Independent connection in action during inflation

This paper asks whether the inflationary epoch can be explained without introducing an ad hoc inflaton sector, by instead using a dynamical affine connection that is independent of the metric (a key feature of metric-affine gravity). The motivation is observational: recent Atacama Cosmology Telescope (ACT) results, when combined with other datasets (Planck and DESI baryon acoustic oscillations), favor a higher scalar spectral index than earlier Planck and BICEP/Keck (BK18) determinations. In particular, the combined analysis labeled P-ACT-LB is reported to give a relatively high value of the scalar tilt, with the paper quoting

$n_s=0.9743\pm 0.0034$.

Inflation models that matched the older Planck/BK18 constraints can become strained under this shift, so the author’s goal is to show that a geometrical mechanism—where a component of the connection behaves as the inflaton—can reproduce the ACT-favored region of $n_s$, while also predicting the running ${\alpha }_s$, the scalar amplitude $P_R$, and the tensor-to-scalar ratio $r$.

The core theoretical setup is a minimal metric-affine action in which the relevant inflationary dynamics comes from curvature invariants built from the affine connection ${\mathcal{A}}_{\mu \sigma }^{\rho }$. The action used for inflation is

$$S_I=\int d^{4}x\sqrt{-g}\left(\alpha \mathcal{R}+\beta {\mathcal{R}}^{\prime }+c{\mathcal{R}}^{\prime 2}\right),$$

where $\mathcal{R}$ is the double contraction of the affine curvature and ${\mathcal{R}}^{\prime }$ is a parity-odd invariant constructed using the Levi-Civita tensor (a Holst-type invariant). In the GR limit, the connection reduces to the Levi-Civita connection, making ${\mathcal{R}}^{\prime }$ vanish; thus ${\mathcal{R}}^{\prime }$ encodes extra connection degrees of freedom beyond standard GR. The parameters $\alpha$ and $\beta$ have dimensions of energy squared and $c$ is dimensionless. The author then performs an auxiliary-field transformation introducing a field $\omega$ that trades the curvature-squared term for a linear-in-curvature form. After solving the affine connection equations of motion exactly, the action becomes equivalent to GR coupled to a scalar field $\omega$:

$$S_I=\int d^{4}x\sqrt{-g}\left[\alpha R-\frac{(\partial \omega {)}^2}{2}-\mathcal{U}(\omega )\right].$$

The resulting potential $\mathcal{U}(\omega )$ is a function of $\omega$, $\beta$, and $c$, with a parameter $X(\omega )$ defined in terms of $\omega$ and $\beta$. The author fixes $\alpha =M_P^2/2$ to recover the Einstein-Hilbert term. The scalar $\omega$ is identified as a pseudoscalaron: it has odd parity because it shares the symmetry properties of ${\mathcal{R}}^{\prime }$. The potential has a stationary point at $\omega =0$, and the mass is computed as

$$m_{\omega }^2=\frac{16{\beta }^2+M_P^4}{48c{M}_P^2}.$$

Stability requires $c>0$ to avoid tachyons. For $\sqrt{|\beta |}\gtrsim M_P$, the potential develops a plateau, which is the typical ingredient for slow-roll inflation.

Methodologically, the paper uses the slow-roll approximation, defining the standard slow-roll parameters

$$ϵ\equiv \frac{M_P^2}{2}{\left(\frac{1}{\mathcal{U}}\frac{d\mathcal{U}}{d\omega }\right)}^2\ll 1,\eta \equiv \frac{M_P^2}{\mathcal{U}}\frac{d^2\mathcal{U}}{d{\omega }^2}\ll 1.$$

The number of e-folds is computed via an integral over the potential and its derivative, and inflationary observables at horizon exit are given by

$$n_s=1-6ϵ+2\eta ,r=16ϵ,P_R=\frac{\mathcal{U}/ϵ}{24{\pi }^2{M}_P^4}.$$

The running of the spectral index is also included:

$${\alpha }_s=16ϵ\eta -24{ϵ}^2-2{\xi }_2,$$

with

$${\xi }_2\equiv \frac{M_P^4}{{\mathcal{U}}^2}\frac{d\mathcal{U}}{d\omega }\frac{d^3\mathcal{U}}{d{\omega }^3}.$$

A key additional analysis addresses initial conditions. The author notes that slow-roll implicitly assumes the initial kinetic energy ${\Pi }^2={\dot{\omega }}^2$ is small compared to the potential energy on the plateau. While some generic arguments suggest that large ${\Pi }^2$ decreases, the paper explicitly demonstrates an attractor mechanism in this model by studying the dynamical system for homogeneous and isotropic metrics. The system is written in terms of derivatives with respect to e-fold time $N_e$:

$${\Pi }^{\prime }=-3\Pi -\frac{1}{H}\frac{d\mathcal{U}}{d\omega },{\chi }^{\prime }=\frac{\Pi }{H},H^2=\frac{{\Pi }^2+2\mathcal{U}}{6M_P^2},$$

(where $\chi$ tracks the field evolution). Phase-space behavior is shown (via a figure) to illustrate that even if ${\Pi }^2$ starts comparable to or larger than $\mathcal{U}$, the dynamics drives $\Pi$ to negligible values when $\omega$ reaches the plateau, thereby justifying slow-roll without fine-tuned initial momentum.

