Skip to main content
Sandbox Physics
Redshift-space distortions

How a redshift moves a galaxy

Sandbox Physics ·

A galaxy's redshift records its distance plus its velocity along the line of sight, so the map built from redshifts shifts each galaxy by (1+z) v∥/H(z)(1+z)\,v_\parallel/H(z) along that line. This chapter derives that shift, shows why coherent infall boosts each density wave by a factor 1+fμ21+f\mu^2, and works through the numbers at redshift 0.5.

The chapter follows Why do galaxy maps point at us?. You need the Hubble law, Taylor expansion to first order and complex exponentials. The aims are to derive the redshift-space mapping, connect the infall velocity to the growth rate, derive the linear Kaiser boost for one plane wave and test it numerically.

From redshift to comoving distance

In an expanding universe the comoving distance to a source at cosmological redshift zz is

χ(z)=∫0zc dz′H(z′).\chi(z)=\int_0^z\frac{c\,dz'}{H(z')}.

A galaxy with a peculiar velocity v∥v_\parallel along the line of sight has an extra Doppler shift. For speeds much smaller than light, the observed redshift is

1+zobs=(1+z)(1+v∥c).1+z_{\mathrm{obs}}=(1+z)\Bigl(1+\frac{v_\parallel}{c}\Bigr).

The survey assigns the distance χ(zobs)\chi(z_{\mathrm{obs}}). Expanding to first order in v∥/cv_\parallel/c, with zobs−z=(1+z)v∥/cz_{\mathrm{obs}}-z=(1+z)v_\parallel/c and dχ/dz=c/Hd\chi/dz=c/H, gives

s∥=x∥+(1+z) v∥H(z).s_\parallel=x_\parallel+\frac{(1+z)\,v_\parallel}{H(z)}.

Here x∥x_\parallel is the true comoving distance and s∥s_\parallel the distance in redshift space. In vector form, with n^\hat{\mathbf{n}} the unit vector along the line of sight, s=x+(1+z)(v⋅n^)n^/H\mathbf{s}=\mathbf{x}+(1+z)(\mathbf{v}\cdot\hat{\mathbf{n}})\hat{\mathbf{n}}/H. Hamilton's review writes the same mapping as s=r+vs=r+v, with distances measured in velocity units. Hamilton 1998, §4.2, eq. 4.27

Two approximations are hidden here. The line of sight is treated as one fixed direction for the whole region, the plane-parallel approximation, which holds when the region is small compared with its distance. And the expansion is first order in the velocity, which is excellent because galaxy speeds are about a thousandth of the speed of light.

Worked example: 300 kilometres per second at redshift 0.5

Take a flat universe with matter fraction Ωm=0.31\Omega_m=0.31 today. Then

H(z)H0=Ωm(1+z)3+1−Ωm=0.31×3.375+0.69=1.318\frac{H(z)}{H_0}=\sqrt{\Omega_m(1+z)^3+1-\Omega_m}=\sqrt{0.31\times3.375+0.69}=1.318

at z=0.5z=0.5. Writing H0=100 h km s−1 Mpc−1H_0=100\,h\ \mathrm{km\,s^{-1}\,Mpc^{-1}}, the expansion rate is 131.8 h km s−1 Mpc−1131.8\,h\ \mathrm{km\,s^{-1}\,Mpc^{-1}}. A line-of-sight speed of 300 km s−1300\ \mathrm{km\,s^{-1}} therefore moves the galaxy by

1.5×300131.8=3.41 h−1Mpc.\frac{1.5\times300}{131.8}=3.41\,h^{-1}\mathrm{Mpc}.

With h=0.68h=0.68 that is 5.0 megaparsecs. A rich cluster with a velocity dispersion of 700 km s−1700\ \mathrm{km\,s^{-1}} is smeared over about 8 h−1Mpc8\,h^{-1}\mathrm{Mpc}, several times its true size.

