ABSTRACT
We calculate the constraints on dark energy and cosmic modifications to gravity achievable with upcoming cosmic microwave background (CMB) surveys sensitive to the Sunyaev–Zel'dovich (SZ) effects. The analysis focuses on using the mean pairwise velocity of clusters as observed through the kinematic SZ effect (kSZ), an approach based on the same methods used for the first detection of the kSZ effect, and includes a detailed derivation and discussion of this statistic's covariance under a variety of different survey assumptions. The potential of current, Stage II, and upcoming, Stages III and IV, CMB observations are considered, in combination with contemporaneous spectroscopic and photometric galaxy observations. A detailed assessment is made of the sensitivity to the assumed statistical and systematic uncertainties in the optical depth determination, the magnitude and uncertainty in the minimum detectable mass, and the importance of pairwise velocity correlations at small separations, where nonlinear effects can start to arise. In combination with Stage III constraints on the expansion history, such as those projected by the Dark Energy Task Force, we forecast 5% and 3% for fractional errors on the growth factor, . The results suggest that kSZ measurements of cluster peculiar velocities, obtained from cross-correlation with upcoming spectroscopic galaxy surveys, could provide robust tests of dark energy and theories of gravity on cosmic scales.
Export citation and abstract BibTeX RIS
1. INTRODUCTION
The accelerating expansion of the universe continues to be one of the most puzzling problems in cosmology. The background evolution of the universe is constrained by measurements of the cosmic microwave background (CMB; e.g., Hinshaw et al. 2012; Story et al. 2012; Ade et al. 2013a; Calabrese et al. 2013), baryon acoustic oscillations (BAOs) in the galaxy two point correlation function (e.g., Eisenstein et al. 2005; Percival et al. 2007, 2009; Kazin et al. 2014), as well as type Ia supernovae (SNe Ia; e.g., Riess et al. 1998; Perlmutter et al. 1999).
There is significant interest in differentiating between alternative explanations of cosmic acceleration by extending beyond the expansion history to dark energy's impact on the growth of structure (see Uzan 2010; Jain et al. 2013; Joyce et al. 2014 for reviews). This approach is key if a modification of gravity on astrophysical scales is responsible for cosmic acceleration. Large scale structure observations provide two complementary probes of the properties of gravity: the bending of light due to a gravitational potential and the effect of gravity on the motions of non-relativistic objects. The latter manifests as the peculiar velocities of galaxies imprinted in redshift space distortions (RSDs) in the galaxy correlation function (Huterer et al. 2013) as well as cluster motions as observed through the kinematic Sunyaev–Zel'dovich (kSZ) effect (Sunyaev & Zeldovich 1980). Upcoming surveys such as the Dark Energy Survey (DES; Abbott et al. 2005), HyperSuprimeCam,3 the Large Synopic Survey Telescope (LSST; Abell et al. 2009) and the Euclid (Laureijs et al. 2011) and WFIRST (Spergel et al. 2013) space telescopes, will provide gravitational lensing surveys out to redshift 2, and beyond. Concurrently, spectroscopic surveys such as eBOSS (Comparat et al. 2013), DESI (Levi et al. 2013) and spectroscopy from Euclid and WFIRST, will provide both BAO and redshift space clustering measurements of overlapping epochs and survey areas. Each of those probes, though having the potential to constrain gravity, are affected by systematic effects. Cosmological measurements using weak gravitational lensing will require precise photometric redshift and point-spread function calibrations along with characterization of intrinsic alignment contamination of shear correlations, (e.g., LSST Dark Energy Science Collaboration 2012), that can bias and dilute dark energy constraints (Joachimi & Bridle 2010; Kirk et al. 2011; Laszlo et al. 2011). Accurate modeling of redshift space clustering into the nonlinear regime requires precise descriptions of the galaxy clustering correlations beyond the Kaiser formula (Kwan et al. 2012). Clusters are high density environments that are highly affected by the underlying theory of gravity. The peculiar velocities of clusters provide an alternative, complementary measurement of the cosmological gravitational potential field that has different systematic uncertainties. Considering these as part of a multiple tracer approach will provide the clearest picture of gravity's properties.
Cluster motions leave a secondary imprint in the CMB known as the kSZ effect (Sunyaev & Zeldovich 1980), the process of CMB photons passing through a cluster and being Doppler shifted due to the cluster's peculiar velocity relative to the CMB rest frame. This provides a potentially powerful complementary measurement of gravity's influence on cosmic structure to the peculiar motions of individual galaxies (Zhang et al. 2004; Diaferio et al. 2005; DeDeo et al. 2005; Hernandez-Monteagudo et al. 2006; Bhattacharya & Kosowsky 2007; Fosalba & Dore 2007; Kosowsky & Bhattacharya 2009; Mak et al. 2011). Despite its potential, the kSZ has been hard to measure; the signal is small when compared to the thermal SZ effect and emission from dusty galaxies, and does not have a distinct frequency dependence. Observational efforts to constrain the cluster peculiar velocities have come from multi-band photometry in combination with X-ray spectra (Holzapfel et al. 1997; Benson et al. 2003; Kitayama et al. 2004; Mauskopf et al. 2012; Mroczkowski et al. 2012) and spectroscopy around the thermal SZ null frequency (Zemcov et al. 2012). Recent work extracted the kSZ signature from individual clusters by combining sub-millimeter, X-ray and sub-arcminute resolution CMB data to respectively remove dusty galaxy emission, estimate electron density and fit thermal and kSZ templates (Sayers et al. 2013). Data from the WMAP and Planck satellites have been used to place upper limits on the bulk flows and statistical variation in cluster peculiar velocities (Osborne et al. 2011; Ade et al. 2013b), while South Polar Telescope (SPT) data (Reichardt et al. 2012; Zahn et al. 2012) and Atacama Cosmology Telescope (ACT) data (Sievers et al. 2013) have been used to place limits on the kSZ signal from the epoch of reionization.
