(1973)]{MTW}
716: Misner, C.~W., Thorne, K.~S., \& Wheeler, J.~A. \textit{Gravitation}
717: (Freeman, San Francisco, 1973)
718:
719: \bibitem[Mouri \& Taniguchi(2002a)]{MT02a}
720: {Mouri}, H., \& {Taniguchi}, Y. 2002a, ApJ, 566, L17
721:
722: \bibitem[Mouri \& Taniguchi(2002b)]{MT02b}
723: {Mouri}, H., \& {Taniguchi}, Y. 2002b, ApJ, 580, 844
724:
725: \bibitem[O'Leary et al.(2006)]{OL06}
726: {O'Leary}, R.~M., {Rasio}, F.~A., {Fregeau}, J.~M., {Ivanova}, N.,
727: \& {O'Shaughnessy}, R. 2006, ApJ, 637, 937
728:
729: \bibitem[Amaro-Seoane et al.(2007)]{Pau}
730: Amaro-Seoane, P., Gair, J.~R., Freitag, M., Miller, M.~C., Mandel, I.,
731: Cutler, C.~J., Babak, S. 2007, submitted to CQG, arXiv:astro-ph/0703495
732:
733: \bibitem[Taniguchi et al.(2000)]{Tan00}
734: {Taniguchi}, Y., {Shioya}, Y., {Tsuru}, T.~G., \& {Ikeuchi},
735: S. 2000, PASJ, 52, 533
736:
737: \end{thebibliography}
738: \end{document}
739: