Modeling the 3-point correlation function Felipe Marin Department of Astronomy & Astrophysics University of Chicago arXiv: Felipe Marin Department of Astronomy & Astrophysics University of Chicago arXiv:
06/01/2007Great Lakes Cosmology Workshop 8 Collaborators: Josh Frieman (KICP-Chicago & Fermilab) Josh Frieman (KICP-Chicago & Fermilab) Bob Nichol (ICG, Portsmouth) Bob Nichol (ICG, Portsmouth) Risa Wechsler (KICP-Chicago, now Stanford) Risa Wechsler (KICP-Chicago, now Stanford)
06/01/2007Great Lakes Cosmology Workshop 8 Correlation functions on LSS Galaxy surveys show us that the (luminous) matter does not distribute very smoothly in the Universe From cosmological N-body simulations, we can see that is also not the case for the dark matter How do we get a more quantitative insight? Can we infer DM clustering using galaxies? Galaxy surveys show us that the (luminous) matter does not distribute very smoothly in the Universe From cosmological N-body simulations, we can see that is also not the case for the dark matter How do we get a more quantitative insight? Can we infer DM clustering using galaxies? A. Kravtsov M. Tegmark
06/01/2007Great Lakes Cosmology Workshop 8 N-point statistics One way to achieve this is using spatial N-point correlation functions: measure how more likely is to have certain configurations of N-points in a particular field than in a random distribution. For instance, we can describe the probability that two objects (galaxies, dark matter particles, DM halos, etc.) are found at a distance r: One way to achieve this is using spatial N-point correlation functions: measure how more likely is to have certain configurations of N-points in a particular field than in a random distribution. For instance, we can describe the probability that two objects (galaxies, dark matter particles, DM halos, etc.) are found at a distance r: This defines the two-point correlation function. Along with its Fourier counterpart, the Power Spectrum, have been measured in simulations and galaxy surveys, CMB,etc.
06/01/2007Great Lakes Cosmology Workshop 8 The need for a more complete description It is possible that two distributions have the same 2-point statistics, but they look completely different! 2-point statistics just describe completely only Gaussian Fields. They do not take into account non-spherical morphologies It is possible that two distributions have the same 2-point statistics, but they look completely different! 2-point statistics just describe completely only Gaussian Fields. They do not take into account non-spherical morphologies Sefusatti & Scoccimarro 2005
06/01/2007Great Lakes Cosmology Workshop 8 The three-point correlation function The next order correlation is the three-point correlation function (3PCF): The probability to find 3 objects in a certain triangle configuration: 1 3 2 The value of the 3PCF depends on the overall scale of the triangle, as well as on its shape. Useful to define the reduced 3PCF: The value of the 3PCF depends on the overall scale of the triangle, as well as on its shape. Useful to define the reduced 3PCF: u r r
06/01/2007Great Lakes Cosmology Workshop 8 N-body simulations & mock galaxy catalogs We want to measure & compare the 3PCF of dark matter and galaxies in real & redshift space. We use high-resolution N-body simulations run using ART code (Kravtsov et al. ’97,’04) with concordance LCDM parameters. Can detect dark matter halos of galactic size Two boxes: L120 with 120 Mpc/h on the side & L200 with 200 Mpc/h on the side Redshift space: long-distance observer approximation: peculiar velocities distortions We want to measure & compare the 3PCF of dark matter and galaxies in real & redshift space. We use high-resolution N-body simulations run using ART code (Kravtsov et al. ’97,’04) with concordance LCDM parameters. Can detect dark matter halos of galactic size Two boxes: L120 with 120 Mpc/h on the side & L200 with 200 Mpc/h on the side Redshift space: long-distance observer approximation: peculiar velocities distortions Kravtsov et al 04
06/01/2007Great Lakes Cosmology Workshop 8 From DM to galaxies Kravtsov et al (2004), Conroy, Wechsler & Kravtsov (2006) : V max, of a DM halo is a good indicator of the stellar mass and henceforth, of the luminosity of a galaxy. In order to get luminosities, for both L120 & L200 boxes, the r-band SDSS luminosity function is matched to the cumulative velocity function at the redshift of observation n(>V max,now ) Colors are assigned using the observed relation between local density and SDSS color (Wechsler et al. 2004, Tasitsiomi et al 2004). Kravtsov et al (2004), Conroy, Wechsler & Kravtsov (2006) : V max, of a DM halo is a good indicator of the stellar mass and henceforth, of the luminosity of a galaxy. In order to get luminosities, for both L120 & L200 boxes, the r-band SDSS luminosity function is matched to the cumulative velocity function at the redshift of observation n(>V max,now ) Colors are assigned using the observed relation between local density and SDSS color (Wechsler et al. 2004, Tasitsiomi et al 2004). Conroy,Wechsler & Kravtsov 06
