Next Contents Previous


What are the challenges and opportunities for galaxy alignment studies in the nascent era of precision cosmology with deep and large galaxy surveys? Increasingly large datasets with high-quality imaging superior to what the SDSS provided will be produced by on-going surveys 11, and by the even larger projects of the 2020s 12. Not only will these surveys allow for galaxy alignment measurements with improved precision and accuracy, they will also feature weak lensing as a key cosmological probe and thus raise the bar for the performance of intrinsic alignment mitigation. They will be complemented by new spectroscopic redshift surveys 13, which will enable alignment studies (which usually require precise three-dimensional positions of galaxies) for the samples of early-type and emission-line galaxies that are targeted. The advancements on the observational side will be matched by a continued development of algorithms for simulations of structure formation, paired with the rapid evolution of available computational power.

Based on this framework, it is reasonable to expect that the currently good understanding of early-type alignments is going to be consolidated using fainter galaxy samples at higher redshift. Galaxy groups and clusters can be viewed as a high-mass extension to the range of objects for which the tidal stretching paradigm should constitute an accurate description of alignment processes. Also for these comparatively rare objects we anticipate better constraints as data and simulations provide denser sampling of large cosmological volumes, which allows us to test models over a continuous mass range of about three decades from Milky Way-size haloes to massive clusters.

While the linear alignment model works remarkably well on large scales, its extensions to non-linear scales will require further scrutiny. A halo model of alignments as proposed by Schneider & Bridle (2010) is a promising framework, but it is yet to be tested if the assumption that all matter is bound in haloes breaks down on intermediate scales (van Daalen & Schaye 2015), where filamentary structure may play a decisive role. Once a successful model for describing intrinsic alignments down to megaparsec scales exists, it will be desirable to measure higher-order alignment statistics, such as three-point functions. They have to be self-consistently predicted by the model and may in addition help to disentangle intrinsic alignment from weak lensing signals (Semboloni et al., 2008). The current and future high signal-to-noise alignment detections for early-type galaxies, groups, and clusters could in principle be interesting probes of the properties of dark matter and gravity (e.g. Chisari & Dvorkin 2013). However, signals of interest would have to manifest themselves in the redshift dependence or the scale dependence on moderately large scales, whereas the non-linear regime and the overall amplitude of the alignment signals is likely to always be dominated by the effects of highly non-linear physics, baryons, and stochastic processes.

Only in the last few years have hydrodynamic simulations begun to cover cosmological volumes with high enough resolution to allow for measurements of shapes and their alignments (see Kiessling et al., 2015 for details). They will be key to unravelling the link between the alignments of dark matter structures and the visible distribution of stars, as well as elucidating correlations with other observables, such as colour, size, and dynamical state. A work plan for the near future has to include tests of the sensitivity of simulated galaxy shapes and alignments to the choice of simulation code, and to what is referred to as sub-grid physics, i.e. effective descriptions of small-scale processes below the simulation resolution like the impact of supernova explosions and active galactic nuclei on the temperature, distribution, and chemical composition of gas within the galaxy. The implementation of the physics behind these processes will have to mature to a degree that hydrodynamic simulations can simultaneously predict galaxy alignments and basic observables such as the number density of galaxies as a function of their luminosity and the mass-size relation (perhaps via calibration techniques; see e.g. Schaye et al., 2015, Crain et al., 2015). Recent results have highlighted the sensitivity of alignment signals to implementation details of the simulations (Velliscig et al., 2015), but have also yielded promising results in that they show quantitative agreement between observations and simulations (Tenneti et al., 2015). By being able to predict the correct amplitude of alignments, one can begin to answer key issues, such as if alignments are frozen in at the time of galaxy formation, or generated or reset in major merger events, and how these alignments are transferred from the dark matter to the stellar distribution.

For the foreseeable future hydrodynamic simulations will be too computationally expensive to be run in boxes that cover full surveys, and with many realisations. Hence we anticipate that a substantial effort will go into developing statistical or analytic prescriptions (Heymans et al., 2006a, Joachimi et al., 2013a, 2013b) for galaxy alignments in order to paste galaxy properties into pure dark mater simulations, informed by results from smaller hydrodynamic simulations. This is analogous to what is routinely done to include photometric properties of galaxies in mock catalogues based on N-body simulations (Baugh 2006). Since galaxy morphology and colours are strongly correlated, one should expect that incorporating galaxy shapes into the formalism will improve the overall model, which can additionally be tested against observed galaxy shape distributions and alignments. See Kiessling et al., (2015) for a detailed discussion of requirements on future simulations in relation to galaxy alignment studies.

While the observational prospects for elliptical galaxies are good, the situation for spiral galaxies is much more uncertain. To date, there are few (if any) convincing, highly significant detections for any kind of alignment involving the shape of disc galaxies or their spin. Simulations do see signatures of tidal torque theory, or intriguing alternative models (e.g. Libeskind et al., 2013), but these are largely eradicated in observational data by a combination of projection effects and stochastic misalignments between dark matter and stars (Heymans et al., 2006a). Nor is it entirely clear whether elements of large-scale structure like voids can be identified from future high-redshift spectroscopic galaxy samples in a sufficiently clean manner to yield constraints on spin alignments that are readily interpretable.

