1: %\documentclass[prd,twocolumn]{revtex4}
2: %\input epsf
3:
4:
5: \documentstyle[epsf,prd,aps,epsf]{revtex}
6: %\documentstyle[aps,prd,twocolumn, epsf]{revtex}
7: %\documentstyle[preprint,prd,aps,epsfig]{revtex}
8: %\tighten
9:
10:
11:
12: \begin{document}
13:
14:
15: %\baselineskip 8mm
16:
17: \newcommand{\gsim}{\gtrsim}
18: \newcommand{\lsim}{\lesssim}
19: \newcommand{\psim}{\mbox{\raisebox{-1.0ex}{$~\stackrel{\textstyle \propto}
20: {\textstyle \sim}~$ }}}
21: \newcommand{\vect}[1]{\mbox{\boldmath${#1}$}}
22: \newcommand{\lmk}{\left(}
23: \newcommand{\rmk}{\right)}
24: \newcommand{\lnk}{\left\{ }
25: \newcommand{\nn}{\nonumber}
26: \newcommand{\rnk}{\right\} }
27: \newcommand{\lkk}{\left[}
28: \newcommand{\rkk}{\right]}
29: \newcommand{\lla}{\left\langle}
30: \newcommand{\p}{\partial}
31: \newcommand{\rra}{\right\rangle}
32: \newcommand{\beq}{\begin{equation}}
33: \newcommand{\eeq}{\end{equation}}
34: \newcommand{\beqa}{\begin{eqnarray}}
35: \newcommand{\eeqa}{\end{eqnarray}}
36: \newcommand{\lab}{\label}
37: \newcommand{\sol}{M_\odot}
38: \newcommand{\mch}{{\cal M}}
39: \newcommand{\vex}{{\vect x}}
40: \newcommand{\vel}{{\vect \Omega}_l}
41: \newcommand{\ven}{{\vect \Omega}_s}
42: \newcommand{\vep}{{\vect p}}
43: \newcommand{\veq}{{\vect q}}
44: \newcommand{\veo}{{\vect \Omega}}
45:
46: %\if0
47: \draft
48: \title{Effects of finite arm-length of LISA on analysis of gravitational waves from
49: MBH binaries}
50: \author{Naoki Seto
51: %\footnote{e-mail: seto@vega.ess.sci.osaka-u.ac.jp}
52: }
53: \address{Department of Earth and Space Science, Osaka
54: University, Toyonaka 560-0043, Japan}
55:
56: \maketitle
57:
58: \begin{abstract}
59: Response of an interferometer becomes complicated for gravitational wave
60: shorter than the arm-length of the detector, as nature of wave
61: appears strongly. We have studied how parameter estimation for
62: merging massive black hole binaries are affected by this complicated
63: effect in the case of LISA. It is shown that three dimensional positions
64: of some binaries might be determined much better than the past
65: estimations that use the long wave approximation. For equal mass binaries
66: this improvement is most prominent at $\sim 10^5\sol$.
67: \end{abstract}
68:
69: %\pacs{PACS number(s): 95.55.Ym 04.80.Nn, 98.62.-g }
70: %\fi
71:
72:
73:
74:
75: %%%%%%%%%%%%%%%%%
76: \section{Introduction}
77: %%%%%%%%%%%%%%%%5
78:
79: The {\it Laser Interferometer Space Antenna} (LISA), a joint project of NASA
80: and ESA, is planed to be launched in 2011 \cite{lisahttp}. LISA
81: is sensitive to low frequency gravitational waves at
82: $10^{-4}{\rm Hz}\lsim f \lsim 10^{-1}$Hz, and would directly confirm
83: gravitational radiation from some known Galactic binaries
84: \cite{lisa}. Thousands of closed white dwarf binaries would be also
85: detected by LISA.
86: They can be
87: regarded as insurances of the project, but other exciting phenomena might
88: be observed \cite{lisa}.
89: Coalescence of a massive black hole (MBH) binary is the most
90: spectacular and energetic event.
91: Gravitational wave from a MBH binary is in itself
92: very interesting for studies of general relativity, but would also
93: bring significant
94: impacts on astrophysics and cosmology \cite{lisa}, even though event
95: rate is highly
96: unknown at present \cite{Haehnelt:wt}. Parameter estimation
97: errors are
98: basic measures for discussion of gravitational wave
99: astronomy. For example, with small error box for the position of a MBH
100: binary we might specify its host galaxy and investigate it with various
101: observational tools.
102:
103: LISA is constituted by three space crafts that keep a triangle
104: configuration with side (arm-length) $l=5.0\times 10^6$km and trail
105: $\sim 20^\circ$ behind the Earth \cite{lisa}.
106: Response of a detector becomes complicated for gravitational waves with
107: wave-length $\lambda\lsim l$, as the nature of wave appears
108: strongly \cite{Larson:1999we,Schilling:id,Cornish:2002rt}.
109: This interesting feature might
110: be an advantage for signal analysis but has not been taken in properly
111: so far. In
112: this paper we discuss how the parameter estimation errors are changed by
113: this effect.
114:
115: This paper is organized as follows. In Section II we study the effect of
116: a finite arm-length on the phase shift $\delta \phi$ of the detector,
117: and introduce
118: the past analysis that is based on the simple picture using the
119: variations of the arm-lengths $\delta l$ under the long wave
120: approximation. In Sec III A gravitational wave forms from MBH binaries are
121: briefly discussed. Noise spectrum for both the variations $\delta l$
122: and the phase sifts $\delta \phi$ are presented in Sec III B. The
123: spectrum for the latter is simply reproduced from that for the former.
124: In Sec IV our numerical results are shown, and signal to noise ratio
125: and the parameter estimation errors are compared for the two methods (with
126: and without the long wave approximation). Though we mainly study
127: gravitational waves from MBHs, results for nearly
128: monochromatic sources (from {\it e.g.} Galactic binaries) are also presented.
129: Sec V is devoted to discussion. In Appendix A we make same kind of
130: comparisons for the noise canceling combination of data streams.
131: In Appendix B we present explicit expressions for detector's response
132: to
133: gravitational wave from a binary.
134:
135:
136:
137:
138: %%%%%%%%%%%%%%%%%
139: \section{Detectors with finite arm-length}
140: %%%%%%%%%%%%%%%%5
141:
142: We name the three vertexes (space crafts) of LISA as A, B and C.
143: Cutler \cite{Cutler:1998ta} studied parameter estimation of binaries
144: from variations of arm-lengths in the following forms
145: \beq
146: \delta l_{AB}(t)-\delta l_{AC}(t),~~~
147: \lnk(\delta l_{BA}(t)-\delta l_{BC}(t))-(\delta l_{CB}(t)-\delta
148: l_{CA}(t))\rnk/\sqrt{3}.\label{ldata}
149: \eeq
150: The normalization factor $1/\sqrt{3}$ in the second expression is
151: explained later.
152: Here $\delta l_{XY}$ is the variation of the arm-length between two space
153: crafts $X$ and $Y$ ($X,Y\in\{A,B,C\},~X\ne Y$). This simple picture for detector's
154: response is valid
155: only at long wave limit where gravitational wave-length is much larger
156: than the detector's arm-length (see also
157: \cite{Hughes:2001ya,Moore:1999zw}).
