Redshift-Space Distortions

A galaxy redshift survey does not hand us a map of the Universe; it hands us angles on the sky and a redshift for each object, and we must turn that redshift into a distance. The conversion assumes pure Hubble flow, $cz = H_0 s$ , but every galaxy carries a peculiar velocity on top of the expansion, and the line-of-sight piece of it contaminates the inferred distance. The map we reconstruct in "redshift space" is therefore a distorted image of the true "real-space" map — stretched and squashed along the one direction the survey singles out, the line of sight. That anisotropy looks at first like an error to be scrubbed away. It is instead one of the sharpest tools in cosmology. The reason is dynamical. Peculiar velocities are not random noise; in linear theory they are sourced by the very gravitational potential that grows structure, and the continuity equation ties the divergence of the velocity field directly to the logarithmic growth rate $f$ . Measure how strongly clustering is distorted along the line of sight and you have measured how fast structure is growing, independent of the shape of the power spectrum. This is the Kaiser effect, and on large scales it compresses overdensities radially. On small scales the sign flips: inside virialised halos, random orbital motions of hundreds of km/s smear galaxies into the radial "Fingers of God." Modern surveys — BOSS, eBOSS, DESI, Euclid — decompose the anisotropic clustering into Legendre multipoles and read off the degeneracy-robust combination $f\sigma_8(z)$ redshift bin by redshift bin, turning a nuisance into a direct, near-model-independent probe of the growth of structure and a test of general relativity on the largest scales.

Key takeaways

Galaxy positions inferred from redshifts blend the Hubble flow with line-of-sight peculiar velocities, distorting the observed clustering relative to real space — anisotropically, because only the radial velocity enters.
On large linear scales, coherent infall compresses structures along the line of sight (the Kaiser effect), boosting radial power by a factor set by $\beta = f/b$ .
On small scales, random virial motions inside collapsed halos stretch structures radially into Fingers of God, exponentially damping high- $k$ radial power.
The linear anisotropy encodes the growth rate $f = \mathrm{d}\ln D/\mathrm{d}\ln a \approx \Omega_m(a)^{0.55}$ , a dynamical probe of structure growth and of gravity.
Surveys report the bias-robust combination $f\sigma_8(z)$ , extracted from the multipoles $P_0, P_2, P_4$ ; ΛCDM predicts $f\sigma_8(0) \approx 0.43$ .
RSD complements Baryon Acoustic Oscillations in the same data: BAO measures distances, RSD measures growth, and together they separate dark energy from modified gravity.
Key idea: line-of-sight peculiar velocities distort redshift-inferred positions; in linear theory the amplitude of that coherent distortion is set by the growth rate $f$ , so the anisotropy of galaxy clustering measures $f\sigma_8(z)$ .

Where this lives today

Redshift-space distortion is now a headline driver for the largest spectroscopic surveys ever built. SDSS's BOSS and eBOSS measured $f\sigma_8(z)$ across $z \approx 0.15$ – $1.5$ to a few percent; DESI, with tens of millions of galaxy and quasar redshifts, is sharpening that into a percent-level growth history, and Euclid adds a slitless spectroscopic sample out to $z \sim 2$ . The analyses fit the anisotropic clustering — the multipoles $P_0, P_2, P_4$ or the correlation-function moments — pairing the linear Kaiser term to a nonlinear treatment of the Fingers of God and mode coupling, increasingly from the effective field theory of large-scale structure rather than a phenomenological Gaussian damping. Because RSD and the Alcock–Paczynski effect both produce line-of-sight anisotropy, the fits deliver $f\sigma_8(z)$ , the angular-diameter distance $D_A(z)$ , and $H(z)$ together, jointly with the Baryon Acoustic Oscillation feature in the same catalogue. The stakes are gravity itself: a growth history undershooting the ΛCDM prediction $f \approx \Omega_m(a)^{0.55}$ — hinted at by the mild " $\sigma_8$ / $S_8$ tension" with the Planck-normalised expectation — would point either to a lower clustering amplitude than the CMB implies or to a growth index $\gamma \neq 0.55$ , the smoking gun of modified gravity on cosmological scales.

