Abstract
The Chinese Space Station Optical Survey (CSS-OS) is a planned full sky survey operated by the Chinese Space Station Telescope (CSST). It can simultaneously perform the photometric imaging and spectroscopic slitless surveys, and will probe weak and strong gravitational lensing, galaxy clustering, individual galaxies and galaxy clusters, active galactic nucleus, and so on. It aims to explore the properties of dark matter and dark energy and other important cosmological problems. In this work, we focus on two main CSS-OS scientific goals, i.e., the weak gravitational lensing (WL) and galaxy clustering surveys. We generate the mock CSS-OS data based on the observational COSMOS and zCOSMOS catalogs. We investigate the constraints on the cosmological parameters from the CSS-OS using the Markov Chain Monte Carlo method. The intrinsic alignments, galaxy bias, velocity dispersion, and systematics from instrumental effects in the CSST WL and galaxy clustering surveys are also included, and their impacts on the constraint results are discussed. We find that the CSS-OS can improve the constraints on the cosmological parameters by a factor of a few (even one order of magnitude in the optimistic case), compared to the current WL and galaxy clustering surveys. The constraints can be further enhanced when performing joint analysis with the WL, galaxy clustering, and galaxy–galaxy lensing data. Therefore, the CSS-OS is expected to be a powerful survey for exploring the universe. Since some assumptions may be still optimistic and simple, it is possible that the results from the real survey could be worse. We will study these issues in detail with the help of simulations in the future.
Export citation and abstract BibTeX RIS
1. Introduction
Understanding the nature of dark matter and dark energy, and the formation and evolution of the cosmic large-scale structure (LSS), is essential for the study of cosmology. A number of powerful observational tools, such as weak gravitational lensing (WL; e.g., Kaiser 1992, 1998), baryon acoustic oscillations (e.g., Eisenstein 2005; Eisenstein et al. 2005), and redshift-space distortion (RSD; e.g., Jackson 1972; Kaiser 1987), have been applied for solving these issues. A few Stage IV ground- and space-borne telescopes, e.g., the Large Synoptic Survey Telescope (LSST;11 Ivezic et al. 2008; Abell et al. 2009), Euclid space telescope12 (Laureijs et al. 2011), and Wide Field Infrared Survey Telescope,13 have been planned to perform these measurements. These powerful surveys are expected to make great improvements on related scientific objectives. The Chinese Space Station Optical Survey (CSS-OS) is another of this kind of sky survey.
The CSS-OS is a major science project established by the space application system of the China Manned Space Program, which will be performed by the Chinese Space Station Telescope (CSST; Zhan 2011, 2018; Cao et al. 2018). The CSST is a 2 m space telescope in the same orbit as the China Manned Space Station, and is planned to be launched at the end of 2022. The CSS-OS will cover 17,500 deg2 sky area in about 10 years with field of view (FOV) 1.1 deg2. It will simultaneously perform both photometric imaging and slitless grating spectroscopic surveys with high spatial resolution ∼015 (80% energy concentration region) and wide wavelength coverage. There are seven photometric imaging bands and three spectroscopic bands covering 255–1000 nm (see Table 1 and Figure 1). A few important cosmological and astronomical objectives will be explored by the CSS-OS, such as the properties of dark matter and dark energy, cosmic LSS, galaxy formation and evolution, galaxy clusters, active galactic nucleus (AGNs), etc. Comparing to other next generation surveys, the CSS-OS has several advantages, such as wider wavelength coverage, larger number of filters, smaller spatial resolution, better image quality, and simultaneous photo+spec performance. Therefore, the CSS-OS is expected to be a powerful survey for probing the universe, which is comparable and even more robust in some aspects than other Stage IV surveys (Zhan 2011).
Table 1. Designed and Assumed Parameters of the CSST Photometric and Spectroscopic Surveys
Survey Characteristics | |||||||
Telescope | 2 m primary, off-axis TMA, orbit ∼400 km, duration ∼10 yr | ||||||
Survey mode | Photometric imaging + slitless spectroscopic joint survey | ||||||
Wide survey | 15,000 deg2 ( 20°) + 2500 deg2 ( 20°) | ||||||
Deep survey | 400 deg2 (selected areas) | ||||||
Field of view | ≳1.1 deg2 | ||||||
Photometric wide survey | |||||||
Bands | NUV | u | g | r | i | za | ya |
Wavelength ( nm) | 255–317 | 322–396 | 403–545 | 554–684 | 695–833 | 846–1000 | 937–1000 |
Exposure time | 150 s ×4 | 150 s ×2 | 150 s ×2 | 150 s ×2 | 150 s ×2 | 150 s ×2 | 150 s ×4 |
Sensitivity (point/extendedb) | 25.4/24.2 | 25.4/24.2 | 26.3/25.1 | 26.0/24.8 | 25.9/24.6 | 25.2/24.1 | 24.4/23.2 |
PSF size (≤REE80/FWHMc) | 015/020 | 015/020 | 015/020 | 015/020 | 016/021 | 018/024 | 018/024 |
Spectroscopic wide survey (gratings) | |||||||
Bands | GU | GV | GIa | ||||
Wavelength (λ−90 − λ+90 nm) | 255–420 | 400–650 | 620–1000 | ||||
Exposure time | 150 s ×4 | 150 s ×4 | 150 s ×4 | ||||
Sensitivityd (mag/(erg s−1 cm−2)) | 20.5/1.0 × 10−15 | 21.0/4.0 × 10−16 | 21.0/2.6 × 10−16 | ||||
Spectral resolution Re | ≥200 | ≥200 | ≥200 |
Notes.
