Scientific Potential of DECIGO Pathfinder and Testing GR with Space-Borne Gravitational Wave Interferometers
DECIGO Pathfinder (DPF) has an ability to detect gravitational waves from galactic intermediate-mass black hole binaries. If the signal is detected, it would be possible to determine parameters of the binary components. Furthermore, by using future space-borne gravitational wave interferometers, it would be possible to test alternative theories of gravity in the strong field regime. In this review article, we first explain how the detectors like DPF and DECIGO/BBO work and discuss the expected event rates. Then, we review how the observed gravitational waveforms from precessing compact binaries with slightly eccentric orbits can be calculated both in general relativity and in alternative theories of gravity. For the latter, we focus on Brans-Dicke and massive gravity theories. After reviewing these theories, we show the results of the parameter estimation with DPF using the Fisher analysis. We also discuss a possible joint search of DPF and ground-based interferometers. Then, we show the results of testing alternative theories of gravity using future space-borne interferometers. DECIGO/BBO would be able to place 4–5 orders of magnitude stronger constraint on Brans-Dicke theory than the solar system experiment. This is still 1–2 orders of magnitude stronger than the future solar system mission such as ASTROD I. On the other hand, LISA should be able to put 4 orders of magnitude more stringent constraint on the mass of the graviton than the current solar system bound. DPF may be able to place comparable constraint on the massive gravity theories as the solar system bound. We also discuss the prospects of using eLISA and ASTROD-GW in testing alternative theories of gravity. The bounds using eLISA are similar to the LISA ones, but ASTROD-GW performs the best in constraining massive gravity theories among all the gravitational wave detectors considered in this article.
- 1 Introduction
- 2 DECIGO Pathfinder
- 3 DECIGO/BBO
- 4 Gravitational Waves from Compact Binaries
- 5 Observed PN Inspiral Waveforms
- 6 Brans-Dicke Theory
- 7 Massive Gravity Theories
- 8 Observing GWs from IMBHs with DPF
9 Testing Alternative Theories of Gravity with Gravitational Waves
- 9.1 Brans-Dicke
- 9.2 Massive Gravity
- 9.3 Other Theories
- 10 Conclusions
- A Lengthy Formulas
1.1 Gravitational Waves in General
In general relativity (GR), gravitational fluctuations propagate as “ripples” in the spacetime at the speed of light, known as gravitational waves (GWs). Since gravity is weak compared to other fundamental forces, the interaction of GWs with other matters is very faint. This means that GWs can escape from highly dense or optically thick regions that cannot be probed by electromagnetic (EM) waves (e.g. very early universe before the last scattering surface). Therefore, GWs have potentials to open up a novel astronomy and cosmology.
The existence of GWs has been proved indirectly by measuring the orbital decay rate of binary pulsars, caused by the energy loss through gravitational radiation. This can be calculated analytically within the framework of GR, and this matches with the measurements beautifully [1, 2]. However, this indirect GW measurement only detects the radiated energy via GWs and not the perturbation of the spacetime itself. Therefore, the direct detection of GWs has been awaited for a long time to detect this latter effect. Currently, several ground-based detectors that are aiming for the direct detection are shifting from initial phases (first generation) to advanced phases (second-generation) such as adv. LIGO [3, 4], adv. VIRGO [5, 6], GEO-HF [7, 8] and KAGRA (formerly called LCGT) [9, 10]. These are GW interferometers that consist of two arms. When GWs pass through this detector, the length of each arm changes, modifying the interferometric pattern. KAGRA is novel and sometimes called 2.5th-generation detector since it will be cryogenically-cooled and is buried underground to reduce the thermal and seismic noises. These detectors have their best sensitivities at around 100–1000Hz. Basically, the high-frequency parts of these detectors are limited by the shot noises while the low-frequency parts are limited by the radiation pressure and the seismic noises.
Sources and sciences from GW interferometers have been summarized in Refs. [11, 12]. Promising sources for these second-generation detectors are signals from compact binaries (see Secs. 3.4 and 3.5 of Ref.  for a theoretical overview and Refs. [13, 14, 15, 16] for observational results) whose event rates are estimated as /year . (See Ref.  for the review on the electro-magnetic counterparts of GW sources.) Binary signals are divided into three phases, inspiral, merger and ringdown. During the inspiral phase, the binary separation gradually shrinks due to the energy loss via GW radiation. The inspiral waveform has been studied well under the post-Newtonian (PN) formalism [19, 20, 21]. Merger phase is highly non-linear but there has been a great progress thanks to numerical relativity [22, 23, 24]. Ringdown phases are also well-studied using black hole (BH) perturbation method . GWs from a merging compact binary is often called chirp signals since their frequencies get higher and higher as they head towards merger. Since the inspiral waveform modeling has been performed well using PN approach , it can be used to estimate binary parameters  and to probe cosmology as standard sirens . Also, neutron-star (NS) binaries can be used to constrain equation of state (EoS) of NSs [28, 29, 30, 31, 32, 33, 34, 35, 36, 37]. Other interesting sources are burst signals (see Sec. 3.2 of Ref.  for a theoretical review and Refs. [38, 39, 40, 41] for the current observational results) including supernovae , magnetars and cosmic strings [43, 44], and continuous signals (see Sec. 3.3 of Ref.  for a theoretical overview and Refs. [45, 46, 47] for observational results) from e.g. (newly-born) neutron stars. There are also stochastic GW backgrounds (see Sec. 3.6 of Ref.  for a theoretical review and Ref.  for observational results) as potential candidates for GW sources such as the ones from cosmic-strings , pre-big-bang models  and preheatings . GW backgrounds associated with inflation may be detected by ground-based detectors if the spectrum has a sufficient blue tilt. There is also a proposed third-generation project called Einstein Telescope (ET) [52, 53]. The sensitivity is increased roughly by one order of magnitude compared to the second-generation ones and ET has interesting sciences which have been summarized in Ref. .
1.2 Space-Borne Interferometers
Ground-based interferometers have difficulty in detecting GW signals lower than 1–10Hz due to the seismic noise. To overcome this problem, space-borne interferometers have been proposed. Among them, the Laser Interferometer Space Antenna (LISA) has been proposed by ESA and NASA with its optimal sensitivity around GW frequency of 1mHz [55, 56]. Classic LISA consists of three spacecrafts from which laser is emitted to one another. This triangular-shaped interferometer has armlengths of km. Due to this extremely long armlength, LISA is a transponder-type (rather than a Michelson or a Fabry-Perot-type) interferometer. (I.e. Instead of reflecting the incoming laser directly, each spacecraft once receives it and emits the laser with corresponding phase.) The triangular cluster follows the same orbit as the Earth, keeping its position 20 behind the latter. The expected targets are supermassive black hole (SMBH) binaries and white dwarf (WD) binaries. The former detection may help in clarifying the mechanism of how SMBHs are formed. A system consisting of a small compact object orbiting a SMBH is called an extreme mass ratio inspiral (EMRI) , and GWs from this system encodes the information of the structure of the strong-field spacetime [58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76]. Unfortunately, GWs from WD binaries may mask [77, 78] other signals including primordial GW background [79, 80]. As a prototype mission, LISA path finder (LPF) will be launched before LISA .
Although, the classic LISA was planned and developed by NASA and ESA, the collaboration between these 2 agencies terminated in 2011. The new version of LISA (called eLISA) led by ESA alone is now under consideration. The arm-length would be shortened by 5 times compared to the classic one and the number of the arms would be reduced to 2. The satellites are slowly drifting away from Earth to save propellant. The expected operation period is 2-5 yrs. See Refs. [82, 83, 84, 85, 86, 87] for expected sources and sciences with eLISA.
Another space-borne interferometer, the Deci-Hertz Interferometer Gravitational Wave Observatory (DECIGO) has been proposed by Seto, Kawamura and Nakamura .
It consists of four clusters, each having a triangular interferometer.
The major differences between DECIGO and LISA are that the armlengths are shortened to km and the former is a Fabry-Perot-type interferometer whereas the latter is a transponder-type as previously mentioned.
