Cosmic microwave background anisotropy (CMB anisotropy)

Cosmic microwave background anisotropies are direction-dependent perturbations of the relic radiation temperature and polarization.

A finite visibility width averages photon sources over emission time and radial position. For a Gaussian visibility profile, a plane-wave radial phase has the displayed damping factor; its power has the square of this amplitude factor. An acoustic source also varies with frequency $k{c}_s$, so its rapid oscillations are averaged. This averaging is distinct from Silk damping, which smooths the perturbations by photon diffusion before decoupling. Angular projection and source evolution prevent a universal isotropic damping coefficient based only on radial width.

A statistically isotropic Cosmic microwave background anisotropy has $⟨a_{{\ell }_1{m}_1}{a}_{{\ell }_2{m}_2}{a}_{{\ell }_3{m}_3}⟩={\mathcal{G}}_{m_1{m}_2{m}_3}^{{\ell }_1{\ell }_2{\ell }_3}{b}_{{\ell }_1{\ell }_2{\ell }_3}$, where $\mathcal{G}$ is the integral of the three spherical harmonics. The reduced CMB bispectrum removes this angular geometric factor and retains the primordial and transfer-function dependence. A Sachs-Wolfe projection of a constant bispectrum is an analytic large-angle example.

For an isotropic, full-sky temperature map with a fixed bispectrum template $b=f_{\mathrm{NL}}{b}^1$, the inverse-variance-weighted cubic statistic has normalization $F=\frac{1}{6}\sum (b^1{)}^2{\mathcal{G}}^2/(C_1{C}_2{C}_3)$ over ordered triples. It is unbiased under the linear template relation and has Gaussian cosmic variance $1/F$. The factor $1/6$ counts the six cross-triple Wick contractions. A removed monopole cancels the nine internal-pairing terms; incomplete sky coverage or anisotropic noise generally require a linear correction.

In a full-sky cubic bispectrum estimator, pairwise Gaussian covariance within one triple gives a sum of Gaunt integrals with opposite $m$ indices. The spherical harmonic addition theorem turns their pair into the constant $(2\ell +1)/(4\pi )$; its integral against $Y_{LM}$ vanishes for $L>0$. A zero temperature monopole removes $L=0$. Thus only the six pairings connecting the two triples survive in the Gaussian variance; isotropic weights are essential for the cancellation.

Linear cosmological transfer maps the primordial bispectrum to the reduced CMB bispectrum. A Fourier representation of the momentum delta function and three Rayleigh plane-wave expansions separate the radial transfer integrals from a Gaunt integral. The six angular factors and three Fourier measures give $(4\pi {)}^6/(2\pi {)}^9=(2/\pi {)}^3$. The observer-position phase cancels by momentum conservation; the remaining radial position is an auxiliary integration variable.

For a constant primordial bispectrum and large-angle transfer function ${\Delta }_{\ell }(k)=j_{\ell }(kR)/5$, the spherical Bessel product integral reduces the radial projection to ${\int }_0^1{r}^{L+2}dr+{\int }_1^{\infty }{r}^{-L-1}dr$. For $L>0$ this equals $1/(L+3)+1/L$, giving the displayed reduced CMB bispectrum. All powers of the distance $R$ cancel. Under common large-multipole scaling it behaves as ${\ell }^{-4}$ at fixed shape; this is angular scale invariance, not a constant angular bispectrum. The all-monopole case $L=0$ has a logarithmically divergent radial tail and is excluded.

The CMB lensing potential is the line-of-sight projection of gravitational potential that remaps observed CMB directions by the deflection angle $\nabla \psi$.

The CMB angular power spectrum decomposes the variance of temperature or polarization anisotropy by angular multipole. Its large-angle plateau and acoustic peaks encode primordial perturbations and photon-baryon evolution.

For a scale-invariant potential spectrum with ${\mathcal{A}}_{\Phi }=k^3{P}_{\Phi }/(2{\pi }^2)$ constant, the matter-era Sachs-Wolfe effect gives $C_{\ell }=(4\pi {\mathcal{A}}_{\Phi }/9)\int j_{\ell }(x{)}^{2}dx/x$. The logarithmic spherical Bessel square integral then gives $\ell (\ell +1)C_{\ell }=2\pi {\mathcal{A}}_{\Phi }/9$. This plateau applies to large-angle temperature anisotropies, neglecting acoustic evolution and late evolving potentials; it is not a constant $C_{\ell }$ spectrum.

Cosmic variance is the sampling uncertainty caused by observing only one realization of a cosmological random field. For a Gaussian isotropic full-sky temperature field and an ideal noiseless estimate, $\mathrm{Var}({\hat{C}}_{\ell })=2C_{\ell }^2/(2\ell +1)$ because there are only $2\ell +1$ independent real angular degrees of freedom.

Thomson scattering of a local radiation quadrupole generates linear cosmic microwave background polarization. Its acoustic phase follows the photon velocity and is shifted by one quarter-period relative to the monopole temperature oscillation.

The parity-odd part of linear sky polarization is represented by the pseudoscalar potential $P_B$. At linear order an axisymmetric scalar cosmological perturbation gives no physical B mode. Tensor or vector sources and lensing conversion of E-mode polarization are outside that scalar-source statement.

The parity-even part of linear sky polarization can be represented by a scalar potential $P_E$ through $Q\pm iU={ð}_{\pm }^2(P_E\pm i{P}_B)$. It is naturally separated from B-mode polarization by the parity of its spherical multipoles. Linear scalar cosmological perturbations generate E modes; the potential has an unobservable low-multipole kernel.

For a scalar photon quadrupole emitted at distance ${\chi }_{\ast }$, the meridian-basis source is $Q\pm iU\propto {\Theta }_2{e}^{-ik{\chi }_{\ast }\mu }(1-{\mu }^2)$. Integrating its scalar-potential equation and removing the low-multipole kernel of polarization potentials yields $P_E\propto {\sum }_{\ell \ge 2}(-i{)}^{\ell }(2\ell +1){\Theta }_2{j}_{\ell }(k{\chi }_{\ast })P_{\ell }(\mu )/(k{\chi }_{\ast }{)}^2$, with no scalar $P_B$. The apparent $k^{-2}$ divergence lies in discarded low multipoles; $j_{\ell }(x)=O(x^{\ell })$ makes the physical small-$x$ limit regular.

Twice-applied spin-raising or lowering annihilates the ordinary spherical-harmonic multipoles $\ell =0,1$. Polarization potentials therefore have a four-dimensional monopole/dipole ambiguity on the sphere. For an axisymmetric mode it is $A+B\cos \theta$. Setting these components to zero defines the observable $\ell \ge 2$ potentials without changing $Q$ or $U$.

For Fourier wavevector $\mathbf{k}$ and photon direction $\hat{\mathbf{p}}$, the temperature perturbation is expanded as

For a fixed angular order $m$, write $\Theta ={\sum }_{\ell }{C}_{\ell }{q}_{\ell }{Y}_{\ell m}$. Streaming by $ik\mu$ contributes $ik(C_{\ell +1}/C_{\ell })a_{\ell +1,m}{q}_{\ell +1}+ik(C_{\ell -1}/C_{\ell })a_{\ell ,m}{q}_{\ell -1}$. Thus rescaling the expansion coefficients also rescales the hierarchy couplings and source amplitudes. Replacing $C_{\ell }=(-i{)}^{\ell }\sqrt{(2\ell +1)/(8\pi )}$ by ${\tilde{C}}_{\ell }=(-i{)}^{\ell }\sqrt{2\pi /(2\ell +1)}$ requires ${\tilde{q}}_{\ell }=(2\ell +1)q_{\ell }/(4\pi )$; the temperature field is unchanged. This prevents mixing incompatible photon temperature multipole conventions.

Let $⟨F⟩=(4\pi {)}^{-1}\int Fd\Omega$ and $f=\bar{f}(ϵ)-ϵ{\bar{f}}^{\prime }(ϵ)\Theta (\mathbf{e})$ with comoving energy $ϵ=aE$. If ${ϵ}^4\bar{f}$ vanishes at both integration endpoints, integration by parts gives $-\int {ϵ}^4{\bar{f}}^{\prime }dϵ=4\int {ϵ}^3\bar{f}dϵ$. The kinetic stress-energy tensor then gives ${\delta }_{\gamma }=4⟨\Theta ⟩$, $v_{\gamma }^i=3⟨\Theta e^i⟩$, $\delta P=\bar{\rho }{\delta }_{\gamma }/3$, and ${\pi }^{ij}=4\bar{\rho }⟨\Theta (e^i{e}^j-{\delta }^{ij}/3)⟩$. If $\Pi =-\pi$ is used, the last expression changes sign. Frequency independence of $\Theta$ is essential to this common temperature description.

A direction-dependent fractional photon temperature change. For a thermal spectrum, $f(ϵ)=\bar{f}(ϵ)-\Theta ϵ{\bar{f}}^{\prime }(ϵ)$ is its linear expansion at fixed comoving energy. For a nonthermal spectrum the same ansatz is a brightness dilation, rather than necessarily a thermodynamic temperature. In unweighted Legendre polynomial multipoles, $\Theta ={\sum }_{\ell }(-i{)}^{\ell }{\Theta }_{\ell }{P}_{\ell }$, the photon density, velocity and anisotropic stress are ${\delta }_{\gamma }=4{\Theta }_0$, $v_{\gamma }=-{\Theta }_1$ and ${\Pi }_{\gamma }=-3{\Theta }_2/5$.

The photon monopole is the direction-averaged fractional temperature perturbation. In Newtonian gauge its gravitationally observable effective temperature is the Sachs-Wolfe combination ${\Theta }_0+\Psi$.

In Newtonian gauge, ${\Theta }_0+\Psi$ combines the intrinsic photon temperature perturbation with the gravitational redshift at emission. With constant potentials and negligible baryon loading it oscillates as $\cos (k{r}_s)$ for adiabatic initial conditions.

For a constant growing adiabatic mode with curvature convention $\mathcal{R}=-\varphi -\mathcal{H}({\varphi }^{\prime }+\mathcal{H}\varphi )/[4\pi G{a}^2(\bar{\rho }+\bar{P})]$, ${\varphi }_{\mathrm{rad}}=-2\mathcal{R}/3$ and ${\varphi }_{\mathrm{mat}}=-3\mathcal{R}/5$. The superhorizon photon continuity equation conserves ${\Theta }_0-\varphi$. With radiation initial value ${\Theta }_0=\mathcal{R}/3$, it gives ${\Theta }_{0,\mathrm{mat}}=2\mathcal{R}/5$ and $({\Theta }_0+\psi {)}_{\mathrm{mat}}=-\mathcal{R}/5$ when $\psi =\varphi$.

The Sachs-Wolfe effect is the large-angle CMB temperature anisotropy produced by intrinsic photon-temperature perturbations and gravitational redshift at last scattering. For an adiabatic mode during matter domination, ${\Theta }_0+\Psi =\Psi /3$ on superhorizon scales.

The Doppler CMB anisotropy is the line-of-sight velocity contribution from the last-scattering plasma. With the convention used here it contributes $-\hat{\mathbf{n}}\cdot {\mathbf{v}}_e$ to the observed fractional temperature perturbation.

The integrated Sachs-Wolfe effect is the CMB temperature shift produced when the gravitational potentials traversed by a photon evolve with time. Its line-of-sight source is proportional to ${\Phi }^{\prime }+{\Psi }^{\prime }$.

The Rees-Sciama effect is the microwave temperature shift produced by time-dependent gravitational potentials in nonlinear structure. A photon gains energy entering a potential well and loses it leaving; the shifts do not cancel when the well changes during transit. To leading order in its effect on the photon spectrum, this is a frequency-independent thermodynamic temperature shift, unlike the thermal Sunyaev-Zeldovich effect. A nonlinear potential can evolve even during matter domination, when linear growing-mode potentials are constant.

The photon dipole represents the bulk velocity of the radiation. In tight coupling to baryons, $v_b=-3{\Theta }_1$ under the convention used here.

The photon quadrupole is the leading anisotropy that sources linear polarization through Thomson scattering.

Before recombination, frequent Thomson scattering couples photons, electrons, and baryons into an acoustic fluid.

For the velocity convention $v_{\gamma }=-{\Theta }_1$ and $\dot{\tau }<0$, momentum-conserving Thomson scattering gives ${\dot{v}}_b+\mathcal{H}{v}_b+k\psi =(\dot{\tau }/R)(v_b-v_{\gamma })$. Here $R$ is the baryon loading parameter. Its collision force is opposite to the photon force when both are weighted by their enthalpies.

The inertia ratio of the photon-baryon fluid is $R=3{\bar{\rho }}_b/(4{\bar{\rho }}_{\gamma })$. Photon momentum is weighted by the radiation enthalpy $4{\bar{\rho }}_{\gamma }/3$, so this ratio controls momentum exchange, the common sound speed and the offset of the acoustic oscillator. Background conservation gives $R\propto a$ and $\dot{R}=\mathcal{H}R$.

Thomson scattering is the low-energy elastic scattering of electromagnetic radiation by a free charged particle. Before cosmological recombination, repeated photon-electron scattering couples photons to the baryon velocity and damps higher photon multipoles.

The tight-coupling approximation expands the photon-baryon Boltzmann equations in powers of $k/\Gamma$ and $\mathcal{H}/\Gamma$, where $\Gamma$ is the Thomson scattering rate.

With Fourier convention $\mathbf{v}=i\mathbf{k}\theta$, the velocity divergence is $-k^2\theta$. Thomson drag in the velocity equation is the conformal scattering rate $q=a{n}_e{\sigma }_T$ times the velocity difference. Taking its divergence and dividing the whole divergence equation by $-k^2$ leaves the same rate multiplying the difference of velocity potentials; an additional $k^{-2}$ is not retained. Weighting the photon Euler equation by $R=4{\bar{\rho }}_{\gamma }/(3{\bar{\rho }}_b)$ and adding the baryon equation cancels the drag. In tight coupling the remaining equation is $(1+R){\theta }^{\prime }+\mathcal{H}\theta +R({\delta }_{\gamma }/4-{\sigma }_{\gamma })=0$, with sound speed $R/[3(1+R)]$.

For $\Theta (\mu )={\sum }_{\ell }(-i{)}^{\ell }{\Theta }_{\ell }{P}_{\ell }(\mu )$ and a Thomson collision operator omitting polarization feedback, the quadrupole relaxes with rate $(9/10){\Gamma }_T$, where ${\Gamma }_T=-{\tau }^{\prime }>0$. Balancing its leading streaming source gives ${\Theta }_2=20k{\Theta }_1/(27{\Gamma }_T)$. This differs from photon quadrupole in tight coupling with polarization; the two formulas expand different collision operators.

With $v_b=v_{\gamma }$ at leading order and $c_s^2=1/[3(1+R)]$, the photon continuity equation and total Euler equation give ${\ddot{\delta }}_{\gamma }+\dot{R}{\dot{\delta }}_{\gamma }/(1+R)+k^2{\delta }_{\gamma }/[3(1+R)]=4\ddot{\varphi }+4\dot{R}\dot{\varphi }/(1+R)-4k^2\psi /3$. The baryon loading parameter changes inertia and shifts the gravitational equilibrium. The leading equations omit diffusion damping.

A damped mode $\delta =e^{-k^2/k_D^2}(A\cos q+B\sin q)$ does not alone imply oscillations in its ensemble power spectrum. Averaging its modulus squared gives the displayed formula. Equal independent initial variances, $P_A=P_B$ and $P_{AB}=0$, erase the oscillatory variance through ${\cos }^{2}q+{\sin }^{2}q=1$. A coherent initial phase, as supplied by a regular growing adiabatic mode, selects one common oscillator phase and retains acoustic peaks. Its leading density and velocity oscillations are a quarter-period apart, and angular projection also preserves their correlated contribution.

For baryon-loading ratio $R=3{\rho }_b/(4{\rho }_{\gamma })$, the photon-baryon sound speed is $c_s^2=1/[3(1+R)]$.

The velocity slip is the difference between the bulk velocities of photons and baryons. Thomson scattering suppresses it at a rate $\Gamma (1+R^{-1})$, where $R=3{\rho }_b/(4{\rho }_{\gamma })$ and $\Gamma =-\dot{\tau }>0$. Its first correction in the tight-coupling approximation produces heat-conduction damping of acoustic waves.

For the expansion $\Theta (\mu )={\sum }_{\ell }(-i{)}^{\ell }{\Theta }_{\ell }{P}_{\ell }(\mu )$ without a $(2\ell +1)$ factor, the leading polarization relation is $E_2=-\sqrt{6}{\Theta }_2/4$. The photon Boltzmann hierarchy then gives ${\Theta }_2=-8k{\Theta }_1/(9\dot{\tau })$. This small photon quadrupole supplies the shear-viscosity part of Silk damping.

When gravity and expansion can be neglected and $R$ is constant, the photon monopole satisfiesThe first damping contribution comes from photon-baryon velocity slip; the second comes from the photon quadrupole. This is a local form of Silk damping.

For the simplified quadrupole equation ${\sigma }_{\gamma }^{\prime }+4k^2{\theta }_{\gamma }/15=-{\sigma }_{\gamma }/{\tau }_c$, neglecting its time derivative and gravitational driving gives ${\sigma }_{\gamma }=-{\tau }_c{\delta }_{\gamma }^{\prime }/5$. A radiation-dominated inertia then obeys ${\delta }_{\gamma }^{\prime \prime }+4{\tau }_c{k}^2{\delta }_{\gamma }^{\prime }/15+k^2{\delta }_{\gamma }/3=0$. On slowly varying subhorizon scales the amplitude is damped by $\exp [-(2/15)k^2\int {\tau }_{c}d\tau ]$. This coefficient belongs to that truncated collision model, rather than the polarization-inclusive Silk damping equation.

Removing the first derivative from the truncated photon diffusion equation gives the displayed transformed oscillator. Constant mean free time changes its oscillation frequency only at second order. For a slowly varying mean free time and subhorizon wavelengths, derivative corrections to the phase and prefactor are small; the leading solution is an acoustic sine/cosine multiplied by $\exp [-k^2/k_D^2]$, where $k_D^{-2}=2\int {\tau }_{c}d\tau /15$. Dropping the derivative correction requires slow variation as well as a small mean free time; it is not an exact solution for arbitrary time dependence.

The comoving sound horizon is the distance traveled by an acoustic wave,

For last scattering well inside matter domination, $c_s=c/\sqrt{3}$ gives physical sound radius $R_s=2c{a}_{\ast }^{3/2}/(\sqrt{3}{H}_0\sqrt{{\Omega }_m})$. Neglecting the early conformal time in the observer distance, an open dust cosmology has transverse comoving last-scattering distance $r_{\ast }=2c/(H_0{\Omega }_m)$. Thus ${\theta }_s=R_s/(a_{\ast }{r}_{\ast })=\sqrt{{\Omega }_m{a}_{\ast }/3}$ and $\ell \sim \pi /{\theta }_s$. Radiation, baryon sound-speed evolution and acoustic phase shifts refine this estimate.

// Target: cosmology.bigb

Cosmic microwave background diffusion damping is the suppression of small-angular-scale anisotropy as photons random-walk out of overdense regions during the finite-width recombination epoch. It produces an approximately exponential damping tail in the Cosmic microwave background power spectrum at multipoles $\ell \gtrsim 1000$.

Adiabatic initial conditions perturb every species by the same local time shift, so their relative number-density ratios are initially unperturbed. Photon acoustic modes then begin predominantly as displacement modes with cosine phase.

Equivalently, $\delta {\rho }_i/{\dot{\bar{\rho }}}_i$ is independent of the component $i$. The non-adiabatic pressure perturbation then vanishes, and the comoving curvature perturbation is conserved on superhorizon scales when anisotropic stress and gradient terms can be neglected.

For separately conserved barotropic components, an adiabatic cosmological perturbation has equal ${\delta }_N/(1+w_N)$. Pressureless matter and a $w=-1/3$ string component therefore have ${\delta }_s=2{\delta }_c/3$ on superhorizon scales. In the supplied perfect-fluid synchronous equations, neglecting spatial gradients gives $S^{\prime \prime }+2\mathcal{H}{S}^{\prime }=0$ for $S={\delta }_s-2{\delta }_c/3$. Both $S$ and its independent decaying-mode amplitude must vanish to select the growing adiabatic solution. At finite wavenumber, pressure gradients generate departures from this leading long-wavelength relation.

Cosmic microwave background acoustic peaks arise from photon-baryon oscillations caught at successive extrema at recombination. Their approximate phase is set by $k{r}_s=n\pi$ and their angular positions measure the sound horizon relative to the distance to last scattering.

For an outward sightline, Fourier velocity $v_b^i=i{k}^i{\theta }_b$, and sudden last scattering, intrinsic photon temperature projects with a Spherical Bessel function, while the Doppler term projects with its derivative. Density and velocity acoustic sources are approximately a quarter cycle out of phase. Their amplitude is suppressed by photon diffusion, giving a power damping envelope $e^{-2k^2/k_D^2}$. Projection maps the sound scale to peak spacing approximately $\pi D_{\ast }/r_s$ and the damping scale to angular multipoles of order $k_D{D}_{\ast }$.

Adiabatic initial conditions select coherent cosine phases of the photon-baryon acoustic oscillator. Its extrema at last scattering satisfy $k{r}_s\simeq n\pi$ and project to $\ell \simeq k{\chi }_{\ast }$. Thus ${\ell }_n\simeq n\pi {\chi }_{\ast }/r_s({\eta }_{\ast })$, with comoving sound horizon $r_s=\int d\eta /\sqrt{3(1+R)}$. Projection, driving and finite last-scattering width modify the exact peak locations.

In the absence of collisions, a photon phase-space density is constant along its geodesic. Linearizing this statement gives a first-order transport equation for the direction-dependent temperature perturbation.

The photon brightness perturbation is the fractional frequency-integrated directional energy perturbation. For a small blackbody temperature shift $\Theta =\delta T/T$, $f_1=-q{f}_0^{\prime }(q)\Theta$. Integration by parts gives $\Delta =4\Theta$. Its direction average is the photon density contrast. This temperature identification assumes a perturbation of blackbody form; the integrated brightness definition itself is more general.

// Target: cosmology.bigb

For collisionless photons in synchronous gauge, their perturbed momentum obeys $q^{\prime }=-q{h}_{ij}^{\prime }{n}^i{n}^j/2$. Terms in $q^{\prime }{\partial }_q{f}_1$ and direction deflection times angular derivatives of $f_1$ are second order. Integrating the remaining linear Collisionless Boltzmann equation over $q^{3}dq$ gives the displayed equation, because $\int q^4{f}_0^{\prime }dq=-4\int q^3{f}_{0}dq$.

// Target: cosmology.bigb

For synchronous gauge metric $d{s}^2=a^2[-d{\tau }^2+({\delta }_{ij}-h_{ij})d{x}^{i}d{x}^j]$, conserved-background comoving momentum obeys $q^{\prime }=q{h}_{ij}^{\prime }{n}^i{n}^j/2$. In the linear Collisionless Boltzmann equation, insert $\Delta =-4f_1/(q{f}_0^{\prime })$ to obtain the displayed source. Using ${\delta }_{ij}+h_{ij}$ instead reverses both signs; it does not change the physics. The scalar contraction is $h_{ij}^{\prime }{n}^i{n}^j=(h^{\prime }-h_s^{\prime })/3+{\mu }^2{h}_s^{\prime }$.

For a photon in $d{s}^2=a^2[-d{\tau }^2+({\delta }_{ij}+h_{ij})d{x}^{i}d{x}^j]$, the null geodesic condition cancels the homogeneous $2\mathcal{H}{p}^0$ term in the derivative of $q=a^2{p}^0$. The remaining first-order geodesic equation gives the displayed momentum change. Its direction is constant at zeroth order, so first-order redshift and brightness sources can be evaluated on the unperturbed ray; direction deflection multiplied by an already first-order anisotropy is second order.

Expanding the direction-dependent photon temperature in Legendre polynomials turns its transport equation into an infinite hierarchy. Free streaming couples each multipole ${\Theta }_{\ell }$ only to its neighbours ${\Theta }_{\ell -1}$ and ${\Theta }_{\ell +1}$.

With $B_i^{(\pm )}=i{B}^{(\pm )}{m}_i^{(\pm )}/\sqrt{2}$ and the same convention for baryon velocity, choose ${\Theta }^{(\pm )}={\sum }_{\ell \ge 1}(-i{)}^{\ell }\sqrt{2\pi /(2\ell +1)}{\Theta }_{\ell }^{(\pm )}{Y}_{\ell ,\pm 1}$. The spherical-harmonic streaming recurrence givesThis is the temperature-only Thomson scattering hierarchy without polarization feedback, with optical depth to the observer satisfying $\dot{\tau }<0$. A photon multipole normalization change changes the denominators and the dipole-source prefactor. The lower coupling vanishes at $\ell =1$, so no vector monopole occurs.

In Newtonian gauge in cosmology, the Free-streaming photon Boltzmann equation is $\dot{\Theta }+\mathbf{e}\cdot \nabla \Theta =\dot{\varphi }-\mathbf{e}\cdot \nabla \psi$. Angular averaging and the photon angular temperature moments give the displayed continuity equation. Its dipole yields the photon Euler equation, with the sign of the anisotropic stress term determined by the stress convention.

The dipole member of the photon Boltzmann hierarchy is the photon Euler equation. In Newtonian gauge and without collisions,

The line-of-sight solution integrates the gravitational source along the unperturbed photon trajectory. In Fourier space the integrating factor $e^{ik\mu \eta }$ accounts for free streaming between emission and observation.

In the negative-spatial-perturbation convention, a matter-era scalar solution $h=h_s+C=A{k}^{1/2}{\tau }^2/4$ gives the displayed exact photon brightness perturbation. Integration gives emission bracket ${\Delta }_{\ast }+iA{k}^{-1/2}\mu {\tau }_{\ast }-A{k}^{-3/2}$ times the propagation phase, plus observer monopole $A{k}^{-3/2}$ and dipole $-iA{k}^{-1/2}\mu {\tau }_0$. Under ordinary adiabatic growing-mode initial conditions and $k{\tau }_{\ast }\ll 1$, the emission gravitational term dominates higher angular multipoles. Arbitrary freely specified initial brightness need not obey this dominance.

With instantaneous photon decoupling and tight-coupling initial brightness, integrate the synchronous photon brightness equation along the unperturbed ray $\mathbf{x}(\tau )={\mathbf{x}}_0-\mathbf{n}({\tau }_0-\tau )$. The direction $\mathbf{n}$ here is photon propagation, opposite to the observer-to-source sky direction. Emission terms are evaluated at the emitter, not the observer. The synchronous metric integral combines endpoint gravitational redshifts and evolving-potential effects when expressed in Newtonian gauge; it is not solely the Integrated Sachs-Wolfe effect.

// Target: cosmology.bigb

Use $h_{ij}{n}^i{n}^j=h/3+({\mu }^2-1/3)h_s$. Two integrations by parts of the anisotropic source give the exact Cosmic microwave background anisotropy amplitude $W_{\ast }[{\delta }_{\gamma }/4+3i\mu {\delta }_{\gamma }^{\prime }/(4k)+i\mu (h^{\prime }-h_s^{\prime })/(2k)+h_s^{\prime \prime }/(2k^2){]}_{\ast }+[i\mu h_s^{\prime }/(2k)-h_s^{\prime \prime }/(2k^2){]}_0-{\int }_{\ast }^{0}W[(h^{\prime }-h_s^{\prime })/6-h_s^{\prime \prime \prime }/(2k^2)]d\tau$. The emission velocity was eliminated using ${\delta }_{\gamma }^{\prime }+4ik\cdot v/3+2h^{\prime }/3=0$. The local observer bracket is a monopole plus a dipole and can be discarded only after explicitly restricting to the measured higher angular multipoles. This identity fixes endpoint signs and integral factors without importing a different scalar-metric convention.

At linear order in Newtonian gauge, free streaming, gravitational redshift, and Thomson scattering combine into a first-order transport equation for the photon temperature perturbation. The collision term drives the radiation toward its monopole plus the electron bulk-velocity dipole.

The optical depth from conformal time $\eta$ to observation at ${\eta }_0$ isThe factor $e^{-\tau (\eta )}$ is the probability that a photon travels from $\eta$ to the observer without another scattering.

The visibility functionis the probability density for the conformal time of a CMB photon's last scattering.

This probability density describes last scattering during cosmological recombination and is built from the optical depth, rather than naming the recombination process itself.

The CMB line-of-sight solution separates sharply localized last-scattering sources, weighted by the cosmological visibility function, from integrated gravitational sources weighted by $e^{-\tau }$ along the photon trajectory.

Ignoring the last-scattering angular source and later scattering, the tensor-induced temperature anisotropy is $\Theta =-\frac{1}{2}{\int }_{{\eta }_{\ast }}^{{\eta }_0}{\dot{h}}_{ij}(\eta ,{\mathbf{x}}_0-({\eta }_0-\eta )\mathbf{e})e^i{e}^{j}d\eta$. An instantaneous cosmological visibility function alone leaves a scattering source at last scattering; leading tensor tight coupling supplies the additional reason to neglect it.