![Arxiv:2108.11151V1 [Gr-Qc] 25 Aug 2021 Aee Siain 2,27]](http://data.docslib.org/img/28112cb8951fce0f85be6f267c96f34d-1.webp)
arXiv:2108.11151v1 [gr-qc] 25 Aug 2021 aee siain 2,27]. pa- [26, cosmological estimations and rameter [22–25], [17]. observations detections well binary SGWB (MBH) as [18–21], hole bring black 2030s to massive studied the for been merits in has network GWs LISA-TAIJI joint the The observe and length rpsdt eaLS-iemsinwt 3 is a with TAIJI mission detector, LISA-like of a GW stability be space to the proposed maintain another to As plane +60 constellation. ecliptic be the to to designed respect plane is formation constellation The the 16]. of re- [15, vehicle perturbation launch gravity and duction, capabilities, telemetry the ance risteErhb 20 by Earth the trails ytresaerf ih2 with formed milli-Hz is spacecraft the interferometer three in LISA GW The by detect [15]. to band targeting frequency and 2030s the in [14]. to detection yet SGWB found a however, been claim has process, evidence stochastic significant American a statistically no of North evidence re- (the strong has Waves) ported NANOGrav Gravitational for The Ad- on- Observatory actively Nanohertz from are SGWB [8–13]. runs Virgo for going Advanced searches and the LIGO the vanced back- about and information GW Universe, the stochastic early encode may the signals another (SGWB) As ground Ad- source, all [2–7]. targeting and and coalescences important binary O1-O3a, LIGO compact runs Advanced from are observing the Virgo during vanced detected been ∗ † lentv IATIIntok:dtcaiiyt isotro to detectability networks: LISA-TAIJI Alternative e-ioHn [email protected] Han: Wen-Biao [email protected] [email protected], Wang: Gang h pc iso,LS,i ceue ob launched be to scheduled is LISA, mission, space The have events (GW) wave gravitational fifty than More 2 3 aghuIsiuefrAvne td,Uiest fChine of University Study, Advanced for Institute Hangzhou colo srnm n pc cec,Uiest fChines of University Science, Space and Astronomy of School lc oebnre bevtos h AJmcudb recog be could p TAIJIm the LISA. with the Considering observations observations, joint signals. binaries LISA-TAIJIp of hole to parameters black identical specifi the also some determining is for by LISA-TAIJIm network LISA-TAIJIp of the capability tha with shows competitive investigation This is co performanc LISA-TAIJIp. the the and for LISA-TAIJIm examine especially we (SGWB) c background work, wave capable this gravitational most In the observation. as identified binary been has network LISA-TAIJIm ln) AJp(edn h at y20 by Earth the (leading TAIJIp plane), ihLS)wt epc oLS iso taln h at b Earth the (trailing mission LISA to respect with LISA) with IATIIntok,TII laigteErhb 20 by Earth the (leading TAIJIm networks, LISA-TAIJI 1 npeiu ok[] he AJ ria elyet aeb have deployments orbital TAIJI three [1], work previous In hnhiAtooia bevtr,CieeAaeyo Sci of Academy Chinese Observatory, Astronomical Shanghai .INTRODUCTION I. ◦ nahloeti ri obal- to orbit heliocentric a on . 5 × 10 6 msprto and separation km agWang Gang ◦ × nlndwith inclined 10 Dtd uut2,2021) 26, August (Dated: aebackground wave 6 marm km 1, ∗ ◦ n e-ioHan Wen-Biao and n +60 and 1.AdteLS n ieetTIIobtlconfigura- follows. orbital as TAIJI specified different are and tions LISA the And functions reduction [1]. overlap and con- binaries, polarizations, MBH on their straints for localizations investigated sky on and performances networks LISA-TAIJI construct ewr ihteLS detector. joint a LISA construct the to with candidates deploymentnetwork qualified the optimal TAI- be and an would TAIJIm JIp be and considerations, not comprehensive should There- for TAIJIc noises. the correlated similar fore, cause to and subject side, environments be other may net- space TAIJIc the two and On LISA other colocated observation. the the binaries than MBH worse for performance much works the is but LISA-TAIJIc function LISA, of with reduction op- cross-correlation overlap an achieves timal TAIJIc higher The a network. LISA-TAIJIm have than can they worse observation, but binary slightly MBH is for network LISA-TAIJIp LISA-TAIJIm of than detec- detectability two The the between tors. function the reduction shows overlap and lower observations binary MBH best 1. for three the performance demonstrates in LISA-TAIJIm Fig. the results, LISA-TAIJIc), in and investigation LISA-TAIJIp, shown (LISA-TAIJIm, networks previous LISA-TAIJI are the deployments Restating orbital their And ◦ b AJm hc ed h at by Earth the leads which TAIJIm, (b) d AJc hc risteErhby Earth the trails which TAIJIc, (d) a IA hc risteErhby Earth the trails which LISA, (a) c AJp hc ed h at by Earth the leads which TAIJIp, (c) npeiu ok epooe he AJ risto orbits TAIJI three proposed we work, previous In eAaeyo cecs aghu302,China 310124, Hangzhou Sciences, of Academy se nlnd,TII clctdadcoplanar and (colocated TAIJIc inclined), ◦ and cdm fSine,Biig104,China 100049, Beijing Sciences, of Academy e h citcplane. ecliptic the naino omto ln s+60 is plane formation of entation olnrwt LISA. with coplanar − +60 60 prsno w lgbeconfigurations, eligible two of mparison − ie sa pia pint fulfill to option optimal an as nized ftrentok otestochastic the to networks three of e ◦ ◦ romne nSW n massive and SGWB on erformances osprt h GBcomponents SGWB the separate to e rpsdt ops alternative compose to proposed een GBseta hps n the And shapes. spectral SGWB c h eetblt fLISA-TAIJIm of detectability the t 20 y 60 ngrto o asv lc hole black massive for onfiguration nlndorientation. inclined nlndorientation. inclined ne,Saga 000 China 200030, Shanghai ences, ,2 3, 2, 1, ◦ ◦ nlndwt epc oecliptic to respect with inclined n +60 and † i tcatcgravitational stochastic pic ◦ nlnd.Adthe And inclined). ∼ ◦ 20 ihrsetto respect with ∼ ∼ ∼ ◦ n t ori- its and , 20 20 20 ◦ ◦ ◦ iha with , iha with , n is and 2
In this work, we further investigate the detectability II. LISA-TAIJI NETWORKS AND ASSUMED of three LISA-TAIJI networks on the SGWB observation SGWB SPECTRAL SHAPES especially for the comparison of LISA-TAIJIm and LISA- TAIJIp. The SGWB is supposed to be composed of two A. The deployments of LISA and TAIJI kinds of sources: astrophysical origin and cosmological origin. The former one could be yielded by the over- The LISA and TAIJI are proposed space GW missions lapped GWs from abundant unresolved compact binary to be launched in the same epoch, and targeting to de- systems [10, 28–30]. The cosmological sources could be tect the GW in the frequency band 0.1 mHz 0.1 Hz produced by multiple mechanisms in the early Universe [15, 17]. A GW interferometer network formed− by prop- such as time-varying scalar fields, preheating, phase tran- erly deployed detectors will increase the observation ef- sitions, and cosmic strings, etc [31–35]. One of the target- ficiency (for instance, SNR cumulation, sky localization, ing sources for space-borne detectors is the cosmological intrinsic parameter determination, and etc) comparing to origin SGWB which may significantly improve our un- single detector. In previous work [1], we proposed three derstandings of the early Universe and particle physics alternative TAIJI orbits to construct networks with LISA beyond the standard