Two out of four clusters are situated on the same site so that the correlation analysis can be performed to detect primordial GW background, while the other two are placed far apart so as to increase the angular resolutions of the sources [90, 91, 92]
(see Fig. 1).
It is most sensitive at around 0.1–1Hz.
Similar interferometer, the Big Bang Observatory (BBO) has been suggested as a follow-on mission to LISA [93, 94].
Since it is a transponder-type interferometer, it can be considered as a mini-LISA with its configuration similar to the one of DECIGO.
(The noise curves of adv. LIGO, ET, LISA and DECIGO/BBO are shown in Fig. 2
Since the WD/WD binary signals has a high cutoff frequency at Hz , the main target of DECIGO/BBO is the primordial GW background (PGWB) [89, 95, 96]. NS/NS foreground signals might mask PGWB instead of WD/WD signals, but it is likely that BBO would have enough sensitivity to subtract sufficient amount [97, 98]. See Refs. [89, 94, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114] for other cosmological and astrophysical scientific potentials of DECIGO and BBO.
As for the prototype mission, DECIGO Pathfinder (DPF) [115, 91, 116] is hoped to be launched in 2016–2017. The main goals are to test the key technologies for the space mission and to carry out observations of GWs and Earth gravity. The first space-borne GW detector, SWIM, has already been launched in 2009 [117, 116]. It is a torsion-bar type space antenna. The ground-based torsion-bar antennae have been used to place upper limits on the energy density of stochastic GWs at 0.1–1Hz [118, 119]. The improved version of the torsion-bar antenna (TOBA) has been proposed by Ando et al. .
DPF has a sensitivity of at –. The main GW sources for DPF are the intermediate mass BH (IMBH) binaries. There are evidences that stellar-mass BHs and SMBHs exist, but there is no unambiguous detection of individual IMBHs. IMBH detection may reveal the formation mechanism of SMBHs. DPF has enough sensitivity to detect GWs from galactic IMBH binaries that might exist at the centers of globular clusters and massive young clusters (see Refs. [120, 121] for reviews on IMBHs and Ref.  for the possibility of IMBH binary formation). For the observation of the Earth gravity, DPF has an ability to perform complementary operation compared to other missions such as CHAMP , GRACE  and GOCE .
ASTROD-GW (Astrodynamical Space Test of Relativity using Optical Devices optimized for GW detection) is another space-based GW interferometer mission . It consists of 3 satellites that are placed at near Lagrange points L3, L4 and L5, respectively, forming a large triangluar interferometer similar to LISA. Since its armlength is 52 times larger than that of LISA, it is more sensitive than LISA at lower frequency.
1.3 Testing Alternative Theories of Gravity using GWs
Among all the sciences that have been mentioned above, there is yet another very interesting and fundamentally important science that can be best probed by using GWs. It is the tests on alternative theories of gravity [127, 128, 129]. In order to solve problems like dark energy , dark matter and inflation [131, 132] within the context of GR, usually we need to introduce unknown matters or fields, but there are possibilities that these problems can be explained naturally by modifying gravitational theories. Also, if the classical gravitational theory is to be realized at the low energy limit of more fundamental theory like superstring theory [133, 134], the classical one does not necessarily reduce to GR due to the additional fields (e.g. dilatons) that couple to gravity.