aCut off by detector quantum efficiency. b5σ AB mag, and assuming galaxies with 03 half-light radius for extended sources. In the CSST deep survey, the sensitivity can be 1 mag deeper at least than the wide survey. The corresponding SNR cut is ∼5 for the i band. The magnitude limits shown here are the updates of the values given in Cao et al. (2018). cAssuming Gaussian-like PSF profile, and REE80 is the radius of 80% energy concentration. This PSF size indicates that the peak of galaxy size distribution is ∼02, and it has a cut around ∼013. d5σ AB mag per resolution element for point sources. In the CSST deep survey, the sensitivity can be 1 mag deeper at least than the wide survey. Note that because the region of emitting lines in an emission line galaxy (ELG) is usually small compared to the full size of galaxy, we treat the ELGs as point sources here for simplicity. eR = λ/Δλ.Download table as: ASCIITypeset image
In this work, we predict the measurements of weak lensing and galaxy clustering for the CSS-OS, and investigate the constraint accuracy of the cosmological parameters. We make use of two catalogs from the real observations, i.e., COSMOS and zCOSMOS surveys, as the mock catalogs for the CSST photometric and spectroscopic surveys, respectively. These two catalogs have similar magnitude limits as the CSS-OS, and can well represent the CSST observations. For the weak lensing survey, we derive the galaxy redshift distribution from the mock catalog, and divide it into several photometric redshift (photo-z) bins to calculate the auto and cross convergence power spectra. The galaxy intrinsic alignments and systematics (multiplicative and additive) due to point-spread function, photometry offsets, instrumental noise, etc., are included when estimating the errors. We also evaluate the redshift-space galaxy clustering power spectra in the CSST spectroscopic survey with slitless gratings. The multipole power spectra are calculated in the spectroscopic redshift (spec-z) bins, and the effects of frequency resolution, galaxy bias, velocity dispersion, and systematic errors are considered in the error estimate. Then we compute the observed CSS-OS angular cross power spectra of the weak lensing and galaxy clustering, i.e., galaxy–galaxy lensing power spectra. The mock data of weak lensing, galaxy clustering, and their cross correlation are used in the constraints on the cosmological parameters. The Markov Chain Monte Carlo (MCMC) technique is adopted to illustrate the probability distributions of the parameters.
The paper is organized as follows: the weak lensing, galaxy clustering, and galaxy–galaxy lensing power spectra surveys are discussed in Sections 2–4, respectively. In Section 5, we show the details of the fitting process using the MCMC method. The constraint results are shown in Section 6. We finally summarize the conclusions in Section 7. Throughout the paper, we assume the flat ΛCDM cosmology with Ωm = 0.3, Ωb = 0.05, σ8 = 0.8, ns = 0.96, and h = 0.7 as the fiducial model.
2. Weak Lensing Survey
The CSS-OS covers 17,500 deg2 survey area in the photometric imaging survey with survey depth i ≃ 26 AB magnitude (5σ detection for point sources; Cao et al. 2018). It has seven filters, i.e., NUV, u, g, r, i, z, and y bands, which covers the wavelength range 255–1000 nm (see Figure 1). The point-spread function (PSF) of the CSS-OS is designed to be as small as 015 within 80% energy concentration, and it requires the maximum PSF ellipticity to be less than 0.15 at any position of the FoV, and the average value is less than 0.05. Hence it is expected to obtain excellent galaxy shape measurements for the weak lensing study. In this section, we discuss the predicated galaxy redshift distribution and shear power spectra for the CSS-OS.
2.1. Galaxy Photometric Redshift Distribution
Following Cao et al. (2018), we adopt a galaxy redshift distribution n(z) derived from the COSMOS catalog for the CSS-OS photometric survey (Capak et al. 2007; Ilbert et al. 2009). This catalog contains about 220,000 galaxies in 2 deg2, and has a similar magnitude limit as the CSS-OS with i ≤ 25.2 for galaxy observation. Although the survey area of the CSS-OS is much larger, it can represent the redshift and magnitude distributions, and galaxy types observed by the CSS-OS. The CSS-OS redshift distribution derived from this catalog is shown by the dotted line in Figure 2. We can find that the redshift distribution has a peak around z = 0.6, and can extend to z ∼ 4.
Download figure:
Standard image High-resolution imageIn order to extract more information from the weak lensing data, we divide the redshift range into different photo-z tomographic bins, and study the auto and cross power spectra of these bins. As shown in Figure 2, for instance, we divide the redshift range into four photo-z bins (gray vertical lines). The first three bins have equal intervals with Δz = 0.6, and the last bin occupies the rest of the redshift range of redshift distribution.14 Note that the fitting results of the cosmological parameters also depend on the number of photo-z bins (e.g., Huterer 2002). Under the assumptions of the systematics as discussed in Section 2.2, we find that although a greater number of bins may further improve the constraint results, the improvement is not very significant (averaging a factor of ∼1.3 and ∼1.5 on the standard deviations of cosmological parameters for the five and six photo-z bins cases, respectively). Considering the purpose of this work, a four-bin division is adequate for our study. A greater number of bins can be used in the real data analysis, which may help to improve the constraints on the cosmological parameters.