model [15]. as shown in Fig. 1, and the main differences between the To examine the capabilities of the LISA-TAIJI net- networks are the separation between two constellations works, four signal models are assumed to represent the and their orientation of the constellation plane. possible SGWB spectrum. A power-law shape is em- ployed to describe the astrophysical origin SGWB which is yielded by the unresolved BH and neutron star bina- ries [10] (stochastic foreground generated by the galactic binaries is not included). For the cosmological SGWB, three spectral shapes (flat, broken power-law, and sin- gle peaked) are selected to depict the possible SGWBs from processes in the early Universe [31–34]. As the first step, the sensitivity of each networks’ cross-correlation to the SGWB is evaluated. The LISA-TAIJIp shows better sensitivity than LISA-TAIJIm in frequencies lower than 1 mHz because of the stronger correlation between two detectors. The LISA-TAIJIm becomes more sensitive than LISA-TAIJIp in the frequency band [1, 8] mHz. To quantify the performances, the SNRs from each pairs’ cross-correlation are calculated by tuning the parame- ters in the signal models. The SNRs of LISA-TAIJIm are competitive to the LISA-TAIJIp network in the se- lected parameter spaces, and LISA-TAIJIm could surpass LISA-TAIJIp for the assumed fiducial cases. We fur- ther examine the abilities of three networks to determine the SGWB parameters and separate the astrophysical and cosmological components. Both LISA-TAIJIm and LISA-TAIJIp could improve the parameter resolutions by a factor of 1.8 compared to single LISA mission, and LISA-TAIJIc∼ could promote this factor to 2.3-3.2. Therefore, considering the gains for MBH binaries∼ obser- FIG. 1. The diagram of LISA and TAIJI missions’ orbital vation, the TAIJIm could be a better choice than TAIJIp configurations. The upper panel shows the LISA (trailing the to assemble its observations with LISA. Earth by ∼20◦ and +60◦ inclined with respect to the ecliptic plane) and the TAIJIp (leading the Earth by ∼20◦ with an This paper is organized as follows. In Sec. II, we inclination +60◦). The lower panel shows TAIJIm (leading introduce three LISA-TAIJI network configurations and the Earth by ∼20◦ and −60◦ inclined) and TAIJIc (colocated the assumed SGWB spectral shapes. In Sec. III, we and coplanar with LISA). The angle between the LISA and evaluate the sensitivities of each mission and their joint TAIJIp formation planes is ∼34.5◦, and the angle for LISA cross-correlation for SGWB observation, and compare and TAIJIm constellation is ∼71◦. the SNRs obtained from each LISA-TAIJI network for SGWB signals with different parameters. In Sec. IV, by For their performances on MBH binary observations employing Fisher matrix analysis, we examine the capa- as we investigated, the large separation between LISA 8 bility of each network to determine SGWB parameters and TAIJIp/TAIJIm (Dsep 1 10 km) yields sig- from the combined astrophysical and cosmological sig- nificant improvement on source∼ sky× localizations than nals. We recapitulate our conclusions in Sec. V. (We set LISA and LISA-TAIJIc network [1]. Benefiting from the G = c = 1 in this work except otherwise stated). more misaligned formation plane ( 71◦), LISA-TAIJIm ∼ 3 could achieve a higher parameter resolution than LISA- noise are considered to be the dominant noises for GW TAIJIp network (with a 34.5◦ misaligned angle). On observation, and they are included to evaluate the sensi- the other side, subjecting∼ to a larger misaligned forma- tivity of a mission to SGWB. The budges of acceleration tion plane, the overlap reduction function between LISA noise for LISA and TAIJI are treated as same [15, 46], and TAIJIm is lower than LISA-TAIJIp especially for the 2 4 frequencies lower than 1 mHz. Furthermore, the larger fm2/s4 0.4 mHz f S =9 1+ 1+ . separated distance between two detectors lowers down a acc Hz f 8 mHz " # " # critical frequency by fcrit = c/(2Dsep)=1.5 mHz (c is (3) speed of light) as shown in Appendix Fig. 9. To further And their optical path noise budgets are slightly different compare the performance of the LISA-TAIJIm and LISA- as TAIJIp networks, We resume the investigation to explore their detectabilities to isotropic SGWB. Although LISA- pm2 2 mHz 4 Sop,LISA = 100 1+ , (4) TAIJIc network is supposed to be ineligible considering Hz " f # its low improvement on MBH binary observation, our ex- 4 aminations also include it for comparison and integrity. pm2 2 mHz Sop,TAIJI = 64 1+ . (5) The LISA is a nominal 4 years mission and could be Hz " f # extensible to 10 years [15]. In a realistic scenario, the ob- servation would be interrupted by antenna repositioning The power spectrum density (PSD) of noise in a TDI and other operations, and only 75% scientific duty cycle channel Sn,TDI is calculated by using the numerical can be expected [33]. Therefore, only 3 years of data is method, and we refer to our previous works for detailed suppose to be effective in the 4 years observation. In the algorithm [44, 47]. following investigation, observation time Tobs = 3 years is preset for LISA and TAIJI joint observation. C. SGWB models
B. Time-delay interferometry and noise budgets The isotropic SGWB could be characterized as the variation of the energy density in the frequency domain, Time-delay interferometry (TDI) will be employed for 1 dρGW both LISA and TAIJI to suppress the laser frequency ΩGW = , (6) noise and achieve targeting sensitivity. The principle of ρc d ln f the TDI is to combine multiple time-shifted interfero- 2 2 where ρ = 3H0 c is the critical density of the Universe metric links and obtain an equivalent equal path for two c 8πG [48, 49], and H 2.185 10−18 Hz is the Hubble con- interferometric laser beams. The GW response of TDI is 0 ≃ × combined by the response of every single link, and for- stant [50]. The SGWB could be yields by astrophysical mulation has been specified in [36–40]. sources and cosmological mechanisms, and kinds of cos- For a LISA-like with six laser links, three optimal TDI mological SGWB signals are predicated during the pro- channels (A, E, T) could be constructed from three first- cesses of the early Universe [31, 32, 51]. Four SGWB generation Michelson TDI configuration (X, Y, Z), spectral shapes are assumed to perform our investigation as follows [33, 34, 52]. Z X X 2Y+Z X+Y+Z A= − , E= − , T= , (1) 1) Power-law (PL) SGWB signal, √2 √6 √3 f α0 Ω =Ω , (7) where Y and Z are obtained by circulating of the space- PL 0 f craft indexes in the X channel, and the Michelson-X chan- ref nel is where Ω0 is the amplitude of the SGWB energy density, and α is the index of the power law. This X =[ η + η + η + η ] 0 D31D13D21 12 D31D13 21 D31 13 31 power law background is expected to be generated [η21 + 21η12 + 21 12η31 + 21 12 31η13], by abundant unresolved astrophysical BH and NS − D D D D D D (2) binaries. The fiducial signal is selected to be Ω0 = −12 where ij is a time-delay operator, ij η(t)= η(t Lij ), 4.446 10 and α =2/3 at reference frequency D D − × 0 ηji are the Doppler measurement from S/Cj to S/Ci. fref = 1 mHz [10]. By adopting new designs in [41–43], and the expression 2) Flat SGWB signal, of measurement ηij composited by GW signal and noises are referred to our recent work [20, 44]. And the specific Ωflat =Ω1. (8) formulation for GW response are reiterated in Appendix A. These joint optimal channels represent the eventual The flat SGWB is a simply assumed cosmological detectability of a mission [38, 45]. SGWB which permeates all frequency range with By assuming laser frequency noise is sufficiently sup- a constant amplitude [53]. And we choose Ω1 = pressed in TDI, the acceleration noise and optical path 1 10−11 as the fiducial value for this shape. × 4
