Cosmology and Primordial Black Holes

Cosmology and primordial-black-hole research joins evidence across enormous ranges of scale, but the join is always model dependent. This field contains six dated assessments organized around three questions: when a cosmological discrepancy supports new gravity; when an early-Universe or stochastic signal has a primordial source; and when heterogeneous primordial-black-hole constraints can be turned into a population statement. The pages share a discipline: keep data releases separate, carry formation and transfer assumptions into every inference, and distinguish a surviving parameter region from positive evidence.

Two linked evidence maps

The first map runs from cosmological observations to physical interpretation. Cosmological Parameter Tensions and Gravity asks whether persistent discrepancies survive calibration, covariance, model, and prior changes and whether any predictive gravity theory is selected. Cosmological Backreaction and Apparent Acceleration tests a more specific mechanism: can nonlinear inhomogeneity and light-cone effects reproduce acceleration without dark energy under gauge-robust, observation-matched evolution?

The second map runs from primordial perturbations and compact objects to present messengers. Can Primordial Black Holes Be All the Dark Matter? poses the global population-existence question. Combining Primordial-Black-Hole Constraints supplies the statistical rules needed before heterogeneous curves can answer it.

Two gravitational-wave records provide source and transfer tests. Scalar-Induced Gravitational Waves and PBHs asks whether one scalar probability distribution predicts both a stochastic spectrum and an allowed PBH abundance. Primordial Tensor Backgrounds broadens the view to inflation, phase transitions, defects, and other early-Universe sources across CMB, pulsar-timing, space, and ground-based bands.

Choose an entry route

Inference and public data. Start with the distance-likelihood project if you can calculate FLRW distances and explain a correlated Gaussian likelihood. Its reading order is the local inference background, the DESI distance definitions and inference method, and only then the dark-energy interpretation. Bring a reproduced baseline contour and a list of shared inputs to an adviser. Posterior plotting is accessible on a laptop; new survey likelihoods need collaboration with people who understand calibration, selection, and mock validation.

Formation physics and compact-object populations. Start with the single-constraint recast, then the mass-function-tail project. Derive the normalization and response formula before selecting a mass window. This route is a good fit for students who enjoy probability tails, numerical integration, and comparing assumptions across experiments. Reliable formation predictions require collapse and early-Universe expertise; a scan of a convenient distribution is preparation for that work, not a formation theory.

Perturbation theory and gravitational waves. The abundance-bridge project asks you to extract a scalar spectrum, window, transfer function, and collapse prescription from one analysis. The tensor-constraint project instead starts from a released marginalized likelihood. Choose according to whether you want to understand the forward physics or the inference pipeline; eventually a joint claim needs both.

Relativistic computation. The averaging project starts with an integrability identity and a homogeneous control. Move to full simulations only with constraint-satisfying data, numerical-relativity supervision, and resources for convergence and ray tracing. A thesis proposal should name one operational observable and one controlled comparison before promising an explanation of acceleration.

In each route, the first product is a reproduction or diagnostic. The next scientific question becomes viable when its inputs are available, its uncertainty can be bounded, and a literature check establishes what remains unanswered. The dated assessments below help choose that question; they do not certify a proposed extension as novel.

How the questions connect

Cosmological tensions supply possible motivation for changing the late-time model, but they do not identify the mechanism. The backreaction record takes one candidate explanation and raises the evidentiary bar: an averaged term must become a gauge-robust light-cone observable and fit the same supernova, BAO, CMB, and growth data. Thus the tensions page prevents a fit improvement from being called gravity, while the backreaction page prevents a formal mechanism from being called an observational explanation.

The PBH pages form a dependency chain. Formation physics produces a mass function with tails, spin, clustering, and a history. Each observational probe responds to that population under its own nuisance model. The constraint-synthesis page asks whether those responses are linear and independent enough to combine. Only then can the all-dark-matter page ask whether a normalized present fraction of one survives. Reading the all-dark-matter verdict without the synthesis rules encourages the false inference that the lower edge of a plotted envelope is a joint likelihood.

Scalar-induced waves connect PBH formation to the primordial-background record. A stochastic spectrum probes a convolution of scalar modes; PBH abundance probes an exponentially sensitive rare tail. Similar wave spectra can therefore coexist with very different PBH fractions. The primordial-tensor page adds a second safeguard: nanohertz, CMB, and ground-band results cannot be joined by drawing one power law unless a source mechanism and thermal transfer actually predict it.

The observational products also overlap. PTA source templates reuse the same timing data; LVK early-Universe translations reuse the same cross spectra; DESI analyses can combine different releases and tracers. Independence lives at the level of data, calibration, likelihood, and theory inputs—not at the level of paper titles.

Shared standards for cosmological evidence

Every numerical result carries its release, calibration, selection, likelihood, priors, and cutoff. A parameter inferred under ΛCDM is not a direct observation; a likelihood improvement in a flexible extension is not source identification; a mock-biased estimator should not survive because it is physically interesting. Current products must not inherit numbers from an older release by name alone.

Population claims require the same care. A monochromatic PBH curve cannot be applied to an extended distribution without its response assumptions. A background fit does not establish PBHs, because waves and collapse sample different parts of the scalar statistics. A remaining sensitivity gap is permission, not evidence. Strong conclusions arise when independent observations test additional predictions of the same fixed model.

Reproduction also has layers. Public chains permit posterior and plotting checks, while raw timing, strain, map, or survey products are needed to reproduce the likelihood. Simulations require initial data, code revision, resolution and box convergence, gauge control, and a forward observable. Each leaf states which layer was not rerun, so a reader can choose the next technically meaningful step without mistaking artifact availability for independent confirmation.