The real galaxy redshift distribution in the ith photoz bin can be expressed as (e.g., Ma et al. 2006)
where and are the lower and upper limits of the ith photo-z bin, n(z) is the total redshift distribution, and is the photo-z distribution function given the real redshift z. We assume it takes the form of
where Δz and σz are the redshift bias and scatter, respectively, which vary as functions of redshift. In this work, we assume that they are constants in different photo-z bins, and treat them as free parameters when fitting the data. Then the ni(z) in Equation (1) can be reduced to
where erf(x) is the error function, and
Note that we always have , no matter what form ni(z) takes.
In Figure 2, the solid blue, green, orange, and red lines show the ni(z) for the four photo-z bins with Δz = 0 and σz = 0.05. In the COSMOS catalog, we find that there are about 208,000 galaxies with σz ≤ 0.05, which take about 95% of the whole sample (Cao et al. 2018). This number can be used, as a reference, to estimate the galaxy number density in the CSS-OS weak lensing survey. We will adopt the ni(z) shown in Figure 2 and the estimated number density in the following WL discussion.
2.2. Shear Power Spectra
Considering intrinsic alignments and systematics, the measured shear power spectrum at a given multipole ℓ for the ith and jth tomographic bins can be estimated by (e.g., Huterer et al. 2006; Amara & Refregier 2008)
where is named as the signal power spectrum, which is composed of three components (e.g., Hildebrandt et al. 2016; Joudaki et al. 2017; Troxel et al. 2017)
Here is the convergence power spectrum, which is the desired galaxy shear power spectrum for cosmological analysis. The and are Intrinsic–Intrinsic (II), and Gravitational-Intrinsic (GI) power spectra, respectively. They are accounting for the intrinsic galaxy alignment effects, that "II" denotes the correlation of the intrinsic ellipticities between neighboring galaxies, and "GI" means the correlation between the intrinsic ellipticity of a foreground galaxy and the gravitational shear of a background galaxy (see, e.g., Joachimi et al. 2016). Assuming Limber and flat-sky approximations (Limber 1954), we have
where χ is the comoving radial distance, χH = χ(z = 5) denotes the horizon distance, and r(χ) is the comoving angular diameter distance. Pm is the matter power spectrum, which is calculated by the halo model (Cooray & Sheth 2002). qi(χ) is the lensing weighting function in the ith tomographic bin
where H0 = 100 h is the Hubble constant, c is the speed of light, and a = 1/(1 + z) is the scale factor. ni(χ) is the normalized source galaxy distribution of the ith tomographic bin, and . The intrinsic–intrinsic and gravitational-intrinsic power spectra are given by
and
Here Fi(χ) is written as
where , ρc is the present critical density, D(χ) is the linear growth factor normalized to unity at z = 0, and z0 = 0.6 and L0 are pivot redshift and luminosity, respectively. AIA, ηIA, and βIA are free parameters in this model. For simplicity, we fix βIA = 0 here, i.e., we do not consider luminosity dependence, because the change of the average luminosity can be ignored across different tomographic bins (Hildebrandt et al. 2016; Joudaki et al. 2017). The fiducial values of AIA and ηIA are set to be −1 and 0, respectively, when producing mock data.
In the shot-noise term of Equation (5), is the shear variance per component caused by intrinsic ellipticity and measurement error, and is the average galaxy number density in a given tomographic bin per steradian. According to the mock CSS-OS catalog, we assume ∼100,000 galaxies with σz ≤0.05 per deg2 (with ∼9.0% outlier fraction) can be observed by the CSS-OS. This corresponds to a total density ∼28 arcmin−2, and , 11.5, 4.6, and 3.7 arcmin−2 for the four redshift bins. Note that the number density estimated here can be smaller in the real observations, considering the complicated instrumental and astrophysical uncertainties. We need to perform realistic simulations to evaluate these effects in future work.
Besides, we also consider the systematic errors in the CSS-OS weak lensing measurements. In Equation (5), mi accounts for the effect of the multiplicative error in the ith redshift bin, which is averaged over all directions and galaxies in that bin. We assume it varies independently in different tomographic bins, and treat it as a free parameter in a given photo-z bin in the fitting process (Troxel et al. 2017). We set mi = 0 as a fiducial value in each bin when generating mock data.
The in Equation (5) is the additive error, which can be generated by the anisotropy of the PSF (Huterer et al. 2006). In principle, could vary at different scales and in different redshift bins, and also can appear in the correlations between bins (Huterer et al. 2006; Amara & Refregier 2008; Amara et al. 2010). For simplicity, we adopt an average constant over all scales of all auto and cross shear power spectra for different tomographic bins (Zhan 2006). Based on the estimates of STEP (the Shear Testing Programme) and GREAT10 (the Gravitational Lensing Accuracy Testing 2010), the can be controlled within 10−10 with S/N ∼ 10 in the Stage IV weak lensing surveys (Heymans et al. 2006; Massey et al. 2007, 2013; Kitching et al. 2012). We find that the S/N of most CSS-OS galaxies is around 7 for the most important g, r, and i bands (Cao et al. 2018). Although the are expected to achieve ∼10−10 when adding up the flux of all seven CSST imaging bands, we would take a conservative estimate with as a moderate value in this work. We will discuss and 10−10 as pessimistic and optimistic cases.