3) Broken power-law (BPL) SGWB signal, A. The sensitivities of LISA-TAIJI networks
(α2−α1)/∆ f α1 f ∆ The cross-correlation of data streams from two in- ΩBPL =Ω1 1+0.75 . terferometers in Fourier domain could be described as f f ref " ref # [48, 49, 53], (9) The broken power-law SGWB may produced dur- +∞ +∞ C df df ′δ (f f ′)d˜ (f)d˜∗(f ′)Q˜(f ′), (14) ing the first-order phase transition [31, 32, 52, 54, ≃ T − i j and reference therein]. The fiducial parameters are Z−∞ Z−∞ presumed to be α1 = 3, α2 = 4, and ∆ = 2, ˜ −−9 where di =s ˜i +n ˜i is the data in TDI i channel,s ˜i is and the amplitude Ω1 = 1 10 and reference sin(πfTobs) × the SGWB signal,n ˜i the detector noise, δT = , frequency fref = 10 mHz. πf and Q˜ is a filter function. Assuming the noises are not 4) Single peaked (SP) SGWB signal, correlated between two TDI channels from different mis- sions, and there is also no correlation between noises and 2 signal, the average of C will be (log10(f/fref)) ΩSP =Ω1 exp 2 . (10) "− ∆ # C = si,sj + si,nj + ni,sj + ni,nj h i h i h i h i h i = s ,s h i j i A single peaked SGWB signal could be a cosmo- +∞ +∞ logical source and yielded by during particular pro- = df df ′δ (f f ′) d˜ (f)d˜∗(f ′) Q˜(f ′) T − i j cesses in the early universe, and the typical values Z−∞ Z−∞ −11 +∞ D E of parametersareset to be ∆ = 0.2, Ω1 =1 10 , T × = obs dfP (f)γ (f)Q˜(f). and fref = 3 mHz [33, 34]. 2 h ij Z−∞ (15) The energy spectral densities of four SGWB models with The correlated SGWB signal in two TDI channels is im- their respective fiducial values are plotted in the upper plemented in the last row of Eq. (15), panel of Fig. 2. And the conversion from relative energy density to PSD will be 1 d˜ (f)d˜∗(f ′) = δ (f f ′)P (f)γ (f), (16) i j 2 T − h ij 3H2 D E P (f)= 0 Ω (f). (11) h 4π2f 3 GW where γij is the overlap reduction function between LISA TDI i channel and TAIJI TDI j channel, For the LISA-like mission, considering the antenna pat- 1 tern of an interferometer, the PSD of SGWB in a TDI γ (f)= d2nF (f, n)F (f, n), (17) ij 4π i,LISA j,TAIJI channel is expected to be Z where Fi,LISA/TAIJI is the response function from TDI S (f)= P (f) (f), (12) h,TDI h RTDI channel i (i=A/E/T) of LISA/TAIJI as calculated by Eqs. (1) and (A5). The overlap reduction functions for where is the averaged response function of TDI RTDI each LISA-TAIJI pair have been calculated in previous channel in the frequency domain, work [1], and the normalized overlap reduction functions
π of three LISA-TAIJI networks are plotted in Fig. 9. 2π 2 π 1 h n 2 The SNR ρ from two detector cross-correlations are TDI(f)= 2 FTDI(f, ) cos βdψdβdλ, R 4π π | | expected to be [30, 48, 49, 53, and references therein], Z0 Z− 2 Z0 (13) 2 2 h C C and FTDI is calculated by using Eqs. (1) and (A5). ρ2 h i = h i . (18) ≡ N 2 C2 C 2 h i h i − h i And the variance is III. DETECTING SGWB WITH CROSS-CORRELATION BETWEEN LISA AND N 2 = C2 C 2 TAIJI − h i T +∞ (19) = obs dfM(f) Q˜(f) 2, 4 | | In this section, we will investigate and compare the de- Z−∞ tectability of LISA-TAIJI configurations for four kinds of by implementing the power spectral density of noise in SGWB models. Similar to the ground-based GW detec- TDI channel, tors, the sensitivity of the LISA-TAIJI networks to the stochastic signals will depend on the cross-correlation be- 1 n˜ (f)˜n∗(f ′) = δ (f f ′)S (f), (20) tween the interferometers. h i i i 2 T − n 5 where -TAIJIm M =Sn,i(f)Sn,j (f)+ Sn,i(f)Sh,j (f)+ Sh,i(f)Sn,j(f) LISA-TAIJIp LISA-TAIJIc 2 2 + Sh,i(f)Sh,j(f) γij (f)Ph (f). LISA − TAIJI (21) PL (SGWB) An optimal filter Q˜(f) is applied to maximize the SNR, flat (SGWB) BPL (SGWB) SP (SGWB)
Ph(f)γij (f) Q˜(f)= . (22) M(f) Therefore, the SNR for SGWB observation from LISA- TAIJI correlation will be +∞ 2 2 2 γij (f)Ph (f) ρ = T df (23) obs 1 −∞ M(f) 1 1 1 1 1 Z (Hz For a weak-signal (S S ), the SNR could be ap- n,i ≫ h,i proximated as -TAIJIm +∞ 2 2 LISA-TAIJIp γij (f)P (f) ρ2 T df h LISA-TAIJIc obs LISA ≃ −∞ Sn,i(f)Sn,j (f) i,j=A,E,T Z TAIJI X (24) +∞ 2 2 γij (f)Ph (f) = 2Tobs df , Sn,i(f)Sn,j (f) i,j=A,E,T 0 Z
X Ω where Tobs = 3 years is the observation time. To characterize the sensitivity of cross-correlation to SGWB, the equivalent energy density could be evaluated as −1/2 2 3 2 4π f γij (f) 1 1 1 1 1 1 Ωcross(f)= 2 LISA| TAIJI| . 3H0 Sn,i (f) Sn,j (f) i,j=AX,E,T (25) FIG. 2. The assumed SGWB signals and sensitivities (up- The sensitivities of three LISA-TAIJI networks are shown per plane) and power-law integrated sensitivities in 3 years by the solid curves in the upper panel of Fig. 2. As we for ρth = 10 (lower panel) for LISA/TAIJI/LISA-TAIJI. The can expect, the LISA-TAIJIc shows an optimal sensitiv- sensitivity from LISA-TAIJIp cross-correlation is better than ity in the three networks. For the LISA-TAIJIm and LISA-TAIJIm for the frequencies lower than 1 mHz, and LISA-TAIJIp pairs, subjecting to the critical frequency LISA-TAIJIm is more sensitive than LISA-TAIJIp for fre- and instrument noises, their most sensitive frequency is quency band [1, 8] mHz. LISA-TAIJIc could achieve the best sensitivity in three networks. around fcrit =1.5 mHz. The LISA-TAIJIp is more sensi- tive than LISA-TAIJIm for frequencies lower than 1 mHz, while LISA-TAIJIm becomes more sensitive to SGWB in the frequency band [1, 8] mHz. observation time Tobs and SNR threshold ρth, the power- For a single LISA-like∼ mission with three optimal TDI law sensitivity (PLS) is calculated as channels, the equivalent energy density of noise could be −1/2 evaluated as, 2 fmax 2 2αi ρth 4π γij (f/fref) Ωi = df | | , −1 √2T 3H2 f 6SLISA STAIJI 4π2f 3 (f) obs 0 i,j=A,E,T Z0 n,i n,j Ω (f)= i , (26) X mission 2 missionR αi 3H0 Sn,i (f) f i=AX,E,T ΩPLS = max Ωi , for αi [ 8, 8]. αi fref ∈ − where is the averaged response function of TDI i chan- (27) Ri nel. The corresponding sensitivities of LISA and TAIJI By assuming ρth = 10, the ΩPLS for each LISA-TAIJI mission are shown by dashed curves in the upper panel networks are shown in Fig. 2 lower panel, as well as the of Fig. 2. PLS curves for LISA and TAIJI mission. As we can see, To illustrate the detectability of a detector to a power- for a power-law SGWB signal, the LISA-TAIJIp will have αi law SGWB signal, Ωh = Ωi(f/fref) , a power-law in- better sensitivity than LISA-TAIJIm for the frequencies tegrated sensitivity is proposed in [55]. Based on given lower than 1 mHz, and LISA-TAIJIm will be more sen- 6
-law sitive for the frequency band [1, 8] mHz. 1 LISA-TAIJIm LISA-TAIJIp LISA-TAIJIc B. SNR to SGWB
1 For assumed SGWB model listed by Eqs. (7)-(10), the SNR is calculated by implementing Eq. (23) to evalu- ate the detectability of each LISA-TAIJI network. The motivation to use Eq. (23) instead of Eq. (24) is that 1 weak signal approximation becomes inaccurate when the amplitude of energy density of SGWB is large compared to detector noise. To compare the detectability of net- 1 1 works in different SGWB parameter spaces, one or two 1 1 1 11 1 1 1 parameters in Eqs (7)-(10) are selected to be tunable. Ω For a power-law SGWB signal with a fixed power in- dex α0 =2/3, the SNR of three LISA-TAIJI networks is calculated with varying amplitude Ω in Eq. (7), and the 1 0 -TAIJIm results are shown in the upper plot of Fig. 3. The LISA- LISA-TAIJIp TAIJIc network has the best capability in three networks LISA-TAIJIc as expected, and its SNR is almost one order higher than the other two networks. The LISA-TAIJIm shows a bet- 1 ter detectability than LISA-TAIJIp when Ω0 is smaller −10
than 10 , and LISA-TAIJIp will obtain a higher SNR −10 −12 for Ω > 10 . For the fiducial value Ω0 =4.446 10 , 1 the LISA-TAIJIm would be a better option to detect× this astrophysical SGWB. The SNRs of three LISA-TAIJI networks varying with 1 amplitude Ω1 of flat SGWB signal are shown in Eq. (8). 1 The flat signal is actually a special power-law with a 1 1 1 11 1 1 1 power index equal to zero. Similar to the power-law case, Ω1 the SNR from LISA-TAIJIm is higher than LISA-TAIJIp for Ω < 5 10−11, and SNR of LISA-TAIJIp surpass LISA-TAIJIm’s× for Ω > 5 10−11. FIG. 3. The SNRs from three LISA-TAIJI networks vary- × ing with amplitudes Ω0 for power-law (upper panel) and Ω1 For a broken power-law SGWB signal, the amplitude for flat (lower panel) SGWB spectral shapes. Compared Ω1 and reference frequency fref in Eq. (9) are selected to to LISA-TAIJIp configuration, LISA-TAIJIm achieves higher be variables to calculate SNRs of LISA-TAIJI networks, SNR when the amplitude of power-law SGWB Ω0 is smaller −12 −10 and the parameter space are chosen to be Ω1 [10 , than 10 and amplitude of flat SGWB Ω1 is less than 10−9] and f [0.3, 30] mHz. Other parameters∈ are 5 × 10−11. ref ∈ fixed to be their fiducial values, (α1 = 3, α2 = 4, and ∆ = 2). As contour plot shown by upper left− panel of Fig. 4 shown, LISA-TAIJIm could achieve its high- For all three networks, the larger Ω1 and larger ∆ val- est SNR around f 2 mHz at a given Ω . And ues will yield a higher SNR. The SNR from LISA-TAIJIc ref ≃ 1 LISA-TAIJIp has the highest SNR at fref 1 mHz at cross-correlation is one order higher than another two a fixed amplitude as the upper right plot shown.≃ The networks. The SNR ratios of LISA-TAIJIp and LISA- optimal reference frequency moves to 4 mHz for LISA- TAIJIm are shown in the lower right panel of Fig. 5. In TAIJIc network as shown in the lower left panel, and the selected parameter space, the LISA-TAIJIp is more the SNR is significantly larger than the other two net- sensitive than LISA-TAIJIm only in a small region with works. To compare the detectability of LISA-TAIJIm both large Ω1 and large ∆ values, and LISA-TAIJIm and LISA-TAIJIp in the selected parameter space, their could achieve higher SNR than LISA-TAIJIp in most of SNR ratio is made in the lower right panel. The LISA- the parameter space. If the reference frequency fref is the TAIJIm could achieve higher SNR than LISA-TAIJIp in second variable instead of ∆, we could intuitively expect the region fref & 0.9 mHz, while the LISA-TAIJIp be- from Fig. 2 that the LISA-TAIJIm would yield higher comes more sensitive when reference frequency is lower SNR than LISA-TAIJIp for fref < 1 mHz, and LISA- than 0.9 mHz. TAIJIp is more sensitive to single peaked SGWB when −12 −9 The parameter space (Ω1 [10 , 10 ], ∆ reference frequency lower than 1 mHz. [0.005, 1]) in Eq. (10) are explored∈ to evaluate the de-∈ Comparing the results from LISA-TAIJIm and LISA- tectability of three networks for single peaked SGWB TAIJIp for four assumed SGWB signals, the LISA- signal with a fixed reference frequency fref = 3 mHz. TAIJIm achieves higher SNRs than LISA-TAIJIp for the 7
-TAIJIm -TAIJIp 1 1
1.8e+02 1.8e+02
1.0e+02 1 1 1.0e+02 1 1 5.6e+01 102 102 5.6e+01 3.2e+01 1 1
Ω 3.2e+01 SNR Ω SNR 1.8e+01
1.8e+01 1 11 1 11 1.0e+01 1.0e+01
1 1 101 1 1 101 1 1 1 1 f f
-TAIJIc 1 1
1.8e+03 1
1 1.0e+03 1
1
3 1 1 5.6e+02 10 1 1 1 1 3.2e+02
1 1
1.8e+02 1 1 1
Ω SNR Ω 1.0e+02
5.6e+01 102 1 1 11 1 11 3.2e+01 1.8e+01 1.0e+01
1 1 101 1 1 1 1 1 1 f f
FIG. 4. The SNR from LISA and TAIJI cross-correlations for broken power-law SGWB signal with different amplitude Ω1 and reference frequency fref . The SNR contours for three LISA-TAIJI networks are shown in upper and lower left plots. As the SNR ratios are shown in the lower right plot, the correlation between LISA and TAIJIm would yield higher SNR than LISA-TAIJIp for fref & 0.9 mHz, and LISA-TAIJIp configuration becomes more sensitive to broken power-law signals for reference frequency lower than 0.9 mHz. most selected parameter spaces, and a decisive factor is the compositions and retrieving the shapes of the sig- that the LISA-TAIJIm has a better sensitivity at the fre- nals. By using a single LISA mission, Adams and Cornish quencies around 2 mHz which is also the most sensitive [56, 57] developed a method to separate the stochastic band for these two networks. On the other side, the GW background from the instrument noise and Galac- SGWB spectral shapes tend to have stronger signals at tic foreground, Caprini et al. [33] and Flauger et al. [34] a frequency higher than 1 mHz in selected parameter demonstrated the reconstruction of the spectral shapes of spaces. Although the LISA-TAIJIp∼ may have a better SGWB, and Boileau et al. [58] implemented the SGWB detectability for other signal assumption(s), the LISA- spectral separation by using simulated data from LISA TAIJIm will be a better configuration for the power- Mock Data Challenge. In this section, we will exam- law SGWB inferred from Advanced LIGO and Advanced ine the performance of three LISA-TAIJI networks on Virgo runs [10, 11]. SGWB components separation.
IV. DISCRIMINATING SGWB COMPONENTS A. Fisher matrix analysis BY USING LISA-TAIJI NETWORKS The Fisher information matrix (FIM) is employed As introduced in Section II C, the SGWB includes as- to determine the parameters describing SGWB spectral trophysical and cosmological origins. It will be essential shapes [51, 52, 58–60, and references therein]. Two parts to decipher the information in SGWB by discriminating are expected to contribute to the FIM calculation, the 8
-TAIJIm -TAIJIp 0Δ 0Δ
1.8e+02
3.2e+02 3.2e+02 1.8e+02 0Δ 0 0Δ 0 102 102
5.6e+01
Ω SNR Ω SNR 1.0e+02 1.0e+02 0Δ 0Δ 5.6e+01 3.2e+01
3.2e+01 1.8e+01
1.8e+01 1.0e+01 1.0e+01 0Δ 101 0Δ 101 0 0 0 0 0 0 0 0 0 0
-TAIJIc 0Δ 0Δ Δ Δ
5.6e+03
3.2e+03 3 0 10 0 0Δ 0 0Δ 0 1.8e+03
0 0 Ω 1.0e+03 SNR Ω
102 0Δ 5.6e+02 0Δ 3.2e+02 0
1.8e+02 0 0 5.6e+01 1.0e+02 0 0Δ 101 0Δ 0 0 0 0 0 0 0 0 0 0 0
FIG. 5. The SNR from LISA and TAIJI cross-correlation for the single peaked SGWB signals in parameter space Ω1 ∈ [10−12, 10−9] × ∆ ∈ [0.005, 1]. The SNR contours for three LISA-TAIJI networks are shown in upper and lower left plots. The SNR ratios of LISA-TAIJIp and LISA-TAIJIm are shown in the lower right plot, the LISA-TAIJIm achieves higher SNR than LISA-TAIJIp configuration for most selected parameters except for higher Ω1 and ∆ at the upper right corner.
first one is from an individual mission’s three optimal optimal channels [58, 59], channels, and the second part is the cross-correlation be- tween two missions.
fmax ∂Sh,i ∂Sh,i ∂θa ∂θb Fab,self = 2Tobs 2 df 0 [Sn,i + Sh,i] i=AX,E,T Z 2 2 fmax 2 ∂Ωtot,i ∂Ωtot,i 3H0 i ∂θa ∂θb = 2Tobs R df, 2 6 2 4π 0 f [Sn,i + Sh,i] i=AX,E,T Z (28) For LISA or TAIJI mission with six laser links, three where Ωtot is energy density of SGWB combining the orthogonal optimal channels (A, E, and T) could be astrophysical and cosmological components, θa is the formed. And the T channel is expected to be a null data SGWB parameter to be determined, and Sn,i and Sh,i stream and could be utilized to characterize the noise of are the PSD of noise and SGWB in TDI i channel, re- the instruments, and two science data channels, A and E, spectively. could be used to detect SGWB signals [56–58]. In a real- istic case, the T channel will not a fully null data stream The cross-correlations between two missions are also due to the inequality of arms [19, 44, 47, 56]. And the could used to determine the parameters of SGWB. By parameters of SGWB could be inferred from joint three assuming the signal is weak compared to the detector 9 noises, the corresponding FIM could be calculated as [51], When SGWB is described by the broken power-law model, three parameters (Ω1, α1, and α2) are selected fmax 2 ∂Ph ∂Ph γij to be determined, and the partial derivatives of Ω = F = 2T ∂θa ∂θb df tot ab,cross obs LISA TAIJI Ω +Ω to three parameters will be 0 Sn,i Sn,j PL BPL i,j=AX,E,T Z 2 ∂Ωtot ∂Ωtot 2 fmax γ2 ∂Ωtot 3H0 ij ∂θa ∂θb =Ω ln 10, (36) = 2Tobs df, BPL 4π2 f 6SLISASTAIJI ∂ log10 Ω1 i,j=A,E,T 0 n,i n,j X Z ∂Ω f (29) tot =Ω ln (37) ∂α BPL 10 mHz where γij is the overlap reduction function between 1 LISA’s i channel and TAIJI’s j channel as calculated in 1 f ∆ Eq. (17), P is the PSD of SGWB defined in Eq. (11). ln 1+0.75 , h −∆ 10 mHz The FIM of the joint LISA-TAIJI network is obtained by !# summing up the FIM from two missions’ cross-correlation ∂Ω 1 f ∆ tot =Ω ln 1+0.75 . (38) and two individual missions, ∂α BPL ∆ 10 mHz 2 ! TAIJI m Fab,joint = Fab,cross + Fab,self (30) If the cosmological SGWB is a single peaked shape, m=LISA Ωtot = ΩPL + ΩSP, the derivatives of Ωtot to Ω1 and X ∆ are The variance-covariance matrix of the parameters will be ∂Ω ≫ tot −1 −1 ρ 1 −1 =ΩSP ln 10, (39) δθiδθj = Fab,joint + (ρ ) Fab,joint . ∂ log Ω1 h i ij O ≃ ij 10 2 (31) ∂Ωtot 2 [log10(f/3 mHz)] =ΩSP . (40) The standard deviations σi of the parameter i is ∂∆ ∆3
ρ≫1 −1 For each SGWB combination, the FIM from a LISA- σi Fab,joint . (32) ≃ ii TAIJI network is obtained by implementing Eq. (30). r And the distribution of parameters could be made by
B. determining SGWB parameters Fab 1 T p(∆θ~)= det exp δθ~ Fabδθ~ . (41) s 2π −2 The total SGWB signal is composed of astrophysical and cosmological parts, Ωtot = Ωastro + Ωcosmos. The The corner plot for the Ωtot =ΩPL +Ωflat case is shown shape of the astrophysical component is assumed to be a in Fig. 6, and the uncertainties of three parameters power-law shape in Eq. (7), and the cosmological part is from different mission configurations are listed in Table represented by flat, broken power-law, or single peaked I. As Table I showed, the constraints on parameters from shapes as specified in Eqs. (8)-(10). Therefore, three LISA-TAIJIm and LISA-TAIJIp are comparable and bet- combinations, (ΩPL +Ωflat, ΩPL +ΩBPL, ΩPL +ΩSP), ter than the LISA mission by a factor of 1.8. The LISA- are constructed for SGWB observed by LISA-TAIJI net- TAIJIc could further narrow down uncertainties∼ by a fac- works, and the constraints on parameters describing the tor of 2.5 compared to LISA. spectral shapes are investigated. And the fiducial values ∼ are utilized to characterise the shapes of each selected SGWB signal. TABLE I. The 1 − σ uncertainties of parameters of SGWB For the power-law signal, the parameters, Ω0 and α0 model Ωtot = ΩPL + Ωflat from LISA and LISA-TAIJI config- in Eq. (7) are selected to be determined, and the partial urations. derivatives of SGWB with respect to these two parame- parameter LISA LISA-TAIJIm LISA-TAIJIp LISA-TAIJIc ters are, δ log10 Ω0 2.24e-2 1.18e-2 1.16e-2 8.65e-3 δα0 1.74e-2 9.04e-3 8.90e-3 6.71e-3 ∂ΩPL δ log Ω1 1.27e-2 6.77e-3 6.64e-3 4.99e-3 =ΩPL ln 10, (33) 10 ∂ log10 Ω0 ∂ΩPL f =ΩPL ln . (34) The corner plot and uncertainties of parameters for ∂α0 1 mHz Ωtot =ΩPL +ΩBPL scenario are shown in Fig. 7 and Ta- When the flat is assumed to be the cosmological SGWB ble II, respectively. Compared to PL+flat results in Table shape, the partial derivative of Ωtot =ΩPL +Ωflat to the I, the uncertainties of the amplitude of the energy density amplitude of energy density Ω1 will be, Ω0 and Ω1 could be narrowed in the PL+BPL case which means the PL and BPL components could be better sepa- ∂Ωtot =Ωflat ln 10. (35) rated. For performance between mission configurations, ∂ log10 Ω1 the capabilities of the LISA-TAIJIm and LISA-TAIJIp 10
TABLE III. The 1 − σ uncertainties of the parameters of SGWB model Ωtot = ΩPL +ΩSP from LISA and LISA-TAIJIc networks. parameter LISA LISA-TAIJIm LISA-TAIJIp LISA-TAIJIc δ log10 Ω0 7.60e-3 4.10e-3 4.08e-3 3.20e-3 δα0 7.75e-3 4.15e-3 4.13e-3 3.28e-3 δ log10 Ω1 4.00e-3 2.21e-3 2.22e-3 1.58e-3 δ∆ 2.24e-3 1.26e-3 1.27e-3 8.47e-4
is largely contributed by the two missions’ individual ob- servations which correspond to the term Fab,self in Eq. (30). The reason is that comparing the curves in Fig. 2, the sensitivity for LISA or TAIJI obtained from Eq. (26) is much better than the sensitivities for LISA-TAIJIm or LISA-TAIJIp cross-correlation from Eq. (25). There- fore, we can expect the FIM from Fab,self (from Eq. (28)) should contribute more constraints on parameter mea- surements than Fab,cross (from Eq. (29)). For the LISA- TAIJIc network, the FIM from cross-correlation could be FIG. 6. The parameter corner plot of SGWB signal model comparable to a single LISA mission. However, a caveat Ωtot = ΩPL + Ωflat with single LISA (green), LISA-TAIJIm is that a weak signal approximation is applied by using (blue) and LISA-TAIJIc (magenta) configurations. The un- Eq. (29), and the FIM may overestimate the contribu- certainties of parameters from LISA and LISA-TAIJI net- tion of cross-correlation from two detectors for a strong works are shown in Table I. (The result from LISA-TAIJIp SGWB signal. is not shown in the plot because it is highly overlapped with LISA-TAIJIm result). V. CONCLUSIONS are still comparable and could determine the parameters with a better accuracy than LISA by a factor of 1.8. In this work, we investigate the detectability of three The parameter resolutions from the LISA-TAIJIc∼ net- LISA-TAIJI networks to the isotropic SGWB. For the work could be 2.5-3.2 times better than LISA for the colocated and coplanar LISA-TAIJIc network, it can de- selected five parameters.