158: With quantities more close to
159: detector's output \cite{lisa} these
160: two data should be replaced by
161: \beq
162: \delta\phi_{AB}(t)-\delta\phi_{AC}(t),~~~
163: \lnk(\delta\phi_{BA}(t)-\delta\phi_{BC}(t))-(\delta\phi_{CB}(t)-\delta\phi_{CA}(t))\rnk/\sqrt{3},\label{data}
164: \eeq
165: where $\delta\phi_{XY}$ represents the single-arm phase shift of the laser beam
166: that leaves a detector (vertex of the triangle) X for Y and then
167: returns back to
168: X. The
169: quantity $\delta\phi_{AB}-\delta\phi_{AC}$ is an interferometer
170: signal obtained at the detector A.
171:
172:
173: In Ref.\cite{Cutler:1998ta} it is assumed
174: that the
175: noise for the time variations $\delta l_{XY}$ is
176: stationary, Gaussian and symmetric among three arms.
177: We make same assumption for the corresponding phase shifts $\delta \phi_{XY}$
178: In this case
179: noises of the above two data (\ref{data}) (or (\ref{ldata})) do not
180: correlate to each other.
181: %As in past studies for parameter estimation of binaries
182: %\cite{Cutler:1998ta,Hughes:2001ya,Moore:1999zw}, we do not use the noise
183: %canceling combinations of data (see {\it e.g.} \cite{Armstrong:uh}).
184:
185:
186:
187: Response of the phase shift $\delta\phi_{AB}$ for
188: gravitational wave $h$ incoming from
189: a direction $\veo_s$ is expressed as follows
190: \cite{Larson:1999we,Schilling:id,Hellings:jm}
191: \beq
192: \frac1{2\pi \nu_0}\frac{d\delta\phi_{AB}(t)}{dt}=\frac12 \cos2\psi_{AB}
193: [(1-\cos\theta_{AB})h(t,A)+2\cos\theta_{AB}
194: h(t-\tau,B)-(1+\cos\theta_{AB})h(t-2\tau,A)],\label{pt}
195: \eeq
196: where $\theta_{AB}$ is the
197: angle between the source direction $\ven$ and the arm
198: $\vex_A-\vex_B$, and $\nu_0$ is the fundamental frequency of the
199: laser. $\psi_{AB}$ is the principle polarization angle of the quadrupole
200: gravitational wave $h$ for the arm $\vex_A-\vex_B$. We denote the
201: polarization basis tensor $H_{ab}$
202: of the wave as $H_{ab}=p_ap_b-q_aq_b$ using two orthogonal vectors $\vep$
203: and $\veq$ with $\vep\cdot\ven=\veq\cdot\ven=\vep\cdot\veq=0$ and $|\vep|=|\veq|=1$. Then the angle $\psi_{AB}$
204: is given by $\tan \psi_{AB}=({\veq}\cdot(\vex_A-\vex_B))/({\vep}\cdot(\vex_A-\vex_B))$ \cite{Hellings:jm}.
205: The quantity
206: $h(t,A)$ is value of the gravitational wave at point A and
207: time $t$, and can be
208: expressed as $h(t+\veo_s\cdot\vex_A(t)/c)$ in the present case.
209: We have denoted the propagation time of light for the
210: arm $l$ by $\tau\equiv l/c$,
211: and neglected very small relative motions between the vertexes in the
212: time scale $\tau$. In Appendix B we present explicit expression of
213: Eq.(\ref{pt}) for a binary source.
214:
215:
216: When the gravitational wave-length is much larger
217: than the arm-length $l$, namely $f\ll (2\pi\tau)^{-1}\equiv f_{arm}$,
218: structure of wave is irrelevant at the spatial scale $l$ and the wave $h$
219: causes variations $\delta \phi_{AB}$ or $\delta l_{AB}$ in very simple
220: manners. In this long wave limit
221: the phase
222: shift is directly proportional to
223: the variation of the arm-length $\delta l_{XY}$ as $\delta \phi_{XY}/2\pi\nu_0=
224: \delta l_{XY}/2c$, and two data sets (\ref{ldata}) and
225: (\ref{data}) are equivalent. We
226: have the
227: following relation by perturbative evaluation of Eq.(\ref{pt})
228: (see also Ref.\cite{Hellings:jm})
229: \beq
230: \frac{\delta \phi_{AB}}{2\pi \nu_0\tau}=
231: \sin^2\theta_{AB} \cos 2\psi h(t)=\frac{\delta l_{AB}}{2l}\label{lw}.
232: \eeq
233: LISA has arm-length $l=5.0\times 10^6$ km, corresponding to the
234: critical
235: frequency
236: $f_{arm}=0.01$Hz.
237: Note that the frequency $f$ does not appear in the
238: above relation
239: (\ref{lw}).
240: In this long wave limit the angular pattern function
241: (usually
242: denoted as $F^{+,\times}$ \cite{thorne}) contains complete information of the
243: angular dependence
244: of the interferometer signal ({\it e.g.}
245: $\delta\phi_{AB}-\delta\phi_{AC}$). Two data
246: streams (\ref{ldata}) are equivalent to responses of two
247: $90^{\circ}$-interferometers
248: rotated by $45^\circ$ (see discussion in \cite{Cutler:1998ta}). The factor $1/\sqrt{3}$
249: in the second expression of (\ref{ldata}) or (\ref{data}) is given for
250: this purpose.
251:
252:
253:
254:
255: When the gravitational wavelength $c/f$ is comparable to or shorter
256: than the arm-length $l$, the phase shifts $\delta\phi_{XY}$ show
257: complicated response that
258: depends strongly on the frequency $f$ and the source direction relative to
259: the detector (see Appendix B). As discussed earlier, this troublesome but interesting
260: effects have been cut down so far in analysis of parameter
261: estimation errors for binary sources.
262: Past analysis was
263: performed in the following manner
264: \cite{Cutler:1998ta,Hughes:2001ya,Moore:1999zw}.
265:
266: (i) The angular averaged transfer
267: function $T(f)$ between the phase shift $\delta \phi$ and wave amplitude
268: $h$ was used $\delta \phi\propto T(f)h$ to include amplitude modulation
269: caused by finiteness of the arm-length $l\ne 0$.
270: Its effect can be effectively
271: absorbed in the noise curve $\sqrt{S_h(f)}$ for the wave amplitude
272: $h$.
273:
274: (ii) Then the simple analysis
275: with the variation of the arm-length $\delta l_{XY}$ (valid only at the
276: long wave limit) was
277: applied with the angular pattern function $F^{+,\times}$ for
278: the angular dependence of the
279: response (and also Doppler phase caused by velocity of detectors).
