Spatial Curvature of the Universe

Spatial curvature is, unusually for cosmology, a nearly closed observational question. The flat prior in standard $\Lambda$CDM analyses rests on the CMB acoustic scale: Planck's first temperature peak near multipole $\ell\approx220$, with baryon acoustic oscillations in galaxy surveys (BOSS, eBOSS, DESI), gives $\Omega_{k,0}$ consistent with zero at the $10^{-3}$ level. The debate is subtler than that. The CMB primary anisotropies alone actually prefer a mildly closed geometry ($\Omega_k\approx-0.04$) because of a lensing-amplitude anomaly in the Planck spectrum; adding BAO and CMB lensing pulls it firmly back to flat, and that tension is one thread in the broader concordance discussion alongside the Hubble tension. DESI's BAO program now measures the transverse and radial rulers across $0.1<z<3.5$ precisely enough to constrain curvature independently of the CMB. On the theory side, measured flatness remains the empirical anchor of inflation: a confirmed nonzero curvature near the $10^{-4}$ cosmic-variance floor would be a landmark, matching inflation's tiny generic residual or breaking the simplest models.

Freeze the Universe at one instant and strip the expansion away. What remains is a three-dimensional space of uniform curvature, and there are only three kinds. On a sphere — positive curvature — a triangle's angles sum to more than $180^\circ$, initially parallel great circles eventually cross, and the space has finite volume yet no edge. On the flat plane the angles sum to exactly $180^\circ$ and parallels never meet. On a saddle — negative curvature — the angles fall short of $180^\circ$, parallels diverge, and the space runs on forever. The spatial slices of the cosmos are the three-dimensional versions of these surfaces, and the sign of $k$ tells you which.

Because the geometry is uniform, one length characterises it: the curvature radius $R_0=a_0/\sqrt{|k|}$. As the Universe expands that radius stretches with $a$, so any fixed intrinsic curvature is diluted and space looks locally flatter with time — which is why the present near-flatness is a puzzle rather than a coincidence. And the chain is causal: mass-energy curves spacetime, so the energy density relative to the expansion rate sets whether space is closed, flat, or open — curvature reads off the total energy budget. The observational statement is sharper still. Curvature bends the null geodesics that carry light, changing how large a distant object of known physical size appears. If a standard ruler at high redshift looks bigger than flat-space geometry predicts, $S_k$ has been compressed by positive curvature and space is closed; if smaller, $S_k$ has been stretched by negative curvature and space is open. This is precisely how the microwave-background acoustic scale becomes the tightest constraint we have on the shape of the cosmos.

Everything begins with the metric. The Friedmann–Lemaître–Robertson–Walker line element is the unique form for a homogeneous, isotropic spacetime, with the curvature constant $k$ inside its spatial part. In reduced-circumference coordinates $r$ is chosen so a sphere at comoving radius $r$ has proper area $4\pi a^2 r^2$, and curvature appears as the factor $1/(1-kr^2)$ on the radial term alone.

where:

Reading: only the radial term is modified. At fixed $r$ angles and transverse areas are Euclidean; all the curvature is in how radial distance accumulates. An equivalent, cleaner form uses the comoving geodesic distance $\chi$ as radial coordinate, so the radial term is pure $\mathrm{d}\chi^2$ and all curvature sits in the transverse function $S_k(\chi)$ — the form in which distances are transparent.

where:

Reading: $S_k(\chi)$ is the comoving size that one radian subtends at distance $\chi$. In flat space it is just $\chi$; curvature bends it away from that line. The transverse function is the heart of the matter: sine for positive curvature (sizes converge, recontracting past the antipode), the identity for flat space, and hyperbolic sine for negative curvature (sizes run away exponentially).

where:

Reading: for $\chi\ll R_0$ all three collapse to $S_k\approx\chi$ — small patches look flat; deviations grow only as the distance probed approaches $R_0$. Now the dynamics. The first Friedmann equation, from Einstein's equations for the FLRW metric, carries the curvature term explicitly as $-kc^2/a^2$ beside the density term — the bridge from geometry to matter.

where:

Reading: fix $H$; a larger $\rho$ must be balanced by a more negative curvature term, i.e. by $k>0$. Density and curvature are two sides of one equation. Setting the curvature term to zero defines the density that produces exact flatness — the critical density, to which everything is then referred as a ratio.

where:

Reading: with $H_0\approx67.4$ km s⁻¹ Mpc⁻¹, $\rho_{c,0}\approx8.5\times10^{-27}$ kg m⁻³ — about five hydrogen atoms per cubic metre, a stunningly low bar the Universe meets. Dividing the Friedmann equation by $H_0^2$ today turns $k$ into the observable density parameter $\Omega_0\equiv\rho_0/\rho_{c,0}$ — the most useful equation on the page, making the sign of $k$ equal the sign of $(\Omega_0-1)$.

where:

Reading: $\Omega_0>1\Leftrightarrow k>0$ (closed); $\Omega_0=1\Leftrightarrow k=0$ (flat); $\Omega_0<1\Leftrightarrow k<0$ (open). Measure the density budget and you measure the geometry. It is convenient to promote the curvature term to a density parameter of its own, $\Omega_{k,0}$, so the closure relation reads as a clean sum equal to one.

where:

Reading: curvature behaves as if it were a fluid with $\Omega_{k,0}=1-\Omega_0$; flatness means the real components ($\Omega_m+\Omega_r+\Omega_\Lambda$) already sum to one. Finally, the curvature parameter is not static: at any epoch it scales as $1/(a^2H^2)$, and because $H$ falls faster than $1/a$ in the radiation and matter eras, curvature grows in importance with time — the seed of the flatness problem.

where:

Reading: radiation domination ($H^2\propto a^{-4}$) gives $\Omega_k\propto a^2$; matter domination ($H^2\propto a^{-3}$) gives $\Omega_k\propto a$. Either way it was minuscule early and swells late — unless dark energy freezes it out.

The three geometries and the dynamics of curvature, side by side:

1 · How curved could space still be? (fully worked)

Planck with BAO gives $\Omega_{k,0}=0.0007\pm0.0019$. Take the largest curvature still allowed at $1\sigma$, $|\Omega_{k,0}|=0.0019$, and find the minimum curvature radius $R_0$. Compare it to the observable Universe. Use $H_0=67.4$ km s⁻¹ Mpc⁻¹.

Step 1. Equation (7) rearranges to the curvature radius $R_0=a_0/\sqrt{|k|}=(c/H_0)/\sqrt{|\Omega_{k,0}|}$ — so the smaller the allowed $|\Omega_{k,0}|$, the larger the minimum $R_0$.

Step 2. The Hubble distance is $c/H_0 = 299\,792.458/67.4 = \mathbf{4448\ Mpc}$.

Step 3. Divide by $\sqrt{|\Omega_{k,0}|}=\sqrt{0.0019}=0.0436$:

Step 4. The comoving radius of the observable Universe is about $14$ Gpc, so $R_0\gtrsim100$ Gpc is roughly seven times larger than everything we can see. Over the observable patch $\chi/R_0\lesssim0.14$, so $S_k(\chi)$ departs from $\chi$ by at most $\sim1\%$. The data force any curvature radius to dwarf the observable Universe, which is exactly why space looks flat.

2 · The volume of a closed universe

If the Universe were closed at the $1\sigma$ edge, $R_0\approx100$ Gpc, what would its total spatial volume be? Derive the closed-universe volume first.

Step 1. From metric (2) with $a=a_0$, the volume element is $\mathrm{d}V=a_0^3\,S_k^2(\chi)\sin\theta\,\mathrm{d}\chi\,\mathrm{d}\theta\,\mathrm{d}\phi$. The angular integral is the standard solid angle, $\int_0^{2\pi}\mathrm{d}\phi\int_0^\pi\sin\theta\,\mathrm{d}\theta=4\pi$.

Step 2. For $k>0$ the radial coordinate runs to the antipode, $\chi\in[0,\pi R_0]$, where $S_k$ returns to zero and space closes. Absorbing $a_0$ into $R_0$, the radial integral is

Step 3. Combine: $V = 4\pi\cdot\tfrac{\pi}{2}R_0^3 = \mathbf{2\pi^2 R_0^3}$ — the 3-volume of a 3-sphere of radius $R_0$, finite where flat and open are infinite.

Step 4. Plug in $R_0=100$ Gpc: $V=2\pi^2(100)^3\ \mathrm{Gpc}^3 \approx \mathbf{2.0\times10^{7}\ Gpc^3}$ — enormous but finite, a universe you could in principle circumnavigate. $V=2\pi^2R_0^3\approx2\times10^7$ Gpc³.

Riemann showed space need not be Euclidean; Einstein made curvature dynamical; and Alexander Friedmann first realised the Universe could be any of the three curved geometries, expanding in time.

Texts. Ryden, Introduction to Cosmology, 2nd ed., ch. 3–4 (geometry and the Friedmann equation) and ch. 10 (the flatness problem) · Dodelson, Modern Cosmology, 2nd ed., ch. 2 (FLRW) and ch. 6 (inflation) · Carroll & Ostlie, An Introduction to Modern Astrophysics, 2nd ed., ch. 29 · Mo, van den Bosch & White, Galaxy Formation and Evolution, ch. 3 · Binney & Tremaine, Galactic Dynamics, 2nd ed., cosmology appendix.

Provenance. A verifier should reproduce: (1) the $S_k(\chi)$ branches by integrating $\mathrm{d}r/\sqrt{1-kr^2}$ and inverting; (2) the sign rule $\Omega_0-1=kc^2/(a_0^2H_0^2)$ by dividing the Friedmann equation by $H_0^2$; (3) the closed volume $V=2\pi^2R_0^3$ using $\int_0^\pi\sin^2u\,\mathrm{d}u=\pi/2$; (4) the scalings $\Omega_k\propto a^2$ (radiation), $\propto a$ (matter); (5) the minimum curvature radius $R_0=(c/H_0)/\sqrt{|\Omega_{k,0}|}\approx102$ Gpc for $|\Omega_{k,0}|=0.0019$ with $c/H_0=4448$ Mpc; (6) $\rho_{c,0}=3H_0^2/8\pi G\approx8.5\times10^{-27}$ kg m⁻³. Numerical anchors $\Omega_{k,0}=0.0007\pm0.0019$ and $100\,\theta_*=1.04109\pm0.00030$ are Planck 2018 (A&A 641, A6, 2020).

Prerequisites. The FLRW Metric · The Friedmann Equations · The Cosmological Principle · Critical Density and Density Parameters. Leads to. Critical Density and Density Parameters · The Friedmann Equations · Cosmological Redshift.

Ask the tutor: Ask me to walk through how the CMB acoustic peak turns a single measured angle into the sub-percent bound $\Omega_{k,0}=0.0007\pm0.0019$, or to derive the closed-universe volume $2\pi^2R_0^3$ step by step.