The tests on alternative theories of gravity have been performed with great accuracies under the solar system experiments and binary pulsar observations [128, 2] and no deviation from GR has been reported so far. The former can probe only the weak field limit of the theory, while the latter can probe the strong field gravity  such as the effacement principle and the scalarization [136, 137, 138] in scalar-tensor theory. However, since the typical velocity of the binary component is , the system is not so dynamical. We are here interested in testing GR in the strong and dynamical field regime. The best way to perform these tests in the near future is to use GWs (especially GWs from compact binaries) since they directly contain gravitational information in the strong-field regions. Also, GWs lead to remarkably high precision GR tests in the following sense. Usually, GWs are so weak that they are buried under the detector noises. In order to dig them out from the noises, we need to perform the matched filtering analysis [139, 26]. This is nothing but taking correlations between the observed signals and the template waveforms so that if the former match with the latter, we can extract the signals out of the noises. Therefore, parameter estimations using this matched filtering technique are sensitive to deviations in GW phase since the correlations are remarkably reduced even if the phases between GWs and the templates are different only slightly. Now, it is often the case that alternative theories of gravity modify the phases of GWs from compact binaries (see Fig. 3). Let us say that we observe a signal at a frequency Hz for observation period yr. Then, the number of GW cycles (the number of phases) that we see is roughly . This shows that we would be able to detect the deviations from GR if they modify the phase at least by ! There are many works calculating how accurately we can probe alternative theories of gravity using GWs. (See e.g. Ref.  and references therein, and see Refs. [141, 142, 143, 144, 140, 145, 146, 147, 148, 149, 150, 151, 152] for model-independent tests. See Refs. [153, 154, 155, 156, 157] for possible tests on GR using ringdown signals.)
Two of the most important characteristics of GR are (i) it has only two tensor gravitational degrees of freedom and (ii) the graviton is massless (in other words, GWs propagate at the speed of light). In this review, we focus on simple extensions of each of the above points by (i) introducing an additional scalar field and (ii) giving a finite mass to graviton.
As for the first point, we consider Brans-Dicke (BD) theory  which is the simplest representative of the scalar-tensor theory . This theory is parameterized by the so-called Brans-Dicke parameter which gives the (inverse) coupling between the additional (massless) scalar field and the matter fields. If we take the decoupling limit of , this theory reduces to GR. The current strongest constraint has been obtained from the solar system experiment by Shapiro time delay measurements using the Cassini satellite  as . There is a future space mission called ASTROD which is expected to be launched in 2021 . It would perform very precise tests of GR in the solar system and the constraint on would be 3 orders of magnitude stronger than the Cassini bound. One of the remarkable points about considering gravitational radiation in this theory is that due to the additional scalar field, there exists scalar dipole radiation [162, 163, 164]! Because of this additional energy release, the binary evolution is modified compared to GR, which further makes a change in GW phase. The orbital decay of the binary pulsar PSR J1738-0333 has put a slightly weaker constraint than the solar system one . There have been several works calculating the possible bounds on using future GW interferometers [166, 167, 168, 169, 170]. (See Refs. [171, 172] for the constraint on massive BD theory, Ref.  for the one on generic scalar-tensor theories and Ref.  for the bound on generic dipolar gravitational radiation.) Unfortunately, these results show that the bounds from future GW observations using adv.LIGO and LISA would not be able to exceed the one from the solar system experiment. In this review, we will show that DECIGO/BBO would be able to place 4 orders of magnitude stronger constraint than the Cassini bound .
For the second point, there exist many theories in which graviton acquires a finite mass [176, 177, 178, 179, 180, 181, 182]. (See e.g. Refs. [183, 184] for recent reviews.) These theories are called massive gravity or massive graviton (MG) theories. From a pioneering work by Fierz and Pauli  in 1939, most of massive gravity theories have suffered from problems like Boulware-Deser ghost modes appearing in the curved background  or violating the Lorentz invariance [178, 179]. Only recently a self-consistent massive gravity is proposed [181, 182]. Independent of these specific massive gravity theories, the gravitational potential is modified to the one of Yukawa type. This means that the effective gravitational constant depends on the distance from the source, which modifies the Kepler’s third law. This has been tested in the solar system and the constraint has been put on the graviton Compton wavelength as cm . As for GWs, the propagation speed is modified from the speed of light which again makes a change in GW phase. GWs can also perform a model-independent tests on the graviton mass. Compared to the solar system experiment, ground-based detectors can only put comparable (or slightly stronger) constraints [187, 188, 189] whereas LISA bounds should be stronger by four orders of magnitude [187, 168, 169, 188, 190, 170, 189, 191, 145, 192, 147, 193]! This is because the deviation in the propagation speed of GWs is larger for lower frequency GWs. DECIGO/BBO would be able to place slightly weaker constraint than LISA  and DPF may be able to place comparable constraint to the solar system bound .