Multi-band methods do not yet provide a practical approach to extract the kSZ signal from thousands of clusters as desired for large scale cosmological correlations. Cross-correlating arcminute resolution CMB maps with cluster positions and redshifts determined by a spectroscopic large scale structure survey can enable extraction of the pairwise kSZ signal (Bhattacharya & Kosowsky 2007; Kosowsky & Bhattacharya 2009; Li et al. 2014). Indeed, the first detection of the kSZ effect in the CMB spectrum was made by combining CMB measurements from the ACT (Fowler et al. 2007) with the SDSS BOSS spectroscopic survey (Comparat et al. 2013) to measure the mean pairwise momentum of clusters, using luminous red galaxies (LRGs) as a tracer for clusters (Hand et al. 2012). The pairwise approach for extracting the kSZ signal measures the difference in peculiar velocities of nearby clusters as a function of the comoving distance between the clusters. This approach minimizes contributions from the CMB, thermal SZ, and foregrounds, which can be treated as approximately constant on these scales, and by averaging over many clusters pairs any effects independent of the separation will cancel. CMB surveys such as ACTPol (Niemack et al. 2010), SPTPol (Austermann et al. 2012), Advanced ACTPol (Calabrese et al. 2014), SPT-3 G (Benson et al. 2014), and a next-generation, so-called Stage IV CMB survey (Abazajian et al. 2013) in combination with overlapping galaxy surveys, such as those described above, can improve upon this detection and enable the use of mean pairwise velocities as a cosmological probe.
In this paper, we study the constraints on dark energy and cosmic modifications to gravity expected from analyzing the mean pairwise velocity of clusters observed through the kSZ effect by upcoming CMB observations in combination with spectroscopic large scale structure redshift surveys. In Section 2 the analytical formalism used to construct statistics and associated covariances for cluster velocity correlations is summarized. The analysis approach and findings are presented in Section 3, with conclusions and implications for future work discussed in Section 4. A detailed derivation of key results in Section 2 is presented in the
2. FORMALISM
We consider the mean pairwise velocity of clusters derived from the kSZ effect as a probe for dark energy models and modifications to general relativity (GR). Section 2.1 outlines how the growth of structure can be used to constrain modified gravity, Section 2.2 summarizes the halo model approach to analytically calculate the mean pairwise velocity of clusters, and Section 2.3 presents the formalism to estimate the covariance matrix of the mean pairwise velocity. In Sections 2.4 and 2.5 we discuss the fiducial cosmological model and survey assumptions.
2.1. Cosmic Structure and Modified Gravity
Even though on large scales the universe appears homogenous and isotropic, initial local matter overdensities form galaxies and galaxy clusters and evolve into the large scale structure of the universe. The growth of these structures depends on the underlying physical theory and can therefore be used to constrain cosmological models.
According to linear theory, the matter over-density, , is related to the velocity of dark matter particles,
, which connects the time evolution of the perturbations to the dark matter velocity. Any tracer of the underlying dark matter velocity distribution can be used to constrain cosmology and in particular modified gravity models. In a variety of modified gravity scenarios the evolution of the density perturbations can be quite different from standard gravity even though the background expansion of the universe is undistinguishable from a
, factorizing the spatial and temporal dependency, with Da being the growth factor. We can define the growth rate at a given scale factor, a, as
![Equation (1)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn1.gif)
to parametrize the growth of structure. The growth rate is well approximated by with the growth index
for standard gravity (Wang & Steinhardt 1998; Linder 2005). Pairwise velocity statistics can be used to constrain the cosmological model of the universe and the underlying theory of gravity (Davis & Peebles 1977; Ferreira et al. 1999; Juszkiewicz et al. 1999).
2.2. Motion of Clusters as a Probe of Cosmology
We analytically model the expected large scale motion of clusters under cosmological gravitational interactions by considering the properties of dark matter particles, in linear theory, and then using a halo model to infer the velocity statistics of gravitationally bound halos, which we use as proxies for galaxy clusters.
Following the formalism outlined in (Sheth et al. 2001), we assume linear theory to describe the mean pairwise streaming velocity, v, between two dark matter particles, at positions and
, in terms of their comoving separation
,
![Equation (2)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn2.gif)
where the volume averaged correlation function, respectively defined as,
![Equation (3)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn3.gif)
![Equation (4)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn4.gif)
with being the dark matter power spectrum and
is the zeroth order spherical Bessel function.
The properties of dark matter halos of mass M, relative to the dark matter distribution, can be modeled using a halo bias
![Equation (5)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn5.gif)
where ,
is the average cosmological matter density, the critical overdensity is taken to have the standard
, and the zeroth order moment of the mass distribution squared is
![Equation (6)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn6.gif)
with a top-hat window function .
Surveys will generally include cluster halos over a range of masses above some limiting mass threshold, . To analyze the mass statistics we consider a mass averaged cluster pairwise velocity statistic,
, for pairs of clusters separated by a comoving distance r
![Equation (7)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn7.gif)
which has an analogous expression to that in (2) (Sheth et al. 2001; Bhattacharya & Kosowsky 2008), with
![Equation (8)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn8.gif)
![Equation (9)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn9.gif)
The mass-averaged halo bias moments, , are given by
![Equation (10)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn10.gif)
where is the number density of halos of mass M, given by the Jenkins mass function, and the Gaussian window function,
, with
.
In Figure 1 we show the mean pairwise velocity, as a function of cluster separation r for a number of cosmological models at z = 0.15 for a survey with limiting mass
, assuming all other survey specifications are fixed (left panel) and for various assumptions on the limiting mass (right panel). The figure suggests that, as with other linear growth rate related statistics, the Equation of state, w0, and growth exponent,
Figure 1. [Left panel] The mean pairwise cluster velocity, , for different values of the dark energy equation of state parameter, w0, and the modified gravity parameter,
. A more negative w0 leads to an increase in
. A decreased value of
whereas a higher value of
than changing w0. [Right panel] The mean pairwise cluster velocity,
, for different minimum mass cut-offs at redshift of z = 0.15. Note that changing
changes the shape as well as the amplitude of
. Higher
leads to an increase in mean pairwise velocity since the more massive clusters tend to have higher streaming velocities. [Lower panels] Ratio of the mean pairwise velocity,
, for the different scenarios to that for the fiducial model,
.
Download figure:
Standard image High-resolution imageFigure 2 shows the effect of different window function when calculating the mass-averaged halo moments . For large minimum mass, a top-hat filter induces sharp features in
due to edge effects of the window function at small scales, whereas a Gaussian filter leads to smoother results. Therefore, in the following we assume a Gaussian window function.
Figure 2. [Upper panel] The mean pairwise cluster velocity, , for a different window function in the
calculation at z = 0.15 and assuming a limiting mass of
with the black line corresponding to a Gaussian filter. A top-hat (red dashed line) window function leads to sharp features due to edge effects at small scales, while no filtering (blue dotted line) shows a slight excess at small scales compared to a Gaussian filter. These edge effects increase for higher masses. [Lower panels] Ratio of the mean pairwise velocity,
, for the different scenarios to the Gaussian filter.
Download figure:
Standard image High-resolution image2.3. Covariance Matrix
Measurements of cluster velocities are subject to a number of statistical and systematic uncertainties. First, discreteness effects need to be taken into consideration; a smooth continuous field is typically assumed to underly a discrete distribution of local objects, which leads to shot noise. For a large sample size the shot noise should be approximately Gaussian resulting in an error proportional to (Eisenstein & Zaldarriaga 1999), where N is the number of objects in the sample. If the number of objects (e.g., clusters) in the sample is not sufficiently large, the Gaussian limit breaks down, and an additional non-Gaussian contribution to the shot noise can become relevant (Cohn 2006).
Second, as in any cosmological survey, the measurement will be subject to cosmic variance due to the finite size of the sample. Third, in addition to the statistical errors we include a velocity measurement error (Bhattacharya & Kosowsky 2008) to account for the accuracy of the measurements and the uncertainty in the optical depth of the clusters. The total covariance for the mean pairwise velocity is therefore a combination of cosmic variance, shot noise, and the velocity measurement error:
![Equation (11)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn11.gif)
A detailed derivation of the covariance terms can be found in the
Defining an estimator for the mean pairwise velocity, , enables the covariance matrix to be calculated using
![Equation (12)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn12.gif)
where is the volume average. For analyzing a survey we include binning, as observations will be combined not at just one radius r but in bins of width
,
![Equation (13)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn13.gif)
assuming spherical symmetry and where a .
The covariance between the mean pairwise velocities of two cluster pairs, with the two pairs separated by r and r' and using bin width , can be expressed as
![Equation (14)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn14.gif)
where Vs(a) is the survey volume, is the number density of clusters, and
![Equation (15)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn15.gif)
![Equation (16)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn16.gif)
The first term in (14) is the Gaussian contribution to the covariance, which includes both cosmic variance () and shot noise (
). The second term is an additional contribution that is often neglected, which arises if the Gaussian limit breaks down; we refer to this term as "Poisson" shot noise as in (Cohn 2006). While we find it is subdominant in comparison to the Gaussian terms for a mass cut-off
(see Figure 4), it can be important for surveys with smaller cluster number densities. The purely Gaussian shot noise contribution on the other hand is not insignificant and should be included.
Figure 4. [Left panel] The relative error on the mean pairwise velocity of clusters at redshift for a separation bin size
assuming a Stage III-like survey (see Table 2). The Poisson shot noise is sub-dominant compared to the other terms, the Gaussian shot noise term, however, cannot be neglected. [Right panel] One over the diagonal terms of the total Fisher matrix relative to the mean pairwise velocity for varying the minimum mass and the measurement error. The effect of the minimum mass on the fisher matrix is more prevailing than the dependency on the measurement error.
Download figure:
Standard image High-resolution imageWe include a contribution to the covariance due to the uncertainty in measuring the velocity given by (Bhattacharya & Kosowsky 2008),
![Equation (17)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn17.gif)
where is the measurement error discussed in more detail in Section 3.4, and
is the number of pairs in each separation bin given by
![Equation (18)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn18.gif)
As shown in Figure 3, the number of cluster pairs increases rapidly with decreased minimum mass. The measurement error term in the covaraince is proportional to and will increase quickly with an increasing number of bins since the number of cluster pairs directly depends on the size of the r-bin.
Figure 3. Number of cluster pairs, , vs. separation in bins of
2 Mpc/h for different mass cutoffs at redshift
. This assumes a Jenkins mass function and a 6000 square degree survey.
Download figure:
Standard image High-resolution imageFigure 4 shows the diagonal elements of the covariance matrix for the different covariance components, for a bin width of . As cluster separation increases, the covariance becomes dominated by cosmic variance, while at smaller separations
, the contributions from each of the terms becomes comparable. As a result of the multiple contributions to the covariance matrix, and their respective sensitivities to bin size and cluster separation, the total covariance matrix slightly depends upon the number of bins. The measurement error and shot noise can be reduced by choosing a coarser binning with the trade-off of decreased resolution and loss of information. On the other hand, the fractional contribution of the cosmic variance will increase as the size of the bins increases. Once the cosmic variance dominates nothing can be gained from a coarser binning. A very coarse binning marginally reduces the constraints, e.g., using
lowers the figures of Merit (FoM) by 30% compared to
, however, any bin size smaller than
leads to equivalent results. Throughout the analysis we assume a bin size of
.
Off-diagonal covariances between cluster pairs of different separations are important. Figure 5 shows the covariance contributions from cosmic variance and shot noise and indicates the comparative importance of off-diagonal terms. The off-diagonal contributions have a notable effect on the Fisher matrix amplitudes as a function of separation, giving rise to the differences between the left and right panels in Figure 4. The right panel shows the effect on the Fisher matrix of changing key model assumptions, the minimum detectable cluster mass and the mean pairwise velocity uncertainty. Altering the mass limit has a larger effect than comparable changes to the the measurement error because the number of clusters and cluster pairs strongly depend on the limiting mass (see Figure 3), changing shot noise as well as the measurement error contribution to the covariance significantly.
Figure 5. 2D contour plots of the cosmic variance [left panel] and the shot noise term [right panel] at redshift assuming a separation bin size of
, a lower mass limit of
, and a sky coverage of 6000 square degrees. Note that both terms have notable non-zero off-diagonal terms that affect the total inverse covariance used in the Fisher analysis, and that while the cosmic variance values are larger, the Gaussian noise term should not be neglected as it can have a significant effect, particularly for small separations.
Download figure:
Standard image High-resolution image2.4. Cosmological Model
For our analysis we consider constraints on 9 cosmological parameters:
![Equation (19)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn19.gif)
where ,
,
and
are the dimensionless baryon, matter, curvature and dark energy densities respectively, h is the Hubble constant in units of 100 km s−1 Mpc−1, w0 and wa are the dark energy equation of state parameters, such that the equation of state is
,
, and ns and As are the spectral index and normalization of the primordial spectrum of curvature perturbations.
Throughout this paper we assume a fiducial model that is a
.
We calculate constraints on cosmological parameters using the Fisher Matrix formalism. The covariance between two parameters and
, from (19), is given by
![Equation (20)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn20.gif)
where is the covariance matrix between two clusters pairs as defined in 2.3, including a redshift bin with mid-point zi and the clusters in each pair having comoving separations of rp and rq. Nz and Nr are the number of redshift and spatial separation bins, respectively.