∼ tect the SGWB with an optimal sensitivity. The TAIJIp and TAIJIm are placed at locations 1 108 km from the LISA detector, compared to the LISA-TAIJIm× network, TABLE II. The FIM uncertainties of parameters of SGWB the LISA and TAIJIp will have stronger correlated data model Ωtot = ΩPL + ΩBPL from LISA and LISA-TAIJI net- streams, and their cross-correlation will be more sensitive works. to SGWB at frequencies lower than 1 mHz. However, parameter LISA LISA-TAIJIm LISA-TAIJIp LISA-TAIJIc the LISA-TAIJIm has a better sensitivity∼ than LISA- δ log10 Ω0 7.24e-3 4.13e-3 4.20e-3 2.60e-3 TAIJIp in the frequency band [1, 8] mHz which will be δα0 1.78e-2 9.07e-3 9.08e-3 7.08e-3 decisive to promote its detectability∼ to SGWB signals. In δ log10 Ω1 5.84e-3 3.36e-3 3.38e-3 1.83e-3 the selected parameter spaces for four assumed SGWB δα1 1.93e-2 1.08e-2 1.09e-2 6.40e-3 shapes, the performance from LISA-TAIJIm is competi- δα2 5.04e-2 2.79e-2 2.81e-2 1.76e-2 tive with LISA-TAIJIp configuration or even better than LISA-TAIJIp for selected fiducial cases. The Fig. 8 and Table III show the parameter con- The capabilities of three LISA-TAIJI networks to de- straints from LISA and LISA-TAIJI networks for SGWB termine parameters and separate the SGWB components signals Ωtot =ΩPL+ΩSP. Similar to the previous two sce- are also examined. By combining two kinds of SGWB narios, both LISA-TAIJIm and LISA-TAIJIp networks signals, the LISA-TAIJIc could resolve the parameters could reduce parameter uncertainties by a factor of 1.8 of SGWB signals with better accuracy than LISA by a compared to LISA, and joint observation from LISA∼ and factor of 2.3-3.2. For LISA-TAIJIm and LISA-TAIJIp TAIJIc could promote the factors to 2.3-2.6 for the se- networks,∼ both of them can determine parameters with lected parameters. ∼ a 1.8 times better resolution than LISA. The higher As we can deduce from these results, for a stochastic parameter∼ resolutions would be used to more precisely signal, LISA-TAIJIm or LISA-TAIJIp joint observation reconstruct the shapes of SGWB and discriminate the could determine the parameters with a better accuracy components. The joint observation is also expected to than LISA by a factor of 1.8. And this improvement better discern Galactic foreground and instrument noises ∼ 11
FIG. 7. The parameter corner plot of SGWB signal model Ωtot = ΩPL + ΩBPL with single LISA (green), LISA-TAIJIm (blue) and LISA-TAIJIc (magenta) configurations. The uncertainties of parameters from LISA and LISA-TAIJI networks are shown in Table II. (The result from LISA-TAIJIp is not shown in the plot because it is highly overlapped with LISA-TAIJIm result). from SGWB, and we commit it as our next work. ACKNOWLEDGMENTS
Considering the previous investigations in [1], although LISA-TAIJIc is optimal for the SGWB detection, it would not be an optimal choice to observe the MBH bi- naries compared to the large separated LISA-TAIJIm or This work was supported by NSFC Nos. 12003059 LISA-TAIJIp networks. Moreover, the colocated detec- and 11773059, the Strategic Priority Research Program tors may be subject to the same space environments and of the Chinese Academy of Sciences under Grants No. cause correlated noises. Compared to the LISA-TAIJIp XDA15021102. This work made use of the High Perfor- configuration, the LISA-TAIJIm could achieve a better mance Computing Resource in the Core Facility for Ad- detectability to the MBH binaries, and its detectability vanced Research Computing at Shanghai Astronomical to the SGWB is competitive with LISA-TAIJIp and even Observatory. The calculations in this work are performed better for the specific cases. And the LISA-TAIJIm net- by using the python packages numpy [62] and scipy [63], work could be recognized as an optimal configuration to and the plots are make by utilizing matplotlib [64] and fulfill the joint observation. GetDist [65]. 12
FIG. 8. The parameter corner plot of SGWB signal model Ωtot = ΩPL + ΩSP with single LISA (green), LISA-TAIJIm (blue) and LISA-TAIJIc (magenta) configurations. The uncertainties of parameters from LISA and LISA-TAIJI networks are shown in Table III. (The result from LISA-TAIJIp is not shown in the plot because it is highly overlapped with LISA-TAIJIm result).
Appendix A: Response formulation of laser link to with GW
For a source locating at ecliptic longitude λ and lati- tude θ (in the solar-system barycentric coordinates), the GW propagation vector will be
kˆ = (cos λ cos θ, sin λ cos θ, sin θ). (A1) − The +, polarization tensors of the GW signal combin- ing source’s× inclination angle ι are
1 0 0 2 T 1+cos ι e+ 1 0 1 0 1 , ≡ O · 0− 0 0 · O × 2 (A2) 0 1 0 T e× 1 1 0 0 1 i( cos ι), ≡ O · 0 0 0 · O × − sin λ cos ψ cos λ sin θ sin ψ sin λ sin ψ cos λ sin θ cos ψ cos λ cos θ − − − − 1 = cos λ cos ψ sin λ sin θ sin ψ cos λ sin ψ sin λ sin θ cos ψ sin λ cos θ , (A3) O − cos−θ sin ψ cos−θ cos ψ − sin θ − 13 where ψ is polarization angle. The response to the GW in laser link from S/Ci to j will be
nˆ e nˆ h p ij p ij ˆ ˆ yij (f)= · · exp(2πif(Lij + k pi)) exp(2πifk pj ) , (A4) 2(1 nˆij kˆ) × · − · P − · h i
wheren ˆij is the unit vector from S/Ci to j, Lij is the Appendix B: Overlap reduction function arm length from S/Ci to j, pi is the position of the S/Ci in the solar-system barycentric ecliptic coordinates. The The overlap reduction function is introduced to indi- response of Michelson-X channel to GW will be cate the cross-correlation between two laser interferome- ters [61]. The overlap reduction function for three LISA- TAIJI networks have been calculated in [1], the colo- cated and coplanar LISA-TAIJIc network yields unity overlap reduction function for frequency lower than 10 mHz which should be an optimal configuration for the h h SGWB observation. The overlap reduction function F (f) =( ∆21 + ∆21∆13∆31)y X − 12 change the sign between two detectors’ characteristic fre- h + ( 1 + ∆13∆31)y21 quencies gap [ c = 50 mHz, c = 60 mHz] (c − (A5) 2LTAIJI 2LLISA h + (∆31 ∆31∆12∆21)y13 is the speed of the light). The most misaligned LISA- − TAIJIm yield the worst overlap function in the three + (1 ∆ ∆ )yh , − 12 21 31 networks for frequencies lower than a critical frequency fcrit c/(2Dsep) 1.5 mHz considering the separation, ≃ 8 ≃ Dsep = 1 10 km, between LISA and TAIJIp/TAIJIm [30]. When× the frequency is higher than the critical fre- quency, the overlap reduction functions of both LISA- TAIJIm and LISA-TAIJIp oscillate and quickly approach where ∆ij = exp(2πifLij ). zero.