In this review, we will first explain how accurately DPF would be able to determine binary parameters by detecting GWs from galactic IMBH binaries. Then, we will describe the ability of testing GR using space-borne GW interferometers, especially DECIGO/BBO and DPF. This article is organized as follows. In Secs. 2 and 3, we explain the basic concepts, noise curves and event rates of DPF and DECIGO/BBO, respectively. In Sec. 4, we derive GWs from compact binaries, starting from the quadrupole, leading contribution for the binary with a circular orbit, and extending it to include higher PN terms and a slightly eccentric orbit. Then, in Sec. 5, we explain how to construct the observed gravitational waveforms for the spin-aligned and precessing binaries. We also introduce the inspiral-merger-ringdown hybrid waveform. In Secs. 6 and 7, we consider BD and MG theories, respectively. We explain how the GW phase is modified from GR, and review current constraints. In Sec. 8, we show the results of the binary parameter estimation using DPF, and in Sec. 9, we show the proposed constraints on BD and MG theories (and also on other theories) using future GW interferometers. Finally, we conclude in Sec. 10.
2 DECIGO Pathfinder
2.1 Basic Designs
DPF [115, 91] is a prototype mission of DECIGO to test the advanced technologies of GW space mission such as (i) a precise position measuring system with Fabry-Perot (FP) cavity, (ii) a highly stabilized laser source and (iii) the drag-free control system which shields external forces due to solar radiation and residual atmosphere. The weight of the satellite is about 350kg and it will be orbiting the earth at the Keplerian velocity with an altitude of 500km. The expected observation period is 1 year. It consists of 1-arm interferometer with armlength 30cm and the laser power of 100mW. FP cavity has not been tested in space up to now and DPF is expected to have better sensitivity than LPF which uses a Mach-Zender interferometer . DPF provides new possibilities for a precise measurement of position and high-stabilized laser in space.
Each mirror is placed inside a module called housing. The relative positions of these mirrors and the frame will be measured by the local sensor and will be fed back to the satellite position using thrusters. LPF will operate at the Lagrange 1 (L1) point where the gravitational environment is stable, while DPF will demonstrate it in an earth orbit. This will open a new window for future space missions.
DPF GW observations aim at the frequency of 0.1–1Hz which is important because the ground-based interferometers and LPF are not sensitive in this frequency range. Also DPF data are expected to be more complicated than the one from ground-based interferometers due to the satellite orbital motion and the effects of the earth, and hence the development of data analysis technique for DPF has a significant meaning.
2.2 Noise Spectrum and Observable Range of DPF
The detected signal can be expressed as the combination of the GW signal and the noise . If the noise is stationary, the noise spectral density can be defined as
where represents the Fourier component of the noise and the angle brackets denote the expectation value. The root noise spectral density of DPF is given as
where the first term corresponds to the acceleration noises while the second term represents the laser frequency noise. It is shown in Fig. 4 as the (red) thick solid curve. We set the lower frequency cutoff at Hz due to the Earth gravity. It may be possible to improve the sensitivity in which the higher frequency part is now limited by the shot noise. We call this the “adv. DPF” whose noise spectral density is given by 
It is shown in Fig. 4 as the (green) thin solid curve.
We also show the lower cutoff frequencies of the adv. LIGO (magenta thick dotted) and KAGRA (blue thin dotted)
In Fig. 5, we show the (sky-averaged) observable range of DPF and adv. DPF as the (red) thick solid curve and the (green) thin solid curve, respectively. We also show the ones of adv. LIGO and KAGRA with the same curves as in Fig. 4. The (black) dashed horizontal line at kpc corresponds to the galactic center. We also plot the possible GW sources at the centers of the globular clusters shown in Table 2.2, assuming that they consist of equal-mass IMBH binaries.
