The Fate of Large Scale White Noise in Second-Order Cosmological Perturbation Theory

The paper The Fate of Large Scale White Noise in Second-Order Cosmological Perturbation Theory by Aurora Ireland re-examines the Large Scale White Noise (LSWN) effect in cosmological perturbation theory, working to second order. The paper confirms that the kurvature density variable ∆ρ does indeed develop LSWN at second order, but it refutes the claim from a previous paper (Ref. [2]) that this white noise is inherited as an infrared (IR) pole in the second-order comoving curvature perturbation R2.

The paper's central findings are:

1. The white noise in ∆ρ is genuine: The paper explicitly computes the quadratic source Q for a perfect fluid during radiation domination, showing it is finite and non-vanishing in the soft limit k → 0. The result is expressed as a constant, scale-independent (white noise) contribution to the large-scale kurvature power spectrum: P LSWN∆ρ2 ≡ lim k→0 P∆ρ2(k) = 4π ∫0∞ (dq/q4) f(η, q)2 P R1(q)2. This is consistent with the claims of Refs. [1, 2].

2. The white noise is not inherited by R2: The paper demonstrates that the apparent IR enhancement of R2 in Ref. [2] was an artifact of using a linear Poisson equation to relate two genuinely second-order quantities. The correct second-order relation is ∆ρ2 = (1/4πGa2)[∇2R2 + Q − ∇2C], where Q ≠ ∇2C. In the soft limit, ∇2C vanishes while Q tends to a finite constant, meaning the white noise content of ∆ρ2 resides solely in Q; the term omitted by the linear relation is not a small correction, but rather the entirety of the white noise.

3. Conservation law protects R2: The paper proves that lim k→0 k2R2 = 0. This follows from a conservation law in the soft limit: the coefficient of the O(ϵ−2) pole in R2, defined as F ≡ P′/H + 3P + Ξ̃0, is an exact constant of motion. The paper shows F′ = 0 identically, and since the initial value of F is vanishing (due to the absence of a primordial second-order mode), it remains zero. This result is robust: it follows directly from the structure of the evolution equation (61) with appropriate boundary conditions, regardless of the explicit form of the quadratic sources Ξ.

4. The bound on k BH is invalid: Since the white noise in ∆ρ2 does not translate to an IR enhancement of the curvature power spectrum, the constraint proposed in Ref. [2] on the scale k BH and its implications for the small-scale primordial power spectrum are invalid. The paper states: The bound on k BH, and by extension its implications for the small-scale power spectrum, do not apply.

The paper also computes the exact weight function W(η, q) for the kurvature power spectrum during radiation domination, showing it grows with wavenumber q, which would have made the integral dominated by small scales for a scale-invariant spectrum. However, this does not lead to the claimed constraint on the primordial power spectrum.

The paper concludes by noting that R2 itself develops a white noise contribution at second order from the quadratic completion term C (which is O(ϵ0)), but this is distinct from the IR enhancement claimed in Ref. [2]. Future directions include examining imperfect fluids, including vector and tensor modes, establishing the mapping to CMB observables at second order, and understanding the physical meaning of the conserved quantity F.

Improvement 1: Physics-Aware Error Detection in Perturbative Calculations

• What the AI can do: Automatically audit second-order cosmological perturbation theory derivations by checking whether a linearized relation (e.g., Poisson equation) is being used to connect two genuinely second-order quantities. It will flag cases where a term like Q (quadratic source) is omitted, and verify if the omitted term is non-negligible in the soft limit (k→0).

• Specific capability: Given a symbolic expression for a second-order variable, the AI can compare the full nonlinear relation against any simplified linear approximation, compute the soft-limit behavior of each term, and warn if a finite white-noise contribution is being incorrectly attributed to an IR pole in a different variable (e.g., R2 vs. ∆ρ).

Improvement 2: Conservation-Law Verification for Gauge-Invariant Quantities

• What the AI can do: Automatically derive and test conservation laws for curvature perturbations in the soft limit. Given the evolution equation for a perturbation (e.g., Eq. 61 in the paper), the AI can identify candidate conserved quantities (like F ≡ P′/H + 3P + Ξ̃0), compute their time derivative symbolically, and verify if it vanishes identically. It will then check whether initial conditions (e.g., absence of primordial second-order modes) force the conserved quantity to zero, thereby proving the absence of IR poles.

