Reheating ACTs on Starobinsky and Higgs inflation

This paper demonstrates that despite recent ACT DR6 data appearing to exclude Starobinsky and Higgs inflation models at the $2\sigma$ level, a Bayesian analysis incorporating non-instantaneous reheating reveals that these models remain consistent with observational data within a significant portion of their parameter space, particularly when allowing for a reheating equation of state greater than $0.5$.

The Gist

The Big Picture: A Cosmic "Glitch" in the Data

Imagine the universe as a giant, expanding balloon. For decades, scientists have had a very popular theory about how this balloon started: Inflation. This is the idea that right after the Big Bang, the universe didn't just grow; it exploded outward faster than the speed of light for a tiny fraction of a second.

Two specific theories about how this inflation happened are the Starobinsky model and the Higgs model. Think of these as two very famous, well-dressed suspects who have been the prime suspects in the "Origin of the Universe" case for years. They fit the evidence perfectly... until recently.

The Problem:
In 2025, a telescope called the Atacama Cosmology Telescope (ACT) released new, super-sharp data (DR6). It measured a specific "fingerprint" of the early universe (called the scalar spectral index, $n_s$). The new number was slightly higher than expected.

When the scientists plugged this new number into the Starobinsky and Higgs models, the math said: "These models are guilty!" The data seemed to rule them out with 95% confidence (a "2-sigma" level). It looked like the case was closed: these two famous theories were dead.

The Twist: The "Reheating" Break

The authors of this paper, however, decided to look closer at the scene of the crime. They realized everyone was making a simplifying assumption that might be wrong.

The Analogy: The Car Engine
Imagine the universe's inflation as a car speeding down a highway.

1. Inflation: The car is on cruise control, zooming incredibly fast.
2. The End of Inflation: The driver hits the brakes.
3. The Big Bang (Hot Phase): The car stops, but the engine is still hot and sputtering, heating up the air inside the cabin. This is called Reheating.

For a long time, scientists assumed the "Reheating" phase happened instantly. They thought the car stopped, and poof, the universe was instantly hot and full of matter. It was like assuming the car engine cooled down and heated the cabin in zero seconds.

The New Idea:
The authors say, "Wait a minute. What if the engine takes a long time to cool down and heat the cabin? What if the transition is slow and messy?"

They modeled this "slow cooling" (non-instantaneous reheating) as a variable in their math. They introduced a new parameter called $R_{reh}$, which is essentially a "Reheating Dial" that controls how long and how hot this transition period is.

The Investigation: Running the Numbers

The team used a powerful computer method called Markov Chain Monte Carlo (MCMC). Think of this as a super-smart detective who runs millions of simulations, tweaking the "Reheating Dial" and the "Inflation Strength" in every possible combination to see which ones match the new telescope data.

They used data from:

• Planck: The previous champion of cosmic maps.
• ACT DR6: The new, sharper telescope data.
• DESI: Data on how galaxies are spaced out (like measuring the spacing of trees in a forest).

The Results: The Suspects Are Cleared!

Here is the exciting part: When they turned on the "Slow Reheating" dial, the Starobinsky and Higgs models suddenly became innocent again.

1. The Time Travel Effect: By allowing the universe to take a "breather" (a slow reheating phase) after inflation, the math shifted. It turned out that the "fingerprint" the telescope saw ($n_s$) wasn't a sign that the inflation model was wrong; it was a sign that the reheating phase was different than we thought.
2. The Verdict: The models are still viable. The tension between the theory and the new data disappears once you account for a realistic, non-instantaneous transition.

What Did They Learn About the "Reheating"?

The paper didn't just save the models; it told us what the universe was actually doing during that transition:

• The Temperature: The universe was surprisingly cold during reheating (relatively speaking). It suggests a long, drawn-out cooling process rather than a sudden explosion of heat.
• The "Stiffness": The matter during this phase acted very strangely. Usually, we think of matter as "dust" (slow) or "radiation" (fast). The data suggests the universe was in a "stiff" state (like a super-dense, rigid jelly) where the pressure was very high. This is a weird state of matter that we don't see in everyday life.
• Information Gain: The new ACT data gave the scientists 75% more information about this reheating phase than they had before. It's like going from looking at a blurry photo of a crime scene to seeing it in high definition.

The Conclusion

Don't cancel the Starobinsky or Higgs models just yet.

The paper argues that the "glitch" in the data wasn't because the inflation theories were wrong, but because we were ignoring the messy, complex "cooling down" period that happens right after inflation.

The Takeaway Metaphor:
Imagine you hear a loud bang in a kitchen.

• Old View: "The chef dropped a heavy pan!" (The Starobinsky model is wrong).
• New View: "Wait, maybe the chef was dropping the pan, but then the floor was covered in a thick, sticky syrup that slowed the fall and changed the sound of the crash." (The Starobinsky model is right, but the "syrup" of the reheating phase changed the outcome).

The authors conclude that we need to study this "syrup" (reheating) much more carefully in the future, because it holds the key to understanding how the universe actually began.

Technical Summary

1. Problem Statement

The Atacama Cosmology Telescope (ACT) released its sixth data release (DR6), reporting a scalar spectral index of $n_s = 0.9743 \pm 0.0034$. This value is significantly higher than previous measurements (e.g., Planck 2018: $n_s \approx 0.965$) and creates a tension with standard inflationary models.