[1] G. Wang, W.-T. Ni, W.-B. Han, P. Xu, [7] R. Abbott et al. (LIGO Scientific, VIRGO), GWTC-2.1: and Z. Luo, Alternative LISA-TAIJI net- Deep Extended Catalog of Compact Binary Coalescences works, Phys. Rev. D 104, 024012 (2021), Observed by LIGO and Virgo During the First Half of the arXiv:2105.00746 [gr-qc]. Third Observing Run, (2021), arXiv:2108.01045 [gr-qc]. [2] B. P. Abbott et al. (LIGO Scientific, Virgo), Obser- [8] B. P. Abbott et al. (LIGO Scientific, Virgo), Up- vation of Gravitational Waves from a Binary Black per Limits on the Stochastic Gravitational-Wave Hole Merger, Phys. Rev. Lett. 116, 061102 (2016), Background from Advanced LIGO’s First Ob- arXiv:1602.03837 [gr-qc]. serving Run, Phys. Rev. Lett. 118, 121101 (2017), [3] B. P. Abbott et al. (LIGO Scientific, Virgo), GW170817: [Erratum: Phys.Rev.Lett. 119, 029901 (2017)], Observation of Gravitational Waves from a Binary Neu- arXiv:1612.02029 [gr-qc]. tron Star Inspiral, Phys. Rev. Lett. 119, 161101 (2017), [9] B. P. Abbott et al. (LIGO Scientific, Virgo), arXiv:1710.05832 [gr-qc]. Search for Tensor, Vector, and Scalar Polar- [4] B. P. Abbott et al. (LIGO Scientific, Virgo), Binary Black izations in the Stochastic Gravitational-Wave Hole Mergers in the first Advanced LIGO Observing Run, Background, Phys. Rev. Lett. 120, 201102 (2018), Phys. Rev. X 6, 041015 (2016), [Erratum: Phys.Rev.X arXiv:1802.10194 [gr-qc]. 8, 039903 (2018)], arXiv:1606.04856 [gr-qc]. [10] B. P. Abbott et al. (LIGO Scientific, Virgo), [5] B. P. Abbott et al. (LIGO Scientific, Virgo), Search for the isotropic stochastic background GWTC-1: A Gravitational-Wave Transient Cat- using data from Advanced LIGO’s second ob- alog of Compact Binary Mergers Observed by serving run, Phys. Rev. D 100, 061101 (2019), LIGO and Virgo during the First and Second arXiv:1903.02886 [gr-qc]. Observing Runs, Phys. Rev. X 9, 031040 (2019), [11] R. Abbott et al. (LIGO Scientific, Virgo, KAGRA), Up- arXiv:1811.12907 [astro-ph.HE]. per Limits on the Isotropic Gravitational-Wave Back- [6] R. Abbott et al. (LIGO Scientific, Virgo), GWTC-2: ground from Advanced LIGO’s and Advanced Virgo’s Compact Binary Coalescences Observed by LIGO Third Observing Run, (2021), arXiv:2101.12130 [gr-qc]. and Virgo During the First Half of the Third [12] R. Abbott et al. (LIGO Scientific, Virgo, KAGRA), Observing Run, Phys. Rev. X 11, 021053 (2021), Constraints on cosmic strings using data from the arXiv:2010.14527 [gr-qc]. third Advanced LIGO-Virgo observing run, (2021), arXiv:2101.12248 [gr-qc]. 14
[22] H. Omiya and N. Seto, Searching for anoma- lous polarization modes of the stochastic gravitational wave background with LISA and Taiji, Phys. Rev. D 102, 084053 (2020), arXiv:2010.00771 [gr-qc]. [23] N. Seto, Gravitational Wave Background Search -TAIJIm by Correlating Multiple Triangular Detectors in LISA-TAIJIp the mHz Band, Phys.Rev.D 102, 123547 (2020), LISA-TAIJIc arXiv:2010.06877 [gr-qc]. [24] G. Orlando, M. Pieroni, and A. Ricciardone, Measur- ing Parity Violation in the Stochastic Gravitational ( Wave Background with the LISA-Taiji network, (2020), arXiv:2011.07059 [astro-ph.CO]. ( [25] A. R. Pol, S. Mandal, A. Brandenburg, and 1 1 1 1 1 1 T. Kahniashvili, Polarization of gravitational waves from helical MHD turbulent sources, (2021), arXiv:2107.05356 [gr-qc]. [26] L.-F. Wang, S.-J. Jin, J.-F. Zhang, and X. Zhang, FIG. 9. The normalized overlap reduction functions of three Forecast for cosmological parameter estimation with LISA-TAIJI networks tensor polarization. The overlap reduc- gravitational-wave standard sirens from the LISA-Taiji tion functions of LISA-TAIJI network are from previous work network, (2021), arXiv:2101.11882 [gr-qc]. [1], and the normalization is implemented to make γij = 1 [27] R. Wang, W.-H. Ruan, Q. Yang, Z.-K. Guo, R.- when two detectors are coplanar and colocated. The LISA- G. Cai, and B. Hu, Hubble parameter estimation via TAIJIc network has the strongest cross-correlation with each dark sirens with the LISA-Taiji network, (2020), other, and the LISA-TAIJIm network yields the worst cross- arXiv:2010.14732 [astro-ph.CO]. correlation in the three configurations for the isotropic SGWB [28] V. Korol, E. M. Rossi, P. J. Groot, G. Nelemans, S. Too- observation. nen, and A. G. A. Brown, Prospects for detection of de- tached double white dwarf binaries with Gaia, LSST and LISA, Mon. Not. Roy. Astron. Soc. 470, 1894 (2017), arXiv:1703.02555 [astro-ph.HE]. [13] R. Abbott et al. (LIGO Scientific, Virgo, KA- [29] N. Cornish and T. Robson, Galactic bi- GRA), Search for anisotropic gravitational-wave back- nary science with the new LISA de- grounds using data from Advanced LIGO’s and Ad- sign, J. Phys. Conf. Ser. 840, 012024 (2017), vanced Virgo’s first three observing runs, (2021), arXiv:1703.09858 [astro-ph.IM]. arXiv:2103.08520 [gr-qc]. [30] J. D. Romano and N. J. Cornish, Detection meth- [14] Z. Arzoumanian et al. (NANOGrav), The ods for stochastic gravitational-wave backgrounds: NANOGrav 12.5 yr Data Set: Search for an a unified treatment, Living Rev. Rel. 20, 2 (2017), Isotropic Stochastic Gravitational-wave Back- arXiv:1608.06889 [gr-qc]. ground, Astrophys. J. Lett. 905, L34 (2020), [31] C. Caprini et al., Science with the space-based in- arXiv:2009.04496 [astro-ph.HE]. terferometer eLISA. II: Gravitational waves from [15] P. Amaro-Seoane, H. Audley, S. Babak, and et al (LISA cosmological phase transitions, JCAP 04, 001, Team), Laser Interferometer Space Antenna, arXiv e- arXiv:1512.06239 [astro-ph.CO]. prints , arXiv:1702.00786 (2017). [32] C. Caprini et al., Detecting gravitational waves from [16] LISA Study Team, LISA (Laser Interferometer Space cosmological phase transitions with LISA: an update, Antenna): A Cornerstone Mission for the Observation JCAP 03, 024, arXiv:1910.13125 [astro-ph.CO]. of Gravitational Waves, Tech. Rep. 11 (ESA-SCI, 2000) [33] C. Caprini, D. G. Figueroa, R. Flauger, G. Nardini, system and Technology Study Report. M. Peloso, M. Pieroni, A. Ricciardone, and G. Tasinato, [17] W.-R. Hu and Y.-L. Wu, The Taiji Program in Space Reconstructing the spectral shape of a stochastic grav- for gravitational wave physics and the nature of gravity, itational wave background with LISA, JCAP 11, 017, Natl. Sci. Rev. 4, 685 (2017). arXiv:1906.09244 [astro-ph.CO]. [18] W.-H. Ruan, C. Liu, Z.-K. Guo, Y.-L. Wu, and R.-G. Cai, [34] R. Flauger, N. Karnesis, G. Nardini, M. Pieroni, A. Ric- The LISA-Taiji network, Nature Astron. 4, 108 (2020), ciardone, and J. Torrado, Improved reconstruction of arXiv:2002.03603 [gr-qc]. a stochastic gravitational wave background with LISA, [19] G. Wang, W.