2.3 Beam Pattern Functions and the Effect of Detector Motion
The beam pattern functions of 1-armed interferometer are given by 
Here, is the angle between the arm and the incoming GW, and represents the polarization angle (see Fig. 6). The sky-averaged values of them are . In Fig. 4, we also show the sky-averaged GW amplitudes of a equal-mass IMBH binary in Centauri (light blue thick dotted-dashed) and a equal-mass one in NGC 6752 (black thin dotted-dashed). We assume the dimensionless spin parameter of these BHs to be whose value is predicted to be the most probable  (see also Ref. ). We use the phenomenological inspiral-merger-ringdown hybrid waveforms [208, 209] (see Sec. 5.3) to estimate these amplitudes. One can see that GW frequency of IMBH binary in Centauri is too low for the ground-based interferometers, and hence its GW signals become unique sources for DPF. On the other hand, the one in NGC 6752 can be detected with both DPF and the ground-based ones. This suggests that it may be possible to perform joint searches between these detectors , which we explain in Sec. 8.1.
For later use, we introduce the celestial coordinates shown in Fig. 7. -axis points the vernal equinox, while -axis points the north celestial pole and is orthogonal to the celestial plane. The source direction and the polarization angle can be re-expressed in terms of the source direction and the direction of the orbital angular momentum as 
2.4 Event Rate
In this subsection, we estimate the event rate for (I) Equal-mass BH binaries and (II) IMRIs using DPF.
Equal-mass BH Binaries
DPF can detect GW signals from galactic IMBH binaries. Two of the candidates where IMBH binaries might exist in our Galaxy are at the centers of globular clusters (GCs) and galactic massive young clusters (GMYCs).
For the former case, we select 21 globular clusters out of 150 globular clusters that have been found. We determine the total mass of the possible IMBH binary by applying the formula obtained by Tremaine et al.  as
We assume the mass ratio of IMBH binary to be 1. The results are listed in Table 2 of Ref. . We plot IMBH binary of each globular cluster in Fig. 5 and count how many of them lie within the observable range of DPF and Adv. DPF. One sees that 2 out of 26 globular clusters (5 shown in Table 2.2 + 21 shown in Table 2 of Ref. ) might contain IMBH binaries detectable with DPF on average. Then, the detection rate of the IMBH binaries in globular clusters is given by
Here, following Ref. , we divide the number of globular clusters by the age of the universe. This is because only one IMBH binary can be formed over its lifetime in each cluster. When we use the adv. DPF, 13 out of 26 IMBH binaries in the galactic globular clusters lie withing the observable reach of the detector. This makes the event rate larger than Eq. (9) by a factor .
For the latter case, currently, more than 10 galactic massive young clusters (GMYCs) have been discovered . Grkan et al.  performed numerical simulations and found that two IMBHs may form in GMYCs if the initial binary fraction is relatively large. After IMBH binary formation, it shrinks due to the dynamical friction with the cluster stars. The timescale of the dynamical encounters would be Gyr . For simplicity, we assume that IMBH binaries are all located at the galactic center. From Fig. 5, one sees that DPF is not sensitive enough to detect GW signals from IMBH binaries in GMYCs. On the other hand, when we use adv. DPF, it has ability to detect all of the IMBH binaries in GMYCs. Following Fregeau et al. , we assume that the number of star clusters massive enough to form IMBH binaries is the same as the one of globular clusters, and 10 of them actually produce IMBH binaries. In our galaxy, about 150 globular clusters  have been observed. This means that at least 15 IMBH binaries are expected to be within the reach of adv. DPF. The detection rate of the IMBH binaries in GMYCs with adv. DPF can be roughly estimated as .
After supernova explosions, many stellar-mass BHs of would form. These would sink to the cluster cores due to the mass segregation, and their number density is expected to be comparable to the one of the main sequence stars, typically pc. In this subsection, we consider IMBH binaries with intermediate-mass ratio inspirals (IMRIs). In Fig. 8, we show observable ranges of DPF (red thick solid) and adv. DPF (green thin solid) with . One sees that DPF is not sensitive enough to detect any GWs from IMRIs with , but adv. DPF has ability to detect these signals from some of GCs and the galactic center.
For GMYCs, from Fig. 8, one sees that even adv. DPF is not sensitive, on average, to an IMRI signal of at the galactic center.