• The Conflict: Under the assumption of instantaneous reheating and a standard number of e-folds ($N_* \approx 50\text{--}60$), the Starobinsky and Higgs inflation models predict $n_s \approx 0.965$, which is excluded by the ACT DR6 data at the $2\sigma$ confidence level.
• The Gap: Standard analyses often treat the reheating phase (the transition from inflation to the radiation-dominated era) as instantaneous. However, the duration and equation of state during reheating affect the number of e-folds ($N_*$) required for the CMB pivot scale to cross the horizon, thereby altering the predicted value of $n_s$.

2. Methodology

The authors perform a Bayesian inference analysis to test the viability of Starobinsky and Higgs inflation models when non-instantaneous reheating is included.

• Models:
  • Starobinsky Inflation: Based on $f(R) = R + R^2/6M^2$ gravity.
  • Higgs Inflation: Standard Model Higgs field non-minimally coupled to gravity ($\xi h^2 R$).
  • Note: Both models yield the same effective potential in the Einstein frame: $V(\phi) = V_0 [1 - \exp(-\sqrt{4/3\pi} \phi/M_{Pl})]^2$. The free parameter is the amplitude $V_0$.
• Reheating Parameterization:
  • Instead of treating the reheating temperature ($T_{reh}$) and effective equation of state ($\bar{\omega}_{reh}$) as independent parameters, the authors use a rescaled reheating parameter $R_{reh}$.
  • $R_{reh}$ combines $T_{reh}$ and $\bar{\omega}_{reh}$ into a single variable that determines the number of e-folds $N_*$ via the relation:
    $$N_* = N_0 + \ln R_{reh} + \ln\left(\frac{H(N_*)}{H_e}\right)$$
  • This approach avoids degeneracies and significantly reduces computational cost compared to sampling $T_{reh}$ and $\bar{\omega}_{reh}$ directly.
• Data Sets:
  • CMB: Planck 2018 (low-$\ell$ and high-$\ell$), ACT DR6 (temperature, polarization, lensing), and BICEP/Keck 2018 (B-modes).
  • BAO: Dark Energy Spectroscopic Instrument (DESI) Data Release 2 (DR2).
• Tools & Techniques:
  • MCMC: Markov Chain Monte Carlo analysis using the cobaya software.
  • Boltzmann Solver: Modified version of CLASS (Cosmic Linear Anisotropy Solving System) to numerically solve for $N_*$ given $R_{reh}$ and $V_0$ within the primordial module.
  • Information Theory: Kullback–Leibler (KL) divergence is used to quantify the information gain provided by the data regarding the reheating parameters.
  • Goodness of Fit: Probability to Exceed (PTE) is calculated to assess model consistency.

3. Key Contributions

1. Re-evaluation of Starobinsky/Higgs Viability: The paper demonstrates that the exclusion of these models by ACT DR6 is an artifact of assuming instantaneous reheating. When non-instantaneous reheating is modeled, the models remain viable.
2. Reparameterization Strategy: The use of $R_{reh}$ allows for efficient Bayesian inference of the reheating stage, which can then be mapped back to physical parameters ($T_{reh}, \bar{\omega}_{reh}$) with minimal computational overhead.
3. Quantification of Information Gain: The authors provide a rigorous quantification of how much the new ACT data constrains the reheating epoch compared to Planck-only data.

4. Key Results

• Viability Restored: By including non-instantaneous reheating, the number of e-folds $N_*$ can be extended to approximately 67 (instead of the standard 50–60). This shift allows the Starobinsky and Higgs models to predict $n_s \approx 0.972$, which is consistent with the ACT DR6 measurement ($0.9743 \pm 0.0034$) within the $1\sigma$ range.
• Parameter Constraints:
  • Inflation Amplitude ($V_0$): Tightly constrained with sub-percent precision ($\ln(V_0/M_{Pl}^4) \approx -29.66$).
  • Reheating Parameter ($R_{reh}$): Constrained within a range of roughly 4 orders of magnitude.
  • Physical Reheating Parameters:
    • Equation of State ($\bar{\omega}_{reh}$): The data shows a strong preference for $\bar{\omega}_{reh} > 0.5$ (mean $\approx 0.88$). This excludes standard dust ($\omega=0$) and relativistic matter ($\omega=1/3$) scenarios at $>2\sigma$ significance, suggesting a "stiff" equation of state during reheating.
    • Temperature ($T_{reh}$): The 95% credible interval allows for a wide range, but the preferred values indicate a relatively low reheating temperature (mean $\approx 11.56$ GeV), implying a long reheating duration.
• Information Gain:
  • The analysis yields 8.66 bits of information about the potential amplitude and 2.52 bits about the reheating parameter.
  • Including ACT DR6 data increases the information gain regarding the reheating stage by 75% compared to analyses using only Planck and BAO data.
• Model Consistency: The combined dataset (Planck + ACT + BAO + BICEP/Keck) yields a PTE of 89%, indicating an excellent fit between the Starobinsky/Higgs models with non-instantaneous reheating and current observational data.

5. Significance

• Resolving Tension: The paper resolves the apparent tension between the ACT DR6 spectral index and the highly successful Starobinsky/Higgs inflationary models. It proves that these models are not ruled out but require a specific reheating history.
• Importance of Reheating Modeling: The results emphasize that neglecting the reheating dynamics (treating it as instantaneous) leads to distorted cosmological inferences. The data now contains sufficient information to constrain the reheating epoch.
• Physical Implications: The preference for a stiff equation of state ($\bar{\omega}_{reh} \approx 0.88$) suggests that the post-inflationary universe may have undergone a phase dominated by a kinetic energy term (kination) or other non-standard physics before thermalization, a topic for future theoretical investigation.
• Future Outlook: The authors note that similar conclusions have been reached in other recent works, reinforcing the necessity of incorporating realistic reheating dynamics in future cosmological analyses to interpret high-precision CMB data correctly.