Coherent infall and the growth rate

On large scales the velocity field follows the density field. Write the density contrast as δ\delta and use the linear continuity equation. In terms of the comoving displacement Ψ\boldsymbol{\Psi} of each mass element from its starting point, linear theory gives δ=−∇⋅Ψ\delta=-\nabla\cdot\boldsymbol{\Psi}, and Ψ\boldsymbol{\Psi} grows in proportion to the growth factor DD. The peculiar velocity is the rate of change of the displacement, v=a dΨ/dt=aHfΨ\mathbf{v}=a\,d\boldsymbol{\Psi}/dt=aHf\boldsymbol{\Psi}, where

f≡dln⁡Ddln⁡af\equiv\frac{d\ln D}{d\ln a}

is the growth rate. Insert this velocity into the mapping. The factor (1+z)/H=1/(aH)(1+z)/H=1/(aH) cancels aHaH, and the redshift-space shift becomes simply

s=x+f (Ψ⋅n^) n^.\mathbf{s}=\mathbf{x}+f\,(\boldsymbol{\Psi}\cdot\hat{\mathbf{n}})\,\hat{\mathbf{n}}.

The shift is the displacement along the line of sight, multiplied by the growth rate. In a flat universe with a cosmological constant, f≃Ωm(z)0.55f\simeq\Omega_m(z)^{0.55}, which gives 0.7570.757 at z=0.5z=0.5. Integrating the linear growth equation directly gives the same value to 0.04%. Guzzo et al. 2008

The Kaiser boost for one plane wave

Take a single density wave δ=Acos⁡(k⋅q)\delta=A\cos(\mathbf{k}\cdot\mathbf{q}) with small amplitude AA. Its displacement is Ψ=−A ksin⁡(k⋅q)/k2\boldsymbol{\Psi}=-A\,\mathbf{k}\sin(\mathbf{k}\cdot\mathbf{q})/k^2, as you can check by taking minus the divergence. Let μ=k^⋅n^\mu=\hat{\mathbf{k}}\cdot\hat{\mathbf{n}} be the cosine of the angle between the wave and the line of sight.

The real-space position is x=q+Ψ\mathbf{x}=\mathbf{q}+\boldsymbol{\Psi} and the redshift-space position is s=q+Ψ+f(Ψ⋅n^)n^\mathbf{s}=\mathbf{q}+\boldsymbol{\Psi}+f(\boldsymbol{\Psi}\cdot\hat{\mathbf{n}})\hat{\mathbf{n}}. The density of points in either space has Fourier amplitude proportional to the average of e−ik⋅xe^{-i\mathbf{k}\cdot\mathbf{x}} over all particles. Expanding the exponential to first order in AA,

⟨e−ik⋅s⟩≈⟨e−ik⋅q(1−i k⋅Ψ−ifμ k Ψ⋅n^)⟩.\bigl\langle e^{-i\mathbf{k}\cdot\mathbf{s}}\bigr\rangle\approx\bigl\langle e^{-i\mathbf{k}\cdot\mathbf{q}}\bigl(1-i\,\mathbf{k}\cdot\boldsymbol{\Psi}-if\mu\,k\,\boldsymbol{\Psi}\cdot\hat{\mathbf{n}}\bigr)\bigr\rangle.

For this wave Ψ⋅n^=μ Ψ⋅k^\boldsymbol{\Psi}\cdot\hat{\mathbf{n}}=\mu\,\boldsymbol{\Psi}\cdot\hat{\mathbf{k}}, so the last term is fμ2f\mu^2 times the middle one. The amplitude of the wave in redshift space is therefore

δs(k)=(1+fμ2) δ(k).\delta_s(\mathbf{k})=\bigl(1+f\mu^2\bigr)\,\delta(\mathbf{k}).

Waves along the line of sight, μ=1\mu=1, are amplified by 1+f1+f. Waves across it, μ=0\mu=0, are untouched. For galaxies that trace matter with a bias bb, the galaxy wave has amplitude b δb\,\delta while the velocity still follows the matter, so δs=(b+fμ2)δ\delta_s=(b+f\mu^2)\delta. This is Kaiser's result, and its square gives the power spectrum (b+fμ2)2P(k)(b+f\mu^2)^2P(k). Hamilton 1998, §4.2.1, eqs. 4.35 and 4.36

A numerical check with particles

The course reference does this test without using the Kaiser formula anywhere. It places 64364^3 particles on a lattice, moves them by one plane wave with A=10−3A=10^{-3}, applies the redshift mapping with f=0.8f=0.8, and sums e−ik⋅se^{-i\mathbf{k}\cdot\mathbf{s}} directly over the particles.