We quote results in terms of the Dark Energy FoM (Albrecht et al. 2006) defined as
![Equation (21)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn21.gif)
![Equation (22)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn22.gif)
is the 2 × 2 submatrix of the inverted Fisher matrix excluding the modified gravity parameter
Throughout this paper we consider results in combination with either simply a Planck-like CMB prior or a Dark Energy Taskforce (DETF; Albrecht et al. 2006) prior that includes CMB, SN, and non-kSZ related LSS constraints on the background cosmological and dark energy parameters. We do not include a prior on the modified gravity parameters unless stated otherwise. For the Planck-like CMB survey, we consider complementary constraints on the cosmological parameters from the temperature (T) and polarization (E) measurements up to l = 3000 as summarized in Table 1.4
Table 1. CMB Survey Specifications
Parameter | Frequency (GHz) | ||
---|---|---|---|
100 | 143 | 217 | |
![]() |
⋯ | 0.8 | ⋯ |
![]() |
10.7 | 8.0 | 5.5 |
![]() |
5.4 | 6.0 | 13.1 |
![]() |
⋯ | 11.4 | 26.7 |
Note. For the sky coverage, , beam size,
, and noise levels per pixel for the temperature and polarization detections at 3 frequencies, for a Planck-like survey.
Download table as: ASCIITypeset image
2.5. Survey Specifications.
We forecast cosmological constraints for three different combinations of surveys: (1) a current (Stage II) CMB survey, such as ACTPol (Niemack et al. 2010), combined with a galaxy sample that includes spectroscopic redshifts, such as SDSS BOSS (Comparat et al. 2013), (2) a near-term (Stage III) survey, such as Advanced ACTPol (Calabrese et al. 2014), also combined with SDSS BOSS, and (3) a longer-term (Stage IV) survey, such as CMB-S4 (Abazajian et al. 2013), combined with a next generation spectroscopic survey, such as DESI (Levi et al. 2013).
The mean cluster pairwise velocity can be measured by cross-correlating the kSZ signal with cluster positions and redshifts. For the cluster sample, we assume that a spectroscopic survey provides redshifts to LRGs over an overlapping area with the CMB survey. Recent studies show that the kSZ signal can be extracted from the CMB maps using LRGs of the BOSS survey as a proxy for clusters (Hand et al. 2011). Using LRGs creates a large, precise positioned sample of tracers to extract the kSZ correlation.
However, there are several factors that need to be considered in using LRGs as cluster tracers. LRGs are not perfect tracers of a cluster's center, with perhaps 40% of bright LRGs and 70% of faint LRGs off-centered, satellite galaxies (Hikage et al. 2012) that may be related to cluster mergers (Martel et al. 2014). The imprecise match between LRGs and clusters could lead to detrimental misalignments, such as trying to extract the kSZ signal from positions that are not associated with clusters or an incomplete cluster catalog if spectroscopic measurements of an LRG near the cluster center were not obtained. The theoretical mean cluster pairwise velocity is an observable averaged over all cluster pairs assuming a complete sample above a limiting minimum mass. While (Hand et al. 2011) optimize the angular size of the CMB sub-map used in the stacking approach to minimize the overall covariance, this does not ensure that the cluster sample obtained from the LRGs is complete. Further studies are needed to quantify the effects of using LRGs as cluster tracers and ensure that no bias is introduced in the analysis before this approach can be used for cosmological constraints. Additional clustering information, such as the halo occupation distribution of the galaxy sample, must be used to quantify the uncertainty of LRGs as accurate cluster tracers and ensure that no bias is introduced in the cosmological constraints. Modeling the number and distribution of galaxies within the dark matter halos could include both full simulations and constraints from large scale structure data (White et al. 2011).
Another issue is that the uncertainty in the minimum mass of the cluster sample associated with the LRGs is difficult to estimate, although, the minimum mass uncertainty could be treated as an additional nuisance parameter in the analysis.
To acknowledge these issues in our forecasts, we assume a scenario that aims to maximize cluster completeness and purity, with a well defined cluster mass cut-off, rather than cluster number density. We select a survey area that has photometric and spectroscopic galaxy catalogs and overlapping CMB kSZ data. Specifically, we consider BOSS and a DESI-like survey, for which we expect photometric catalogs to exist over the survey area. We note that Euclid spectroscopic and imaging surveys, and LSST imaging with overlapping WFIRST imaging and spectroscopy would also provide future valuable data sets at higher redshifts. The uncertainties in the cosmological parameters evolve as the square root of the sky coverage. Requiring spectroscopic redshifts, e.g., from BOSS, limits the survey area, but provides confidence that the comoving cluster separation can be accurately calculated as in (Hand et al. 2012). Photometric information allows cluster detection, and mass estimates, using algorithms, such as the friends-of-friends, as used in redMaPPer (Rykoff et al. 2014), to maximize the completeness and purity of the cluster sample, with the drawback of a limited number of clusters and a volume-limited catalog. A study of using only photometric information to extract the kSZ signal can be found in (Keisler & Schmidt 2013).
The survey specification assumed in our analysis for the CMB and large scale structure Stage II, III, and IV surveys are given in Table 2. We assume a BOSS-like spectroscopic survey for Stages II and III and a DESI-like Stage IV survey with redshift ranges that are determined by the redshift coverage of the LRG sample and assume joint photometric survey data. We assume Stages II and III have access to the same or comparable LRG surveys so retain the same limiting mass, but do slightly increase CMB overlap with these data due to the larger survey area planned for Advanced ACTPol (Calabrese et al. 2014). For Stage IV we assume a deeper LRG survey that provides lower minimum mass, higher z, and larger overlap. Our minimum mass assumptions are conservative, and will likely be improved upon at each respective stage. As an example, the LSST survey projects that the minimum detectable cluster mass at will be lower than
after a single visit image in all bands, and be better than
in all bands in the complete 10 years survey (Abell et al. 2009). Similarly the SDSS-derived MaxBCG Catalog already achieves 90% purity and >85% completeness for clusters of masses exceeding
(Koester et al. 2007).