The covariance matrix of the shear power spectra can be estimated by Hu & Jain (2004), Huterer et al. (2006), and Joachimi et al. (2008)
where is the observed shear power spectrum given by Equation (5), and fsky is the sky coverage fraction of the survey. The CSS-OS can cover 17,500 deg2, but after removing masked area (covering image defects, reflections, ghosts, etc.), we assume an effective area of ∼15,000 deg2 can be used in the data analysis (Hildebrandt et al. 2016; Abbott et al. 2017; Troxel et al. 2017).
In Figure 3, we show the power spectra discussed above for the one photo-z bin case (i.e., no tomographic bins). We assume that the shot-noise and additive terms can be eliminated in the data analysis,15 and thus the mock data points of the can be derived as shown in blue circles with error bars. A random Gaussian distribution derived from the covariance matrix is added to each data point. The for the four photo-z bins are shown in Figure 4. To avoid the nonlinear effects, we only take account of the data at ℓ < 3000.
Download figure:
Standard image High-resolution imageDownload figure:
Standard image High-resolution image3. Galaxy Clustering Survey
In addition to the photometric imaging survey, the CSS-OS also can simultaneously perform the spectroscopic survey using slitless gratings. The CSST spectroscopic survey covers the same survey area (17,500 deg2) and similar wavelength range (255–1000 nm) as the photometric imaging survey. It contains three bands, i.e., GU, GV, and GI (see Figure 1), with AB magnitude 5σ limit ∼21 mag per resolution element for point sources with spectral resolution R ≥ 200 (see Table 1). In this section, we will discuss the galaxy clustering power spectrum with the effect of RSD measured by the CSST spectroscopic survey.
3.1. Galaxy Spectroscopic Redshift Distribution
We adopt the zCOSMOS catalog (DR3 release) to simulate the CSST spectroscopic survey result. The zCOSMOS redshift survey observes in the COSMOS field using the VIMOS spectrograph mounted at the Melipal Unit Telescope of the VLT (Lilly et al. 2007, 2009). It covers 1.7 deg2 with a magnitude limit IAB ≃ 22.5, which is close to survey depth of the CSS-OS. The spectral coverage is 5550–9450 Å with a spectral resolution R ∼ 600. There are about 20,000 sources in this catalog, and after selecting high-quality data suggested by the zCOSMOS team, we obtain about 16,600 sources (80% of the total) with reliable spectroscopic redshifts.
The redshift distribution of this mock CSST spectroscopic catalog is shown in Figure 5. We can see that it has a peak at z = 0.3 − 0.4, and can extend to z ∼ 2.5. Since there are not many galaxies at high redshifts, we only consider the sources at z < 1.5 in the galaxy clustering analysis (∼40 galaxies are out of this range). In order to study the evolution of the equation of state (EoS) of dark energy and other cosmological parameters, we divide the redshift range into five tomographic bins with Δz = 0.3 assuming no spec-z outliers for simplicity. The galaxy number fractions in the five spectroscopic bins are 0.17, 0.37, 0.35, 0.10, and 0.01, respectively. Note that a greater number of redshift bins can be used in the real data analysis, depending on the number density of observed galaxies.
Download figure:
Standard image High-resolution image3.2. Redshift-space Galaxy Power Spectrum
The redshift-space galaxy power spectrum in (k, μ) dimensions can be measured by the CSST spectroscopic survey, which can be expanded in Legendre polynomials (Ballinger et al. 1996; Taylor & Hamilton 1996)
where (s) denotes the redshift space, μ = k∥/k is the cosine of the angle between the direction of wavenumber k and the line of sight, is the Legendre polynomials that only the first few nonvanishing orders ℓ = (0, 2, 4) are considered in the linear regime, and is the multipole moments of the power spectrum. Considering the Alcock–Paczynski effect (Alcock & Paczynski 1979), the galaxy multipole power spectra are given by
Here and are the scaling factors in the transverse and radial directions, respectively, and the superscript "fid" means the quantities in the fiducial cosmology. The and are the apparent wavenumber and cosine of angle, where and . Assuming there is no peculiar velocity bias, the apparent redshift-space galaxy power spectrum can be written as
where is the apparent real-space galaxy power spectrum, bg is the galaxy bias, Pm is the matter power spectrum, and β = f/bg where is the growth rate. is the damping term at small scales, which is given by
Here , where σv = σv0/(1 + z) is the velocity dispersion (Scoccimarro 2004; Taruya et al. 2010), and we assume σv0 = 7 Mpc/h for the measurements of emission line galaxies in the CSS-OS (Blake et al. 2016; Joudaki et al. 2017). It will be set as a free parameter in the fitting process. The σR = c σz/H(z) is the smearing factor for the small scales below the spectral resolution of spectroscopic surveys, where H(z) is the Hubble parameter, and (Wang et al. 2009). Based on the instrumental design of the CSST, we set as the accuracy of the spectral calibration. Note that this accuracy could be larger in the real observation, and we need to run simulations in the future for further confirmation.