Table 2.1 Redshift-space amplitude of one plane wave over its real-space amplitude, for f=0.8f=0.8.
Wave indexμ\muMeasured ratio1+fμ21+f\mu^2
(0, 0, 4)11.800001.8
(4, 0, 4)0.7071.400001.4
(4, 0, 0)01.000001
(2, 3, 5)0.8111.526321.52632

The largest relative error is 3×10−73\times10^{-7}, the size of the neglected second-order terms. The derivation and the code agree.

What the boost does to a map

Every wave along the line of sight grows, so structures look denser along that direction: a wall seen edge-on becomes thinner and sharper. Galaxies in the Lab sit in a synthetic catalogue with b≈2b\approx2 and f=0.76f=0.76. The boost along the line of sight is then (b+f)2/b2=1.9(b+f)^2/b^2=1.9 in power, and the squashing is visible in the correlation map as contours flattened along the vertical axis.

Lab tasks

  1. Choose Coherent infall only and watch a filament that lies across the line of sight. Describe how it changes when the loop slides galaxies into redshift space.
  2. Turn the line of sight by dragging the observer to 4040 degrees. Check that galaxies still move only along the new line of sight.
  3. Raise the random motion to 700 km s−1700\ \mathrm{km\,s^{-1}} and estimate the length of the longest finger in megaparsecs from the scale bar. Compare with the worked example.

Exercises

  1. At redshift 1 with Ωm=0.31\Omega_m=0.31, how far does a line-of-sight speed of 300 km s−1300\ \mathrm{km\,s^{-1}} move a galaxy?
  2. Why does sideways motion not distort the map at all?
  3. Show that a wave at μ=0.5\mu=0.5 is amplified by 1+f/41+f/4, and evaluate it for f=0.76f=0.76.
  4. For a biased tracer with b=2b=2 and f=0.76f=0.76, by what factor is the power of a wave along the line of sight multiplied?
Answers

For the first exercise, H(1)/H0=0.31×8+0.69=1.780H(1)/H_0=\sqrt{0.31\times8+0.69}=1.780, so the shift is 2×300/178.0=3.37 h−1Mpc2\times300/178.0=3.37\,h^{-1}\mathrm{Mpc}. The factor 1+z1+z and the larger expansion rate nearly cancel between redshift 0.5 and 1.

For the second, the redshift only measures motion along the line of sight. Sideways motion leaves the redshift unchanged, and the angular position on the sky does not depend on distance, so neither coordinate moves.

For the third, μ2=1/4\mu^2=1/4 gives 1+fμ2=1+f/41+f\mu^2=1+f/4, which is 1.191.19 for f=0.76f=0.76.

For the fourth, the amplitude grows from bδb\delta to (b+f)δ(b+f)\delta, so the power grows by (2.76/2)2=1.90(2.76/2)^2=1.90.

Sources

  1. Hamilton 1998, §4.2, eq. 4.27
  2. Guzzo et al. 2008

Keep reading

What this page links to directly

See this series on the content map →

Redshift-space distortions · Reading path

From a moving galaxy to a growth measurement

  1. ArticleWhy do galaxy maps point at us?
  2. You are hereHow a redshift moves a galaxy
  3. ArticleMultipoles of a squashed map
  4. ArticleGrowth or a distance error?
  5. ArticleMeasure multipoles from a frozen catalogue
  6. ArticleHow Guzzo et al. measured growth from redshift distortions in 2008
Something broken, confusing, or missing on this page?Tell us
What kind of feedback?

Sent with this page’s address, its Lab ID, the site language and your device type. Please leave out personal details. Feedback is kept for up to 12 months.Privacy policy