280:
281:
282: At
283: high frequency limit $f\gg f_{arm}$ the
284: transfer function becomes $T(f) \propto f^{-1}$ due to cancellation
285: of wave within arms. The measurement sensitivity of LISA for the
286: phase shift
287: $\delta \phi$ is $\sqrt{S_{\delta\phi}(f)}\propto f^0$ (in
288: units of ${\rm Hz}^{-1/2}$) at $10^{-2}{\rm Hz}\lsim f\lsim 10^0{\rm Hz}$ where the shot
289: noise is dominant source of noise. Thus LISA has
290: $\sqrt{S_{h}(f)}\propto \sqrt{S_{\delta\phi}(f)}T(f)^{-1}\propto f$
291: at high
292: frequency region \cite{lisa,Larson:1999we,Schilling:id}.
293:
294:
295: Here we discuss basic aspects of signal analysis with paying attention
296: to the finiteness of the arm-length $l$.
297: The most important quantity for detection of gravitational wave is the signal
298: to noise ratio (SNR) \cite{thorne}. For its estimation the above mentioned method with
299: the
300: angular averaged transfer function would be effective for LISA
301: considering its rotation and frequency average of chirping signals. But
302: parameter estimation errors ({\it
303: e.g.} error box for direction of a source) might become much smaller
304: than the previous
305: analysis when we include the complicated response of the phase shift
306: as it is.
307: Gravitational wave signal measured by a detector (as expression
308: (\ref{data})) is given by values of gravitational wave $h$ at
309: different positions and times. Time delay
310: between them would also affect parameter estimation.
311: These two effects, time delay and cancellation within the arms, are
312: natural outcome of
313: the finiteness of the arm-length and appear in a similar form, namely, linear
314: combination of terms like $\exp[2\pi i f \veo_s\cdot
315: (\vex_A-\vex_B)/c]$ in Fourier space of the time variable $t$.
316: Our primary aim in this paper is to show how the parameter estimation would
317: be changed if we simply
318: replace the arm-length variation $\delta l_{XY}$ (as in Eq.(\ref{ldata})) with the corresponding
319: phase shifts $\delta \phi_{XY}$ (as in Eq.(\ref{data})) without using the long wave
320: approximation.
321:
322: \section{Analysis of Gravitational Waves from Merging Massive Black Hole
323: Binaries}
324: \subsection{Gravitational Waveform}
325:
326: We investigate the parameter estimation errors expected in matched
327: filtering method \cite{Finn:1992wt}.
328: When the wave signal contains fitting parameters
329: $\lnk \gamma_i\rnk$, their estimation errors
330: (variances) for signal analysis are evaluated
331: with using the Fisher's
332: information matrix $\Gamma_{ij}$ as $\lla \Delta\gamma_i \Delta\gamma_j
333: \rra=\Gamma^{-1}_{ij}$.
334: We analyze the
335: phase shifts (\ref{data}) by extending earlier studies
336: to include effects caused by the arm-length, but other points are almost
337: identical to Ref.\cite{Cutler:1998ta} that uses the variations
338: (\ref{ldata}) under the long wave approximation.
339: %The inner product $(a|b)$ of two data $a(t)$ and $b(t)$ is
340: %defined as $(a|b)=4{\rm Re} \int_0^\infty df \frac{{a^*(f)b(f)}}{S(f)}$
341: Detailed analysis for the phase shift is given in Appendix B.
342:
343: %For multiple signal $h_m$ with uncorrelated noise the
344: %above inner product $(h|h)$ are replaced by summation $\sum_m
345: %(h_m|h_m)$.
346:
347: For gravitational wave emitted by a binary we adopt the
348: restricted post-Newtonian approach, but higher-order harmonics could
349: become important in some cases \cite{Hellings:2002si}. We use the 1.5PN phase for a
350: circular orbit \cite{Cutler:1993tc}
351: \beq
352: \Psi(f)=2\pi f t_c-\phi_c-\frac\pi4+\frac34(8\pi G c^{-3} \mch_z f)^{-5/3} \lkk
353: 1+\frac{20}9 \lmk \frac{943}{336}+\frac{11\mu_z}{4M_z}\rmk
354: x+(4\beta-16\pi) x^{3/2} \rkk, \label{pev}
355: \eeq
356: where $t_c$ and $\phi_c$ are integration constants, $\beta$ is the
357: spin-orbit coupling term, $\mu_z$, $M_z$ and $\mch_z$ are the reduced
358: mass, the total mass and the chirp mass of the binary. All of the mass
359: parameters are multiplied by the factor $(1+z)$ ($z$: redshift of the
360: binary) with suffix $z$. The total mass $M_z$ is expressed by other two
361: masses as $M_z=
362: \mch_z^{5/2} \mu_z^{-3/2}$.
363: The post-Newtonian expansion parameter $x$ is defined as $x\equiv \lnk
364: G \pi c^{-3}
365: M_z f
366: \rnk^{2/3} $.
367: We have used the stationary phase approximation for the Fourier
368: transformed waveform.
369: In our analysis total number of the fitting parameters is 10 as
370: $\mch_z,\mu_z,\beta,\phi_c, t_c, \ln D$(luminosity distance) and
371: $(\theta_s,\phi_s),(\theta_l,\phi_l)$. The latter four parameters define
372: the direction $\ven$ and the orientation $\vel$ of the binary in
373: a fixed
374: frame at the barycentre of the solar system. In this paper we use the
375: error ellipse
376: \beq
377: \Delta \veo_i\equiv 2\pi \sin \theta_i\sqrt{\lla
378: \Delta\theta_i^2\rra \lla \Delta\phi_i^2\rra -\lla
379: \Delta\theta_i\Delta\phi_i\rra^2 },\label{angl}
380: \eeq
381: for the angular
382: parameters ($i=s,l$) \cite{Cutler:1998ta} and $\Delta V\equiv \Delta D \Delta \ven$ for
383: the three dimensional position error of a source. We also fix $\beta=0$ for
384: the
385: true value of the parameter $\beta$.
386:
387:
388: Unless stated explicitely, we put the upper cut-off frequency $f_{cut}$
389: of signal integration at $f_{isco}$
390: when
391: the binary separation becomes
392: $r=6M_z/c^2(1+z)$, roughly corresponding to the inner most stable circular
393: orbit.
394: The observed wave frequency at this separation becomes
395: \beq
396: f_{isco}=\frac{6^{-3/2}c^3}{G\pi M_z}=0.022\lmk \frac{M_z}{2\times10^5
397: M_\odot}\rmk^{-1}
398: {\rm Hz}\label{fcut}.
399: \eeq
400: We start integration of the wave signal from the time when
401: the binary is 1yr
402: before its coalescence in the observer's frame.
