⚛️ phenomenology

Prospects of Detecting Cosmological Domain Walls in the Cosmic Microwave Background

This paper proposes that dark, gravitationally interacting cosmological domain walls leave distinct, non-Gaussian imprints on the Cosmic Microwave Background which can be detected using advanced curvelet/wavelet analyses and neural network-based inference, potentially allowing future South Pole Telescope-like surveys to probe symmetry-breaking scales as low as 10–100 MeV.

Original authors: Berkin Gürbüz, Robert Brandenberger, Yitian Sun

Published 2026-10-26

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Berkin Gürbüz, Robert Brandenberger, Yitian Sun

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The universe is filled with invisible structures that shape the cosmos, from the vast clusters of galaxies to the faint afterglow of the Big Bang known as the cosmic microwave background. This background radiation is a snapshot of the infant universe, a sea of light that has traveled for billions of years to reach our telescopes. In the standard story of cosmology, this light is remarkably smooth, with only tiny, random ripples caused by the initial conditions of the universe. However, some theories suggest that the early universe underwent dramatic phase transitions, similar to water freezing into ice, which could have left behind permanent scars in the fabric of space-time. These scars are called topological defects. While some theories predict point-like flaws or string-like tears, this research focuses on a specific type of defect: domain walls. These are vast, two-dimensional sheets that could have formed if a fundamental symmetry of the universe was broken in a specific way. If they exist, they would not be made of ordinary matter but would be regions of trapped energy, interacting with the rest of the universe only through gravity.

For decades, physicists believed that if these domain walls existed, they would have been so heavy and energetic that they would have crushed the universe, causing it to collapse long before stars and galaxies could form. This is known as the overclosure problem. However, a new possibility has emerged: if these walls formed at a much lower energy scale than previously thought, they might be light enough to survive without destroying the cosmos, yet heavy enough to leave a detectable trace. The question then becomes: how do we find something that is invisible, interacts only with gravity, and might be hiding in the vastness of space? The answer lies in how these walls would distort the light from the cosmic microwave background as it passes by them.

A team of researchers at McGill University has taken a deep dive into this possibility, simulating exactly what would happen if a dark domain wall were to pass between us and the cosmic microwave background. They found that these walls act as a unique kind of gravitational lens, but with a twist. Unlike a normal lens that focuses light, the repulsive gravity of a domain wall pushes matter away from it. As the universe expanded, this repulsive force would have cleared out a massive, empty void of matter right next to the wall, while piling up a dense shell of matter around the edges of that void. When photons of light from the early universe travel through this region, they do not just pass straight through; they are deflected. The light rays are pushed away from the center of the void and pulled slightly toward the dense shell, creating a distinct pattern of distortion on the sky.

The researchers calculated that this distortion would leave a very specific fingerprint on the cosmic microwave background. It would not look like a random smudge or a simple blur. Instead, it would create a coherent, line-like feature where the temperature of the sky appears to pinch or squeeze. This pattern is highly unusual because it is not random; it is a structured, non-Gaussian signal that stands out against the natural, random fluctuations of the early universe. The strength of this signal depends on how massive the wall is and how far away it is. The team discovered that for walls with a specific, relatively low energy scale, the distortion would be strong enough to be seen by modern telescopes, provided the wall is not tilted too far away from our line of sight. If the wall were perfectly face-on, the effect would vanish, but if it is tilted, the signal becomes visible.

To find these subtle signals, the researchers developed new ways of looking at the data. Standard methods used to analyze the cosmic microwave background look for patterns that are the same in every direction, but a domain wall creates a pattern that is stretched in one direction and compressed in another. The team used advanced mathematical tools called curvelets and wavelets, which are designed to spot these kinds of elongated, directional features. They also employed a sophisticated technique called simulation-based inference, where they trained a computer program to recognize the specific shape of a domain wall signal by showing it millions of simulated examples. This computer learned to distinguish the unique "pinch" of a domain wall from the background noise of the universe with high precision.

