In the standard cosmological paradigm, the initial seeds of cosmic structure are modeled as a Gaussian random field. Under this assumption, the formation of small-scale structures—such as dark matter halos and galaxies—is locally determined by the immediate matter density. On very large scales, this relationship simplifies to a constant proportionality factor known as the linear clustering bias ($b$). This means that on large scales, the spatial distribution of halos simply mirrors the underlying matter distribution up to a constant multiplier.
However, if the early universe experienced a period of inflation that departed from this minimal setup, it would leave behind primordial non-Gaussianity (PNG). Local-type PNG introduces a profound twist to structure formation: it physically couples long-wavelength (large-scale) fluctuations to short-wavelength (small-scale) fluctuations. This mode coupling breaks the scale-independence of cosmic bias, causing the halo bias to vary dramatically across large scales.
Understanding why and how this scale-dependent bias manifests is a primary objective of modern large-scale structure analysis.
The Intuition of Mode Coupling
To understand why local PNG induces a scale-dependent bias, consider how halos form. Halos collapse from small-scale density peaks that exceed a critical density threshold. In a perfectly Gaussian universe, a long-wavelength background density mode merely shifts the local baseline of these peaks up or down, acting as a uniform background.
Local PNG (often parameterized by $f_{NL}$) changes this by making the small-scale fluctuations explicitly dependent on the large-scale primordial gravitational potential, $\phi$. Instead of a simple baseline shift, a long-wavelength potential mode modulates the amplitude (the variance) of the small-scale fluctuations. Where $\phi$ is positive, small-scale fluctuations are amplified, making density peaks higher and more frequent, thereby accelerating halo collapse. Where $\phi$ is negative, fluctuations are suppressed.
$$\delta_{\text{local}} \propto \phi_{\text{long}}$$
Because the cosmic density fluctuation field ($\delta$) is related to the gravitational potential ($\phi$) through the Poisson equation, they scale differently in Fourier space:
$$\delta(k) \propto k^2 \phi(k)$$
Consequently, a modulation that is physically proportional to the potential $\phi$ translates into a density bias that scales as $1/k^2$. At exceptionally large scales (where the wavenumber $k \to 0$), this $1/k^2$ term grows rapidly. As noted by Eiichiro Komatsu et al. in Non-Gaussianity as a Probe of the Physics of the Primordial Universe and the Astrophysics of the Low Redshift Universe, this unique scale-dependent clustering signature allows cosmologists to constrain early-universe physics directly from large-scale galaxy surveys without computing computationally expensive higher-order correlation functions.
The Peak-Background Split and Stochastic Barriers
The mathematical foundation for calculating this effect is the peak-background split formalism. In this framework, the density field is separated into long-wavelength background modes and short-wavelength peak modes.
Anže Slosar et al. extended this formalism to non-Gaussian initial conditions in Constraints on local primordial non-Gaussianity from large scale structure, demonstrating how local $f_{NL}$ modifies the halo mass function across large spatial patches. This theoretical framework establishes a direct consistency relation: the scale-dependent halo bias is proportional to the derivative of the halo mass function with respect to the amplitude of initial fluctuations. This relation was explicitly verified using N-body simulations by Matteo Biagetti et al. in Verifying the consistency relation for the scale-dependent bias from local primordial non-Gaussianity.
Gaussian Universe:
[Long-wave mode] ---> Simple baseline shift ---> Constant Bias (b)
Local PNG Universe:
[Long-wave potential φ] ---> Modulates short-wave variance ---> 1/k² Scale-Dependent Bias
However, standard peak-background split models rely on simplified, deterministic assumptions—such as assuming that a halo collapses as soon as a density peak crosses a flat, fixed threshold. Real cosmic collapse is non-spherical and highly complex.
To capture this physics, Matteo Biagetti et al. implemented a stochastic moving barrier within the excursion set peak approach in Scale-dependent bias from an inflationary bispectrum: the effect of a stochastic moving barrier. They discovered that incorporating a realistic, stochastic barrier significantly alters the predicted amplitude of the non-Gaussian bias, particularly for lower-mass dark matter halos. This indicates that simpler, deterministic models can systematically miscalculate the expected PNG signal.
The Complication of Halo Assembly Bias
A major challenge in extracting clean cosmological signals from galaxy surveys is that galaxies do not map perfectly to halo mass. Halos of the exact same mass can exhibit different clustering behaviors based on their formation history, concentration, or accretion rate—a phenomenon known as halo assembly bias.
When measuring scale-dependent bias, assembly bias introduces significant systematic shifts. Boryana Hadzhiyska et al. demonstrated in Modest set of simulations of local-type primordial non-Gaussianity in the DESI era that the PNG-induced bias parameter ($b_\phi$) shows a strong response to halo concentration and accretion rates. If a survey targets a specific population of galaxies that preferentially occupy highly concentrated halos, the observed scale-dependent bias will deviate from pure mass-based predictions, potentially distorting constraints on $f_{NL}$.
