De Sitter Space: Definition, Geometry, and Applications,

• De Sitter space is the maximally symmetric Lorentzian spacetime with positive constant curvature of the embedding hyperboloid in Minkowski space, characterized by compact global spatial sections, cosmological horizons, and observer-dependent particle notions. Essential in inflationary cosmology, late-time accelerated expansion, and quantum field theory.
• Its defining properties include the hyperboloid equation $-(X^0)^2+rac{1}{ ext{d}+2}egin3mbsum\_{i=1}^{d+1}(X^i)^2=H^{-2}$ within Minkowski space, positive cosmological constant, and the isometry group $SO( ext{d}+1,1).
• Global coordinates are applied to demonstrate the topology and causal structure of de Sitter space, while specific coordinate patches like the Poincaré and static patches offer simplified local descriptions of physical observables. ”

De Sitter space is the maximally symmetric Lorentzian spacetime with positive constant curvature and positive cosmological constant. In $(d+1)$ dimensions it is realized as the hyperboloid

$-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$

in $(d+2)$-dimensional Minkowski space, where $H^{-1}$ is the curvature radius. Its curvature tensors are

$$R_{\mu\nu\rho\sigma} =H^2\left(g_{\mu\rho}g_{\nu\sigma}-g_{\mu\sigma}g_{\nu\rho}\right), \qquad R_{\mu\nu}=dH^2g_{\mu\nu}, \qquad R=d(d+1)H^2,$$

and

$\Lambda=\frac{d(d-1)}{2}H^2.$

The connected isometry group is $SO(d+1,1)$. De Sitter space is central to inflationary cosmology, late-time accelerated expansion, quantum field theory in curved spacetime, semiclassical gravity, holography, and representation theory. Its distinctive features include compact global spatial sections, cosmological horizons, the absence of a globally timelike Killing vector, observer-dependent particle notions, and pronounced infrared effects.

1. Geometry, embeddings, and coordinate systems

Global geometry

Global coordinates exhibit the topology and causal completeness of de Sitter space:

$ds^2=-dT^2+\frac{1}{H^2}\cosh^2(HT)\,d\Omega_d^2,$

with

$\mathrm{dS}_{d+1}\simeq \mathbb R\times S^d.$

The spatial sections are compact $d$-spheres. Their scale factor reaches a minimum at $-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$0 and grows exponentially as $-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$1. In four dimensions,

$-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$2

The global conformal coordinate $-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$3 is defined by

$-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$4

so that

$-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$5

Thus global de Sitter space is conformal to a finite strip of the Einstein static universe. Its past and future conformal boundaries, $-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$6 and $-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$7, are each $-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$8.

The embedding-space invariant

$-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$9

determines de Sitter-invariant two-point functions. Correlators in a de Sitter-invariant state depend on $(d+2)$0 rather than independently on the two spacetime points. The antipodal point is $(d+2)$1, for which $(d+2)$2 (Akhmedov, 2013).

Poincaré patch

The expanding Poincaré patch has metric

$(d+2)$3

The scale factor is

$(d+2)$4

and future infinity is approached as $(d+2)$5. Equivalently,

$(d+2)$6

The patch covers only part of global de Sitter space. Its past boundary has arbitrarily large physical blueshift for fixed comoving momentum, while its future boundary is the regime of super-horizon modes with $(d+2)$7.

The contracting Poincaré patch is obtained by reversing the sign of the exponential scale factor. The expanding and contracting patches meet at a null boundary. Global de Sitter contains both patches, so analyses restricted to one patch do not automatically capture the full global state or its infrared behavior (Akhmedov, 2013).

Static patch

A static observer uses

$(d+2)$8

where $(d+2)$9 and $H^{-1}$0. The cosmological horizon is at

$H^{-1}$1

Euclidean regularity requires

$H^{-1}$2

giving the Gibbons–Hawking temperature

$H^{-1}$3

The horizon entropy obeys the area law,

$H^{-1}$4

The static-patch time translation is timelike only inside the observer’s causal diamond. Consequently, it defines a local Hamiltonian and local particle notion rather than a global de Sitter energy.

Generalized hypersurfaces

Replacing the constant de Sitter radius by a function produces generalized de Sitter hypersurfaces in Minkowski space. With

$H^{-1}$5

the induced metric is

$H^{-1}$6

Writing

$H^{-1}$7

the hypersurface is timelike when $H^{-1}$8, null when $H^{-1}$9, and spacelike when $$R_{\mu\nu\rho\sigma} =H^2\left(g_{\mu\rho}g_{\nu\sigma}-g_{\mu\sigma}g_{\nu\rho}\right), \qquad R_{\mu\nu}=dH^2g_{\mu\nu}, \qquad R=d(d+1)H^2,$$0. For non-null cases it is a warped product with warping function

$$R_{\mu\nu\rho\sigma} =H^2\left(g_{\mu\rho}g_{\nu\sigma}-g_{\mu\sigma}g_{\nu\rho}\right), \qquad R_{\mu\nu}=dH^2g_{\mu\nu}, \qquad R=d(d+1)H^2,$$1

The ordinary de Sitter geometry corresponds to $$R_{\mu\nu\rho\sigma} =H^2\left(g_{\mu\rho}g_{\nu\sigma}-g_{\mu\sigma}g_{\nu\rho}\right), \qquad R_{\mu\nu}=dH^2g_{\mu\nu}, \qquad R=d(d+1)H^2,$$2, for which $$R_{\mu\nu\rho\sigma} =H^2\left(g_{\mu\rho}g_{\nu\sigma}-g_{\mu\sigma}g_{\nu\rho}\right), \qquad R_{\mu\nu}=dH^2g_{\mu\nu}, \qquad R=d(d+1)H^2,$$3 and the curvature is constant. Nonconstant $$R_{\mu\nu\rho\sigma} =H^2\left(g_{\mu\rho}g_{\nu\sigma}-g_{\mu\sigma}g_{\nu\rho}\right), \qquad R_{\mu\nu}=dH^2g_{\mu\nu}, \qquad R=d(d+1)H^2,$$4 generally produces non-Einstein geometries with time-dependent curvature and, in some families, asymptotic curvature singularities (Pham, 2012).

2. Isometries, stationary vacua, and observer dependence

The de Sitter group is $$R_{\mu\nu\rho\sigma} =H^2\left(g_{\mu\rho}g_{\nu\sigma}-g_{\mu\sigma}g_{\nu\rho}\right), \qquad R_{\mu\nu}=dH^2g_{\mu\nu}, \qquad R=d(d+1)H^2,$$5. In the embedding space its generators are

$$R_{\mu\nu\rho\sigma} =H^2\left(g_{\mu\rho}g_{\nu\sigma}-g_{\mu\sigma}g_{\nu\rho}\right), \qquad R_{\mu\nu}=dH^2g_{\mu\nu}, \qquad R=d(d+1)H^2,$$6

Unlike Minkowski space or global anti-de Sitter space, de Sitter space has no globally defined timelike Killing vector. Therefore there is no unique global Hamiltonian, globally preferred positive-frequency decomposition, or observer-independent particle concept.

A stationary vacuum is defined relative to a Hamiltonian generating a timelike isometry in an appropriate region. The standard static vacuum is associated with the boost

$$R_{\mu\nu\rho\sigma} =H^2\left(g_{\mu\rho}g_{\nu\sigma}-g_{\mu\sigma}g_{\nu\rho}\right), \qquad R_{\mu\nu}=dH^2g_{\mu\nu}, \qquad R=d(d+1)H^2,$$7

More general stationary generators combine a boost with spatial rotations. In $$R_{\mu\nu\rho\sigma} =H^2\left(g_{\mu\rho}g_{\nu\sigma}-g_{\mu\sigma}g_{\nu\rho}\right), \qquad R_{\mu\nu}=dH^2g_{\mu\nu}, \qquad R=d(d+1)H^2,$$8,

$$R_{\mu\nu\rho\sigma} =H^2\left(g_{\mu\rho}g_{\nu\sigma}-g_{\mu\sigma}g_{\nu\rho}\right), \qquad R_{\mu\nu}=dH^2g_{\mu\nu}, \qquad R=d(d+1)H^2,$$9

while in higher dimensions one may have

$\Lambda=\frac{d(d-1)}{2}H^2.$0

These loxodromic generators define rotating vacua. The corresponding coordinate descriptions take the form of Kerr–de Sitter metrics, although the underlying empty geometry remains de Sitter space (Parikh et al., 2012).

The rotating Hamiltonian becomes spacelike in an observer-dependent ergoregion. In four dimensions, the rotating coordinates have

$\Lambda=\frac{d(d-1)}{2}H^2.$1

so the ergosurface satisfies

$\Lambda=\frac{d(d-1)}{2}H^2.$2

The cosmological horizon remains at $\Lambda=\frac{d(d-1)}{2}H^2.$3 in the dimensionless coordinates used there. Between the ergosurface and the horizon, the stationary Killing vector is spacelike, although suitable constant-time slices can remain spacelike.

Static and rotating vacua can be particle-inequivalent. A static mode

$\Lambda=\frac{d(d-1)}{2}H^2.$4

becomes, under a rotating transformation,

$\Lambda=\frac{d(d-1)}{2}H^2.$5

Thus a mode with $\Lambda=\frac{d(d-1)}{2}H^2.$6 can have negative rotating-frame energy when $\Lambda=\frac{d(d-1)}{2}H^2.$7. The associated Bogoliubov transformation has nonzero $\Lambda=\frac{d(d-1)}{2}H^2.$8-coefficients for suitable angular momenta. Horizon and ergosphere properties are therefore determined not solely by the manifold but also by the chosen stationary time evolution.

3. Quantum fields, vacua, and representation theory

Scalar modes and vacua

A scalar field on global de Sitter space can be expanded in spherical harmonics on $\Lambda=\frac{d(d-1)}{2}H^2.$9. In the principal-series regime, the reduced mode equation has asymptotic oscillatory solutions. For a massive scalar in four dimensions, the late-time behavior in the Poincaré patch is

$SO(d+1,1)$0

where

$SO(d+1,1)$1

For heavy fields,

$SO(d+1,1)$2

and the late-time dimensions are

$SO(d+1,1)$3

For light fields,

$SO(d+1,1)$4

The Bunch–Davies, or Euclidean, vacuum is selected by regularity under analytic continuation to the Euclidean sphere and by matching to the flat-space positive-frequency modes at high physical momentum. There are also in and out vacua, defined by the absence of particles at past and future infinity, respectively. The Bunch–Davies state is generally not empty with respect to both asymptotic particle notions.

Unitary representations

The de Sitter analogue of Poincaré particle classification is the classification of unitary irreducible representations of $SO(d+1,1)$5. A scalar representation can be labeled by

$SO(d+1,1)$6

with quadratic Casimir eigenvalue

$SO(d+1,1)$7

The principal series has real $SO(d+1,1)$8 and describes heavy fields. The complementary series has imaginary $SO(d+1,1)$9 and describes light fields. The massless scalar lies at a discrete or exceptional endpoint in the representation-theoretic classification (Şengör, 2022).

At late times, spatial translations, rotations, dilatations, and special conformal transformations act as the Euclidean conformal group on $ds^2=-dT^2+\frac{1}{H^2}\cosh^2(HT)\,d\Omega_d^2,$0. The future-boundary operators $ds^2=-dT^2+\frac{1}{H^2}\cosh^2(HT)\,d\Omega_d^2,$1 and $ds^2=-dT^2+\frac{1}{H^2}\cosh^2(HT)\,d\Omega_d^2,$2 furnish conformal representations with dimensions $ds^2=-dT^2+\frac{1}{H^2}\cosh^2(HT)\,d\Omega_d^2,$3 and $ds^2=-dT^2+\frac{1}{H^2}\cosh^2(HT)\,d\Omega_d^2,$4. For principal-series fields, the dimensions are complex conjugates and produce oscillatory dependence on $ds^2=-dT^2+\frac{1}{H^2}\cosh^2(HT)\,d\Omega_d^2,$5; for complementary-series fields, they are real and yield power-law late-time behavior.

The absence of a global timelike Killing vector means that these representation labels should not be identified naively with a conserved energy. A recent momentum-space construction instead identifies the de Sitter frequency with the unitary-representation parameter that diagonalizes the quadratic Casimir, while spatial momentum diagonalizes translations. This produces a Kontorovitch–Lebedev–Fourier momentum space in which free equations become algebraic and in-in time integrals become meromorphic frequency integrals (Belrhali et al., 21 Jan 2026).

Plane waves and higher spin

Coordinate-independent de Sitter plane waves are constructed from a null ambient vector $ds^2=-dT^2+\frac{1}{H^2}\cosh^2(HT)\,d\Omega_d^2,$6:

$ds^2=-dT^2+\frac{1}{H^2}\cosh^2(HT)\,d\Omega_d^2,$7

They satisfy

$ds^2=-dT^2+\frac{1}{H^2}\cosh^2(HT)\,d\Omega_d^2,$8

in four dimensions. In the flat limit they reduce to ordinary Minkowski plane waves. Higher-spin modes are obtained by applying polarization operators to the scalar factor:

$ds^2=-dT^2+\frac{1}{H^2}\cosh^2(HT)\,d\Omega_d^2,$9

This construction gives massive vector and spin-two modes and extends representation-theoretically to higher spin (Tanhayi, 2014).

4. Quantum stability, particle production, and infrared dynamics

De Sitter quantum stability is not a single criterion. It may refer to the stability of a particular false vacuum, the absence of particle production in a chosen free-field vacuum, the persistence of de Sitter invariance under loops, or the stability of the full gravitational background.

Eternal manifolds and propagator composition

An “eternal manifold” is defined perturbatively by the validity of ordinary Feynman rules, stability of the vacuum, absence of an imaginary part in vacuum loops, and vanishing of spontaneous particle-creation diagrams in the Schwinger–Keldysh formulation. The criterion can be expressed through a composition principle for propagators:

$\mathrm{dS}_{d+1}\simeq \mathbb R\times S^d.$0

A propagator with both asymptotic phases,

$\mathrm{dS}_{d+1}\simeq \mathbb R\times S^d.$1

generally violates this composition property when $\mathrm{dS}_{d+1}\simeq \mathbb R\times S^d.$2. In the doubled formalism, diagrams containing branch-changing propagators or “spiders” represent spontaneous particle creation. The Bunch–Davies propagator has both asymptotic exponentials for heavy fields, whereas a $\mathrm{dS}_{d+1}\simeq \mathbb R\times S^d.$3-function propagator can have a single asymptotic exponential but develops an antipodal singularity (0709.2899).

The same issue appears in a one-dimensional scattering problem. A mode with late-time behavior

$\mathrm{dS}_{d+1}\simeq \mathbb R\times S^d.$4

has nonzero reflection when $\mathrm{dS}_{d+1}\simeq \mathbb R\times S^d.$5. The proposed eternity condition is therefore zero reflection amplitude.

One-loop particle production

For a free massive scalar in global de Sitter, the in and out vacua are related by Bogoliubov coefficients $\mathrm{dS}_{d+1}\simeq \mathbb R\times S^d.$6 and $\mathrm{dS}_{d+1}\simeq \mathbb R\times S^d.$7, with

$\mathrm{dS}_{d+1}\simeq \mathbb R\times S^d.$8

The mode-by-mode particle number is

$\mathrm{dS}_{d+1}\simeq \mathbb R\times S^d.$9

For even spacetime dimensions, particle production is nonzero mode by mode. For odd spacetime dimensions, the relevant Pöschl–Teller potential is reflectionless and $d$0. The one-loop in/out effective action satisfies

$d$1

analogously to the Heisenberg–Euler action in a constant electric field.

After zeta-function regularization of the angular-momentum sum, the weak-curvature renormalized effective action vanishes in even spacetime dimensions and is finite and real in odd dimensions. This has been interpreted as evidence for one-loop stability of free massive de Sitter space, but the conclusion depends on the in/out vacuum, the heavy-field restriction, and the regularization prescription. It does not address interactions, gravitational backreaction, or infrared effects beyond the specified one-loop calculation (Kim, 2010).

Infrared effects in interacting fields

For nonconformal fields, de Sitter loop corrections are not generically suppressed relative to tree-level contributions. In the expanding Poincaré patch, principal-series fields acquire secular corrections of the form

$d$2

so the ratio of loop to tree contributions contains

$d$3

At sufficiently late times, higher powers of this logarithm become comparable. Complementary-series fields exhibit stronger power-law infrared behavior.

The infrared state dependence is concentrated in the Keldysh propagator. In the principal series, Dyson–Schwinger equations can be reduced to kinetic equations for a comoving particle density and anomalous correlator. Weak perturbations can relax toward a stationary out-state, whereas sufficiently strong perturbations can generate nonlinear self-reproduction and explosive growth of the comoving density. The resulting stress tensor may invalidate the fixed-background approximation (Akhmedov, 2013).

Polyakov’s analysis emphasizes infrared divergences even for massive scalar fields and proposes that particle production, growing occupation numbers, and gravitational backreaction may screen the cosmological constant. The full gravitational instability is not established because tensor modes, gauge fixing, and nonlinear in-in backreaction are not completely computed (0709.2899).

Gravitational infrared effects

Transverse-traceless gravitons in the Poincaré patch obey the same linearized equation as a massless minimally coupled scalar. Their low-momentum propagator behaves as

$d$4

so coincident correlators contain

$d$5

Rajaraman proposes that this divergence is resolved not by a graviton mass or an explicitly non-de Sitter-invariant regulator, but by spontaneous breaking of de Sitter invariance. A deformation

$d$6

regulates the infrared and yields a deformation parameter scaling parametrically as

$d$7

The claim is that exact de Sitter space is not a solution of the quantum-corrected equations, although the work does not determine the complete late-time geometry or demonstrate catastrophic decay (Rajaraman, 2016).

Effective-field-theory organization

Soft de Sitter effective theory separates super-horizon modes into growing and decaying scaling operators with dimensions

$d$8

in four dimensions. Massive fields have irrelevant interactions in the deep infrared after redundant operators are removed. Massless scalars are exceptional because infinitely many composite operators become degenerate and mix. Their renormalization-group flow is equivalent to stochastic inflation, with a Fokker–Planck or more general Kramers–Moyal equation.

For single-clock inflation, nonlinear large spatial diffeomorphisms force the growing curvature perturbation and tensor modes to have zero scaling dimension. Consequently,

$d$9

in the long-wavelength limit to all loop orders, under symmetry-preserving renormalization and the single-field assumptions. This conservation theorem does not resolve the global or nonperturbative problems of eternal de Sitter space (Green, 2022).

5. Euclidean saddles, vacuum decay, and thermodynamics

Coleman–De Luccia and Hawking–Moss configurations

A Coleman–De Luccia instanton is an $-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$00-symmetric Euclidean solution interpolating between false and true vacua. Its fluctuation spectrum contains a negative mode associated with changing the bubble radius. A Hawking–Moss instanton places the scalar at the top of the potential barrier.

A single negative mode is not sufficient to establish exponential decay. It implies decay only when integration along that direction makes the original Euclidean path integral divergent and requires contour deformation into the complex plane. In de Sitter space, the Euclidean continuation is compact $-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$01, and the Euclidean action can be bounded below. A CDL or Hawking–Moss saddle may therefore be a real, subdominant contribution to a convergent path integral rather than a source of an imaginary false-vacuum energy.

For sufficiently long-lived false vacua, characterized parametrically by

$-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$02

an approximately de Sitter-invariant Hartle–Hawking false vacuum can be constructed. Rapid transitions with

$-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$03

can still produce genuine exponential decay. The argument depends on boundedness of the full theory, including backreaction (0911.3142).

Semiclassical free energies

In three-dimensional de Sitter gravity, pure $-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$04 has static metric

$-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$05

with

$-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$06

Under the boundary-term prescription used in the comparison,

$-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$07

Three-dimensional Kerr–de Sitter space has one cosmological horizon at $-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$08. Its Euclidean free energy is

$-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$09

At equal cosmological-horizon temperature,

$-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$10

one obtains

$-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$11

The physically allowed Kerr–de Sitter parameter range excludes the region in which $-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$12. Pure de Sitter therefore dominates the semiclassical partition function among the two saddles considered, and no temperature-driven Hawking–Page-like transition occurs in this setup (Fareghbal et al., 2023).

Holographic thermodynamics

Static de Sitter space can be conformally mapped to $-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$13, which is the conformal boundary of hyperbolic $-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$14. The de Sitter horizon maps to the asymptotic region of $-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$15, while the hyperbolic AdS horizon has temperature

$-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$16

The construction reproduces de Sitter-like energy and entropy dependence but yields a ratio $-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$17 differing by a factor $-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$18 from the pure four-dimensional de Sitter relation. This mismatch reflects that the boundary theory is not four-dimensional dynamical gravity (Li et al., 2011).

For global de Sitter, the equator of the $-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$19 at $-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$20 is the horizon of a north-pole observer and the entangling surface separating northern and southern hemispheres. A bulk minimal surface anchored on the equator gives a Ryu–Takayanagi interpretation of the entropy within the AdS embedding.

In dS/CFT constructions, spherical boundary regions admit complex extremal surfaces. For even boundary dimension $-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$21, their areas contain a universal logarithmic divergence whose coefficient agrees with the logarithmic term in the sphere free energy computed from the semiclassical de Sitter wavefunction. This coefficient is the de Sitter continuation of the conformal-anomaly coefficient. The surfaces are generally complex, the putative dual CFT is Euclidean and nonunitary, and the interpretation as ordinary entanglement entropy remains tentative (Narayan, 2015).

6. Global structures, holography, and specialized constructions

Antipodes and twistors

The antipodal map

$-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$22

exchanges $-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$23 and $-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$24. The quotient

$-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$25

identifies antipodal points and is not globally time-orientable. It also lacks a global distinction between left- and right-handed fields.

Twistors are treated as spinors of $-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$26. At a point $-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$27 they decompose into local Weyl spinors through the projectors

$-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$28

Each projector defines a Riemann sphere in twistor space, so a de Sitter point corresponds to a pair of spheres exchanged by the antipodal map. The scalar Penrose transform integrates over both spheres, while spin-$-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$29 transforms generate chiral massless fields satisfying

$-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$30

or its opposite-chirality counterpart. The twistor construction naturally combines antipodal points and opposite helicities, making $-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$31 a proposed setting for twistor formulations of quantum gravity with positive cosmological constant (Neiman, 2013).

de Sitter entropy and gravitating observers

A two-dimensional JT-gravity construction couples de Sitter space to an asymptotically AdS black hole acting as a gravitating observer. Replica wormholes connecting the de Sitter and AdS regions exist when

$-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$32

For $-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$33, the computed de Sitter entropy vanishes and the effective de Sitter Hilbert space behaves as one-dimensional. Even when $-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$34, weak entanglement can leave the entropy zero. In the strong-entanglement limit, the dominant quantum-extremal surface lies at the AdS black-hole horizon and yields

$-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$35

The nominal de Sitter horizon entropy determines whether the connected saddle exists, whereas the observer’s horizon controls the computed entanglement entropy. This result is conditional on the JT model, boundary sources, replica-wormhole sum, and semiclassical saddle approximation (Balasubramanian et al., 2023).

Complexity

For asymptotically AdS geometries with de Sitter boundaries, holographic complexity can be computed through the volume or Wheeler–DeWitt action prescriptions. The bulk states include horizon-containing ungapped geometries and horizonless bubbles of nothing.

The complexity=volume prescription yields the expected ordering: lower-energy bubbles have smaller complexity and lower complexity-growth rate. The conjectured growth bound is not saturated. The complexity=action prescription gives qualitatively different behavior: lower-energy bubble states can have larger complexity than the higher-energy horizon-containing state, and small bubbles acquire a logarithmic enhancement. These results reveal sensitivity to null-boundary counterterms and to the choice of complexity prescription (Reynolds et al., 2017).

Reflecting cavities

In $-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$36 dimensions, a perfectly reflecting cavity of fixed physical size can isolate a massless scalar from de Sitter thermal correlations. In the vacuum defined by positive frequency with respect to the static cavity time, an Unruh–DeWitt detector has

$-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$37

Even an accelerating detector inside the cavity detects no particles. The state is analogous to a Boulware vacuum: it is regular inside the cavity but becomes singular as the mirrors approach the de Sitter horizon. The total energy contains Casimir-like and curvature-dependent terms and is lower than the corresponding region in the Bunch–Davies state. The result is exact only for the specified two-dimensional massless conformal model and does not establish that de Sitter horizons are universally nonthermal (Davies et al., 2020).

Projective and signature-changing models

A Beltrami–de Sitter model places a Riemannian Beltrami–Klein metric inside a unit ball and a Lorentzian de Sitter metric outside it. In two dimensions the common unit circle is a signature-changing boundary, while the same quadratic metric formula is Riemannian inside and Lorentzian outside:

$-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$38

The exterior light cones are tangent to the unit circle, and geodesics are straight Euclidean lines. The construction extends to higher dimensions and produces point–hyperplane dualities, Radon-like transforms, and split-signature correspondence spaces. In the two-dimensional model’s twistor distribution, a $-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$39 distribution has hidden split-real $-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$40 symmetry. Its proposed cosmological interpretation connects a Lorentzian de Sitter region to a Riemannian inter-eon phase, but its physical suitability remains conjectural (Nurowski, 16 Mar 2025).

Exceptional scalar theories

The special galileon and DBI/conformal-dilaton theories are exceptional scalar effective field theories on de Sitter space. Their special masses are

$-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$41

with $-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$42 for DBI/conformal-dilaton theories and $-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$43 for the special galileon. The special galileon realizes

$-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$44

while the DBI/dilaton theory realizes

$-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$45

Their nonlinear symmetries can be interpreted as broken diffeomorphisms of de Sitter space. Compact and $-(X^0)^2+\sum_{i=1}^{d+1}(X^i)^2=H^{-2}$46-manifest formulations of the special galileon are related by field redefinitions, and brane and dilaton formulations of DBI are likewise strongly indicated to be equivalent. In the flat limit, these maps reduce to familiar galileon dualities and conformal DBI relations (Bonifacio et al., 2021).

De Sitter space therefore supports several complementary descriptions: a constant-curvature hyperboloid, a thermal static patch, a conformally flat cosmological background, a representation-theoretic state space, a boundary conformal geometry, and a semiclassical gravitational saddle. None of these descriptions supplies a universally preferred particle notion or a complete nonperturbative quantum theory. The central unresolved issues concern the status of de Sitter observables, the infrared completion of interacting quantum fields and gravity, the interpretation of horizon entropy, the existence of a complete holographic dual, and the ultimate fate of exact de Sitter invariance in quantum gravity.