403: The gravitational wave
404: frequency $f_0$ at this starting time is given as follows
405: \beq
406: f_0=1.9\times10^{-4} \lmk\frac{\mch_z}{0.87\times 10^5 M_\odot} \rmk^{-5/8}{\rm Hz}. \label{f0}
407: \eeq
408:
409: \subsection{Noise Curve}
410:
411: We make quantitative analysis using noise curve of LISA that is sum of
412: the instrumental noise and the binary confusion noise. For the
413: instrumental noise
414: spectrum given for the wave amplitude $h$ (almost equivalently for
415: $\delta l$) we
416: adopt the
417: following function \cite{Cutler:1998ta}
418: \beq
419: S_i(f)=5.049\times 10^5[\alpha_1(f)^2+\alpha_2(f)^2+\alpha_3(f)^2]~{\rm
420: Hz}^{-1}\label{sp},
421: \eeq
422: where $\alpha_1(f)=10^{-22.79}(f/10^{-3}{\rm Hz})^{-7/3}$ is mainly the acceleration
423: noise, $\alpha_2(f)=10^{-23.04}$ is mainly the shot noise, and
424: $\alpha_3(f)=10^{-24.54}(f/10^{-3}{\rm Hz})$ approximately
425: represents the cancellation effects
426: due to the finiteness of the arm-length. The last term reflects the angular
427: averaged transfer function as explained before, and becomes important at
428: $f\gsim
429: 10^{-2}$Hz (see {\it e.g.} \cite{Larson:1999we}). We also include the
430: binary confusion noise $S_c(f)$ (estimated for
431: one year observation) that is stronger than the instrumental one at
432: $f\lsim 10^{-2.5}$Hz and shows strong dependence on frequency
433: $f$ \cite{Bender:hs}. Around $f\gsim 10^{-2.75}$Hz the confusion noise
434: decreases
435: significantly as the Galactic
436: binaries are resolved at higher frequencies.
437: Its explicite expression is given as follows \cite{Cutler:1998ta}
438: \beq
439: S_c(f)=\cases{
440: 10^{-42.685}f^{-1.9}{\rm Hz^{-1}}, ~~ f\le 10^{-3.15}, \cr
441: 10^{-60.325}f^{-7.5}{\rm Hz^{-1}}, ~~ 10^{-3.15}\le f\le 10^{-2.75}, \cr
442: 10^{-46.85}f^{-2.6}{\rm Hz^{-1}}, ~~ 10^{-2.75}\le f, \cr
443: }
444: \eeq
445: where $f$ is written in units of Hz.
446:
447:
448: Now let us simply recover the noise spectrum $S_{\delta \phi}(f)$ for
449: the phase shift $\delta
450: \phi_{AB}-\delta
451: \phi_{AC}$ from that for the variation $\delta l$.
452: %The above spectra ({\it
453: %e.g.} Eq.(\ref{sp}))
454: %are given for the gravitational wave amplitude $h$ or $(\delta
455: %l_{AB}-\delta l_{AC})/l$, but can be easily
456: %related to the spectra $S_{\Delta\phi}(f)$ of the phase shift by noticing
457: %their behaviors
458: At low frequency region $f\ll f_{arm}
459: $ they are simply related to each other as expressions (\ref{ldata})
460: (\ref{data}) and are essentially equivalent (see Eq.(\ref{lw})). We use the
461: following functional
462: shape for the instrumental noise of
463: the phase shift $S_{\delta\phi}(f)$ as
464: $ S_{\delta\phi}(f)\propto (\alpha_1(f)^2+\alpha_2(f)^2)$ by removing
465: the effect of the angular averaged transfer function from
466: $S_i(f)$. Note
467: that the confusion noise is
468: expected to be much smaller than the instrumental one at the relevant
469: frequencies $f\gsim f_{arm}$.
470: For the second data of expression (\ref{data}) we use the same noise
471: curve as the first one.
472:
473: %Strictly speaking this is not correct but would
474: %not change general behaviors of our results.
475:
476: %The two data streams (\ref{data}) are integrated for the
477: %last 1 year before
478: %coalescence upto the cut off frequency given by Eq.(\ref{fcut}).
479: Our
480: calculation
481: using the phase shift is much more complicated than the past one.
482: But our numerical results should coincide with these
483: by past one when we
484: decrease the arm-length $l$ and use the shape of the noise curve
485: $S_h(f)$ instead of $S_{\delta\phi}(f)$. We have confirmed that the
486: parameter errors $\Delta \gamma_i$ given in table 2 of
487: Ref.\cite{Cutler:1998ta} (obtained
488: by past simple calculation) is reproduced with this procedure.
489:
490:
491:
492: %%%%%%%%%%%%%%%%%
493: \section{results}
494: %%%%%%%%%%%%%%%%5
495: We have analyzed gravitational waves from MBH binaries with 300
496: realizations of random directions $\ven$ and orientations $\vel$.
497: Firstly, their SNRs and estimation errors of the fitting parameters $\lnk
498: \gamma_i \rnk$ are
499: calculated in both (i) past approach (suffix $0$) with angular
500: averaged transfer function under the long wave approximation, and (ii)
501: our method (suffix $L$) based on the
502: phase shifts of expression (\ref{data}). In figure 1 we present
503: distribution for
504: ratios of SNRs by two methods $(SNR)_L/(SNR)_0$ and position errors
505: $\Delta V_L/\Delta V_0$, $(\Delta\ven)_L/(\Delta\ven)_0$ and
506: $\Delta D_L/\Delta D_0$ obtained
507: for equal mass MBH binaries with redshifted masses
508: $M_{z}=10^4+10^4, 10^5+10^5$ and $10^6+10^6\sol$.
509: Note that these ratios do not depend on cosmological parameters or
510: distance $D$ to the MBH binaries. As shown in the bottom panel of figure
511: 1, ratios
512: of SNRs are close to unity and difference of two methods is very small
513: (within $20\%$).
514: However, two methods show considerable difference for the position errors. For $10^5+10^5\sol$ MBH binaries the errors $\Delta V_L $
515: become typically 10 times smaller than the past estimate $\Delta
516: V_0$. For some samples the errors $\Delta V$ are reduced even by a
517: factor $\sim 10^{-2}$. Comparing errors $\Delta \ven$ and $\Delta D$,
518: the former become smaller by the finiteness of the arm-length than
519: the latter. But note that the angular resolution $\Delta \ven$ is,
520: roughly speaking, a product of two errors $\Delta \theta_s$ and $\Delta
521: \phi_s$ as in Eq.(\ref{angl}).
522:
523:
524: Interestingly enough, difference between two methods is not a monotonic
525: function
526: with respect to BH mass and is most prominent around $\sim
527: 10^5\sol$. This mass dependence can be understood as follows. At higher
528: masses $\gg 10^5\sol$ the upper cut-off frequency $f_{isco}$ is smaller than the critical
529: frequency $f_{arm}$. Therefore nothing is different between two methods
530: with expressions (\ref{ldata}) and (\ref{data}). With the quadrupole formula
531: for gravitational radiation
532: the time
533: before coalescence is given by the frequency $f$ and the chirp mass $\mch_z$
534: as follows \cite{thorne}
535: \beq
536: t_{GW}=8.4\times 10^4 \lmk \frac{f}{10^{-2.75}{\rm Hz}}\rmk^{-8/3} \lmk
537: \frac{\mch_z}{0.87\times 10^5 \sol}\rmk^{-5/3} {\rm sec}.
538: \eeq
539: For lower mass BH binaries ($\ll 10^5\sol$) LISA moves longer than the
540: arm-length $l$ during
541: the phase $f\gsim 10^{-2.75}$Hz where signal becomes very strong due to
542: decrease of the binary
543: confusion noise. Thus effective baseline of the detector becomes larger
544: than the arm-length $l=5.0\times 10^6$km for smaller mass BH binaries
545: and impact of the finite arm-length would decrease.