This interplay between internal halo properties and large-scale clustering is a recurring theme:
- Benjamin D. Wibking et al. introduced the environmental parameter $Q_{\text{env}}$ in Emulating galaxy clustering and galaxy–galaxy lensing into the deeply non-linear regime: methodology, information, and forecasts to absorb assembly bias effects down to deeply non-linear scales.
- Sihan Yuan et al. showed in Exploring the squeezed three-point galaxy correlation function with generalized halo occupation distribution models that accounting for halo assembly history is essential for breaking degeneracies when analyzing higher-order clustering statistics like the squeezed three-point correlation function.
- Early empirical work by Anže Slosar et al. noted that the merger histories of highly active tracers, such as quasars, can suppress or modify the scale-dependent bias signal compared to quiescent halos of identical mass.
Non-Local Environments and Covariance Realities
As observations move from ultra-large scales down to intermediate scales (below 60 Megaparsecs), the local density approximation breaks down entirely. Halos are shaped not just by the density at a single point, but by the surrounding gravitational tidal fields.
To model this regime, the bias framework must be expanded to include non-local terms. Matteo Biagetti et al. developed and measured these quadratic non-local terms in Nonlocal halo bias with and without massive neutrinos and Measuring non-local Lagrangian peak bias, demonstrating that accounting for non-local Lagrangian peak bias is necessary to accurately trace halo clustering in complex cosmologies.
Furthermore, because the scale-dependent bias physically originates from the coupling of long and short modes, it is deeply connected to the "squeezed" configuration of the bispectrum (a three-point statistic where one wavenumber is much smaller than the other two). However, analyzing these configurations introduces severe statistical challenges.
In The covariance of squeezed bispectrum configurations, Matteo Biagetti et al. revealed that the common practice of using a simplified Gaussian approximation for the bispectrum covariance matrix fails significantly. Because non-Gaussian correlations introduce strong off-diagonal terms, neglecting them artificially overestimates the precision of $f_{NL}$ constraints by roughly a factor of two.
Alternative Tracers: Voids, Bubbles, and Peaks
While dark matter halos and galaxies are the standard tracers used to map scale-dependent bias, the underlying physics applies to any cosmic structure that forms via threshold crossing. This has led to the exploration of alternative cosmic tracers:
- Cosmic Voids: Voids are vast, underdense regions in the cosmic web. Kwan Chuen Chan et al. demonstrated in Constraint of void bias on primordial non-Gaussianity that cosmic voids exhibit a distinct scale-dependent bias in the presence of PNG, acting as a negative mirror to halo bias. Incorporating voids into a multi-tracer analysis helps break parameter degeneracies and can double the precision of $f_{NL}$ measurements.
- Ionized Bubbles: During the Epoch of Reionization, the first generation of stars and quasars generated expanding bubbles of ionized hydrogen. Hiroyuki Tashiro demonstrated in Constraining primordial non-Gaussianity with CMB–21 cm cross-correlations? that local PNG alters the spatial bias of these ionized regions. This scale-dependent effect amplifies the large-scale cross-correlation signal between the Cosmic Microwave Background (CMB) and the 21 cm emission line, providing an independent high-redshift probe.
- Density Peaks and Dips: Rather than focusing exclusively on bound halos, Adi Nusser et al. showed in Abundance of peaks and dips in 3D mass and halo density fields: a test for cosmology that tracking the raw geometric abundance of local maxima (peaks) and minima (dips) in the continuous density field provides an alternative, highly sensitive method for isolating non-Gaussian bias signatures.
Mapping the Physics of Inflation
The ultimate goal of isolating scale-dependent bias is to constrain the physical mechanisms driving cosmic inflation. Standard single-field inflationary models predict a negligible local non-Gaussianity signal. Conversely, multi-field inflationary models can generate significant, detectable values of $f_{NL}$. As explored by Matteo Biagetti et al. in Testing multifield inflation with halo bias, measuring the scale-dependent clustering of halos provides a direct experimental test capable of ruling out broad classes of multi-field scenarios.
Moreover, more complex inflationary models predict that non-Gaussianity is not a static constant, but rather "runs" (changes its amplitude) depending on the physical scale being observed. To constrain this running, Matteo Biagetti et al. showed in Testing the running of non-Gaussianity through the CMB $\mu$-distortion and the halo bias how large-scale halo bias measurements can be cross-referenced with small-scale CMB spectral distortions.
To fully exploit these multi-scale signatures, Nishant Agarwal et al. argued in Constraining the initial conditions of the Universe using large scale structure for generalizing the standard local ansatz parameterization. Moving beyond rigid templates ensures that future large-scale structure surveys can flexibly detect diverse, scale-dependent shapes of primordial non-Gaussianity, unlocking a definitive fossil record of the early universe.
Key Foundational Work
Constraints on local primordial non-Gaussianity from large scale structure (2008) — contributed 7 concept elements to this direction.
Non-Gaussianity as a Probe of the Physics of the Primordial Universe and the Astrophysics of the Low Redshift Universe (2009) — contributed 1 concept elements to this direction.