06/01/2007Great Lakes Cosmology Workshop 8 Results: DM vs halos Equilateral triangles Reduced 3PCF for DM particles & halos Jack-knife error bars Halos strongly biased w.r.t. DM particles Strong scale dependence in real space, strong redshift evolution on small scales Features of Q very washed out in redshift space Reduced 3PCF for DM particles & halos Jack-knife error bars Halos strongly biased w.r.t. DM particles Strong scale dependence in real space, strong redshift evolution on small scales Features of Q very washed out in redshift space MWFN 2007
06/01/2007Great Lakes Cosmology Workshop 8 Why real-space Q so different from redshift-space Q? Big effect in observation, in Galaxy biasing
06/01/2007Great Lakes Cosmology Workshop 8 Results: DM vs. halos Shape dependence We measured Q(r,u=2, ), for = degrees Blue line: Biased dark matter Q… see later… The amplitude of Q( ) is higher at elongated configurations: U-shape. We measured Q(r,u=2, ), for = degrees Blue line: Biased dark matter Q… see later… The amplitude of Q( ) is higher at elongated configurations: U-shape. MWFN 2007
06/01/2007Great Lakes Cosmology Workshop 8 Luminosities & Colors Kayo et al. (2004): little or no dependence of Q for SDSS galaxies in color and luminosity, for equilateral triangles Our results agree in general: need more volume? Kayo et al. (2004): little or no dependence of Q for SDSS galaxies in color and luminosity, for equilateral triangles Our results agree in general: need more volume? MWFN 2007
06/01/2007Great Lakes Cosmology Workshop 8 Comparison w/SDSS results. We compare the 3PCF of our boxes to the recent measurements of the SDSS 3PCF by Nichol et al (2006) There’s a good agreement within the errors in general with our L120 box results using V max,now Here we use a much lower resolution than in our previous results: then features of Q( ) are severely attenuated. We compare results with bigger resolution: errors do not get much higher. We compare the 3PCF of our boxes to the recent measurements of the SDSS 3PCF by Nichol et al (2006) There’s a good agreement within the errors in general with our L120 box results using V max,now Here we use a much lower resolution than in our previous results: then features of Q( ) are severely attenuated. We compare results with bigger resolution: errors do not get much higher.
06/01/2007Great Lakes Cosmology Workshop 8 Galaxy Biasing with 3PCF The different 3PCF from Dark Matter and galaxies reflect differences in spatial distributions: galaxy bias Higher-order statistics can provide constrains. On large scales, where rms overdensities are small compared to unity, we can adopt a local bias model. This will affect the values of the correlation functions as well: The different 3PCF from Dark Matter and galaxies reflect differences in spatial distributions: galaxy bias Higher-order statistics can provide constrains. On large scales, where rms overdensities are small compared to unity, we can adopt a local bias model. This will affect the values of the correlation functions as well:
06/01/2007Great Lakes Cosmology Workshop 8 Galaxy Biasing: results Adopting c 1 =b 1 and c 2 =b 2 /b 1, using the JK error covariance matrix we get constrains in these parameters from the 3PCF, 2PCF & overdensities. The three methods have a good agreement in real space, giving c 1 ~1.2 & c 2 ~ -0.2 In redshift space the agreement is not as good, but constrains from 3PCF are consistent with 2dF results: c 1 ~0.9& c 2 ~ -0.3 Adopting c 1 =b 1 and c 2 =b 2 /b 1, using the JK error covariance matrix we get constrains in these parameters from the 3PCF, 2PCF & overdensities. The three methods have a good agreement in real space, giving c 1 ~1.2 & c 2 ~ -0.2 In redshift space the agreement is not as good, but constrains from 3PCF are consistent with 2dF results: c 1 ~0.9& c 2 ~ -0.3
06/01/2007Great Lakes Cosmology Workshop 8 Summary The 3PCF for both galaxies and dark matter has a strong dependence on scale and shape. The redshift space 3PCF is strongly attenuated w.r.t. the real space 3PCF. The galaxy reduced 3PCF shows little dependence on luminosity & color. Our model predictions are in good agreement with the last SDSS measurements On scales of order 10 Mpc/h, a local bias scheme is in reasonable agreement with galaxy and DM distributions. The 3PCF for both galaxies and dark matter has a strong dependence on scale and shape. The redshift space 3PCF is strongly attenuated w.r.t. the real space 3PCF. The galaxy reduced 3PCF shows little dependence on luminosity & color. Our model predictions are in good agreement with the last SDSS measurements On scales of order 10 Mpc/h, a local bias scheme is in reasonable agreement with galaxy and DM distributions.