Table 2. Reference Survey Specifications Used to Model Stages II, III, and IV kSZ Cluster Surveys
Survey Stage | ||||
---|---|---|---|---|
Survey | Parameters | II | III | IV |
CMB |
![]() ![]() |
20 | 7 | 1 |
Galaxy |
![]() |
0.1 | 0.1 | 0.1 |
![]() |
0.4 | 0.4 | 0.6 | |
No. of z bins, Nz | 3 | 3 | 5 | |
![]() ![]() |
1 | 1 | 0.6 | |
Overlap | Area (1000 sq. deg.) | 4 | 6 | 10 |
Note. The expected instrument sensitivity of the CMB survey, , along with the assumed optical large scale structure survey redshift range
, redshift binning, and minimum detectable cluster mass,
are shown. We consider an effective sky coverage by estimating the degree of overlap between the respective cmb and optically selected cluster data sets.
Download table as: ASCIITypeset image
The measurement error for the radial peculiar velocity, v, of a cluster is a combination of the instrumental sensitivity as well as the uncertainty in the optical depth,
![Equation (23)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn23.gif)
where is the temperature of the CMB. We estimate the total measurement error by adding those two sources of uncertainty in quadrature as
![Equation (24)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn24.gif)
The accuracy of the instrument is given by
![Equation (25)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn25.gif)
where is the sensitivity of the instrument per pixel and
being the number of pixels of a cluster. We assume that an average size cluster will have
and an instrument sensitivity as summarized in Table 2. The uncertainty in the optical depth is given by
![Equation (26)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn26.gif)
Assumed uncertainties contributing to the measurement error are summarized in Table 3. We use the scatter in the optical depth, , and the mean value of
Table 3.
The Assumed Individual Contribution from Instrument Sensitivity, , and Uncertainty in
Parameter | Survey Stage | Redshift bin | ||||
---|---|---|---|---|---|---|
0.15 | 0.25 | 0.35 | 0.45 | 0.55 | ||
![]() |
3.45 | 2.27 | 1.84 | 1.45 | 1.20 | |
![]() |
⋯ | ⋯ | 0.15 | ⋯ | ⋯ | |
![]() |
⋯ | ⋯ | 120 | ⋯ | ⋯ | |
![]() |
Stage II | 290 | 440 | 540 | ⋯ | ⋯ |
(km s−1) | Stage III | 100 | 150 | 190 | ⋯ | ⋯ |
Stage IV | 15 | 22 | 27 | 34 | 42 | |
![]() |
Stage II | 310 | 460 | 560 | ⋯ | ⋯ |
(km s−1) | Stage III | 160 | 200 | 230 | ⋯ | ⋯ |
Stage IV | 120 | 120 | 120 | 120 | 130 |
Note. The values of , are estimated from simulations assuming a convolution over a 1
3 beam.
is the total measurement uncertainty for the reference case.
Download table as: ASCIITypeset image
3. ANALYSIS
Section 3.1 summarizes and compares the results of each survey. The effect of modeling assumptions on the minimum detectable cluster mass, the minimum cluster separation considered, the measurement error, and the dark energy model are discussed in Sections 3.2–3.5.
3.1. Potential kSZ Constraints on Dark Energy and Modified Gravity
In this section we discuss the potential of upcoming kSZ surveys to constrain dark energy and modified gravity parameters. Figure 1 shows both the equation of state, driving the expansion history, and parameter plane for kSZ in combination with a Planck-like CMB and DETF priors on all parameters excluding
with a CMB prior (which has
alone), and
with DETF Stage III data included (
).
Figure 6. 2D projected likelihoods for the parameter space, showing the 68% and 95% confidence levels for the Stage IV-like survey are shown for two well-separated spectroscopic redshift bins,
(blue) and
(red), and when all five redshift bins,
(green) are considered when combined with Planck-like CMB priors. The inclusion of multiple redshift bins breaks degeneracies between w and
Download figure:
Standard image High-resolution imageFigure 7. 2D projected likelihoods for parameter space, showing the 68% confidence levels for Stage II-(red), III-(green), and IV-(blue) like surveys when combined with Planck-like CMB priors only (dashed) and DETF stage III GR priors (solid, excluding DETF constraints on
Download figure:
Standard image High-resolution imageComplementary, contemporaneous constraints from baryonic acoustic oscillations and SNe Ia will provide significantly tighter constraints on the background expansion history and the equation of state. If we include the impact of a DETF Stage III prior on all parameters, excluding the growth factor, the degeneracy between the equation of state and growth factor is significantly reduced and the projected constraints on ), as well as the
constraints of w0, wa (marginalizing over
Table 4.
[Left Columns] Results for the Reference Survey Assumptions as Summarized in Table 2 Including [Top Rows] Planck Priors and [Lower Rows] Constraints on the Background Cosmological Parameters (Excluding
Data Set | Parameter | Fiducial assumptions | + Uncertainty in ![]() |
+ Lower ![]() |
||||||
---|---|---|---|---|---|---|---|---|---|---|
Stage II | Stage III | Stage IV | Stage II | Stage III | Stage IV | Stage II | Stage III | Stage IV | ||
+CMB |
![]() |
5 | 6 | 26 | 4 | 9 | 19 | 7 | 9 | 154 |
![]() |
7 | 10 | 47 | 6 | 12 | 29 | 10 | 15 | 195 | |
![]() |
33 | 43 | 102 | 28 | 40 | 84 | 50 | 63 | 289 | |
![]() |
0.73 | 0.70 | 0.35 | 0.62 | 0.39 | 0.41 | 0.67 | 0.63 | 0.09 | |
![]() |
2.6 | 2.5 | 1.2 | 2.2 | 1.4 | 1.5 | 2.5 | 2.3 | 0.3 | |
![]() |
0.09 | 0.08 | 0.05 | 0.10 | 0.07 | 0.05 | 0.07 | 0.06 | 0.02 | |
+DETF |
![]() |
119 | 130 | 208 | ⋯ | ⋯ | ⋯ | ⋯ | ⋯ | ⋯ |
![]() |
128 | 139 | 229 | ⋯ | ⋯ | ⋯ | ⋯ | ⋯ | ⋯ | |
![]() |
176 | 191 | 321 | ⋯ | ⋯ | ⋯ | ⋯ | ⋯ | ⋯ | |
![]() |
0.10 | 0.09 | 0.07 | ⋯ | ⋯ | ⋯ | ⋯ | ⋯ | ⋯ | |
![]() |
0.30 | 0.29 | 0.24 | ⋯ | ⋯ | ⋯ | ⋯ | ⋯ | ⋯ | |
![]() |
0.07 | 0.05 | 0.03 | ⋯ | ⋯ | ⋯ | ⋯ | ⋯ | ⋯ |
Note. For reference, the Planck-like Fisher matrix alone has and DETF has
. [Central columns] Results in which the impact of an uncertainty in the exact minimum mass of the cluster sample,
, is included by marginalizing over
as a nuisance parameter with a 15% prior imposed. [Right columns] Results for a more optimistic mass cut-off of
for Stages II and III and
for Stage IV with marginalization over
with a 15% prior imposed as well. Constraints as a function of
are also shown at the top of Figure 8.