Regarding observational comparison, the paper emphasizes that the overall normalization of the potential is controlled by $1/c$, while the shape-dependent slow-roll quantities $ϵ$, $\eta$, and ${\xi }_2$ (and thus $n_s$, $r$, and ${\alpha }_s$) are independent of this normalization. Therefore, the scalar amplitude $P_R$ can always be matched to the observed value by choosing $c$ appropriately; the author quotes an observational target $P_R\simeq (2.10\pm 0.03)\times 10^{-9}$. The main nontrivial comparison is thus in the $(n_s,r)$ plane and in the predicted running.

The paper reports that slow-roll inflation in this minimal model can “remarkably accommodate” ACT bounds for appropriate e-fold numbers $N_e$. It also states that ACT favors values of $\sqrt{|\beta |}$ closer to $M_P$ than the parameter regions favored by earlier Planck and BK18 constraints. Concretely, the author provides approximate lower bounds on $\sqrt{|\beta |}$ to remain within a $1\sigma$ region for two choices of e-folds: for $N_e=60$, $\sqrt{|\beta |}\gtrsim 4M_P$; for $N_e=49$, $\sqrt{|\beta |}\gtrsim 3M_P$. The paper also notes that for these values the slow-roll approximation remains valid (shown in an inset figure).

For the running, the author’s qualitative quantitative claim is that the predicted ${\alpha }_s$ is “so small” that it lies within about $1\sigma$ observational bounds from both referenced datasets, and that ${\xi }_2$ remains small across relevant $\beta$. In the conclusions, the paper makes a sharper falsifiability statement: any evidence for ${\alpha }_s$ significantly above the order of $10^{-4}$ would falsify the model. The tensor-to-scalar ratio $r$ is computed from $ϵ$ and is plotted against $n_s$ in figures comparing ACT, Planck, and BK18 constraints, but the text provided does not include explicit numerical values of $r$ or $n_s$ at best fit.

The author also discusses reheating and the role of the Higgs boson. In the minimal setup, reheating is facilitated by adding a non-minimal coupling between the Higgs field $\varphi$ and the affine-curvature scalar $\mathcal{R}$:

$$S_{\mathrm{nm}}=\int \sqrt{-g}\frac{\xi {\varphi }^2}{2}\mathcal{R}.$$

The paper claims that this allows the pseudoscalaron to decay into two Higgs bosons after inflation and yields a reheating temperature

$$T_{\mathrm{RH}}\gtrsim 10^9|\xi |\text{GeV},$$

for the minimal setup considered and for $|\beta |\gtrsim M_P^2$. It further argues that even for small $|\xi |$, this temperature is well above what observations require, while for $\xi$ not large the Higgs dynamics during inflation is negligible in this minimal model. The paper contrasts this with Higgs inflation scenarios in both metric and metric-affine gravity that typically require $\xi \gg 1$ to flatten the Higgs potential; here, the author suggests that extended models might allow Higgs inflation even at $\xi =0$, but that is outside the minimal analysis.

Limitations are not presented as a formal list, but they are implied by the methodology and scope. The analysis relies on the effective field theory description of the metric-affine action and assumes that the inflationary energy scale ($\sim \mathcal{U}$) stays below the Planck cutoff; the paper notes that for an example choice $\beta =-20M_P^2$, the potential energy is well below the Planck scale. The slow-roll approximation is used throughout, with validity checked for the parameter regions of interest. The observational comparison is qualitative in the provided text (figures are referenced but numerical best-fit values and $p$-values are not quoted). Finally, reheating is treated briefly and depends on introducing the Higgs non-minimal coupling term, so the mechanism is not fully determined within the minimal inflationary sector alone.

Practically, the results matter for model builders and phenomenologists because they propose a concrete geometrical origin for inflation that can shift the predicted $n_s$ toward the ACT-favored higher values while keeping ${\alpha }_s$ very small. Researchers interested in inflationary parameter inference should note the model’s emphasis on the scale $\sqrt{|\beta |}$ being near $M_P$ and on the attractor behavior that reduces sensitivity to initial kinetic energy. Observationally, the paper highlights a testable prediction: the running ${\alpha }_s$ should be at or below the $10^{-4}$ scale, and future CMB polarization experiments such as LiteBIRD are suggested as a way to further test whether $n_s$ lies near the upper observational bound. Overall, the work argues that if ACT’s preference for higher $n_s$ persists, a metric-affine geometrical inflaton (pseudoscalaron) provides a viable and dynamically robust explanation.