-T. Ni, W.-B. Han, S.-C. Yang, JCAP 01, 059, arXiv:2009.11845 [astro-ph.CO]. and X.-Y. Zhong, Numerical simulation of sky [35] N. Christensen, Stochastic Gravitational Wave localization for LISA-TAIJI joint observation, Backgrounds, Rept. Prog. Phys. 82, 016903 (2019), Phys. Rev. D 102, 024089 (2020), arXiv:2002.12628. arXiv:1811.08797 [gr-qc]. [20] G. Wang and W.-B. Han, Observing grav- [36] F. B. Estabrook and H. D. Wahlquist, Response of itational wave polarizations with the LISA- Doppler spacecraft tracking to gravitational radiation., TAIJI network, Phys. Rev. D 103, 064021 (2021), General Relativity and Gravitation 6, 439 (1975). arXiv:2101.01991 [gr-qc]. [37] H. Wahlquist, The Doppler response to grav- [21] K. J. Shuman and N. J. Cornish, Massive Black Hole itational waves from a binary star source., Binaries and Where to Find Them with Dual Detector General Relativity and Gravitation 19, 1101 (1987). Networks, (2021), arXiv:2105.02943 [gr-qc]. 15
[38] M. Vallisneri, J. Crowder, and M. Tinto, Sensitivity arXiv:gr-qc/0103075. and parameter-estimation precision for alternate LISA [54] M. Hindmarsh, S. J. Huber, K. Rummukainen, and D. J. configurations, Class. Quant. Grav. 25, 065005 (2008), Weir, Gravitational waves from the sound of a first order arXiv:0710.4369 [gr-qc]. phase transition, Phys. Rev. Lett. 112, 041301 (2014), [39] M. Vallisneri and C. R. Galley, Non-sky-averaged arXiv:1304.2433 [hep-ph]. sensitivity curves for space-based gravitational-wave [55] E. Thrane and J. D. Romano, Sensitiv- observatories, Class. Quant. Grav. 29, 124015 (2012), ity curves for searches for gravitational-wave arXiv:1201.3684 [gr-qc]. backgrounds, Phys. Rev. D 88, 124032 (2013), [40] M. Tinto and M. E. da Silva Alves, LISA Sensitiv- arXiv:1310.5300 [astro-ph.IM]. ities to Gravitational Waves from Relativistic Met- [56] M. R. Adams and N. J. Cornish, Discriminating be- ric Theories of Gravity, Phys. Rev. D 82, 122003 (2010), tween a Stochastic Gravitational Wave Background arXiv:1010.1302 [gr-qc]. and Instrument Noise, Phys. Rev. D 82, 022002 (2010), [41] M. Otto, G. Heinzel, and K. Danzmann, TDI and clock arXiv:1002.1291 [gr-qc]. noise removal for the split interferometry configuration [57] M. R. Adams and N. J. Cornish, Detecting a of LISA, Class. Quant. Grav. 29, 205003 (2012). Stochastic Gravitational Wave Background in [42] M. Otto, Time-Delay Interferometry Simulations for the the presence of a Galactic Foreground and In- Laser Interferometer Space Antenna, Ph.D. thesis, Leib- strument Noise, Phys. Rev. D 89, 022001 (2014), niz U., Hannover (2015). arXiv:1307.4116 [gr-qc]. [43] M. Tinto and O. Hartwig, Time-Delay [58] G. Boileau, N. Christensen, R. Meyer, and N. J. Interferometry and Clock-Noise Cali- Cornish, Spectral separation of the stochastic bration, Phys. Rev. D 98, 042003 (2018), gravitational-wave background for LISA: Ob- arXiv:1807.02594 [gr-qc]. serving both cosmological and astrophysical [44] G. Wang, W.-T. Ni, W.-B. Han, and C.-F. Qiao, Algo- backgrounds, Phys. Rev. D 103, 103529 (2021), rithm for TDI numerical simulation and sensitivity inves- arXiv:2011.05055 [gr-qc]. tigation, (2020), arXiv:2010.15544 [gr-qc]. [59] T. L. Smith and R. Caldwell, LISA for [45] T. A. Prince, M. Tinto, S. L. Larson, Cosmologists: Calculating the Signal-to- and J. W. Armstrong, The LISA optimal Noise Ratio for Stochastic and Determinis- sensitivity, Phys. Rev. D 66, 122002 (2002), tic Sources, Phys. Rev. D 100, 104055 (2019), arXiv:gr-qc/0209039 [gr-qc]. arXiv:1908.00546 [astro-ph.CO]. [46] Z. Luo, Z. Guo, G. Jin, Y. Wu, and W. Hu, [60] A. Saffer and K. Yagi, Parameter Estima- A brief analysis to Taiji: Science and technology, tion for Tests of General Relativity with the Results in Physics 16, 102918 (2020). Astrophysical Stochastic Gravitational Wave [47] G. Wang, W.-T. Ni, and W.-B. Han, Revisiting time Background, Phys. Rev. D 102, 024001 (2020), delay interferometry for unequal-arm LISA and TAIJI, arXiv:2003.11128 [gr-qc]. (2020), arXiv:2008.05812 [gr-qc]. [61] N. Christensen, Measuring the stochastic gravitational [48] B. Allen, The Stochastic gravity wave background: radiation background with laser interferometric anten- Sources and detection, in Les Houches School of Physics: nas, Phys. Rev. D 46, 5250 (1992). Astrophysical Sources of Gravitational Radiation (1996) [62] C. R. Harris, K. J. Millman, S. J. van der Walt, R. Gom- arXiv:gr-qc/9604033. mers, P. Virtanen, D. Cournapeau, E. Wieser, J. Tay- [49] B. Allen and J. D. Romano, Detecting a lor, S. Berg, N. J. Smith, R. Kern, M. Picus, S. Hoyer, stochastic background of gravitational radia- M. H. van Kerkwijk, M. Brett, A. Haldane, J. F. del R´ıo, tion: Signal processing strategies and sensitivities, M. Wiebe, P. Peterson, P. G´erard-Marchant, K. Shep- Phys. Rev. D 59, 102001 (1999), arXiv:gr-qc/9710117. pard, T. Reddy, W. Weckesser, H. Abbasi, C. Gohlke, [50] N. Aghanim et al. (Planck), Planck 2018 and T. E. Oliphant, Array programming with NumPy, results. VI. Cosmological parameters, Nature 585, 357 (2020). Astron. Astrophys. 641, A6 (2020), [Erra- [63] P. Virtanen, R. Gommers, T. E. Oliphant, M. Haber- tum: Astron.Astrophys. 652, C4 (2021)], land, T. Reddy, D. Cournapeau, E. Burovski, P. Pe- arXiv:1807.06209 [astro-ph.CO]. terson, W. Weckesser, J. Bright, S. J. van der Walt, [51] S. Kuroyanagi, T. Chiba, and T. Takahashi, M. Brett, J. Wilson, K. J. Millman, N. Mayorov, A. R. J. Probing the Universe through the Stochastic Nelson, E. Jones, R. Kern, E. Larson, C. J. Carey, I.˙ Po- Gravitational Wave Background, JCAP 11, 038, lat, Y. Feng, E. W. Moore, J. VanderPlas, D. Laxalde, arXiv:1807.00786 [astro-ph.CO]. J. Perktold, R. Cimrman, I. Henriksen, E. A. Quintero, [52] K. Martinovic, P. M. Meyers, M. Sakellari- C. R. Harris, A. M. Archibald, A. H. Ribeiro, F. Pe- adou, and N. Christensen, Simultaneous estima- dregosa, P. van Mulbregt, and SciPy 1.0 Contributors, tion of astrophysical and cosmological stochas- SciPy 1.0: Fundamental Algorithms for Scientific Com- tic gravitational-wave backgrounds with terres- puting in Python, Nature Methods 17, 261 (2020). trial detectors, Phys. Rev. D 103, 043023 (2021), [64] J. D. Hunter, Matplotlib: A 2D Graphics Environment, arXiv:2011.05697 [gr-qc]. Comput. Sci. Eng. 9, 90 (2007). [53] N. J. Cornish and S. L. Larson, Space mis- [65] A. Lewis, GetDist: a Python package for analysing Monte sions to detect the cosmic gravitational wave Carlo samples, (2019), arXiv:1910.13970 [astro-ph.IM]. background, Class. Quant. Grav. 18, 3473 (2001),