Relaxing Constraints on Inflation Models with Curvaton

The deliverable here is a map: which inflation models survive once a curvaton supplies part of the curvature perturbation. Chaotic monomials with α = 4 and 6 become viable for small curvaton amplitude, natural inflation is liberated over a range of F, and α ≥ 10 plus the Z_n supersymmetric new inflation stay dead. The specific bounds — X ≤ 0.137 for α = 6, X ≤ 0.088 for α = 8 — are new relative to Ref. [6], and the negative results are as useful as the positive ones. The framework is imported from [6] and [8] and the paper says so plainly in §1, which is the right way to do this. The background is evolved numerically through χD, RD1, φD and RD2 rather than assumed, the observational target is external and published (Leach–Liddle), and the conclusions are properly conditioned on Γχ, mφ and Γφ rather than stated unconditionally. That is honest workmanship on a modest question, and the modest question is worth answering for anyone building an inflation model who needs to know whether a curvaton rescues their potential. I would send it out. Three things need fixing first, and one of them is not what your report says it is. Eq. (2.24) as printed is wrong: in the curvaton-dominated limit it collapses to −2ε + 2η, which gives a blue tilt, while §3.1.2 explicitly uses −2ε. The intended form is −2ε + (2η − 4ε)/(1 + f̃²ε), and the code evidently used that. Since this is the paper's central formula, it cannot stay as written. Second, your weakest_assumption is aimed at the wrong place. The upper edge of the allowed region is set by f(X) ∝ 1/X in Eq. (2.22), not by the Ne reduction from second inflation. At X = 0.137 the curvaton energy density is roughly 5×10³⁸ GeV⁴, short of the ~10⁴¹ GeV⁴ needed to dominate while H > mφ = 100 GeV, so no second inflation occurs at the bound and Ne is standard there. The Ne reduction is real around X ~ 10, but it only deepens an exclusion f(X) already produced; disabling that module would not move the edge. The genuine defect is the paper's own sentence in §3.1.1 claiming curvaton contributions grow at large X, which contradicts Eq. (2.22) and reads as a slip for energy density rather than fluctuation. Third, the horizontal cutoff in Figs. 1, 3 and 4 is asserted in the captions and never derived anywhere in the text; a reader cannot tell what physics sets it. The Fig. 6 caption also mislabels the new inflation plot as natural inflation. All of these are one-pass fixes. The physics is sound and the bounds are believable; the prose around the bounds is not yet.

The reader's weakest_assumption is not the load-bearing element. The upper bound on X in Figs. 1, 3 and 4 is set by the 1/X falloff of f(X) in Eq. (2.22), not by the Ne reduction from second inflation. At the quoted bound X = 0.137 (α = 6, m_φ = 10² GeV), φinit = 3.3×10^17 GeV and V_curv = (1/2)m_φ²φinit² ≈ 5.5×10^38 GeV⁴. A second inflation requires the curvaton to dominate while still slowly rolling, i.e. while H > m_φ = 100 GeV, which needs V_curv > 3H²Mpl² ≈ 1.7×10^41 GeV⁴; this fails by a factor ~300. So at the bound the curvaton oscillates before it can dominate, Ne is standard (~50), and the bound is fixed by R = f̃²ε reaching ~1.4, i.e. by f(X) = 4/(9X). The Ne reduction is real at X ~ 10 (I estimate ~23 e-folds there, consistent with the paper's 20–30), but it only deepens an exclusion that f(X) has already produced. Disabling the second-inflation module would not move the upper edge of the allowed region. The genuine defect is the paper's own explanation in §3.1.1: 'in such region, the contributions from the curvaton fluctuation becomes large' contradicts Eq. (2.22), where f ∝ 1/X. The reader flagged this as a slip but did not connect it to the misidentification of the load-bearing mechanism. The numbers stand; the stated reason does not. A secondary point: the flattened Eq. (2.24) reads −2ε + 2η − 4ε/(1+f̃²ε), whose curvaton limit is −2ε + 2η, inconsistent with §3.1.2's use of −2ε; the intended form is presumably −2ε + (2η−4ε)/(1+f̃²ε), and §3.1.2 confirms the code used the latter.