546:
547:
548: Now we investigate various aspects of parameter estimation caused by the
549: finiteness of the arm-length using a specific set of binary parameters.
550: We pick up the binary that has the smallest volume ratio $\Delta
551: V_L/\Delta V_0=0.011$ in 300 realization of figure 1 for $10^5+10^5\sol$
552: MBH binaries.
553: It has angular parameters $\theta_s=2.49, \phi_s=0.03, \theta_l=2.32$
554: and $\phi_l=4.46$. Here we present the estimation errors $\Delta \gamma_i$,
555: not ratios as in figure 1. To normalize the amplitude of the signal we
556: take the redshift of MBH binaries at $z=1$ with cosmological parameters
557: $\Omega_0=0.3, \lambda_0=0.7$ and $H_0=75$km/sec/Mpc. In figure 2 estimation
558: errors $\Delta \ven, \Delta \vel, \Delta D/D$ and $\Delta
559: \mu_z/\mu_z$ are presented as functions of the upper cut-off frequency
560: $f_{cut}$ that was fixed at $f_{isco}$ with Eq.(\ref{fcut}) in the case
561: of figure 1. We fix the lower cut-off frequency at $f_0$ (eq.[\ref{f0}])
562:
563:
564:
565: We have found that the intrinsic binary parameters such as
566: $\mch_z, \mu_z, \beta, t_c$ and $\phi_c$ depend weakly on the cut-off frequency
567: $f_{cut}$.
568: However, errors for the binary position $\Delta\ven$, $\Delta D/D$
569: or its orientation $\Delta \vel$ decrease significantly at
570: $f_{cut}\gsim 0.01$Hz where the long wave approximation breaks down. This shows remarkable contrast to the past analysis
571: shown with thin lines.
572: Our results seem reasonable, as the response of a detector with
573: finite arm-length depends strongly on the direction of the source
574: $\ven$, and information
575: of the distance $D$ or orientation $\vel$ is tightly correlated to them
576: (Appendix B).
577: Significant reduction of position errors $\Delta V_L$ at higher
578: frequencies would give further motivation for studies of nonlinear
579: gravitational dynamics ({\it e.g.} Post-Newtonian approach)
580: that is a very tough problem on general relativity. By analyzing
581: gravitational wave close to the
582: final coalescence we might identify the host galaxy of a MBH
583: binary!
584:
585:
586:
587:
588: Next let us make hypothetical experiments to clarify some interesting
589: points. We use the same set of the parameters as
590: in figure
591: 2 with $f_{cut}$ given by Eq.(\ref{fcut}). As commented before, the finiteness of the arm-length causes two
592: similar effects (i) cancellation of waves within the arm and (ii) time
593: delay between the vertexes of LISA. To extract effects only of the latter we
594: calculate the volume error $\Delta V_{L'}$ using data from three
595: interferometers that exist at three vertexes of LISA but have arm-length
596: $l\to0$ with angular averaged transfer function for sensitivity of
597: $h$. Thus only the
598: positions (separation) of the detectors are different from the past
599: analysis that take the separation $l=0$. We
600: obtain $\Delta V_{L'}/\Delta V_0=0.15$. This result indicates that two
601: effects work cooperatively.
602: Next we stop motion of LISA and keep its position at time $t=t_c$. In
603: this case the volume error $\Delta V_{L''}$ becomes $\Delta
604: V_{L''}/\Delta V_{0}=0.012$ and is very close to $\Delta
605: V_{L}/\Delta V_{0}=0.011$ that includes motion of LISA. In the past
606: analysis (with $l=0$) we use the
607: amplitude modulation through the pattern function and the Doppler phase
608: modulation both caused by motion of LISA \cite{lisa}, and cannot solve degeneracy
609: of sources direction $\ven$ and
610: other variables when LISA stops. The response of a detector with $l\ne
611: 0$ depends strongly both on angular variables and frequency of
612: incoming waves at $f\gsim f_{arm}$ (see Appendix B). Thus we can solve the degeneracy
613: for chirping binaries
614: even without motion of LISA, though there would be two solution for
615: detector's signal due to the symmetry of source-detector configuration.
616: This is a qualitatively interesting point.
617: %and might be also applied for a ground-based detector to
618: %determine the direction of a binary source.
619:
620:
621: Considering the mass dependence of our results, it is expected that
622: the past simple method using expression
623: (\ref{ldata}) under the long wave approximation would be effective
624: for studying galactic compact binaries. These binaries with $f\lsim
625: 0.1$Hz are nearly
626: monochromatic and
627: have
628: long durations $t_{GW}\gg 1$yr in the LISA band due to their small chirp masses \cite{lisa}.
629: We have investigated binaries with $m_1=m_2=1\sol$, and the time to
630: coalescence $t_{GW}=30$yr and
631: 100yr. The wave frequency becomes $f=0.071$Hz at $t_{GW}=30$yr and
632: $f=0.045$Hz at $t_{GW}=100$yr. We fix the observation period at 1yr,
633: thus do not observe the final
634: coalescence in contrast to the analysis for MBH binaries. We calculate
635: parameter estimation errors in this situation with random direction $\ven$ and
636: orientation $\vel$. It is found that the
637: differences between ($\Delta V_L$,$\Delta V_0$) or ($(SNR)_L$,$(SNR)_0$) are less
638: than 15 percent. Recalling that we have made a simple treatment for the
639: effects of
640: the transfer function, this result seems excellent.
641:
642:
643:
644:
645:
646: %%%%%%%%%%%%%%%%%
647: \section{discussion}
648: %%%%%%%%%%%%%%%%5
649:
650: In this paper we have studied how data analysis of gravitational waves
651: from MBH binaries are affected by finiteness of the arm-length of LISA
652: with using the phase shift of detectors.
653: We have numerically confirmed that
654: the
655: past method with the long wave approximation is very effective
656: for estimation of SNRs that are the most important quantities for
657: detection of gravitational waves.
658: However, LISA is able to observe merging MBH binaries with significant SNRs
659: ($\gsim 1000$) even at cosmological distances \cite{lisa,Cutler:1998ta,Hughes:2001ya,Moore:1999zw}, and aspects of
660: gravitational wave astronomy are more relevant for them, rather than
661: SNRs. We have examined the parameter estimation errors $\Delta
662: \gamma_i$ expected in both
663: two methods, and shown that three dimensional position of MBH binaries
664: could be determined much better than the past estimations. In the case of
665: equal mass binaries differences between two methods are most prominent
666: at $\sim 10^5\sol$ and the volume of the error box can decrease
667: significantly. We have shown that (i)
668: this reduction is mainly
669: cause by gravitational waves close to the final coalescence and (ii) position
670: of chirping binaries can be in principle, estimated even without motion of LISA.
671: The former would give further meaning for studies on strong
672: gravitational dynamics, and the latter makes remarkable contrast to
673: former discussions.