Download table as: ASCIITypeset image
These results could provide valuable complementary constraints to those on from measurements at
using OII for a DESI-like survey (Font-Ribera et al. 2014), and comparable from a Euclid-like
survey, for which (Amendola et al. 2012) projected
(assuming a luminosity function Geach et al. 2009) that has since been revised downwards to lower
number counts (Wang et al. 2012; Colbert et al. 2013).
3.2. Dependence on Minimum Mass of the Galaxy Cluster Sample
For cluster abundance measurements knowing the precision with which the minimum mass is known is important. To assess the degree of precision required for the pairwise measurements we consider the impact on the cosmological constraints of marginalizing over the minimum mass, with a 15% prior on . The middle panel of Table 4 shows the effects of this marginalization: the constraints are loosened only slightly compared to the fiducial case, that has no marginalization over the minimum mass. This implies that a precise knowledge of the minimum mass is not crucial to achieve cosmological constraints. An explanation for the comparative insensitivity of the dark energy constraints to uncertainties in Mmin, can be understood with reference to Figure 1. While varying dark energy parameters and Mmin both change the large scale pairwise velocity amplitude the minimum mass also changes the shape of the pairwise velocity function. This means that uncertainties in the minimum mass can be discerned from those in dark energy, and do not translate into a comparable degradation of constraints on w or
The measurement uncertainty on the mean pairwise velocity decreases with the number of clusters used for the cross-correlation. The upper panels of Figure 8 presents the dependence of the FoM and constraints on the assumed minimum observed mass. The increased number density of clusters and cluster pairs arising from a lower mass bound, below
, significantly improves the statistical uncertainties in the pairwise velocity. For our analysis we integrated over a Jenkins mass function using the minimum observed mass as our lower limit. As the number density of clusters drops off quickly for higher masses the constraints deteriorate strongly for a minimum mass above
.
Figure 8. [Left panels] The impact of the assumed minimum mass of the cluster sample, , on the dark energy figures of merit (FoM) and uncertainty on the growth factor,
, for the Stage II (red), III (green), and IV (blue) reference survey specifications (as given in Table 2) with a Planck-like CMB prior on all parameters except
. While including smaller-scale observations below ∼20 Mpc/h would appear to improve both the FoM and
, as discussed in the text, we note that caution must be used in including these scales, with the potential for additional theoretical uncertainties, not included here, as nonlinear effects become important.
Download figure:
Standard image High-resolution imageAssuming that the complications discussed in Section 2.5, in determining LRG centrality and cluster mass estimates, can be controlled, in principle one could achieve much higher number densities and a smaller minimum mass. This would increase the number of pairs in the cluster sample. The window functions for lower mass halos would include additional information in the mass averaged statistics from the power spectrum at smaller scales that would lead to tighter constraints on the cosmological parameters. The right columns of Table 4 show the results assuming a more optimistic mass cut-off than the reference case, for Stages II and III and
for Stage IV. To account for the uncertainty in mass we marginalize over the minimum mass assuming a 15% prior. The GR FoM, with a CMB prior, are improved from
and
for Stages II and III to
and
, compared to the reference scenarios, and by a factor of 1.8 for Stage IV. The uncertainty in
, 0.06, and 0.02 for Stages II, III, and IV.
3.3. Dependence on the Nonlinearity Cut-off
As shown in Figure 4, the inverse covariance rises at lower cluster separations so that the inclusion of cluster pairs at small separation can have a potentially significant effect on improving the dark energy constraints. Simulation show a deviation from the predicted theoretical mean pairwise velocity, however, starting at separations of (Bhattacharya & Kosowsky 2008) so that the nonlinear corrections to the cluster motion needs to be considered. Equation (2) has two major deficiencies: It relies on linear theory to model the underlying dark matter distribution (Juszkiewicz et al. 1999; Sheth et al. 2001) and it assumes a linear, scale-independent bias (Sheth et al. 2001). The former leads to a discrepancy of the dark matter pairwise velocity with linear theory at nonlinear scales around
, the latter introduces deviations at even larger scales. It is worthwhile, therefore, to assess how accurately the mean pairwise velocity of clusters can be modeled in the transition to the nonlinear regime and how the cosmological constraints depend upon the assumed limiting minimum mass.
In Figure 8 we highlight the sensitivity of the FoM and uncertainty in the modified gravity parameter . For a Stage IV-like survey including all scales up to
increases the FoM by 50% compared to an analysis with separations above
excluded and halves the uncertainty on
In this work we chose a moderate approach cutting off our analysis in the mildly nonlinear regime using a minimum separation of Mpc/h. On-going work on using an perturbative approach to model nonlinearities (Okumura et al. 2014) and improved N-body simulations suggests that the formalism will be improved in the near future to fully exploit the mildly nonlinear regime.
3.4. Dependence on the Measurement Error
Central to utilizing the kSZ for cosmology, is the ability to measure the pairwise momentum accurately, and then in turn extract the pairwise velocity, from the momentum, through being able to determine the cluster optical depths. In this section we investigate in more detail the sensitivity of the constraints to these important effects.
As described in Section 2.5, the measurement error of a given cluster is given by the combination in quadrature of the instrument noise and the uncertainty in the optical depth of the cluster. In the fiducial analysis we include an uncertainty in the measurement of
To understand the impact of greater uncertainty in the determination of in the optical depths of the cluster sample and the second is a systematic offset in the
, in each redshift bin that scales the amplitude of the mean pairwise velocity,
, and consider its effect on cosmological constraints when marginalizing over
. Additionally we consider a constant, redshift independent nuisance parameter,
, that scales the amplitude across all clusters. For clarity, when studying the impact of
we remove the
contribution to the covariance and purely parameterize the uncertainty in
.
Figure 9. Impact of modeling assumptions in the determination of , in the pairwise velocity covariance. The upper left panel shows the FoM assuming the growth rate is determined by GR (solid lines) and marginalizing over a freely varying
constraints in the lower left panel. The right panels show the effect of a prior on a systematic offset in the
, and a redshift independent multiplicative bias (dashed lines),
. A detailed discussion of the relative sensitivities is provided in the text.