• Specific capability: The AI can take any second-order perturbation equation, extract the coefficient of the O(ϵ−2) pole, and automatically test if that coefficient is a constant of motion. If yes, it can conclude that the perturbation cannot grow a white-noise tail in the IR, regardless of the explicit form of the source terms.

Improvement 3: Soft-Limit Integral Evaluation with Divergence Diagnostics

• What the AI can do: Evaluate integrals like P LSWN∆ρ2 = 4π ∫0∞ (dq/q4) f(η, q)2 P R1(q)2 with automatic handling of IR/UV divergences. It will identify whether the integral is dominated by small scales (large q) or large scales (small q), and whether the result is truly scale-independent (white noise) or an artifact of a specific weight function.

• Specific capability: The AI can compute the weight function W(η, q) for any given background (e.g., radiation domination) and report its asymptotic behavior. If the weight grows with q, the AI will flag that the integral is UV-dominated and that any resulting constraint on the primordial power spectrum is invalid unless the small-scale spectrum is explicitly bounded—preventing the erroneous inference of a bound on k BH.

Improvement 4: Automated Refutation of Prior Claims via Counterexample Construction

• What the AI can do: When a paper claims an IR enhancement (e.g., in R2), the AI can automatically construct a counterexample by solving the full second-order equations numerically or symbolically for a simple fluid (e.g., radiation domination) and comparing the true soft-limit behavior of R2 against the claimed pole. It will then output a diagnostic report showing whether the pole is real or an artifact of an invalid linearization.

• Specific capability: The AI can take a prior paper’s derivation (e.g., Ref. [2]), identify the step where a linear relation was used, and generate a corrected version with the full quadratic completion term C. It will then show that ∇2C → 0 while Q → const, proving that the white noise resides solely in Q and not in R2.

Improvement 5: Cross-Variable Consistency Check for Cosmological Observables

• What the AI can do: Given a set of second-order perturbation variables (e.g., ∆ρ, R, C, Q), the AI will enforce consistency between their power spectra and cross-correlations. It will check that a white-noise signal in one variable (e.g., ∆ρ) does not automatically imply a white-noise signal in another (e.g., R), unless a valid mapping (e.g., via the full Einstein equations) is provided.

• Specific capability: The AI can generate a consistency matrix for all second-order variables, showing which ones share IR behavior. It will flag any paper that claims a constraint on k BH based on a single variable’s white noise without verifying that the curvature perturbation itself is protected by a conservation law. This prevents the propagation of invalid bounds to CMB observables.

Improvement 6: Automated Extension to Imperfect Fluids and Vector/Tensor Modes

• What the AI can do: Using the paper’s methodology (conservation law + soft-limit analysis), the AI can automatically generalize the proof to imperfect fluids (with viscosity, heat flux, anisotropic stress) and include vector and tensor perturbations. It will recompute the conserved quantity F and the quadratic source Q for these cases, and test whether the white-noise non-inheritance result holds.

• Specific capability: The AI can take a given stress-energy tensor (e.g., with shear viscosity) and automatically derive the second-order evolution equations, identify the soft-limit pole structure, and output a verdict on whether lim k→0 k2R2 = 0 remains true. This extends the paper’s result beyond perfect fluids, providing a general theorem for IR safety of curvature perturbations.

Working to second order in cosmological perturbation theory, we reconsider the Large Scale White Noise (LSWN) effect proposed in [2511.13866] and [2511.15803]. We demonstrate that the kurvature density variable rho does indeed develop LSWN at second order. However, contrary to the claims of [2511.15803], we show that this is not inherited as an infrared (IR) pole in the second-order comoving curvature perturbation R 2. This apparent enhancement was an artifact of using a linear Poisson equation to relate two genuinely second order quantities. Further, we show that R 2 is protected from developing any such IR enhancement: k to 0 (k squared R 2) = 0, which follows from a conservation law in the soft limit. The constraint proposed in [2511.15803] and its implications for the small-scale primordial power spectrum are therefore invalid.