We found that 1.7 of these binaries have SNRs greater than 5 due to the optimal orientations .
On the other hand, the merger rate due to three-body interaction per cluster becomes 
For GCs, we found that the merger rate is limited by the requirement that the merger rate in a single cluster cannot exceed [206, 194], where is the current age of the cluster. Otherwise, the mass of the larger BH has been increased considerably. From Fig. 8, one sees that IMRI signals from about 3 out of 26 globular clusters might be detected by adv. DPF. This leads to the estimate that it can detect IMRI signals from about globular clusters in total, and hence the detection rate of globular clusters in total becomes
3.1 Basic Designs
DECIGO is first proposed by Seto, Kawamura and Nakamura . It consists of four triangular sets of detectors whose configuration is shown in Fig. 1. This effectively corresponds to eight individual interferometers. Each triangular detector has an arm-length of km. Its primary goal is to detect PGWB with the GW energy density of . Since the WD/WD confusion noise will have a cutoff frequency at around 0.2 Hz , DECIGO has an advantage on detecting this source over LISA (see Refs. [97, 98, 214] for the discussions of the NS/NS confusion noise). Therefore two of the four triangular detectors are located on the same site forming a star of David so that correlation analysis [215, 216] can be performed to detect PGWB. The rest of the two detectors are placed far apart to increase the angular resolutions of the source locations.
BBO has almost the same constellation as DECIGO. The main difference is that while DECIGO is a Fabry-Perot type interferometer, BBO is a transponder-type interferometer. BBO has arm-lengths of km.
3.2 Noise Spectrum
It has 20/3 times better sensitivity than the one for the sky-averaged sensitivity . Apart from instrumental noise, there are astrophysical foreground confusion noises. These confusion noise spectral densities and the energy densities of gravitational waves are related as 
Here, is the critical energy density of the Universe and
is the energy density of GWs per log frequency normalized by . The energy density of GWs that originate from extra-galactic WD binaries has been estimated as which leads to the noise spectral density of 
On the other hand, the one from galactic WD binaries has been calculated as 
We multiply and by a factor , which corresponds to the high frequency cutoff for the white dwarf confusion noises. We also have to take into account the confusion noise from NS binaries. Its noise spectral density is estimated as [97, 214]
where denotes current merger rate density of NS/NS binaries. Putting altogether, the total noise spectral density for BBO becomes
Here is the number density of galactic white dwarf binaries per unit frequency, which is given by 
is the average number of frequency bins that are lost when each galactic binary is fitted out. The factor in front of represents our assumption of the fraction of GWs that cannot be removed after foreground subtraction. In this review, we assume that NS/NS foregrounds can be subtracted down to the level below the instrumental sensitivity. The lower and higher frequency ends of the BBO sensitivity band are set as Hz and Hz, respectively. The noise spectrum of BBO is shown as a red thick solid curve in Fig. 2.
DECIGO has been proposed with 3-4 times less sensitive spectrum than BBO. Its instrumental noise spectrum is given by
where with Hz. However, this is not the fixed design sensitivity and there is a project going on to improve the sensitivity to the same level as BBO. The frequency range of DECIGO/BBO is given as
3.3 Event Rate
The promising sources for DECIGO/BBO are inspirals of NS binaries. In this subsection, following Cutler and Harms , we derive the detection rate of these sources. Here, we adopt the model where NS/NS binaries only exist at redshift below . (This is inferred from observations. See the discussion below.) Since DECIGO/BBO has enough sensitivity to detect the farthest NS/NS binaries considered here, the merger rate corresponds to the detection rate. The NS/NS merger rate can be written as
Here, represents the Hubble parameter at redshift . For -CDM cosmology, it is given by
is the merger rate of NS/NS binaries at redshift and it can be re-expressed in terms of the current merger rate and the redshift dependence as . For , we adopt the following piecewise linear fit [97, 218] based on observations discussed in Ref. ;
Then, we finally obtain the detection rate as
The detection rate for BH/NS binaries are expected to be roughly 1/10 of NS/NS detection rate.