After adding the shot-noise term and systematics, we obtain the total multipole power spectra of the ath spec-z bin
where is the galaxy number density in the ath spec-z bin. We take an average value in a given redshift bin, and find that , 1.1 × 10−2, 5.5 × 10−3, 1.2 × 10−3 (Mpc/h)−3, and 7.9 × 10−5 from the mock CSS-OS catalog for the five spec-z bins, respectively. Besides, we consider an effective redshift factor to account for the fraction of galaxies that can achieve the required accuracy of the CSS-OS spec-z calibration with in the slitless grating observations (Wang et al. 2010). We assume the fraction decreases as the redshift increases, i.e., , where we set , 0.5, and 0.7, respectively, as the pessimistic, moderate, and optimistic cases. So the final number density in the ath spec-z bin should be . The systematics due to instrumentation effects of the CSST slitless gratings are assumed to be a constant in different tomographic bins, or it can be seen as the average value over all redshift bins and scales. Combining the three assumptions of , we jointly assume three values, i.e., , 5 × 104, and 104 (Mpc/h)3, as pessimistic, moderate, and optimistic cases in the CSST galaxy clustering surveys.
The errors of the multipole power spectra in the ath bin are then given by
where is the survey volume of the ath bin. We assume a 15,000 deg2 sky coverage for the galaxy clustering spectroscopic survey, which is the same as the photometric imaging survey area used in the weak lensing, after masking the area with bad measurements. For a constant ng in a given redshift bin, the corresponding effective survey volume for a multipole component of power spectrum can be defined as
In Figure 6, we show the mock multipole power spectra P0g, P2g, and P4g. In order to avoid the nonlinear effect, we only consider the data at k < 0.2 h/Mpc. The errors of the mock data are generated by assuming and (Mpc/h)3. In Figure 7, the Veff of the P0g for each redshift bin are shown. The dashed, solid, and dotted curves are for the optimistic, moderate, and pessimistic cases.
Download figure:
Standard image High-resolution imageDownload figure:
Standard image High-resolution imageIn addition to the redshift-space power spectrum, the CSS-OS three-dimensional (3D) galaxy clustering data also can be analyzed using other methods, such as topology (Park & Kim 2010), tomographic Alcock–Paczynski (Li et al. 2016), and three-point correlation function (Takada & Jain 2003), that can extract more cosmological information. We will discuss these methods in future work.
4. Galaxy–Galaxy Lensing Survey
We can also cross-correlate the WL and galaxy clustering surveys to get galaxy–galaxy lensing power spectra for the photo-z and spec-z bins in the CSS-OS. This can help us to derive more cosmological information.
4.1. Angular Galaxy Power Spectrum
We first consider the two-dimensional (2D) angular galaxy power spectrum for the ath and bth spec-z bins, which can be derived from the 3D galaxy power spectrum by assuming Limber approximation (Limber 1954; Kaiser 1992; Hu & Jain 2004)
where bg(k, χ) is the galaxy bias, which depends on the scales and redshifts. For simplicity, we assume it is a constant at different scales and only varies as a function of redshift bg(z) when producing mock data. is the normalized galaxy redshift distribution for the ath spec-z bin, that we have . Incorporating the shot-noise term and systematics, the total angular galaxy power spectrum can be written as
where is the average galaxy number density in the ath spec-z bin per steradian, and it is found to be 0.46, 1.0, 0.94, 0.29, and 0.02 arcmin−2 for the five bins (∼2.7 arcmin−2 totally), respectively. is the systematic noise for the angular galaxy power spectrum, and we assume it is a constant for different scales and redshift bins. We find that the result is not sensitive to the as long as it is less than 10−8, i.e., the shot-noise term (the second term) in Equation (21) is relatively large and dominant.
Note that the cross power spectra vanish in the spectroscopic surveys (e.g., the case we discuss here), because there is no overlapping region between spec-z bins. However, for the photometric surveys, both the auto and cross angular galaxy power spectra are important, and can be used to calibrate the galaxy bias and redshift distributions when cooperating with the weak lensing survey (Hu & Jain 2004; Zhan 2006). The CSS-OS also can probe the angular galaxy power spectrum in the photometric imaging survey, and we will discuss it in our future work. In Figure 8, we show the 2D angular galaxy power spectra for the five spec-z bins.
Download figure:
Standard image High-resolution image4.2. Galaxy–Galaxy Lensing Power Spectrum
Then we can calculate the angular galaxy–galaxy lensing power spectrum, i.e., cross-correlating the galaxy clustering and weak lensing surveys, of the ath galaxy spec-z bin and the ith photo-z bin. Considering the intrinsic alignment effect, it can be expressed as (Joudaki et al. 2017)
where the and are given by
and
The covariance matrix for the galaxy–galaxy lensing power spectra is given by Hu & Jain (2004)
In Figure 9, we show the mock galaxy–galaxy lensing power spectra for the first photo-z and second spec-z bins. The green solid, light blue dashed–dotted, red dashed curves denote the total, galaxy–galaxy lensing, and galaxy intrinsic power spectra, respectively. For comparison, the corresponding convergence (dark blue dashed–dotted) and angular galaxy (orange dashed–dotted) power spectra are also shown. The data points with error bars are obtained by assuming and .