Download figure:
Standard image High-resolution imageTable 2 shows that for a near-term Stage II survey the noise will be dominated by the instrument accuracy, for a more sensitive Stage III both components become comparable, and for a Stage IV survey the velocity accuracy may be limited by the accuracy of between 0 and 1000 km s−1 only minimally changes the constraints on the dark energy FoM and the constraints on
For Stage II and III surveys, conclusions for the effect of on w0 and wa are similar to those for
. The constraints on these dark energy parameters are principally determined by the Planck-like prior, independent of the kSZ constraints, and uncorrelated with
. For the Stage IV survey the kSZ constraints provide additional constraints on the equation of state, increasing their correlation with
, and the prior has a more pronounced effect on improving the FoM once below
. Equivalently, Stages II and III are not affected by the assumptions on the
model without imposing any prior.
For the growth parameter, which is predominantly constrained by the kSZ data, the model assumptions on the , without imposing any prior doubles the uncertainty in
. The difference between this behavior and the FoM constraints (shown in Figure 9 lower panels) indicates that the redshift dependence of the FoM versus
leads to a factor of 5–10 improvement in the parameter constraints for the redshift dependent
model and a factor of 3–4 improvement for a constant
. For the Stage IV survey we find that the multiple-redshift bins and improved covariance reduce the degeneracy between the
Beyond uncertainties in km s−1 on the cluster peculiar motion due to internal motion within the cluster (Nagai et al. 2003; Holder 2004) that will ultimately limit the CMB observations.
3.5. Dependency on the Modified Gravity Parametrization
In the previous sections we parametrized modified gravity models using one extra parameter (Wu et al. 2009), equivalent to the dark energy
model. Table 5 summarizes the FOM as well as the
constraints on {w0,wa,
,
}. Introducing an additional extra parameter loosens the constraints on the parameters with the advantage of imposing a smaller theoretical prior on modified gravity. Figure 10 shows the
and
constraints on
for Stages II, III, and IV.
Figure 10. Marginalized constraints on the parameter space showing the 68% confidence contours for the pairwise velocity constraints in combination with a Planck-like (dashed) or DETF Stage III (solid) prior on all parameters except
and
. The fiducial model assumes GR with
and
.
Download figure:
Standard image High-resolution imageTable 5.
Marginalized Constraints on the Dark Energy and Modified Gravity Parameters in the and
Parameterization
Paramerer | ![]() |
||
---|---|---|---|
Stage II | Stage III | Stage IV | |
![]() |
127 | 137 | 211 |
![]() |
0.10 | 0.09 | 0.07 |
![]() |
0.29 | 0.28 | 0.23 |
![]() |
0.31 | 0.23 | 0.08 |
![]() |
0.69 | 0.51 | 0.14 |
Note. A summary of the dark energy FoM and 1 parametrization for Stages II, III and IV scenarios in combination with a DETF prior on all parameters except
and
.
Download table as: ASCIITypeset image
An even more general approach is to directly constrain the growth rate in redshift bins as a "model-independent" way. This approach is particularly applicable for spectroscopic galaxy surveys which can isolate peculiar velocity data, and hence the growth rate, in precise redshift bins. In Figure 11 we present the forecasts for this parametrization. We find, in combination with the Stage III DETF constraints on the equation of state, an uncertainty in fg of around 2% at for Stage IV surveys. These constraints provide a complementary measurement to the growth rate measurements from redshift space clustering of emission line galaxies and LRGs with a DESI-like survey. On its own, without the additional DETF stage III data, forecasts for DESI project 1.4%–1.6% errors on the growth factor multiplied by
over a comparable redshift range to that considered in our analysis (Levi et al. 2013). DESI, along with Euclid (Amendola et al. 2012) and WFIRST (Spergel et al. 2013) will also provide complementary spectroscopic constraints on the growth rate at higher redshifts,
.
Figure 11. Expected fractional errors on the growth rate, fg, in each redshift bin for Stages II (red), III (green), and IV (blue) when combined with a Planck-like CMB (dashed) or DETF Stage III (solid) prior on all parameters except fg(z).
Download figure:
Standard image High-resolution image4. CONCLUSIONS
Recent analyses have demonstrated that the kSZ can be successfully extracted from sub arcminute resolution CMB maps by cross-correlating them with cluster positions and redshift from spectroscopic large scale structure surveys. In this paper we have considered the potential to apply this technique, in light of planned CMB and LSS surveys with greater sensitivity and larger areas, to constrain dark energy and modifications to gravity on cosmic scales using the mean pairwise velocity of clusters as an observable. We have extended the model presented in (Bhattacharya & Kosowsky 2008) to account for the dependence on the binning in cluster separations, shot noise, and potential contributions to the total covariance matrix due to small number densities, that we show are significant, despite being frequently neglected, and provided a detailed derivation of the covariance components.
The projected constraints are intimately related to not only the quality of future data, determined by the instrumental precision, but also to the modeling of the uncertainties in transforming the kSZ observations into velocity estimates that constrain the large scale structure growth history. We investigate a range of uncertainties, using reasonable assumptions based on simulations and projected survey capabilities. We also study the sensitivity to assumptions by varying theoretical priors to understand and estimate the robustness of the results. We included a study of the effect of survey assumptions on the minimum detectable cluster mass and the minimum cluster separation that could be included, in light of the influence of nonlinear effects in the cluster motions/correlations.
The mean pairwise velocity is modeled assuming linear theory for the underlying matter distribution as well as the halo bias. Variations in the equation of state and the growth rate affect the linear growth factor in similar ways, that leads to degenerate effects on the pairwise velocity amplitude. However, the different redshift dependence of these effects helps to break the degeneracy, and constraints on the expansion history, such as those from SNe Ia, BAO, and CMB geometric constraints, break it further, allowing growth information to be extracted from the kSZ.
The cluster sample's minimum mass has a significant impact on the predicted constraints. A smaller minimum mass leads to an increase of the number of clusters in the catalog (assuming the catalog is nearly complete) and significantly reduces the errors on all cosmological parameters. Assuming an optimistic mass cut-off for the upcoming cluster catalogs leads to an improvement on the FoM (including CMB priors) from to
and a reduction of the 1
In contrast, the uncertainty in the exact minimum mass had only a mild impact on the dark energy and modified gravity constraints. This was understood in terms of the additional effect of the minimum mass on the shape, as a function of cluster pair separation, as well as amplitude, of the pairwise velocity. Marginalizing over the minimum mass, while imposing a 15% prior in our analysis, in a scenario in which the covariance remains unchanged, reduces the FoM for a Stage IV survey from to
and marginally loosens the constrains on
Considering pairwise correlations down to cluster separations of halves the error on
. While extending the analysis to smaller separations could significantly improve the constraints, including scales in the nonlinear regime without accurate modeling could also potentially bias the constraints and introduce more systematic uncertainties.