Since the typical galaxy samples used for cosmic shear surveys are dominated by late-type galaxies, the lack of evidence for intrinsic alignments among them may suggest that a straightforward cosmological analysis is possible without invoking complicated mitigation schemes. However, the decisive quantity in this case is not the signal amplitude, but the level of uncertainty on this signal which, for blue disc galaxies above z ∼ 0.25, is very large. There is currently no clear avenue to change this situation as the forthcoming spectroscopic surveys will target other galaxy types which have more obvious spectral features that facilitate the determination of a redshift. These datasets will only fill in the current gaps for red galaxies, above z ∼ 0.6 and for low luminosities down to Milky Way brightness. This will be different for future observational campaigns required to calibrate photometric redshifts. They will generate spectroscopic samples that are more representative of those found in weak lensing surveys, but are optimised to cover small areas in different parts of the sky to beat down sample variance, which prohibits the measurement of spatial correlations on the scales used for cosmological weak lensing measurements.

Although the data obtained from a weak lensing survey offers the ability to self-calibrate intrinsic alignments (Joachimi & Bridle 2010), it is still highly desirable to have a direct measurement for blue galaxies over relevant redshift and luminosity ranges to either put priors on the intrinsic alignment contamination in the self-calibration analysis or verify that model choices when marginalising over nuisance parameters related to intrinsic alignments are justified. The redshift survey required for this purpose would have to go to similar depths as the weak lensing data, cover a sufficiently large contiguous area to sample galaxy pair separations of tens of megaparsecs with small errors, and obtain an average galaxy number density of the same order as the weak lensing survey to avoid being limited by shot noise on small scales. Realistically, such datasets could only be obtained in the near future by low-resolution spectroscopic surveys 14 or narrow-band photometric surveys 15. These produce redshift estimates with a low catastrophic failure rate and scatter that is an order of magnitude or more smaller than for broad-band photometric surveys, which preserves most of the large-scale alignment signals and avoids confusion with the lensing signal. Figure 16 illustrates how the key survey parameters – area, galaxy number density, and redshift scatter – influence the signal-to-noise achievable for intrinsic alignment correlation functions of the form given in Equation (16).

Figure 16

Figure 16. Relation between survey parameters and signal-to-noise of a galaxy alignment correlation function as in Equation (16). A simple analytic estimate neglecting sample variance was employed. Lines indicate the survey area required to attain the same signal-to-noise as from a spectroscopic galaxy sample akin to BOSS CMASS (in a redshift range of [0.45; 0.65]; position in plot marked by the grey semi-circle) over 1000 deg2. This is shown as a function of the statistical uncertainty in the redshift determination, given by a Gaussian of width σph(1 + z), and of the angular number density, ng, of galaxies with shape measurements of sufficient quantity in the corresponding redshift range. Note the steep rise of the curves above σph ≈ 0.003, which is due to the rapid dilution of information in the alignment statistic because of the increased scatter in redshifts.

The Square Kilometre Array (SKA) 16 will extend galaxy survey astrophysics to radio wavelengths and mark a transition into a new era for large-scale structure cosmology, including the study of galaxy alignments. Galaxy shapes will be determined by the distribution of neutral gas rather than stellar light. The former extends to much larger radii and may therefore not be perfectly correlated with the optical shape, and be subject to a different alignment strength. Both effects can be exploited to separate the lensing and alignment signals in a joint analysis. Moreover, the SKA will produce maps of polarisation and radial velocities across a galaxy which may be used as independent tracers of the gravitational lensing effect (see Kirk et al., 2015 for details). Finally, the full SKA will deliver precise redshift estimates for up to 10 galaxies / arcmin2 over the whole extragalactic sky and will thereby provide exquisite data for alignment measurements among the star-forming disc galaxies to which it is most sensitive.

In any case, the subtle observational signatures of galaxy alignments will remain challenging to measure, and the underlying highly non-linear, baryonic physics-dependent processes challenging to model. However, being at the interface between galaxy formation and evolution on the one side and fundamental cosmology on the other, with potentially considerable impact on both, galaxy alignments are expected to feature prominently in this new era of precision cosmology.


We acknowledge the support of the International Space Science Institute Bern for two workshops at which this work was conceived. We thank E. Brunnstrom for an investigation into alignments in Palomar Sky Survey catalogues, and J. Blazek for stimulating discussions. We are grateful to B. Binggeli, C. Heymans, S. Singh, A. Slosar, A. Tenneti, and I. Trujillo for sharing their data.

BJ acknowledges support by an STFC Ernest Rutherford Fellowship, grant reference ST/J004421/1. MC was supported by the Netherlands organisation for Scientific Research (NWO) Vidi grant 639.042.814. TDK is supported by a Royal Society URF. RM acknowledges the support of NASA ROSES 12-EUCLID12-0004. CS and HH acknowledge support from the European Research Council under FP7 grant number 279396. AK was supported in part by JPL, run under a contract by Caltech for NASA. AK was also supported in part by NASA ROSES 13-ATP13-0019 and NASA ROSES 12-EUCLID12-0004.

11 including the Kilo Degree Survey,; the Dark Energy Survey,; and the Hyper Suprime-Cam Survey, Back.

12 including the Large Synoptic Survey Telescope (LSST Science Collaboration et al., 2009),; the ESA Euclid satellite (Laureijs et al., 2011), and; and the NASA Wide-Field Infrared Survey Telescope (WFIRST, Spergel et al., 2013), Back.

13 for instance with the Dark Energy Spectroscopic Instrument,; the Subaru Prime Focus Spectrograph; as well as Euclid and WFIRST Back.

14 such as PRIMUS, Back.

15 such as PAU,; and J-PAS, Back.

16 Back.

Next Contents Previous