The paper's central quantitative claim — that the curvaton relaxes constraints on chaotic, natural and new inflation, with upper bounds X ≤ 0.137 (α = 6) and X ≤ 0.088 (α = 8) at Γχ = 10^6 GeV, m_φ = 10² GeV, Γφ = 10^-10 GeV — rests on three ingredients: f(X) from Refs. [6, 8] (published), the spectral-index formula Eq. (2.24) (standard and derivable), and the observational constraint from Ref. [9] (published). All three are supported. The reader's weakest_assumption targets the Ne reduction from second inflation, but this is not load-bearing at the quoted bound: at X = 0.137 the curvaton energy density is ~300× too small to dominate while H > m_φ, so no second inflation occurs and Ne is standard. The bound is instead set by f(X) ∝ 1/X, which is cited and standard. The reader's concern would only affect the shape of the exclusion at X ~ 10, not the headline number. The real defect is the paper's stated mechanism in §3.1.1, which contradicts Eq. (2.22) and which the reader flagged only as a slip. This is a correctness issue in the explanation, not in the numbers. The reader's other two items (underived horizontal cutoff; Fig. 6 caption mislabelling new inflation as natural inflation) are real but minor. I therefore keep CONDITIONAL: the central claim survives, but the contradictory sentence and the mislabelled figure should be fixed before the paper is cited for its mechanism. I disagree with the reader's identification of the load-bearing assumption, hence 'partial' agreement.

The paper is a useful, systematic extension of Ref. [6] to several inflation models. The numerical treatment of the background, including the possibility of a curvaton-driven second inflation, is a genuine strength, and the observational benchmark is external and published. The results are falsifiable and the parameter dependence is transparent. The main weaknesses are interpretive: the central formula is misprinted and the physical explanation of the key exclusion mechanism is inconsistent with the paper's own Eq. (2.22).

The paper asks how much of the standard squeeze on inflation models survives if the primordial density ripples come partly from a second field, a curvaton, rather than entirely from the inflaton. In the standard picture the same slow-roll parameters that fix the perturbation amplitude also fix the tilt $n_s-1$ and the tensor-to-scalar ratio $r$, so chaotic inflation with a quartic potential sits at the edge of what the microwave background allows and steeper monomials are already gone. The authors show that adding a curvaton moves both numbers the same way: the scalar spectrum is diluted by the curvaton's share, pushing $n_s$ toward one and shrinking $r$, so models that looked dead can be brought back inside the observed window. The rescue is not universal: it works for quartic and sextic chaotic inflation and for natural inflation over part of its parameter space, but fails for monomial powers above about eight and for the supersymmetric new inflation model, where the inflaton's own tensor contribution is too small for the curvaton to matter.

What carries the argument. The load-bearing object is the combination $f^2 epsilon$, where $X = phi_{init}/M_{pl}$ and $f$ is proportional to $1/X$ for $phi_{init} ll M_{pl}$ and to $1/X$ with a different coefficient for $phi_{init} gg M_{pl}$. This term sits in the denominator of both the scalar amplitude and the tensor-to-scalar ratio, so a large curvaton share dilutes the scalar perturbation, flattens the tilt toward scale invariance and suppresses $r$, exactly the two directions the data push. The second piece of machinery is the numerical determination of the inflaton amplitude at horizon exit, $chi_*$, from the full thermal history: because a curvaton with a large initial amplitude can drive a second inflationary epoch, the e-folding number of the first inflation shrinks by roughly 20 to 30, which is what keeps the large-$X$ corner of parameter space excluded even though the curvaton share is largest there.