Download figure:
Standard image High-resolution imageWe can define the coefficients of the cross power spectra of weak lensing and galaxy clustering between the photometric and spectroscopic redshift bins, which is given by
In Figure 10, we calculate the coefficients of the cross power spectra for the four photometric and five spectroscopic bins. For instance, in the top panel, we find that the bin only has significant correlations with the and bins, because they cover a similar redshift range (see Figures 2 and 5).
Download figure:
Standard image High-resolution image5. Fitting the Mock Data
When generating the mock data, we assume the flat ΛCDM cosmology with the EoS of dark energy w = − 1. In order to explore the constraint of the CSS-OS on the evolution of the dark energy EoS, we adopt the wCDM model with w = w0 +wa(1−a) when fitting the mock data. The HMcode is used to calculate the nonlinear matter power spectrum for the wCDM model (Mead et al. 2015, 2016). The cosmological and systematical parameters for the weak lensing, galaxy clustering, and galaxy–galaxy lensing power spectra in the fitting process are shown in Table 2. In the joint constraint case of the CSST WL and galaxy clustering surveys with four photo-z bins and five spec-z bins, we totally have 31 free parameters in the model.
Table 2. The Fiducial Values and Ranges of the Free Parameters in the Model
Parameter | Fiducial Value | Flat Prior |
---|---|---|
Cosmology | ||
Ωm | 0.3 | (0, 1) |
Ωb | 0.05 | (0, 0.1) |
σ8 | 0.8 | (0.4, 1) |
ns | 0.96 | (0.9, 1) |
w0 | −1 | (−5, 5) |
wa | 0 | (−10, 10) |
h | 0.7 | (0, 1) |
Intrinsic alignment | ||
AIA | −1 | (−5, 5) |
ηIA | 0 | (−5, 5) |
Galaxy bias | ||
(1.15, 1.45, 1.75, 2.05, 2.35) | (0, 4) | |
Velocity dispersion | ||
(6.09, 4.83, 4.0, 3.41, 2.98) | (0, 10) | |
Photo-z calibration | ||
(0, 0, 0, 0) | (−0.1, 0.1) | |
(0, 0, 0, 0) | (−0.1, 0.1) | |
Shear calibration | ||
mi | (0, 0, 0, 0) | (−0.1, 0.1) |
Download table as: ASCIITypeset image
The fiducial value of photo-z bias Δz is set to be zero for each photo-z bin, because the outlier fraction in the photo-z fitting is ignored. Note that our method is actually not quite sensitive to the fiducial value of Δz, which is because we treat it as a free parameter in the fitting process. The fiducial values of bg for the five bins are obtained by , where b0 = 1 and b1 = 1 with the central redshifts of the five spec-z bins zc = 0.15, 0.45, 0.75, 1.05, and 1.35, respectively. The fiducial galaxy velocity dispersion for the five spec-z bins is assumed as σv = σv,0/(1 + zc), where σv,0 = 7 Mpc/h. siz is the stretch factor that adjusts the width of the ni(z) in a given photo-z bin, which can equally change the redshift variance σz. We set the fiducial values of Δzi, , and mi to be 0 in the four photometric bins.
The χ2 statistic method is adopted to fit the mock data, which is defined by
where is the mock data, and is the theoretical power spectrum of the ith redshift bin. Covij is the corresponding covariance matrix. The total χ2 for joint surveys of the WL and galaxy clustering is given by , where , , and are the chi-squares for the galaxy clustering, WL, and galaxy–galaxy lensing power spectra, respectively. The likelihood function then can be calculated by .
We make use of the MCMC technique to constrain the free parameters in the model. The Metropolis–Hastings algorithm is adopted to find the accepting probability of new chain points (Metropolis et al. 1953; Hastings 1970). The proposal density matrix is estimated by a Gaussian sampler with adaptive step size (Doran & Muller 2004). We assume flat priors for the parameters as shown in Table 2. We run 16 parallel chains for each case of systematical assumption, and obtain about 100,000 points for one chain after reaching the convergence criterion (Gelman & Rubin 1992). After the burn-in and thinning processes, we combine all chains together and obtain about 10,000 chain points to illustrate 1D and 2D probability distribution functions (PDFs) of the free parameters.
6. Constraint Results
In this section, we show the constraint results of the cosmological and systematical parameters using the CSS-OS mock data. We compare the results in different cases for the cosmological parameters, and discuss the impact of the systematics on the constraint results.
6.1. Constraints on Cosmological Parameters
In Figure 11, the fitting results of Ωm versus σ8 and w0 versus wa are shown. We explore three cases, i.e., the moderate, optimistic, and pessimistic assumptions (in solid, dashed, and dotted curves), about the systematics of the CSST WL and galaxy clustering surveys. The top, middle, and bottom panels show the constraint results from the WL, galaxy clustering, and joint (WL+galaxy clustering+galaxy–galaxy lensing) surveys, respectively. The gray lines show the fiducial values of the parameters.
Download figure:
Standard image High-resolution imageFor the CSST WL survey, we find that , , , and in 1σ confidence level (C.L.) for the moderate systematic case with . We also show the constraint results of Ωm versus in Figure 12. We find that in the moderate case (solid lines). This constraint result of the cosmological parameters is improved on average by a factor of ∼6 more than that from the Dark Energy Survey (DES) and Kilo Degree Survey (KiDS) (Hildebrandt et al. 2016; Troxel et al. 2017). The improvement can be even larger in the optimistic case (), and at least a factor of ∼3 in the pessimistic case (). This enhancement is due to several advantages of the CSST WL survey, e.g., the large survey area (17,500 deg2), excellent image quality (small and regular PSF),16 and accurate photo-z calibration, which can efficiently suppress both statistical and systematical errors.