Improved constraints on , as well as a constant, redshift independent bias,
. We found that the effect of
on the dark energy FoM was minimal reflecting that the principal constraints come from the external CMB or DETF prior. For Stage IV, in particular, the dispersion in
or
had the biggest impact for Stage II and III surveys for which significant degeneracies exist between the
bias model, the redshift bins and reduced covariance allowed both
and
In addition to a minimal model to modify gravity, in which a modification to the growth rate is parameterized by a single parameter, 1
for Stage III and
constraints for Stage IV when combined with a Stage III DETF constraints on the expansion history.
Potential improvements in the covariance could include taking advantage of multi-frequency information available in upcoming surveys (e.g., Benson et al. 2014; Calabrese et al. 2014) to improve the kSZ signal extraction and reduce the measurement error. Larger LRG catalogs could also be used, such as in the first kSZ detection; however, this increases uncertainty in the minimum mass of the cluster sample. Similarly, with the improvements in cluster photometric redshift uncertainties that are coming from improved algorithms and spectroscopic training sets, it may be feasible to use photometric surveys, without spectroscopic follow up, to significantly enlarge the cluster sample. This will degrade the redshift accuracy, and therefore the measurements of the cluster separation, particularly on small scales; however, the larger sample size will help compensate and might even improve the constraining power.
Measurements of the kSZ effect provide complementary constraints on the growth of structure to weak lensing and RSD measurements by providing measurements on larger physical scales and using a highly complementary, and more massive, tracer of the cosmological gravitational field, that is not dependent upon a characterization of galaxy bias. Having a variety of cosmological probes of dark energy and modified gravity with different systematics is going to be vital for reducing systematic effects and biases in parameter estimation and determining the properties of dark energy and gravity in a variety of epochs and regimes.
The authors would like to thank Nicholas Battaglia, Joanna Dunkley, Kira Hicks, Arthur Kosowsky, Thomas Loredo, Eduardo Rozo, and David Spergel for useful inputs and discussions on pairwise statistics, survey capabilities, and astrophysical uncertainties, and comments on the manuscript.
The work of E.M.M. and R.B. is supported by NASA ATP grants NNX11AI95G and NNX14AH53G, NASA ROSES grant 12-EUCLID12-0004, NSF CAREER grant 0844825, and DoE grant DE-SC0011838.
APPENDIX: DERIVATION OF THE MEAN PAIRWISE VELOCITY COVARIANCE
In this section we provide a detailed derivation of the Gaussian and non-Gaussian contributions to the covariance matrix given in Equation (14) in the main text. The covariance matrix specified in (12) in terms of the volume average of the estimator, of the pairwise velocity
,
![Equation (27)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn27.gif)
Let us first consider a covariance between the pairwise cluster velocities of two cluster pairs, each with respective separations r and r', we will then incorporate the effect of including finite bin sizes in the cluster separations. Using the expression for the mean pairwise cluster velocity, given in (2), the covariance of can be written as
![Equation (28)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn28.gif)
For simplicity in the following derivation, we drop the subscript "h" (denoting halo) from the correlation function, and mass average correlation function,
, denoting them respectively by
. Similarly we use
to denote the halo linear dark matter power spectrum, given in full by
, and the cluster number density ncl is denoted n.
We define an estimator of the volume averaged correlation function equivalently to the estimator of the correlation function
![Equation (29)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn29.gif)
where
![Equation (30)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn30.gif)
The covariance matrix at a given redshift (dropping the subscript a) becomes
![Equation (31)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn31.gif)
The expectation value of the four including noise is
![Equation (32)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn32.gif)
Evaluating the first two terms of the above equation leads to the Gaussian contribution of the covariance matrix
![Equation (33)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn33.gif)
Using the approximation
![Equation (34)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn34.gif)
the Gaussian terms become
![Equation (35)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn35.gif)
The remaining non-Gaussian terms come from the trispectrum (Matarrese et al. 1997)
![Equation (36)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn36.gif)
dropping all the terms proportional to the bispectrum, and four point functions, assuming a Gaussian density distribution.
Evaluating the first term of (36) leads to a non-zero contribution only for a separation with r = 0, proportional to and is therefore not relevant for this work. Similarly, the last term leads to
![Equation (37)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn37.gif)
The only non-zero term, proportional to , gives rise to a non-Gaussian contribution to the covariance, which we will denote as the "Poisson" shot noise term, and can be evaluated using
![Equation (38)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn38.gif)
dropping the term proportional to . Using spherical symmetry
![Equation (39)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn39.gif)
and volume averaging over r and r' leads to
![Equation (40)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn40.gif)
where we have used the integral expression for the Dirac delta function
![Equation (41)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn41.gif)
to calculate the integrals as
![Equation (42)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn42.gif)
The total covariance for the pairwise velocity correlation of cluster pairs of exact separation r and r' can therefore be written as
![Equation (43)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn43.gif)
with
![Equation (44)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn44.gif)
![Equation (45)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn45.gif)
Now we consider the statistics calculated by binning cluster separations in a bin of width . In this case the pairwise velocity estimate is averaged over cluster pairs with separations within the finite bin,
![Equation (46)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn46.gif)
where we again assume spherical symmetry. Volume averaging over a bin of size yields
![Equation (47)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn47.gif)
Using
![Equation (48)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn48.gif)
with
![Equation (49)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn49.gif)
Binning in r translates into replacing the Bessel function with a function related to the bin limits,
![Equation (50)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn50.gif)
Rewriting the volume averaged correlation function in terms of the power spectrum,
![Equation (51)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn51.gif)
and with
![Equation (52)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn52.gif)
the full, angle-averaged covariance for the mean pairwise velocity, excluding measurement error, is given by the sum of a Gaussian cosmic variance and shot noise component plus a Poisson component,
![Equation (53)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn53.gif)
with
![Equation (54)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn54.gif)
![Equation (55)](https://content.cld.iop.org/journals/0004-637X/808/1/47/revision1/apj513614eqn55.gif)
These results are used in Equation (14) in the main text.
Footnotes
- 3
- 4
- 5
N. Battaglia 2015, private communication.