The physical picture

Picture a spherical shell of galaxies falling coherently toward a central overdensity. On the near side the galaxies move away from us and their redshift is nudged up, placing them farther than they are; on the far side they move toward us and are placed nearer. The net effect flattens the sphere along the line of sight — the overdensity appears squashed radially. That compression is the linear Kaiser signature, and its strength grows with how vigorously matter flows, that is, with the growth rate $f$ . Because only the radial velocity shifts a redshift, the distortion depends entirely on the angle between a Fourier mode and the line of sight: transverse modes record the true real-space clustering, radial modes carry the full velocity signal. Now zoom into a single virialised cluster. Its galaxies orbit with random speeds of several hundred km/s, so their redshifts scatter over tens of Mpc even though the cluster is physically only about a megaparsec across. In the map the cluster is drawn out into a thin spike pointing straight back at the observer — a Finger of God. Large scales get compressed, small scales elongated; redshift-space distortion is the competition between these two regimes, and separating them is what lets a survey isolate the linear growth signal.

Figure

In real space an overdensity is round. In redshift space coherent infall squashes it along the line of sight — the Kaiser flattening — while random virial motions stretch the collapsed core into a radial Finger of God pointing straight at the observer.

The pay-off is that the strength of the linear squashing is not a free parameter — it is the growth rate. As dark energy comes to dominate, the matter fraction $\Omega_m(a)$ falls, growth stalls, and the RSD amplitude $f\sigma_8(z)$ declines. Every spectroscopic survey measures one point on that growth curve — a redshift-resolved consistency test of ΛCDM and gravity — which is the quantity the second figure plots.

Figure

ΛCDM growth history fσ₈(z) for Ωm = 0.315 and σ₈ = 0.811: the observable rises from fσ₈ = 0.43 at z = 0 to a broad peak ≈ 0.48 near z = 0.4, then declines toward high redshift as growth was faster but the amplitude σ₈(z) was smaller — each RSD survey pins one point on this curve.

The core equations

The whole subject starts from one admission: the coordinate we assign a galaxy is not its true position. Converting redshift to distance with the Hubble law places the galaxy at an apparent comoving position $\mathbf{s}$ displaced from the true position $\mathbf{r}$ by the line-of-sight peculiar velocity — a coordinate map, not a small perturbation of the density.

\mathbf{s} = \mathbf{r} + \frac{\hat{\mathbf{r}}\cdot\mathbf{v}}{aH}\,\hat{\mathbf{r}}

where: $\mathbf{s}$ — apparent comoving position in redshift space; $\mathbf{r}$ — true comoving position; $\mathbf{v}$ — peculiar velocity; $\hat{\mathbf{r}}$ — unit vector along the line of sight; $a$ — scale factor; $H$ — Hubble parameter at the observed epoch. Reading: only the projection $\hat{\mathbf{r}}\cdot\mathbf{v}$ enters, so the displacement is purely radial — the map singles out one direction and makes clustering anisotropic. The amplitude of the distortion is governed by how fast structure grows. The linear growth factor $D(a)$ amplifies a perturbation as $\delta(a) = D(a)\,\delta_0$ , and its logarithmic derivative — the growth rate $f$ — is captured across ΛCDM history, to remarkable accuracy, by a single power of the matter density parameter.

f \equiv \frac{\mathrm{d}\ln D}{\mathrm{d}\ln a} \approx \Omega_m(a)^{\gamma}, \qquad \gamma = \tfrac{6}{11} \approx 0.55

where: $f$ — logarithmic growth rate; $D$ — linear growth factor; $\Omega_m(a)$ — matter density parameter at scale factor $a$ ; $\gamma$ — growth index, $\approx 0.55$ in general relativity and different in modified gravity. Reading: at early times $\Omega_m \to 1$ and $f \to 1$ ; as $\Lambda$ takes over, $\Omega_m(a)$ and $f$ both fall — the decline dark energy imprints on structure growth. The growth rate enters the distortion through the linear continuity equation $\nabla\cdot\mathbf{v} = -aHf\,\delta_m$ : matter conservation forces the velocity divergence to track $\dot\delta_m = Hf\delta_m$ , so peculiar velocities are sourced by the growing potential in proportion to $f$ , not scattered at random. Feeding this and the map through number conservation (Derivation A) gives the central result of the field: in the plane-parallel limit the linear redshift-space overdensity picks up an angle-dependent boost. With $\mu$ the cosine of the angle between the wavevector $\mathbf{k}$ and the line of sight, and linear bias $b$ through $\delta_g = b\,\delta_m$ ,

\delta_s(\mathbf{k}) = \left(1 + \beta\mu^2\right)\delta_r(\mathbf{k}), \qquad \beta \equiv \frac{f}{b}

where: $\delta_s$ — redshift-space galaxy overdensity (Fourier mode); $\delta_r$ — real-space galaxy overdensity; $\mu$ — cosine of the angle between $\mathbf{k}$ and the line of sight; $b$ — linear bias; $\beta = f/b$ — the redshift-space distortion parameter. Reading: transverse modes ( $\mu = 0$ ) are undistorted; radial modes ( $\mu = 1$ ) get the maximum boost $1+\beta$ . Squaring the overdensity gives the observable, the anisotropic power spectrum. On small scales the coherent Kaiser boost is multiplied by an exponential Fingers-of-God damping from random virial velocities of dispersion $\sigma_v$ — the phenomenological "dispersion model."

P_s(k,\mu) = \left(1 + \beta\mu^2\right)^2 b^2 P_m(k)\, e^{-k^2\mu^2\sigma_v^2}

where: $P_s$ — anisotropic redshift-space power spectrum; $P_m(k)$ — real-space matter power spectrum; $b^2 P_m \equiv P_r$ — real-space galaxy power; $\sigma_v$ — line-of-sight pairwise velocity dispersion (comoving). Reading: the Kaiser factor enhances large-scale radial power, the Gaussian suppresses small-scale radial power — both regimes in one expression. Because $P_s$ depends on both $k$ and $\mu$ , surveys compress its angular structure into Legendre multipoles. Only three are non-zero for the pure Kaiser term, with exact rational coefficients (Derivation B).

P_0 = \left(1 + \tfrac{2}{3}\beta + \tfrac{1}{5}\beta^2\right)P_r, \quad P_2 = \left(\tfrac{4}{3}\beta + \tfrac{4}{7}\beta^2\right)P_r, \quad P_4 = \tfrac{8}{35}\beta^2 P_r

where: $P_r = b^2 P_m$ — real-space galaxy power; $P_0, P_2, P_4$ — monopole, quadrupole, hexadecapole. Reading: the monopole is boosted, the quadrupole is generated entirely by the distortion (vanishing if $\beta = 0$ ), and the hexadecapole is pure $\beta^2$ . The amplitude-independent ratio $P_2/P_0$ estimates $\beta$ , and combining it with the amplitude yields the bias-free observable $f\sigma_8(z) = (\mathrm{d}\ln D/\mathrm{d}\ln a)\,\sigma_8(z)$ that surveys report (Derivation C), with $\sigma_8(z)$ the rms linear density fluctuation in $8\,h^{-1}\mathrm{Mpc}$ spheres.

Derivation ADeriving the Kaiser formula from the coordinate map

The goal is to show, from first principles, how the innocuous coordinate shift $\mathbf{s} = \mathbf{r} + v_z\hat{\mathbf{z}}/aH$ turns into the angle-dependent factor $(1 + \beta\mu^2)$ that dominates all of large-scale RSD. We work in the distant-observer (plane-parallel) approximation, treating the line of sight as a fixed direction $\hat{\mathbf{z}}$ so that "radial" means "along $z$ ." The only physical inputs are conservation of galaxy number, the Jacobian of the map, linearisation, and the continuity relation between velocity and density.

Step 1. Number conservation. Galaxies are neither created nor destroyed by relabelling positions, so $[1 + \delta_s(\mathbf{s})]\,\mathrm{d}^3s = [1 + \delta_r(\mathbf{r})]\,\mathrm{d}^3r$ . Why: the distortion is a coordinate change, so number is conserved and all the physics is in the volume element.

Step 2. Jacobian of the map. Since only the radial coordinate is displaced, $\mathbf{s} = \mathbf{r} + (v_z/aH)\hat{\mathbf{z}}$ , the volume element changes by $\mathrm{d}^3s/\mathrm{d}^3r = 1 + (1/aH)\,\partial v_z/\partial z$ to leading order. Why: only the displacement gradient along the line of sight matters; transverse gradients are higher order in the plane-parallel limit.

Step 3. Linearise. Keeping first order in the small quantities $\delta$ and $v$ gives $\delta_s \simeq \delta_r - (1/aH)\,\partial v_z/\partial z$ . Why: the redshift-space overdensity is the real-space one plus the compression from the radial-velocity gradient — infall piling galaxies up in redshift.

Step 4. Fourier plus continuity. For irrotational flow $v_z(\mathbf{k}) = i\,aHf\,\delta_m\,k_z/k^2$ , so $-(1/aH)\,\partial v_z/\partial z \to (k_z^2/k^2)f\,\delta_m = f\mu^2\delta_m$ , using $\nabla\cdot\mathbf{v} = -aHf\delta_m$ and $\mu = k_z/k$ . Why: this is where the growth rate enters — the velocity divergence is proportional to $f$ , and its radial projection carries the $\mu^2$ dependence.

Step 5. Add bias. Galaxies trace mass with $\delta_r = b\,\delta_m$ while velocities respond to total matter, so $\delta_s = (b + f\mu^2)\delta_m = b(1 + \beta\mu^2)\delta_m$ with $\beta = f/b$ . Why: bias rescales the density but not the velocity, so the measurable combination is the ratio $\beta = f/b$ , not $f$ itself. Squaring the result gives $P_s(k,\mu) = (1 + \beta\mu^2)^2 b^2 P_m(k)$ , the Kaiser power spectrum, in agreement with Kaiser (1987) and Hamilton (1998).

Go deeper · Derivation B · 1 minThe multipole moments of the redshift-space power spectrum

Derivation BThe multipole moments of the redshift-space power spectrum

The aim is to reduce the two-variable observable $P_s(k,\mu)$ to a handful of one-dimensional functions of $k$ by projecting onto Legendre polynomials, and to compute the exact coefficients $\tfrac{2}{3}, \tfrac{1}{5}, \tfrac{4}{3}, \tfrac{4}{7}, \tfrac{8}{35}$ that appear in every RSD analysis. The projection is just the orthogonality of Legendre polynomials applied to the Kaiser factor expanded in powers of $\mu^2$ , and the monopole, quadrupole, and hexadecapole between them fully specify the linear signal.

Step 1. Expand the Kaiser factor. Write $(1 + \beta\mu^2)^2 = 1 + 2\beta\mu^2 + \beta^2\mu^4$ , a degree-two polynomial in $\mu^2$ . Why: it has non-zero projections only onto $\mathcal{L}_0$ , $\mathcal{L}_2$ , and $\mathcal{L}_4$ , which is why exactly three multipoles survive.

Step 2. Project the monopole. With $P_\ell = \tfrac{2\ell+1}{2}\int_{-1}^{1} P_s\,\mathcal{L}_\ell\,\mathrm{d}\mu$ and $\mathcal{L}_0 = 1$ , the averages $\langle\mu^0\rangle = 1$ , $\langle\mu^2\rangle = \tfrac{1}{3}$ , $\langle\mu^4\rangle = \tfrac{1}{5}$ give $P_0 = (1 + \tfrac{2}{3}\beta + \tfrac{1}{5}\beta^2)P_r$ . Why: the monopole is the angle-average, collecting the mean powers of $\mu$ weighted by the Kaiser coefficients.

Step 3. Project the quadrupole. Using $\mathcal{L}_2 = (3\mu^2 - 1)/2$ , with $\int_{-1}^{1}\mu^2\mathcal{L}_2\,\mathrm{d}\mu = \tfrac{4}{15}$ and $\int_{-1}^{1}\mu^4\mathcal{L}_2\,\mathrm{d}\mu = \tfrac{8}{35}$ , we get $P_2 = \tfrac{5}{2}(2\beta\cdot\tfrac{4}{15} + \beta^2\cdot\tfrac{8}{35})P_r = (\tfrac{4}{3}\beta + \tfrac{4}{7}\beta^2)P_r$ . Why: the constant term projects to zero against $\mathcal{L}_2$ , so the quadrupole is generated entirely by the distortion — no distortion, no quadrupole.

Step 4. Project the hexadecapole. Only the $\mu^4$ term survives against $\mathcal{L}_4 = (35\mu^4 - 30\mu^2 + 3)/8$ , with $\int_{-1}^{1}\mu^4\mathcal{L}_4\,\mathrm{d}\mu = \tfrac{16}{315}$ , giving $P_4 = \tfrac{9}{2}\beta^2\cdot\tfrac{16}{315}\,P_r = \tfrac{8}{35}\beta^2 P_r$ . Why: pure $\beta^2$ , so the hexadecapole is the cleanest fingerprint of the quadratic Kaiser term and a strong lever on $\beta$ . Direct numerical Legendre integration reproduces $\tfrac{2}{3}, \tfrac{1}{5}, \tfrac{4}{3}, \tfrac{4}{7}, \tfrac{8}{35}$ exactly, confirming the coefficients used in the core-equations block.

Go deeper · Derivation C · 1 minWhy the growth rate collapses to fσ₈ as the robust observable

Derivation CWhy the growth rate collapses to fσ₈ as the robust observable

The purpose is to explain, algebraically, why RSD papers never quote $f$ , $b$ , or $\sigma_8$ separately but always the product $f\sigma_8(z)$ . The root cause is that galaxy clustering measures amplitudes, and every amplitude in the Kaiser multipoles is a product of the normalisation $\sigma_8$ with either bias or growth — so the individual factors are degenerate and only certain products are observable. We factor the power spectrum into amplitude and shape and track where $\sigma_8$ , $b$ , and $f$ appear.

Step 1. Factor the amplitude. Write $P_m(k) = \sigma_8^2\,\hat{P}_m(k)$ , splitting the rms amplitude $\sigma_8$ from a unit-normalised shape $\hat{P}_m$ that RSD does not constrain. Why: the shape is fixed by the transfer function and inflation, so the only free amplitude the survey sees is $\sigma_8$ , always appearing squared with $b$ or $f$ .

Step 2. Expose the entanglement. The monopole becomes $P_0 \propto (b^2 + \tfrac{2}{3}fb + \tfrac{1}{5}f^2)\sigma_8^2$ , i.e. a sum of the products $(b\sigma_8)^2$ , $(b\sigma_8)(f\sigma_8)$ , and $(f\sigma_8)^2$ . Why: neither $b$ nor $f$ nor $\sigma_8$ can be pulled out alone — the clustering amplitude only knows about the two combinations $b\sigma_8$ and $f\sigma_8$ .

Step 3. Isolate growth with the quadrupole. The quadrupole $P_2 \propto (\tfrac{4}{3}fb + \tfrac{4}{7}f^2)\sigma_8^2$ carries the velocity contribution, so combining $P_0$ , $P_2$ , and $P_4$ solves for $b\sigma_8$ and $f\sigma_8$ jointly. Why: the extra angular information in the quadrupole breaks the monopole's single-amplitude degeneracy, but only down to the two products, not the three separate factors.

Step 4. Report the robust combination. The surviving growth observable is $f\sigma_8(z) = (\mathrm{d}\ln D/\mathrm{d}\ln a)\,\sigma_8(z)$ , independent of the unknown linear bias amplitude. Why: because it is bias-free and directly proportional to the growth rate, $f\sigma_8(z)$ is the near-model-independent quantity that tests ΛCDM and gravity redshift by redshift. This reduces to the Kaiser multipoles when $\sigma_8^2\hat{P}_m = P_m$ , and underlies every BOSS, eBOSS, and DESI growth measurement.

Limiting cases

The redshift-space power spectrum interpolates between physically distinct regimes as the angle $\mu$ , the scale $k$ , and the epoch are varied.

Case	Result	Meaning
Transverse modes ( $\mu \to 0$ )	$P_s(k, \mu{=}0) = P_r(k)$	no radial velocity gradient; recovers the true real-space power
Radial linear modes ( $\mu \to 1$ )	$P_s(k, \mu{=}1) = (1+\beta)^2 P_r(k)$	maximal Kaiser boost; for $\beta = 0.28$ a factor $1.64$
Small-scale nonlinear	$P_s \propto e^{-k^2\mu^2\sigma_v^2}$	Fingers of God — virial dispersion $\sigma_v$ damps high- $k$ radial power
No distortion ( $\beta \to 0$ )	$P_2 = P_4 = 0$ , $P_0 = P_r$	quadrupole and hexadecapole vanish; only the isotropic monopole survives
Matter domination ( $\Omega_m \to 1$ )	$f = \Omega_m(a)^{0.55} \to 1$	growth unsuppressed at high $z$ ; dark energy lowers $f$ and $f\sigma_8$ later

Common trapRedshift-space distortions are just noise to be corrected away

They are not noise, and they do not average out. It is tempting to treat the peculiar-velocity smearing as a nuisance to subtract in order to recover the "real" positions — but the anisotropy is a coherent, deterministic signal sourced by the same potential that grows structure, and it carries the growth rate $f$ that nothing else in the survey measures directly. Two traps hide inside this one. First, the redshift-space overdensity is related to the real-space one by a coordinate transformation, not a small additive correction — you cannot "clean" it without a model. Second, the Kaiser factor involves $\beta = f/b$ , so growth is entangled with bias: a naive reading gives $\beta$ , not $f$ , and only $f\sigma_8$ is robust. And none of it is trustworthy until the small-scale Fingers of God are modelled or marginalised, since their radial damping leaks into the same multipoles that carry the linear signal.

Worked examples

1 · The growth amplitude fσ₈ today (fully worked)

Take the Planck ΛCDM parameters $\Omega_{m,0} = 0.315$ , $\Omega_\Lambda = 0.685$ , and $\sigma_8 = 0.811$ . Compute the growth rate $f$ and the RSD observable $f\sigma_8$ at $z = 0$ , then say how they change by $z = 1$ .

Step 1. Evaluate the growth rate at $z = 0$ , where $\Omega_m(a{=}1) = \Omega_{m,0} = 0.315$ . Using $f = \Omega_m^{\gamma}$ with $\gamma = 0.55$ : $\ln f = 0.55\ln(0.315) = 0.55 \times (-1.155) = -0.635$ , so $f(0) = e^{-0.635} = \mathbf{0.530}$ .

Step 2. Multiply by the amplitude: $f\sigma_8(0) = 0.530 \times 0.811 = \mathbf{0.430}$ — the red marker in the second figure, matching low-redshift survey measurements.

Step 3. Move to $z = 1$ : the matter fraction climbs to $\Omega_m(1) = \Omega_{m,0}(1{+}z)^3/E^2$ with $E^2 = 0.315\times 8 + 0.685 = 3.205$ , giving $\Omega_m(1) = 2.520/3.205 = 0.786$ , so $f(1) = 0.786^{0.55} = \mathbf{0.876}$ — growth was faster when matter dominated.

Step 4. But the amplitude was smaller: $\sigma_8(1) = \sigma_8(0)\,D(1)/D(0) \approx 0.811 \times 0.607 = 0.492$ , so $f\sigma_8(1) = 0.876 \times 0.492 = \mathbf{0.431}$ . Faster growth and lower amplitude nearly cancel between $z = 0$ and $z = 1$ , which is why the $f\sigma_8(z)$ curve is broad and gently peaked rather than steep. f(0) = 0.53, fσ₈(0) = 0.43; f(1) = 0.88, fσ₈(1) = 0.43.

2 · Kaiser boost, multipole ratio, and the Finger-of-God scale

A BOSS-like sample at $z = 0$ has bias $b = 1.9$ in a cosmology with $f = 0.53$ . Find $\beta$ , the monopole Kaiser boost, the quadrupole-to-monopole ratio, and the comoving length over which a cluster of dispersion $\sigma_v = 500\ \mathrm{km\,s^{-1}}$ is smeared (take $H_0 = 67\ \mathrm{km\,s^{-1}\,Mpc^{-1}}$ ).

Step 1. Distortion parameter: $\beta = f/b = 0.53/1.9 = \mathbf{0.279}$ . Because the galaxies are biased ( $b > 1$ ), the fractional distortion is smaller than $f$ alone would suggest.

Step 2. Monopole boost: $1 + \tfrac{2}{3}\beta + \tfrac{1}{5}\beta^2 = 1 + 0.186 + 0.0156 = \mathbf{1.20}$ — the angle-averaged amplitude is lifted about 20% above real space purely by coherent infall.

Step 3. Quadrupole-to-monopole ratio: numerator $\tfrac{4}{3}\beta + \tfrac{4}{7}\beta^2 = 0.372 + 0.0446 = 0.417$ , so $P_2/P_0 = 0.417/1.20 = \mathbf{0.35}$ . A quadrupole of roughly a third of the monopole is the observable anisotropy that fixes $\beta$ .

Step 4. Finger-of-God scale: $\ell = \sigma_v/H_0 = 500/67 = \mathbf{7.5\ Mpc}$ (equivalently $5.0\ h^{-1}\mathrm{Mpc}$ ). A cluster physically $1$ – $2\ \mathrm{Mpc}$ across is stretched into a radial spike $10$ – $15\ \mathrm{Mpc}$ long — the reason linear analysis must be restricted to large scales or the damping explicitly modelled. β = 0.28, monopole boost 1.20, P₂/P₀ = 0.35, FoG length ≈ 7.5 Mpc.