Download figure:
Standard image High-resolution imageIn the CSST galaxy clustering survey, we find that , , , and in 1σ C.L. for the moderate systematic case ( and (Mpc/h)3). This leads to a factor of ∼5 improvement at least compared to the current galaxy clustering surveys, such as the SDSS-III Baryon Oscillation Spectroscopic Survey, WiggleZ, 2-degree Field Lensing Survey (2dFLenS), etc. (Blake et al. 2016; Wang et al. 2016; Zhao et al. 2016; Hinton et al. 2017). The constraint results could be better or worse with an average factor of 2–3 in the optimistic ( and (Mpc/h)3) and pessimistic ( and (Mpc/h)3) cases. The improvement is mainly caused by the CSST spectroscopic survey having a deep magnitude limit (see Table 1) and large effective survey volume (see Figure 7).
Since the CSST photometric and spectroscopic surveys are planned to perform simultaneously and cover the same sky area, their joint constraint would be convenient and powerful, which gives , , , and in 1σ C.L. for the moderate systematic case. For the optimistic and pessimistic cases, the constraints can be tighter or looser by a factor of ∼1.3 and 1.7, respectively. Comparing to current similar joint fitting results, e.g., KiDS-450 + 2dFLenS (Joudaki et al. 2017), we find that the CSST joint survey can enhance the constraints of the cosmological parameters by one order of magnitude when assuming the moderate systematics.
Since the systematics have been included in the analysis, fitting biases with respect to the best fits appear in the constraint results according to the fiducial values of the cosmological parameters (gray dashed lines), especially for the moderate and pessimistic cases. By comparing the results of the three systematical assumptions, we find that well-controlled systematics not only can shrink the probability contours but also efficiently suppress the fitting biases of the cosmological parameters (see dashed contours in each survey). Besides, it seems that the fitting biases are relatively larger or more apparent in the joint constraint results (e.g., see the result of wa). This means that better controlling of the systematics may be required in the CSS-OS joint fits. The detailed discussion of the systematical parameters can be found in the next section.
In Figure 13, we show the comparison of the results from the WL, galaxy clustering, and joint surveys. As can be seen, the CSST WL and galaxy clustering surveys have similar constraint strength on Ωm. On the other hand, the WL is more powerful to constrain σ8 than the galaxy clustering by a factor of ∼2, because the WL survey explores 2D power spectra integrating over large redshift range. On the other hand, the CSST galaxy clustering survey can provide comparable or even a bit more stringent fitting results on w0 and wa than the WL survey. The joint CSST surveys of WL+galaxy clustering+galaxy–galaxy lensing can further improve the fitting results, which give at least ∼2σ enhancement on the constraints of cosmological parameters, compared to the WL only or galaxy clustering only survey.
Download figure:
Standard image High-resolution imageThe constraint results of all cosmological parameters with the mild assumption of the systematics for the WL, galaxy clustering, and joint surveys can be found in the Appendix.
6.2. Constraints on Systematical Parameters
In Figure 14, the 1D PDFs of the intrinsic alignment parameters AIA and ηIA are shown for the WL (in blue curves) and joint (in green curves) surveys. We can find that there is no significant improvement on the constraint of AIA for the joint constraints, while a factor of ∼1.3 tighter for ηIA compared to the WL survey. This indicates that the joint fitting can be helpful to extract the redshift evolution effect of intrinsic alignment.
Download figure:
Standard image High-resolution imageThe 1D PDFs of the redshift calibration bias Δzi, stretch factor szi, and multiplicative error mi, are shown in Figure 15. The top and bottom panels show the results from the WL and joint surveys, respectively. We can see that the best fits of the WL systematic parameters are close to their fiducial values (in 1σ C.L.), which means our fitting process can correctly extract the systematics as free parameters. This is also useful to reduce the effect of the systematics on the constraints of cosmological parameters, and can help suppressing the fitting biases (see Figure 11). In order to retain small fitting biases of the cosmological parameters (keeping the fiducial values in 1σ C.L.), as shown in the top panels of Figure 15, we need to control the systematical parameters in 1σ C.L. of their PDFs at least. This requires , (about 10%),17 and for the CSST WL survey. When performing the joint fitting, the constraint results of the systematics can be improved by a factor of ∼1.5, which implies that it is helpful to include other surveys (here the galaxy clustering survey) to eliminate the WL systematics in the fitting process.
Download figure:
Standard image High-resolution imageIn Figure 16, the 1D PDFs of galaxy bias bg and velocity dispersion σv0 assuming the moderate systematics for the five spec-z bins are shown. For the galaxy clustering survey (top panels), we find that the dispersions of the best fits of bg and σv0 from the fiducial values are within ±0.1, and the 1σ ranges 0.03–0.05 for bg and 0.1–0.2 for σv0. The dispersions and uncertainties of bg and σv0 can be further suppressed in the joint fitting (bottom panel), especially for σv0. We find that the dispersions of bg and σv0 can be restricted within ±0.05, and the 1σ within 0.01–0.02 for bg and 0.02–0.06 for σv0. (A factor of ∼2 and ∼4 improvement, respectively.) This indicates that the CSS-OS can provide accurate constraints on the galaxy bias and velocity dispersion.