Core claim. The claim is that a curvaton, a late-decaying scalar whose own quantum fluctuations are converted into curvature perturbations when it decays, can move observationally excluded inflation models back into the allowed region of the $(n_s, r)$ plane. Because the curvaton's share of the perturbation scales as the inverse of its initial amplitude for amplitudes below the Planck scale, small initial amplitudes give a large share, and the spectral index and tensor-scalar ratio become $n_s-1 = -2 epsilon + 2 eta - 4 epsilon/(1+ f^2 epsilon)$ and $r = 16 epsilon/(1+ f^2 epsilon)$, where $f$ is the rescaled curvaton transfer function. The authors compute $n_s$ and $r$ at $k=0.01$ Mpc$^{-1}$ by numerically evolving the inflaton, the curvaton and the radiation bath through the inflaton-dominated, radiation, curvaton-dominated and final radiation eras, then compare against the 95% C.L. region of Leach and Liddle. For chaotic inflation with $V = lambda M_{pl}^4 (chi/M_{pl})^alpha$ they find viable regions at small $X = phi_{init}/M_{pl}$: with $Gamma_chi = 10^6$ GeV, $m_phi = 10^2$ GeV and $Gamma_phi = 10^{-10}$ GeV the bounds are $X le 0.137$ for $alpha = 6$ and $X le 0.088$ for $alpha = 8$, while $alpha ge 10$ stays excluded because even a dominant curvaton leaves $n_s-1$ near $-alpha/(2N_e)$, too red to fit. Natural inflation is opened up over a range of the decay constant $F$, with the permitted $X$ shrinking as $F$ decreases, and the supersymmetric $Z_n$ new inflation model is not saved at all because its $epsilon$ is of order $10^{-21}$.

The premise it rests on. The exclusion of large curvaton amplitudes rests on the authors' own numerical evolution of the background, which says a curvaton-driven second inflation shortens the first inflationary epoch by 20 to 30 e-foldings and thereby pushes the spectral index too red; that reduction is not cross-checked against an independent calculation.

Quartic chaotic inflation, which sits at $(n_s-1, r)$ near $(-0.06, 0.32)$ for $N_e = 50$ and is on the verge of exclusion, becomes viable once the curvaton supplies most of the perturbation with $X$ below roughly 0.1.

The rescue has a ceiling in the monomial power: $alpha = 6$ and $alpha = 8$ survive with $X le 0.137$ and $X le 0.088$ respectively, but $alpha ge 10$ remains excluded for the decay rates and masses considered.

Natural inflation gains an allowed band in the $F$ versus $X$ plane, with the permitted curvaton amplitude shrinking as the decay constant $F$ falls.

The supersymmetric $Z_n$ new inflation model with a small scale $v$ cannot be saved, because its $epsilon$ is of order $10^{-21}$ and the curvaton enters only through $f^2 epsilon$.

A larger number of e-foldings, obtained from a smaller inflaton decay rate or a larger curvaton decay rate, is the remaining lever for the steepest monomial potentials.

Because the rescue works precisely by suppressing $r$, the curvaton-dominated branch makes a sharp bet: a future detection of a tensor-to-scalar ratio at the $10^{-1}$ level would remove the very parameter space that makes quartic chaotic inflation viable here.

The dilution mechanism should apply to any model whose only defect is a too-red $n_s$ with an acceptably small $epsilon$; the ceiling near $alpha = 10$ is really a statement about how red the tilt can be before dilution can no longer compensate.

If the curvaton also leaves a detectable non-Gaussian signature, the small-$X$ corner where the rescue lives is where to look, since that is where the curvaton's share of the perturbation is largest.

The shortening of the first inflationary epoch by a second inflationary phase is generic to two-field setups, which suggests that constraints quoted at a fixed $N_e = 50$ can mislead when a late-decaying scalar is present.

This review was written by a machine. No person has reviewed this paper on Pith yet. A scientist listed on Pith can review it for a fee, and their review is published here whatever it concludes.

Request a human review