The results of their simulations suggest that future telescopes, such as the South Pole Telescope, could be sensitive enough to detect these walls if they exist within a certain range of energies. The researchers found that the most powerful method, which uses the full shape of the signal rather than just a single number, could detect walls with an energy scale as low as a few tens of millions of electron volts. This is a significant range, covering a band of possibilities that previous studies might have missed. The study confirms that if these domain walls exist and have the right properties, they would leave a clear, non-random signature on the sky that is distinct from any other known cosmic phenomenon.

However, the researchers are careful to note that this is a forecast of what could be found, not a confirmation that these walls have been found. Their work assumes ideal conditions, with no interference from other cosmic sources or instrumental errors. In the real world, finding these walls will be much harder because the signal must be separated from a noisy background. The study serves as a proof of concept, demonstrating that the tools exist to hunt for these elusive structures. If the next generation of cosmic microwave background experiments detects a signal that matches the specific, line-like distortion predicted here, it would be a monumental discovery. It would provide the first direct evidence of these ancient, two-dimensional scars in the universe, offering a rare glimpse into the high-energy physics of the very first moments after the Big Bang. Until then, the domain walls remain a compelling possibility, waiting in the shadows of the cosmic microwave background for a telescope sharp enough to reveal their presence.

Technical Summary: Prospects of Detecting Cosmological Domain Walls in the Cosmic Microwave Background

Problem Statement
Topological defects, specifically domain walls arising from the spontaneous breaking of discrete symmetries, are generic predictions of quantum field theories. While stable domain wall networks typically suffer from the "overclosure problem" (where their energy density scales slower than matter or radiation, dominating the universe's expansion), a phenomenological window exists for walls generated at low symmetry-breaking scales ( $\eta$ ). In this regime, walls can evade the Zel'dovich overclosure bound while remaining cosmologically relevant. The paper investigates the gravitational imprint of such "dark" domain walls (interacting only via gravity) on the Cosmic Microwave Background (CMB). Unlike cosmic strings, which produce line-like discontinuities, domain walls are expected to induce coherent, planar distortions in the gravitational potential, leading to specific non-Gaussian signatures in CMB temperature anisotropies. The study aims to derive these signatures analytically and assess their detectability using current and future CMB survey capabilities, specifically modeling a South Pole Telescope-3G (SPT-3G) like survey.

Methodology
The authors employ a multi-stage approach combining analytical derivation, numerical simulation, and machine learning-based inference:

Analytical Derivation of Density and Deflection:
- Density Profile: Starting from the repulsive gravitational acceleration ( $\alpha = 2\pi G\sigma$ ) exerted by a planar domain wall, the authors calculate the resulting matter density profile. Matter is repelled from the wall, creating an evacuated void bounded by an overdense shell. This profile is derived relativistically, accounting for the causal horizon and the time evolution of the wall network under the scaling solution ( $L \sim H^{-1}$ ).
- Photon Deflection: Using the Born scattering approximation, the authors compute the photon deflection field ( $\vec{\alpha}$ ) caused by the projected surface density contrast of the void and shell. The calculation is performed for walls with arbitrary inclination angles ( $\theta$ ) relative to the line of sight. The deflection field is shown to behave like a diverging (concave) lens, pushing photons away from the void center.
- CMB Signatures: The deflection field is applied to primary CMB temperature maps via coordinate transformation ( $T_{obs} = T_{prim}(\theta - \alpha_{obs})$ ). The resulting signatures are characterized as coherent, line-like, and strongly non-Gaussian distortions, distinct from the Gaussian primordial fluctuations.
Statistical Analysis (Curvelets and Wavelets):
- Recognizing that two-point statistics (e.g., angular power spectra) are inefficient for localized, anisotropic signals, the authors utilize curvelet and wavelet decompositions.
- Wavelets: An orthogonal Daubechies-12 (db12) basis is used to detect localized non-Gaussianity without specific directional tuning.
- Curvelets: A directional multiscale basis is employed to match the elongated, edge-like geometry of the domain wall signal.
- Statistic: The detection statistic is the excess kurtosis of the signed coefficients. The presence of a wall creates heavy tails in the coefficient distribution compared to the Gaussian background.
Simulation-Based Inference (SBI):
- To bypass the intractability of explicit likelihood functions for complex, non-Gaussian signals, the authors use Neural Posterior Estimation (NPE) and Neural Ratio Estimation (NRE).
- NPE: A Convolutional Neural Network (CNN) featurizer combined with a normalizing flow is trained to learn the posterior distribution $p(\Theta | x)$ for parameters: scattering redshift ( $z$ ), symmetry-breaking scale ( $\eta$ ), and inclination ( $\theta$ ).
- NRE: A binary classifier is trained to learn the likelihood ratio $r(x, \Theta) = p(x|\Theta)/p(x)$ , enabling the construction of frequentist upper limits on $\eta$ .