674:
675: \if0
676: Past analysis with using noise curve with angular averaged transfer
677: function might be more or less affected by inquiry for detectability
678: (as represented by SNR) of the wave signal. Gravitational wave
679: astronomy would ripe in this century. We hope that this paper provides an
680: opportunity to take a fresh look at theoretical study of data analysis
681: for gravitational waves.
682: \fi
683:
684:
685: The author would like to thank an anonymous referee for helpful
686: comments to improve the manuscript.
687: This work was supported in part by
688: Grant-in-Aid of Scientific Research of the Ministry of Education,
689: Culture, Sports, Science and Technology No. 0001416.
690:
691: \appendix
692:
693: \section{Noise Canceling Combination}
694: In the main text %we have studied how the parameter estimation changed
695: we have basically followed the model for signal analysis in
696: Ref.\cite{Cutler:1998ta},
697: and simply replace the variations (\ref{ldata}) with the corresponding
698: phase shifts
699: (\ref{data}) without resorting to the long wave approximation.
700: It has been discussed recently that the laser phase noise can be removed
701: well by devising combination of data at different times
702: \cite{Armstrong:uh}. This noise is
703: caused by un-equal arm-lengths of space detectors in contrast to the
704: ground-based ones.
705: In this appendix we discuss parameter estimation using this combination.
706: Let us consider the situation that the space crafts B
707: and C coherently transmit laser beams back to the space craft A.
708: Then the following data stream $X(t)$ is a noise canceling combination
709: \beq
710: X(t)=\delta\phi_{AB}(t)-\delta\phi_{AC}(t)-\delta\phi_{AB}(t-2l_{AC}/c)+\delta\phi_{AC}(t-2l_{AB}/c). \label{xt}
711: \eeq
712: %
713: %In section 4 we also discuss signal analysis with the data stream
714: %$X(t)$.
715: %
716: For quantitative study we take the limit $l_{AB}=l_{AC}=l=5.0\times
717: 10^6$km. Then the Fourier transformation of $X(t)$ is related to that for
718: $\delta\phi_{AB}(t)-\delta\phi_{AC}(t)$ as
719: \beq
720: X(f)=(1-\exp(2\pi i f l/c))(\delta\phi_{AB}(f)-\delta\phi_{AC}(f)).\lab{xf}
721: \eeq
722: In this case the noise curve for the gravitational wave amplitude with data
723: $X$ is identical to that with $\delta\phi_{AB}-\delta\phi_{AC}$
724: \cite{Larson:1999we}, if we only include the acceleration and shot noises.
725:
726:
727:
728: In figure 3 we present the two ratios $(SNR)_L/(SNR)_0$ and $\Delta
729: V_L/\Delta V_0$ for the noise canceling combination $X$ given in
730: Eq.(\ref{xt}). Note that the
731: factor $(1-\exp(2\pi i l f/c))$ in Eq.(\ref{xf}) contains none of our ten fitting
732: parameters. Thus results for the data $X(t)$ is identical to that for
733: the single
734: data stream $\delta\phi_{AB}-\delta\phi_{AC}$. Figure 3 shows same
735: kind of mass dependence as figure 1. Difference between $\Delta
736: V_L$ and $\Delta V_0$ is most prominent at mass $\sim 10^5\sol$ again.
737: But the ratio $\Delta
738: V_L/\Delta V_0$ is generally close to unity and effects of the finite
739: arm-length are smaller.
740:
741:
742: {\bf Note added} After submitting this paper, there appears
743: \cite{Prince:2002hp} that discusses three data streams $A,E,T$
744: whose noises do not correlate. Our study can be easily extend to
745: these data and more realistic results would be obtained.
746:
747: \section{Phase shift of a detector for gravitational wave from binaries}
748: In this appendix we give an explicit expression for the phase shift in
749: the form (eq.[3])
750: \beq
751: \frac1{2\pi \nu_0}\frac{d\delta\phi_{AB}(t)}{dt}. \label{b1}
752: \eeq
753: The shift $\delta\phi_{AB}$ is defined for the laser beam that leaves
754: the detector $A$ for $B$, and then returns back to $A$. For source of
755: gravitational radiation $h$ we consider a chirping binary at luminosity
756: distance $D$, direction $\ven$, and orientation $\vel$. We denote
757: $\vex_Y$ as the
758: position vector of a detector $Y\in \{A,B,C\}$,
759: and use a coordinate system fixed to the barycentre of the solar
760: system. As we do not directly use the angular pattern functions $F_+$ and
761: $F_\times$ (see \cite{thorne}), there is no need to introduce a
762: coordinate system fixed to detectors.
763:
764: The inclination angle $i$ of the binary is given as
765: \beq
766: \cos i=\ven\cdot\vel.\label{b2}
767: \eeq
768: We denote the propagation time of light for the arm-length
769: by $\tau\equiv|\vex_A-\vex_B|/c$, and define the angle $\theta_{AB}$
770: between the direction of a binary $\ven$ and the arm $\vex_A-\vex_B$ as
771: \beq
772: \cos \theta_{AB} =\frac{\ven\cdot (\vex_A-\vex_B)}{c\tau}.
773: \eeq
774:
775: It is convenient to use the principle axes $(\vep,\veq)$ for analysis of
776: gravitational wave from a binary. These two vectors are expressed in
777: terms of direction $\ven$ and orientation $\vel$ as
778: \beq
779: \vep=\frac{\ven\times\vel}{|\ven\times\vel|},~~~\veq=-\ven\times \vep. \label{b4}
780: \eeq
781: With these vectors gravitational wave from a binary is decomposed to two
782: polarization ($+$ and $\times$) modes whose phases differs
783: by $\pi/2$.
784: The plus ($+$) mode has polarization basis tensor
785: $H^+_{ab}=p_ap_b-q_aq_b$
786: and the amplitude $A_+$ at the Newtonian order as
787: \beq
788: A_+=\frac{2G^{5/3}\mch_z^{5/3}}{Dc^4}(\pi f)^{2/3}(1+\cos^2i),\label{b5}
789: \eeq
790: where $\mch_z$ is the redshifted chirp mass of the binary. For this mode
791: the principle polarization angle $\psi_{AB}$ is given by
792: \beq
793: \tan \psi_{AB}=\frac{(\vex_{A}-\vex_B)\cdot \veq}{(\vex_{A}-\vex_B)\cdot\vep}.\label{b6}
794: \eeq
795: The cross ($\times$) mode has the amplitude
796: \beq
797: A_\times=\frac{4G^{5/3}\mch_z^{5/3}}{Dc^4}(\pi f)^{2/3}\cos i,\label{b7}
798: \eeq
799: with the polarization tensor $H^\times_{ab}=p_aq_b+q_ap_b$.
800: Its principle polarization angle differs by $\pi/4$ from the plus
801: mode.
802:
803: Now we can write down the phase shift (\ref{b1}) with various parameters of the
804: binary. For notational simplicities we define a function $U(t,Y)$ that
805: contains information of the phase of the gravitational wave at the detector $Y$ and time $t$
806: (taking its origin appropriately) as
807: \beq
808: U(t,Y)=\exp[2\pi i f (t+\vex_Y\cdot\ven/c)].