Physics of the portrait

Redshift-space distortion converted an observational annoyance into a precision measurement, and it carries the name of the theorist who wrote it down cleanly.

Origins · 1 minNick Kaiser and how this was discovered

Nick Kaiser

1954–2023 · large-scale structure and gravitational lensing

Nicholas "Nick" Kaiser was among the most influential theorists of late-twentieth-century cosmology, a physicist whose short, decisive papers repeatedly turned messy observations into clean statistical probes. Trained at Cambridge, he worked at the Canadian Institute for Theoretical Astrophysics, the University of Hawaii's Institute for Astronomy, and finally the École Normale Supérieure in Paris. His 1984 work on biased galaxy formation explained why galaxies cluster more strongly than the underlying mass — the origin of the bias factor $b$ that RSD must contend with. Three years later, in "Clustering in real and redshift space" (1987), he derived the linear compression $(1 + \beta\mu^2)$ that bears his name, showing that the anisotropy of galaxy clustering is not a defect of redshift surveys but a direct measurement of the growth rate of structure. He went on to co-invent the standard method for measuring weak gravitational lensing (Kaiser–Squires inversion and KSB shear estimation) and to lead the Pan-STARRS survey telescope. The through-line of his work is a single instinct: that the statistical distortions in a survey are signals about gravity and cosmology if you have the wit to model them. He died in 2023, his redshift-space formula still the backbone of every growth-rate measurement from BOSS to DESI.

History

1972

— Jackson describes the radial elongation of clusters in redshift maps, later named Fingers of God

1987

— Kaiser derives the linear compression (1 + βμ²) of clustering in redshift space

1992

— Hamilton recasts the Kaiser effect in the correlation function for direct measurement

2001

— the 2dF Galaxy Redshift Survey measures β ≈ 0.43, turning RSD into a growth probe

2005

— Linder popularizes the growth index γ ≈ 0.55 as a sharp test of gravity

Going further

Texts. Dodelson, Modern Cosmology, 2nd ed., Ch. 9–11 (perturbations, growth, and redshift-space observables) · Mo, van den Bosch & White, Galaxy Formation and Evolution, Ch. 6 and 15 (linear growth, peculiar velocities, and clustering statistics) · Carroll & Ostlie, An Introduction to Modern Astrophysics, 2nd ed., Ch. 30 (structure formation) · Binney & Tremaine, Galactic Dynamics, 2nd ed. (virial velocity dispersions) · Peebles, The Large-Scale Structure of the Universe (foundations of the velocity–density link).

Primary literature. Kaiser, N. (1987), MNRAS 227, 1 ("Clustering in real and redshift space") · Hamilton, A. J. S. (1998), "Linear redshift distortions: a review" · Jackson, J. C. (1972), MNRAS 156, 1P (Fingers of God) · Peacock et al. (2001), Nature 410, 169 (2dF measurement of β) · Linder, E. V. (2005), Phys. Rev. D 72, 043529 (growth index γ).

Provenance. The Kaiser factor $(1+\beta\mu^2)^2$ and the multipole coefficients $\tfrac{2}{3}, \tfrac{1}{5}, \tfrac{4}{3}, \tfrac{4}{7}, \tfrac{8}{35}$ are derived here from the plane-parallel coordinate map, linear continuity, and direct Legendre integration, and were re-verified numerically; they match Kaiser (1987) and Hamilton (1998). The growth index $\gamma = 6/11 \approx 0.545$ follows from the linear growth ODE (Dodelson Ch. 9). The growth-history figure uses $\Omega_{m,0} = 0.315$ , $\sigma_8 = 0.811$ with the growth factor integrated as $D(a) \propto H(a)\int_0^a \mathrm{d}a'/(a'H(a'))^3$ , giving $f\sigma_8(0) = 0.43$ and a peak $\approx 0.48$ near $z \approx 0.4$ — standard ΛCDM values, no survey-specific fit reproduced.

Prerequisites. Density Perturbations and the Density Contrast · Linear Growth of Structure · Transfer Function and the Matter Power Spectrum · Jeans Instability and Cosmological Perturbation Regimes. Leads to. Baryon Acoustic Oscillations · The ΛCDM Concordance Model · Dark Energy and the Cosmological Constant.

Ask the tutor: Ask me to walk through how the quadrupole-to-monopole ratio pins down β for a real survey, or how a measured fσ₈(z) curve would reveal modified gravity.

Why does the distortion depend only on the angle to the line of sight?

Because only the radial component of a galaxy's peculiar velocity changes its redshift, and the redshift is the only thing we convert into a distance. A velocity purely across the sky leaves the inferred distance untouched, so it produces no distortion. A Fourier mode is therefore distorted in proportion to how nearly radial it is: the cosine $\mu$ between wavevector and line of sight controls everything, entering as $\mu^2$ because the effect comes from the gradient of a projected velocity. Transverse modes ( $\mu = 0$ ) record pristine real-space power; radial modes ( $\mu = 1$ ) carry the full $(1+\beta)^2$ boost.

How is the Kaiser effect different from the Fingers of God if both point along the line of sight?

They share a direction but nothing else. The Kaiser effect is large-scale, linear, and coherent: whole regions fall together toward a growing overdensity, and that organised infall compresses structures along the line of sight, enhancing radial power. Fingers of God are small-scale, nonlinear, and random: inside a virialised halo galaxies orbit every which way at hundreds of km/s, and those incoherent motions smear the halo into an elongated radial spike, suppressing radial power. The two push in opposite senses — squashing at large scales, stretching at small scales — with Kaiser compression on tens-of-Mpc scales handing over to Fingers of God on Mpc scales.

What exactly is β, and why is it not the same as the growth rate f?

The distortion parameter is $\beta = f/b$ , the growth rate divided by the linear galaxy bias. It appears rather than $f$ because the Kaiser boost acts on the galaxy density contrast, and galaxies are biased tracers ( $\delta_g = b\,\delta_m$ ), while the velocities respond to the total matter. The density part of the redshift-space overdensity scales with $b$ and the velocity part with $f$ , so their ratio $\beta = f/b$ is what the anisotropy alone measures. Recovering $f$ needs $b$ independently, which clustering amplitude cannot supply — which is why the field settled on the bias-free product $f\sigma_8$ instead.

Why is fσ₈ the quantity everyone quotes rather than f or σ₈ separately?

Because galaxy clustering measures amplitudes, and every amplitude in the redshift-space power spectrum is a product of the fluctuation normalisation $\sigma_8$ with either the bias or the growth rate. The monopole depends on $(b\sigma_8)^2$ , $(b\sigma_8)(f\sigma_8)$ , and $(f\sigma_8)^2$ , the quadrupole on $(fb)\sigma_8^2$ and $f^2\sigma_8^2$ . No manipulation separates $f$ from $\sigma_8$ or $b$ ; the data constrain only the products $b\sigma_8$ and $f\sigma_8$ . Since $f\sigma_8$ is directly proportional to the growth rate and free of the unknown bias amplitude, it is the near-model-independent observable that tests structure growth — and hence dark energy and gravity — redshift by redshift.

How does the growth rate carry information about dark energy?

Through the matter fraction $\Omega_m(a)$ . In general relativity $f \approx \Omega_m(a)^{0.55}$ , so while matter dominates, $\Omega_m \approx 1$ and $f \approx 1$ : structure grows unimpeded. Once dark energy takes over, the expansion accelerates, $\Omega_m(a)$ falls, and structure assembly slows — $f$ and therefore $f\sigma_8$ decline. How $f\sigma_8(z)$ turns over across cosmic time constrains how much dark energy there is and how its density evolves (its equation of state). And because a modified theory of gravity can match the same expansion history while predicting a different growth index $\gamma$ , a growth curve that disagrees with $\gamma = 0.55$ would flag new gravitational physics.

Does the plane-parallel (distant-observer) approximation ever break down?

Yes, and modern surveys go beyond it. The clean $(1+\beta\mu^2)$ derivation treats the line of sight as a single fixed direction, excellent when pair separations are much smaller than the distance to the objects. For wide, deep surveys covering much of the sky, a galaxy pair can straddle angles where "the line of sight" differs between the two, introducing wide-angle corrections and curved-sky geometry; at the largest scales, general-relativistic and lensing contributions to the number counts enter too. Analyses handle these with local-line-of-sight multipole estimators and wide-angle expansions. They are corrections to, not replacements for, the Kaiser picture, which remains the leading-order result.