Download figure:
Standard image High-resolution image7. Summary and Discussion
In this work, we predict the measurements of the CSS-OS on the weak gravitational lensing and galaxy clustering, and explore the constraints on the cosmological parameters with the systematics. We make use of two catalogs, i.e., COSMOS and zCOSMOS catalogs, to simulate the CSST photometric imaging and slitless spectroscopic surveys. We find that the peaks of galaxy redshift distributions are around 0.6 and 0.3 for the CSST photometric and spectroscopic surveys, respectively.
We divide the photometric redshift distribution into four bins, and calculate the auto and cross convergence power spectra of these photo-z bins. The effect of intrinsic alignment, and multiplicative and additive errors are included when generating the mock WL data. In addition to the photometric WL survey, the CSST can simultaneously perform a spectroscopic survey to illustrate clustering of galaxies. We compute the galaxy redshift-space power spectra in five spec-z bins, and obtain the mock data of the multipole power spectra, i.e., P0g, P2g, and P4g. We consider a number of effects when estimating the errors, such as the frequency resolution of the slitless grating, the effective redshift factor accounting for the fraction of galaxies that can achieve the required redshift accuracy in the slitless observations, and the systematic error due to the instrument effect. The cross correlations of the CSST WL and galaxy clustering surveys, i.e., galaxy–galaxy lensing power spectra, are also explored, which can be helpful to further suppress the uncertainties in the joint surveys.
After obtaining the CSS-OS mock data, the MCMC technique is adopted to constrain the cosmological and systematical parameters. We study and compare three cases with different assumptions about the systematics of the WL and galaxy clustering surveys, i.e., the pessimistic, moderate, and optimistic cases, in the constraint process. We find that the CSST WL and galaxy clustering surveys can provide a factor of a few (and optimistically even one order of magnitude) improvement about the cosmological parameters, compared to the current corresponding surveys. The constraints can be further enhanced by ∼2σ in the joint fitting process (WL+galaxy clustering+galaxy–galaxy lensing). We should note that the constraints can be worse in the real survey, because some assumptions made in this work may be simple and optimistic, and more details and uncertainties need to be further considered in future realistic simulations to derive a more realistic result.
The CSS-OS also could provide good fitting about the intrinsic alignment and systematics (in the redshift and shape calibrations) in the WL survey, and galaxy bias and velocity dispersion in the galaxy clustering survey. The joint constraint can further improve the results by a factor of ∼2–4. Particularly, the systematics should be well controlled to avoid large fitting bias on the cosmological parameters in the WL survey, which requires redshift bias , the redshift stretching scale (or the uncertainty of the redshift variance) , and the multiplicative error .
Besides the WL and redshift-space 3D galaxy clustering surveys discussed above, the CSS-OS also can perform 2D angular galaxy clustering photometric survey, strong gravitational lensing survey, galaxy cluster survey, etc. These surveys can offer more valuable information about dark matter and dark energy, the evolution of the LSS, and other important issues in cosmology. Therefore, we can expect that the CSS-OS will be a powerful space sky survey for the studies of our universe.
Y.G. acknowledges the support of NSFC-11822305, NSFC-11773031, NSFC-11633004, the Chinese Academy of Sciences (CAS) Strategic Priority Research Program XDA15020200, the NSFC-ISF joint research program No. 11761141012, and CAS Interdisciplinary Innovation Team. X.K.L. acknowledges the support from YNU Grant KC1710708 and NSFC-11803028. Z.H.F. acknowledges the support of NSFC-11333001. X.Z. and H.Z. were partially supported by China Manned Space Program through its Space Application System and by the National Key R&D Program of China grant No. 2016YFB1000605.
Appendix
The constraint results of seven cosmological parameters, i.e., Ωm, Ωb, σ8, ns, w0, wa, and h, from the CSST WL, galaxy clustering, and joint surveys with moderate systematic assumption are shown in Figures 17–19.
Download figure:
Standard image High-resolution imageDownload figure:
Standard image High-resolution imageDownload figure:
Standard image High-resolution imageFootnotes
- 11
- 12
- 13
- 14
There are several binning strategies that can be adopted in the weak lensing tomographic studies, such as equal binning, Δz ∼ 1 + z binning, etc. We find that the results would not be sensitive to the binning strategy, as long as there is not much difference (in one order of magnitude) of the average source number densities between the bins.
- 15
When fitting the real data, the shot-noise and additive terms can be fitted as free parameters in each tomographic bin in the fitting process. It could make the constraint results looser on the cosmological parameters than assuming an eliminable case, because more parameters are included.
- 16
We should note that there are a number of factors that can affect the PSF shape in the real observation, coming from the optical system (e.g., optical alignment and thermal stability) and the pointing and steering stability of the telescope. Thus the PSF shape can change in flight, and it could affect the relevant shape measurements in the weak lensing survey and make the systematics larger. Hence, we need to calibrate it to the stated accuracy in the orbit. We will discuss this in detail in future work.
- 17