809: \eeq
810: Then the quantity $V$ is the real part of the following expression (Eq.(3))
811: \beqa
812: & &
813: \frac12 \lmk\cos2\psi_{AB} A_+ +i\sin 2\psi_{AB} A_\times \rmk \nonumber\\
814: & \times& \lkk (1-\cos
815: \theta_{AB}) U(t,A)+2\cos\theta_{AB} U(t-\tau,B)-(1+\cos\theta_{AB}) U(t-2\tau,A) \rkk.
816: \eeqa
817: The expression in the square bracket of the above result can be written as
818: \beq
819: u(t,A) R(\theta_{AB},f\tau),
820: \eeq
821: where we have defined the factor $R$ as
822: \beq
823: R(\theta_{AB},f\tau) \equiv \lkk (1-\cos\theta_{AB})+2\cos\theta_{AB} \exp\lnk2\pi i f \tau
824: (-\cos\theta_{AB}-1) \rnk -(1+\cos\theta_{AB})\exp(-4\pi i f \tau) \rkk .\label{b11}
825: \eeq
826: It is a simple task to obtain
827: the Fourier transformation of the phase shift $\delta \phi_{AB}$.
828: In a similar manner we can make the expression for
829: the phase shift of
830: another arm {\it e.g.} $\delta \phi_{AC}$. Then we obtain the signals such as $\delta \phi_{AB}-\delta \phi_{AC}$.
831:
832: When gravitational wave-length is much smaller than the arm-length
833: ($f\tau\gg 1$), the factor
834: $R$ shows complicated response that depends strongly on both the angle
835: $\ven$ and the frequency $f$. In the long wave approximation ($f\tau\ll 1$) we
836: can perturbatively expand Eq.(\ref{b11}), and obtain
837: \beq
838: R=4\pi i f \tau \sin^2\theta_{AB}. \label{lla}
839: \eeq
840: Now the factor $R$ becomes very simple.
841:
842: Parameters such as $\mch_z$, $\mu_z$, $\beta$ are closely related to the
843: time evolution of the frequency (chirp signal) due to gravitational
844: radiation reaction as given in Eq.(\ref{pev}). Thus the complicated response of the factor $R$ has
845: weaker impact on estimation of these parameters than the direction of the
846: source $\ven$ (see figure 2). Distance $D$ or the orientation $\vel$ of the
847: binary are
848: determined by using the
849: information of the amplitudes $\cos2\psi_{AB} A_+$ and $\sin2\psi_{AB}
850: A_\times$.
851: As seen in the expressions (\ref{b2})(\ref{b4})(\ref{b5})(\ref{b6})(\ref{b7}), they are strongly correlated to the
852: direction $\ven$ of the binary. Thus their estimation errors are also
853: improved by the effect of the factor $R$.
854:
855: %\if0
856: %%We have used rather
857: %As commented in the main text,
858: % the
859: %parameter estimation errors $\Delta \gamma_i$ given in table 2 of
860: %Ref.\cite{Cutler:1998ta} (obtained with using the angular pattern
861: % function $F^+$ and $F^\times$) agree with our results by the long wave
862: % approximation with eq.(\ref{lla}) for the factor $R$.
863: %\fi
864:
865: %\input{ref.tex}
866:
867: \begin{thebibliography}{DUM}
868:
869:
870:
871:
872: \bibitem{lisahttp}
873: See http://lisa.jpl.nasa.gov
874:
875: %\cite{lisa}
876: \bibitem{lisa}
877: P.~L.~Bender {\it et al.},
878: {\it LISA Pre-Phase A Report,} Second edition, July 1998.
879:
880:
881:
882: %\cite{Haehnelt:wt}
883: \bibitem{Haehnelt:wt}
884: M.~G.~Haehnelt,
885: %``Low Frequency Gravitational Waves From Supermassive Black Holes,''
886: Mon.\ Not.\ Roy.\ Astron.\ Soc.\ {\bf 269}, 199 (1994)
887: %[arXiv:astro-ph/9405032].
888: %%CITATION = ASTRO-PH 9405032;%%
889:
890:
891:
892:
893: %\cite{Larson:1999we}
894: \bibitem{Larson:1999we}
895: S.~L.~Larson, W.~A.~Hiscock and R.~W.~Hellings,
896: %``Sensitivity curves for spaceborne gravitational wave interferometers,''
897: Phys.\ Rev.\ D {\bf 62}, 062001 (2000)
898: %[arXiv:gr-qc/9909080].
899: %%CITATION = GR-QC 9909080;%%
900:
901:
902:
903: %\cite{Schilling:id}
904: \bibitem{Schilling:id}
905: F.~B.~Estabrook and H.~D.~Wahlquist, Gen.Relativ. Gravit. {\bf 6}, 439,
906: (1975); R.~W.~Hellings,
907: {\it Gravitational Radiation (Les Houches 1982)} edited by
908: N. ~Deruelle and T. ~Piran
909: (North-Holland, Amsterdam, 1983), p.485;
910: R.~Schilling,
911: %``Angular And Frequency Response Of Lisa,''
912: Class.\ Quant.\ Grav.\ {\bf 14} (1997) 1513.
913: %%CITATION = CQGRD,14,1513;%%
914:
915:
916:
917: %\cite{Cornish:2002rt}
918: \bibitem{Cornish:2002rt}
919: N.~J.~Cornish and L.~J.~Rubbo,
920: %``The LISA response function,''
921: arXiv:gr-qc/0209011.
922: %%CITATION = GR-QC 0209011;%%
923:
924:
925:
926: %\cite{Cutler:1998ta}
927: \bibitem{Cutler:1998ta}
928: C.~Cutler,
929: %``Angular resolution of the LISA gravitational wave detector,''
930: Phys.\ Rev.\ D {\bf 57}, 7089 (1998).
931: %%CITATION = GR-QC 9703068;%%
932:
933:
934: %\cite{Hughes:2001ya}
935: \bibitem{Hughes:2001ya}
936: S.~A.~Hughes,
937: %``Untangling the merger history of massive black holes with LISA,''
938: Mon.\ Not.\ Roy.\ Astron.\ Soc.\ {\bf 331}, 805 (2001);
939: %%CITATION = ASTRO-PH 0108483;%%
940: A. Vecchio and C. Cutler for in {\it Laser Interferometer
941: Space Antenna}, edited by W.M. Folkner (American Institute
942: of Physics, NY, 1998), p.101.
943:
944: %\cite{Moore:1999zw}
945: \bibitem{Moore:1999zw}
946: T.~A.~Moore and R.~W.~Hellings,
947: %``The angular resolution of space-based gravitational wave detectors,''
948: Phys.\ Rev.\ D {\bf 65}, 062001 (2002)
949: %[arXiv:gr-qc/9910116].
950: %%CITATION = GR-QC 9910116;%%
951:
952: %\cite{Hellings:jm}
953: \bibitem{Hellings:jm}
954: R.~W.~Hellings, in
955: %``Gravitational Wave Detection At Low And Very Low Frequencies,''
956: {\it The detection of gravitational waves}, edited by D. G. Blair (Cambridge, England, 1991), pp.453-475.