4 Gravitational Waves from Compact Binaries
In this section, we focus on the GWs from inspiral binaries composed of compact stars such as BHs and NSs within the context of GR. In Sec. 4.1, we first derive the GWs from a circular binary at the leading order. Then, In Sec. 4.2, we extend the result to higher order terms. Finally, in Sec. 4.3, we consider binaries with small eccentricity.
4.1 Quadrupole Gravitational Radiation from Binaries with Circular Orbit
In this section, we derive the leading order gravitational waves from compact binaries, namely the quadrupole radiation. The binary circularizes quickly during its orbital decay due to the gravitational radiation reaction, and hence we restrict our attention to the binary with circular orbit in this section. Also, we assume that the velocity of each body is sufficiently small compared to the speed of light.
Let us assume that we have two compact objects with masses and in the - plane with orbital separation at a distance from the detector. We define the total and the reduced mass as and , respectively. The binary components obey the Kepler’s Law at the leading order with orbital angular velocity defined as
As shown in Fig. 9, we denote the positions of these objects in the center of mass frame as and , which can be expressed as
By taking the time derivative of the quadrupole moment twice, we obtain
Here, we have defined
where is the unit orbital angular momentum vector and is the unit vector pointing towards the direction of GW propagation. From this result, we see that the monochromatic GW frequency can be written as at the leading order.
Next, we take the effect of radiation reaction into account and estimate the GW phase . First, the total energy of the binary system is given as
Next, by using the quadrupole formula
we have the radiated energy due to GWs as
By using the balancing equation
the evolution of the binary separation can be expressed as
which can be solved to yield
where is the coalescence time. This can be turned into the frequency evolution as
where is the chirp mass. This equation can be solved as
By integrating this by once, we get the phase as
with representing the coalescence phase.
Next, we derive the waveform in the Fourier domain defined as
Since we have assumed that the velocity of each binary component is much smaller than the speed of light, the amplitude and the phase satisfy the following conditions: and . Under this situation, we can apply the stationary phase approximation, where the Fourier component of a function becomes 
Therefore, we obtain
where the amplitude and phases are given as
respectively. Here, is defined as
and is the luminosity distance given as
4.2 Higher Post-Newtonian Corrections
In the previous section, we derived the waveform at the leading order, where we used the quadrupole formula for the radiated energy flux and assumed that the binary orbit is of Newtonian. In this section, we extend the previous analysis to higher orders. Since we go beyond Newtonian, this expansion is called the post-Newtonian (PN). Our expansion parameter is where is the typical velocity of the binary component, and the terms that are proportional to are of PN orders. In this section, we consider the extension up to 2PN order .
Furthermore, we neglect the PN corrections to the amplitude. This is because when we take the correlation between two different waveforms, it is more sensitive to the deviation in the phase than the one in the amplitude. This type of waveform is called the restricted PN waveform.
In Ref. , expressions for , and are shown to 2PN order. By using them, up to this PN order, becomes
where is the symmetric mass ratio with PN expansion parameter . The spin-orbit and the spin-spin couplings are given as
where the dimensionless spin parameter is defined as with denoting the spin angular momentum of the -th compact object. The gravitational wave phase in the Fourier domain can be calculated as
4.3 Gravitational Waves from Binaries with Slightly Eccentric Orbit
In this section, we consider gravitational waves from a binary that have a small eccentricity . We can express the eccentric orbit as
where the orbital angular velocity can be written as
By using the quadrupole formula, the radiated energy flux can be estimated as
and similarly, the radiated angular momentum flux is calculated as
Next, we derive the evolution of the semi-major axis and the eccentricity . Since the orbital energy of the binary is given in Eq. (36) and the angular momentum is given as
From these equations, we obtain
Therefore, by integrating once, we obtain the relation between and as
where and denote the initial semi-major axis and eccentricity, respectively.
where is the frequency at . Especially, when , we have
so that up to , the asymptotic eccentricity invariant,
The evolution the frequency is given by
for . By substituting Eq. (72) into the above equation, we have