Anti-de Sitter Space

The solution: empty space with Λ < 0

Einstein's vacuum field equations with a cosmological constant are

R_μν − ½ g_μν R + Λ g_μν = 0

For Λ < 0, the maximally symmetric solution is anti-de Sitter space. The Riemann tensor is fully determined by Λ:

R_μνρσ = (Λ / 3) (g_μρ g_νσ − g_μσ g_νρ)

This is constant negative curvature. The Ricci scalar evaluates to R = −d(d−1)/L², where L = √(−3/Λ) is the AdS curvature radius. In our universe Λ > 0, so we don't live in AdS — but as a vacuum solution of GR with negative Λ, it is just as valid mathematically as de Sitter or Minkowski.

Coordinates and metrics

AdS_(d+1) admits several useful coordinate patches:

Patch	Metric	Covers	Use
Global	ds² = L²(−cosh²ρ dτ² + dρ² + sinh²ρ dΩ²_(d−1))	Entire manifold	Embedding, boundary structure
Poincaré	ds² = (L²/z²)(−dt² + dx_i² + dz²)	Half of manifold (Poincaré patch)	AdS/CFT computations
Static	ds² = −(1 + r²/L²)dt² + (1 + r²/L²)⁻¹ dr² + r² dΩ²_(d−1)	One static patch	Killing-vector physics
Conformal cylinder	ds² = L²/cos²ρ (−dτ² + dρ² + sin²ρ dΩ²_(d−1))	Compactified, ρ ∈ [0, π/2]	Conformal diagrams

The Poincaré patch is the workhorse of AdS/CFT. The conformal factor L²/z² blows up as z → 0 — this is the conformal boundary, a Minkowski-style spacetime where the dual CFT lives. As z → ∞, the conformal factor vanishes — this is the 'Poincaré horizon', the boundary of the Poincaré patch within the full AdS manifold.

Visualising AdS: the Poincaré disk plus time

The cleanest visualisation of AdS₃ (the simplest non-trivial AdS) is to take a spatial slice — a copy of two-dimensional hyperbolic space H² — and attach a time direction. The Poincaré disk model of H² maps the entire infinite hyperbolic plane onto the interior of a finite Euclidean disk. The disk's boundary is the 'circle at infinity'; in H² geometry it is infinitely far away, but visually it sits at finite Euclidean radius. Geodesics in the Poincaré disk are circular arcs perpendicular to the boundary; M.C. Escher's Circle Limit woodcuts are the most famous popular visualisations of this geometry.

To build AdS₃ from H², imagine stacking copies of the Poincaré disk along a time axis: AdS₃ = H² × time, with a specific Lorentzian metric. The result is a solid cylinder whose interior is curved hyperbolically and whose boundary is the timelike surface where the conformal field theory of AdS₃/CFT₂ lives.

For AdS₄ and higher, the spatial slices are higher-dimensional hyperbolic spaces (H³, H⁴, ...). The visualisation is harder, but the structure is the same: a hyperbolic spatial geometry crossed with time, with a timelike boundary at infinity that has dimension one less than the bulk.

The timelike boundary — and why it changes everything

The defining feature of AdS, the one that makes it dramatically different from Minkowski or de Sitter, is its boundary at infinity. Take a radial null geodesic in the Poincaré patch: ds² = 0 means dz = ±dt. Starting at z = z_0 at t = 0, the light ray reaches z = 0 (the boundary) at t = z_0 — finite coordinate time. By construction the boundary is also at finite affine parameter. Light can reach the boundary and come back; massive geodesics oscillate radially with a finite proper period.

The boundary is timelike: at every point of the boundary, the metric induces a Lorentzian signature. It has the structure of d-dimensional Minkowski space (or its conformal compactification). This boundary is not a part of the manifold per se — it is a 'conformal compactification' — but it nevertheless plays a crucial physical role.

Why does this matter?

• Not globally hyperbolic. A wave equation in AdS does not have a well-posed Cauchy problem on a single spatial slice — you must also specify boundary conditions, because waves can reach the boundary in finite time.
• Effective box. Reflective boundary conditions make AdS behave like a box: waves bounce back and forth, the spectrum is discrete, particles are confined. This is why AdS is useful for studying confinement.
• Conformal symmetry on the boundary. The boundary metric is only defined up to a conformal factor, so its symmetry group is the conformal group of d-dimensional Minkowski space — the same SO(d, 2) as the AdS isometries.
• Holographic encoding. The matching of isometry groups (AdS bulk) and conformal group (boundary) is what makes AdS/CFT possible. The boundary CFT carries the full content of bulk gravitational physics.

AdS/CFT: the conjecture that re-shaped physics

In November 1997, then-postdoc Juan Maldacena posted a paper conjecturing that Type IIB string theory on AdS₅ × S⁵ is exactly dual to N = 4 supersymmetric Yang-Mills theory in 4D Minkowski space. The two theories — one a quantum gravity theory in ten dimensions, the other a gauge theory in four dimensions — would give the same answer for every physical observable, with the dictionary translating between them.

The key parameters match:

L⁴ / α'² = g_YM² N_c  =  λ  (the 't Hooft coupling)
g_s = g_YM²  / 4π

Strong coupling on the CFT side corresponds to large radius and classical gravity on the AdS side. So a classical-gravity computation in AdS computes a strongly coupled gauge-theory observable — turning hard QFT into solvable geometry.

The conjecture has been tested in a vast range of contexts and survives. Specific quantitative predictions — Wilson loop expectation values, scaling dimensions of operators, entanglement entropy via Ryu-Takayanagi, transport coefficients in strongly coupled QGP plasma — match between the two descriptions to many decimal places. As of 2026 the AdS/CFT paper has roughly 25 000 citations, one of the most cited theoretical-physics papers ever.

de Sitter vs anti-de Sitter: the contrast

Property	de Sitter (Λ > 0)	Anti-de Sitter (Λ < 0)
Sign of Λ	+	−
Curvature R	+ constant	− constant
Isometry group	SO(d, 1)	SO(d−1, 2)
Spatial slices	Sphere S^d or flat R^d	Hyperbolic H^d
Scale factor	Exponentially growing a(t) ∝ e^Ht	Not flat-slicing friendly
Boundary at ∞	Spacelike (in past + future)	Timelike
Globally hyperbolic?	No (cosmological horizon)	No (timelike boundary)
Light reaches infinity in	Infinite time	Finite time
Observer in vacuum sees	Thermal radiation T = ℏH/2πk_B	No analogue (different geometry)
Holographic dual	dS/CFT (conjectured, hard)	AdS/CFT (well-established)
Observed in nature?	Yes — future of our universe	No

The hyperboloid embedding

AdS_(d+1) admits a clean embedding in (d+2)-dimensional flat space with signature (−,−,+,+,...,+). The defining equation is

−X₀² − X_(d+1)² + X₁² + X₂² + ... + X_d² = −L²

This is a single-sheeted hyperboloid in (d+2)-dimensional space with two timelike directions. The induced metric from the flat metric η = diag(−,−,+,+,...,+) gives exactly the AdS_(d+1) metric. The isometry group SO(d, 2) is the group of linear transformations of the flat space that preserve this hyperboloid.

Compare this to de Sitter, which embeds in (d+2)-dimensional Minkowski with signature (−,+,+,...,+) as a single-sheeted hyperboloid. Both are 'hyperboloids in one higher dimension' — the difference is the signature. dS lives in one-time, (d+1)-space; AdS lives in two-time, d-space.

Black holes in AdS

AdS admits black hole solutions. The Schwarzschild-AdS metric in (d+1) dimensions is

ds² = −f(r) dt² + dr²/f(r) + r² dΩ²_(d−1)
f(r) = 1 + r²/L² − r_h^(d−2) / r^(d−2)

where r_h is the horizon radius. Unlike asymptotically flat Schwarzschild, large Schwarzschild-AdS black holes are thermodynamically stable: their Hawking temperature increases with size, so they have positive specific heat and don't evaporate away. This makes them very different from astrophysical black holes.

Through AdS/CFT, a large black hole in the AdS bulk corresponds to a thermal state of the CFT on the boundary, with the Hawking temperature equal to the CFT temperature. This is the cleanest statement of black hole thermodynamics anyone has produced: the area-law entropy of the Schwarzschild-AdS horizon equals the thermal entropy of the dual gauge plasma, derived from first-principles statistical mechanics of the CFT. Strominger and Vafa's 1996 microscopic counting of black hole microstates used a related construction.

History: from 1917 hypothetical to 1997 revolution

Anti-de Sitter space was first constructed mathematically in the 1910s, shortly after Einstein introduced the cosmological constant. Cornelius Lanczos, in 1924, was among the first to analyse maximally symmetric solutions of GR with negative Λ. For most of the 20th century it was a mathematical curiosity, mostly of interest to general-relativity textbook writers documenting all possible vacuum solutions.

Two developments changed that. First, in the 1980s, string theory on AdS backgrounds appeared as a natural object of study, particularly in the context of compactifications and gauge-gravity duality precursors. Second, in November 1997, Juan Maldacena's paper 'The Large-N Limit of Superconformal Field Theories and Supergravity' (Adv. Theor. Math. Phys. 2, 231) made the connection precise. Within months, follow-up work by Witten, by Gubser-Klebanov-Polyakov, and by many others fleshed out the holographic dictionary. The 'AdS/CFT correspondence' became a major branch of theoretical physics, with applications stretching from black-hole information to QCD plasma physics to condensed-matter superconductivity.

The 2008 Heineman Prize in Mathematical Physics was awarded to Maldacena 'for his pioneering proposal' of the correspondence. He has since been elected to the National Academy of Sciences and has shared the Fundamental Physics Prize. As of 2026 AdS/CFT remains the central tool in non-perturbative quantum gravity, and the AdS geometry — once an obscure mathematical artifact — sits at the heart of how theorists think about the holographic principle.

Common pitfalls

• Confusing AdS with hyperbolic space. Hyperbolic space H^d is a Riemannian manifold (all spatial). AdS is Lorentzian — it has one time direction. A spatial slice of AdS_(d+1) is hyperbolic, but AdS itself is a spacetime.
• Assuming AdS is observed in our universe. Our cosmological constant is positive. AdS is not the geometry we live in — it is a theoretical workbench. The cleanest holographic dualities happen to require the wrong sign of Λ; reconciling this with cosmological reality is an open problem.
• Treating the boundary as a regular spacetime point. The conformal boundary of AdS is at infinite affine parameter only in the original metric. After conformal compactification, light reaches it in finite parameter, but the boundary is not part of the original manifold — it is a structure added on the conformally rescaled diagram.
• Confusing global with Poincaré coordinates. Global AdS is topologically R × ball and is well-defined for all time. The Poincaré patch covers only half the manifold and has its own 'Poincaré horizon' — different coordinates, different physics manifests.
• Conflating AdS/CFT with full quantum gravity in our universe. AdS/CFT is the most precise quantum gravity statement we have, but in the wrong universe. Whether its lessons (holography, entanglement, black hole information recovery) extend to cosmologically realistic spacetimes is far from settled.