Key Contributions and Results

Analytical Deflection Field: The paper provides an analytical computation of the photon deflection field for domain walls, demonstrating a characteristic "two-peak" structure in the deflection strength set by the relativistic displacement scale. The signal vanishes for face-on configurations ( $\theta = 90^\circ$ ) and is maximized for edge-on configurations.
Non-Gaussian Signatures: The study confirms that domain walls imprint coherent, planar distortions (demagnification and shear) on the CMB. These signatures are highly non-Gaussian and anisotropic, making them poorly captured by standard power spectrum analyses.
Sensitivity Forecasts:
- Curvelet/Wavelet Thresholds: Statistical thresholds based on kurtosis place the detectability of domain walls in a symmetry-breaking scale band of $\eta \sim 10\text{--}100$ MeV over the accessible redshift range. Curvelets outperform wavelets due to their directional sensitivity to the wall's geometry.
- Neural Ratio Estimator (NRE): The NRE, which exploits the full map morphology rather than compressing the data to a single moment, achieves significantly higher sensitivity. At redshifts $z \gtrsim 20$ , the NRE reaches a factor of $1.3\text{--}1.9$ lower in $\eta$ (or $\sim 2\text{--}7$ in wall tension) compared to the kurtosis-based methods.
- Noise Floor: The NRE sensitivity curve closely tracks the per-resolution-element noise floor of an SPT-3G-like survey, suggesting that the detection limit for the best methods is set by instrumental noise rather than angular resolution or statistical degeneracies.
Parameter Degeneracy: The analysis identifies a partial degeneracy between the symmetry-breaking scale $\eta$ and the scattering redshift $z$ (a stronger wall at higher $z$ mimics a weaker wall at lower $z$ ). However, the location of the overdense shell (set by the horizon) helps break this degeneracy, and the inclination angle $\theta$ is recovered independently.

Significance and Claims
The paper claims to provide a proof-of-concept that CMB lensing can be used to detect cosmological domain walls, specifically those in the low-energy regime that evade overclosure constraints. The significance lies in:

Methodological Advancement: Demonstrating that simulation-based inference (specifically NRE) can extract information from complex, non-Gaussian CMB signatures more effectively than traditional moment-based statistics (kurtosis) or two-point functions.
Parameter Space Exploration: Defining the idealized sensitivity reach for domain walls in the $(z, \eta)$ plane, identifying a viable detection window around $\eta \sim 10\text{--}100$ MeV for next-generation experiments.
Realistic Constraints: The authors explicitly state that their results are "idealized sensitivity forecasts." They do not claim to exclude the existence of domain walls in the real universe, as their simulations omit $\Lambda$ CDM lensing, foregrounds, point sources, and beam systematics. The curves represent the theoretical reach of the methods under optimal conditions (a single, optimally aligned wall on a Gaussian sky).
Future Direction: The paper positions these methods as a necessary precursor for applying similar pipelines to real data from experiments like SPT-3G, the Simons Observatory, CMB-S4, and LiteBIRD to either set genuine bounds or achieve a detection.

The work concludes that while the detection of such walls is currently limited by statistical power for moment-based methods and by the noise floor for the most advanced methods, the distinct non-Gaussian morphology of domain walls offers a viable avenue for discovery in upcoming high-resolution CMB surveys.

More like this