957:
958: %\cite{thorne}
959: \bibitem{thorne}
960: K.~S.~Thorne, in
961: {\it 300 Years of Gravitation,} edited by S. ~W. ~Hawking and W. ~Israel
962: (Cambridge, England, 1987), pp.330-458.
963:
964:
965:
966:
967:
968:
969: %\cite{Finn:1992wt}
970: \bibitem{Finn:1992wt}
971: L.~S.~Finn,
972: %``Detection, measurement and gravitational radiation,''
973: Phys.\ Rev.\ D {\bf 46}, 5236 (1992)
974: %[arXiv:gr-qc/9209010].
975: %%CITATION = GR-QC 9209010;%%
976: %\cite{Cutler:1994ys}
977: %\bibitem{Cutler:1994ys}
978: C.~Cutler and E.~E.~Flanagan,
979: %``Gravitational waves from merging compact binaries: How accurately can one extract the binary's parameters from the inspiral wave form?,''
980: Phys.\ Rev.\ D {\bf 49}, 2658 (1994).
981: %%CITATION = GR-QC 9402014;%%
982:
983:
984: %\cite{Hellings:2002si}
985: \bibitem{Hellings:2002si}
986: R.~W.~Hellings and T.~A.~Moore,
987: %``The information content of gravitational wave harmonics in compact binary inspiral,''
988: arXiv:gr-qc/0207102.
989: %%CITATION = GR-QC 0207102;%%
990:
991:
992:
993: %\cite{Cutler:1993tc}
994: \bibitem{Cutler:1993tc}
995: C.~Cutler {\it et al.},
996: %``The Last three minutes: issues in gravitational wave measurements of coalescing compact binaries,''
997: Phys.\ Rev.\ Lett.\ {\bf 70}, 2984 (1993);
998: %%CITATION = ASTRO-PH 9208005;%%
999: %\cite{Blanchet:1995ez}
1000: %\bibitem{Blanchet:1995ez}
1001: L.~Blanchet, {\it et al.},
1002: %``Gravitational radiation damping of compact binary systems to second postNewtonian order,''
1003: Phys.\ Rev.\ Lett.\ {\bf 74}, 3515 (1995);
1004: %%CITATION = GR-QC 9501027;%%
1005:
1006: %\cite{Bender:hs}
1007: \bibitem{Bender:hs}
1008: P.~L.~Bender and D.~Hils,
1009: %``Confusion Noise Level Due To Galactic And Extragalactic Binaries,''
1010: Class.\ Quant.\ Grav.\ {\bf 14}, 1439 (1997).
1011: %%CITATION = CQGRD,14,1439;%%
1012:
1013:
1014:
1015: %\cite{Armstrong:uh}
1016: \bibitem{Armstrong:uh}
1017: %\bibitem{Tinto:yr}
1018: M.~Tinto and J.~W.~Armstrong,
1019: %``Cancellation Of Laser Noise In An Unequal-Arm Interferometer Detector Of Gravitational Radiation,''
1020: Phys.\ Rev.\ D {\bf 59}, 102003 (1999);
1021: %%CITATION = PHRVA,D59,102003;%%
1022: J.~W.~Armstrong, F. B. Estabrook and M.~Tinto, {Astrophys. J.} {\bf
1023: 527}, 814 (1999)
1024: %%CITATION = PHRVA,D58,042002;%%
1025: J.~W.~Armstrong, F.~B.~Estabrook and M.~Tinto,
1026: %``Sensitivities Of Alternate Lisa Configurations,''
1027: Class.\ Quant.\ Grav.\ {\bf 18}, 4059 (2001);
1028: %%CITATION = CQGRD,18,4059;%%
1029: %\bibitem{Hellings:2000cc}
1030: R.~W.~Hellings,
1031: %``Elimination of clock jitter noise in spaceborn laser interferometers,''
1032: Phys.\ Rev.\ D {\bf 64}, 022002 (2001)
1033: [arXiv:gr-qc/0012013].
1034: %%CITATION = GR-QC 0012013;%%
1035:
1036: %\cite{Prince:2002hp}
1037: \bibitem{Prince:2002hp}
1038: .~A.~Prince, M.~Tinto, .~L.~Larson and .~W.~Armstrong,
1039: %``The LISA optimal sensitivity,''
1040: arXiv:gr-qc/0209039.
1041: %%CITATION = GR-QC 0209039;%%
1042:
1043:
1044:
1045: \end{thebibliography}
1046:
1047:
1048:
1049: \begin{figure}[h]
1050: \begin{center}
1051: \epsfxsize=14.cm
1052: \begin{minipage}{\epsfxsize} \epsffile{fig1.eps} \end{minipage}
1053: \end{center}
1054: \caption[]{ Distribution of relative magnitude of the
1055: three dimensional error box $\Delta V$, $SNR$, the angular resolution
1056: $\Delta\ven$, and the distance error $\Delta D/D$. We compare
1057: results from two data streams (with suffix $L$) and those obtained by
1058: past simple method (with suffix $0$). The solid lines are
1059: results for MBH binaries with redshifted masses
1060: $10^6+10^6 \sol$, the doted lines for $10^5+10^5 \sol$, and dashed
1061: lines for $10^4+10^4 \sol$. We have analyzed 300 binaries
1062: with random directions and orientations.}
1063: \end{figure}
1064:
1065:
1066:
1067: \begin{figure}[h]
1068: \begin{center}
1069: \epsfxsize=14.cm
1070: \begin{minipage}{\epsfxsize} \epsffile{fig2.eps} \end{minipage}
1071: \end{center}
1072: \caption[]{ Dependence of the parameter estimation errors
1073: $\Delta\gamma_i$ on
1074: the upper cut-off frequency $f_{cut}\le f_{isco}$. We star integration
1075: of the signal from $f_0=1.9\times 10^{-4}$Hz. The MBH binary has
1076: redshifted
1077: masses $10^5+10^5\sol$, and exists at $z=1$ with
1078: direction
1079: $\theta_s=2.49, \phi_s=0.03$ and orientation $\theta_s=2.32, \phi_s=4.46$.
1080: The thin lines are the past estimations and thick ones are the new estimations. The solid lines represent for $\Delta D/D$, the long dashed lines for $\Delta \vel$, the short-dashed lines for $\Delta \ven$, and the dash-dotted lines for the reduced mass $\Delta \mu_z/\mu_z$. SNR becomes $\sim 1054$ for $f_{cut}=7.0\times 10^{-3}$Hz and $\sim 1140$ for $f_{cut}=2.0\times 10^{-2}$Hz.}
1081: \end{figure}
1082:
1083:
1084: \begin{figure}[h]
1085: \begin{center}
1086: \epsfxsize=14.cm
1087: \begin{minipage}{\epsfxsize} \epsffile{fig3.eps} \end{minipage}
1088: \end{center}
1089: \caption[]{ Same as figure 1, but for the noise canceling combination $X$.}
1090: \end{figure}
1091:
1092:
1093:
1094: \end{document}
1095:
1096: