Swiss Cheese Model in Cosmology & String Theory

Updated 19 June 2026

Swiss Cheese Model is a framework that partitions spacetime into a homogeneous 'cheese' phase and inhomogeneous spherical 'holes', enabling analysis of local inhomogeneities.
It employs exact solutions such as FLRW with LTB or Kottler metrics to study light propagation, averaging methods, and the resulting impacts on cosmic expansion.
Extensions into quantum gravity and string theory highlight its role in moduli stabilization and phenomenological studies, despite limitations in fully mimicking dark energy effects.

The Swiss Cheese Model—originally a safety engineering paradigm—is a technically precise framework with multiple independent implementations in mathematical physics and string phenomenology. Its usage is particularly prominent in theoretical cosmology, where it designates a class of exact solutions to Einstein’s field equations comprised of a homogeneous “cheese” phase and embedded “holes” representing spherical inhomogeneities. Swiss Cheese models facilitate controlled investigations of backreaction, light propagation, and observational implications of cosmic structure, and serve as archetypes for examining how local inhomogeneities couple to the global evolution of spacetime. In string theory, a “Swiss Cheese” structure refers to moduli space geometries, notably in Calabi–Yau compactifications, where the total volume is dominated by a large cycle punctuated by small cycles—mimicking the classic “holes-in-the-cheese” metaphor.

1. Construction and Mathematical Formulation

Swiss Cheese cosmological models consist of a partitioned manifold, with a homogeneous Friedmann–Lemaître–Robertson–Walker (FLRW) or Einstein–de Sitter (EdS) metric (“cheese”) and exact spherically symmetric inhomogeneous “holes.” In canonical implementations:

The cheese is modeled as a FLRW spacetime:

$ds^2_{\text{FLRW}} = -dt^2 + a^2(t) \left[ \frac{dr^2}{1-Kr^2} + r^2 d\Omega^2 \right]$

with $K=0,\,\pm1$ controlling spatial curvature (0710.5505, Fleury, 2014, Grenon et al., 2011).

The holes are typically constructed using Lemaître–Tolman–Bondi (LTB) dust solutions:

$ds^2_{\text{LTB}} = -dt^2 + \frac{[R'(r, t)]^2}{1+2E(r)} dr^2 + R^2(r, t) d\Omega^2$

where $E(r)$ and $M(r)$ are arbitrary functions dictating spatial curvature and enclosed mass; matching across interfaces is imposed via the Darmois–Israel conditions (0710.5505, Fleury, 2014, Grenon et al., 2011).

In other refinements, holes may use the Schwarzschild–de Sitter (Kottler) or Szekeres metrics, the latter introducing non-spherical, anisotropic inhomogeneities (Bolejko et al., 2010, Fleury, 2014).

Contiguity demands the induced metric and extrinsic curvature must match across each spherical boundary $\Sigma$ , fixing the relation between the mass of the hole and the mean background density (Fleury, 2014, Grenon et al., 2011).

2. Light Propagation, Averaging, and Physical Consequences

Swiss Cheese models provide an exact playground to study light propagation through a clumpy universe, circumventing the ambiguities of perturbative and averaging schemes.

The path of a photon is determined by solving the null geodesic equations through alternating FLRW and LTB/Kottler domains.
Distances, redshifts, and lensing are then computed using the Sachs optical equations. Ricci focusing dominates in FLRW regions, while Weyl (tidal) focusing is significant, but typically subleading, in vacuum (Kottler) holes (Fleury, 2014, Szybka, 2010).
Light-cone averaging of density and expansion scalar along the photon path, with affine-parameter weighting $A_{\text{SC}}(r)$ , captures the cumulative effect of under- and overdense structures (0710.5505):

$\langle S \rangle_{EO} = \left[ \int dr\, \frac{Y'}{W} \right]^{-1} \int dr\, S(r, t(r)) \frac{Y'}{W}$

where $Y(r,t)$ encodes the areal radius and $W(r) = \sqrt{1+2E(r)}$ .

A central result is that light-cone averaging in such models reveals that the expansion scalar is generally unaffected by spherically symmetric inhomogeneities (leading to no backreaction on the FLRW scale factor), but the average energy density sampled by photons can be significantly lower than the global mean. This effect arises because photon worldlines preferentially traverse large underdense voids as opposed to minimal shell-crossing high-density regions, leading to apparent signatures akin to dark energy in effective, phenomenological fits (0710.5505).

3. Observational Implications and Limitations

Swiss Cheese models have been exploited to interpret supernovae dimming, cosmic microwave background (CMB) anisotropies, and weak lensing effects.

When the holes are voids of scale $K=0,\,\pm1$ 0– $K=0,\,\pm1$ 1, the model can mimic the effective equation of state parameters $K=0,\,\pm1$ 2 seen in $K=0,\,\pm1$ 3CDM cosmology, but realistic inhomogeneities at $K=0,\,\pm1$ 4 scales fail to generate the observed supernova dimming (0710.5505, Bolejko et al., 2010, Szybka, 2010).
Null geodesic randomization (i.e., photons following generic, not strictly radial, trajectories) suppresses the cumulative effect on distance–redshift relations, reducing any apparent acceleration to statistical insignificance (Szybka, 2010, Lavinto et al., 2015).
CMB photons propagating through random LTB holes in a $K=0,\,\pm1$ 5CDM background display negligible mean shifts in angular-diameter distance ( $K=0,\,\pm1$ 6 at 95% confidence), with temperature anisotropies orders of magnitude below the integrated Sachs–Wolfe signal (Lavinto et al., 2015).

These findings robustly demonstrate that Swiss Cheese inhomogeneity, unless radically engineered, cannot eliminate the need for dark energy in cosmological fits to distance–redshift data.

4. Analytical Equivalence to the Dyer–Roeder Approximation

The Swiss Cheese model has been shown to have an analytical correspondence with the Dyer–Roeder (DR) approximation for light propagation in inhomogeneous cosmologies. Specifically, in the limit of small, opaque clumps and negligible Weyl focusing, the angular-diameter distance $K=0,\,\pm1$ 7 in a Swiss Cheese universe is governed by the Dyer–Roeder differential equation:

$K=0,\,\pm1$ 8

where $K=0,\,\pm1$ 9 quantifies the smoothness fraction of matter in the beam. This formalism confirms that, up to $ds^2_{\text{LTB}} = -dt^2 + \frac{[R'(r, t)]^2}{1+2E(r)} dr^2 + R^2(r, t) d\Omega^2$ 0 (with $ds^2_{\text{LTB}} = -dt^2 + \frac{[R'(r, t)]^2}{1+2E(r)} dr^2 + R^2(r, t) d\Omega^2$ 1), the Dyer–Roeder distance is a precise effective description for the Swiss Cheese scenario (Fleury, 2014).

5. Quantum and String-Theoretic Swiss Cheese Realizations

In quantum gravity and string phenomenology, the Swiss Cheese model generalizes beyond its original cosmological context.

Quantum Swiss Cheese Cosmology: By applying loop quantum cosmology (LQC) techniques, one obtains a quantum Swiss Cheese spacetime where the “hole” is a region described by a deformed Schwarzschild solution with Planck-scale modifications to the metric:

$ds^2_{\text{LTB}} = -dt^2 + \frac{[R'(r, t)]^2}{1+2E(r)} dr^2 + R^2(r, t) d\Omega^2$ 2

with mass gap $ds^2_{\text{LTB}} = -dt^2 + \frac{[R'(r, t)]^2}{1+2E(r)} dr^2 + R^2(r, t) d\Omega^2$ 3. The resulting causal structure is characterized by repeated trapped, anti-trapped, and wormhole regions. The white-hole interpretation of the big bang emerges naturally, allowing for connections between quantum bounces and cosmic structure formation, and suggesting possible dark matter signatures (Lewandowski et al., 2022).

Calabi–Yau Swiss Cheese Compactifications: In large-volume string scenarios, the term denotes manifolds where the overall Kähler modulus is dominated by a “big” four-cycle punctuated by several “small” cycles (holes). The canonical volumetric form is

$ds^2_{\text{LTB}} = -dt^2 + \frac{[R'(r, t)]^2}{1+2E(r)} dr^2 + R^2(r, t) d\Omega^2$ 4

where $ds^2_{\text{LTB}} = -dt^2 + \frac{[R'(r, t)]^2}{1+2E(r)} dr^2 + R^2(r, t) d\Omega^2$ 5, $ds^2_{\text{LTB}} = -dt^2 + \frac{[R'(r, t)]^2}{1+2E(r)} dr^2 + R^2(r, t) d\Omega^2$ 6 are the volumes of the big and small divisors, respectively. This structure enables simultaneous moduli stabilization, realization of Standard Model-like sectors on D7-branes wrapping the big divisor, and the emergence of “split supersymmetry” scenarios (Misra, 2010).

6. Matching and Generalization: Technical and Physical Constraints

The physical realization of Swiss Cheese models imposes strict technical conditions:

Comoving Boundary Requirement: Under the assumption of dust on both sides, the junction boundary must remain strictly comoving; evolution of the interface requires invoking surface layers (i.e., thin shells with stress–energy), pressure, or additional degrees of freedom. Introducing pressure or anisotropy allows for more realistic, dynamical structure formation but deviates from the pure “hard-edged” Swiss Cheese idealization (Grenon et al., 2011).
Extension to Anisotropic/Non-Spherical Inhomogeneities: Implementing Szekeres geometries in holes accommodates non-spherical, less symmetric inhomogeneities. Although this modifies quantitative predictions (e.g., slightly larger Hubble diagram residuals), the qualitative need for large-scale voids to mimic acceleration persists (Bolejko et al., 2010).