Constraining Dark Energy Dynamics in Curved Spacetime with Current Observations

Technical Summary: Constraining Dark Energy Dynamics in Curved Spacetime with Current Observations

Problem Statement
The standard $\Lambda$CDM cosmological model, while successful, faces theoretical challenges (coincidence and fine-tuning problems) and observational tensions, most notably the discrepancy in the Hubble constant ($H_0$) between early-universe (Planck) and late-universe (SH0ES) measurements. Furthermore, recent analyses suggest that assuming a spatially flat universe may bias constraints on dark energy (DE). While some studies indicate a preference for a closed or open universe, the interplay between spatial curvature ($\Omega_{k0}$) and dynamical dark energy remains under-explored. This paper investigates whether a smoothly evolving DE component within a non-flat spacetime framework produces observable signatures distinct from the standard flat $\Lambda$CDM model and examines how data reconstruction techniques influence these constraints.

Methodology
The authors propose a specific parametrization for the DE equation of state (EoS), $w(z)$, characterized by a single parameter $\alpha$. This model reduces to $\Lambda$CDM when $\alpha=0$ and allows for dynamical DE evolution for $\alpha \neq 0$. The analysis is conducted within a Friedmann-Lemaître-Robertson-Walker (FLRW) metric allowing for non-zero spatial curvature ($k = -1, 0, +1$).

To constrain the model parameters ($H_0, \Omega_{m0}, \Omega_{k0}, \alpha$), the study employs a Bayesian Markov Chain Monte Carlo (MCMC) approach using two distinct datasets:

1. Original Observational Data: A combination of 31 Cosmic Chronometer (CC) $H(z)$ measurements, 1701 Pantheon+SH0ES (PPS) Type Ia supernova data points, and 9 DESI BAO DR2 measurements.
2. Reconstructed Data: The same datasets are processed using an Artificial Neural Network (ANN) with a feed-forward architecture (two hidden layers, ReLU activation). A bootstrap resampling method is applied to incorporate observational uncertainties, generating a smoothed, model-independent representation of the expansion history.

The study compares the proposed model against the standard flat $\Lambda$CDM and the non-flat $\Lambda$CDM+$\Omega_k$ scenarios using statistical criteria: $\chi^2_{min}$, reduced chi-square ($\chi^2_r$), Akaike Information Criterion (AIC), and Bayesian Information Criterion (BIC).

Key Contributions and Results
The analysis yields the following constraints at 68% confidence level (CL):

• Parameter $\alpha$ and DE Dynamics:

  • Original Data: $\alpha \approx 0.35 \pm 0.09$.
  • Reconstructed Data: $\alpha \approx 0.56 \pm 0.09$.
  • The positive values of $\alpha$ in both cases indicate a deviation from the cosmological constant ($w=-1$), suggesting a quintessence-like behavior ($w > -1$). The current EoS value is estimated as $w_0 \approx -0.975$.

• Spatial Curvature ($\Omega_{k0}$):

  • Original Data: $\Omega_{k0} = 0.068 \pm 0.029$, suggesting a slight preference for an open universe, though consistent with flatness at higher confidence levels.
  • Reconstructed Data: $\Omega_{k0} = -0.131 \pm 0.032$, indicating a statistically significant preference for a closed universe.
  • The authors note that the reconstruction method is highly sensitive to curvature, shifting the mean value significantly.

• Matter Density ($\Omega_{m0}$) and Hubble Constant ($H_0$):

  • Original Data: $H_0 = 68.92 \pm 0.32$ km s$^{-1}$ Mpc$^{-1}$ and $\Omega_{m0} = 0.277 \pm 0.027$. These values align well with Planck CMB measurements.
  • Reconstructed Data: $H_0 = 68.80 \pm 0.36$ km s$^{-1}$ Mpc$^{-1}$ (consistent with original) but $\Omega_{m0} = 0.395 \pm 0.035$. The reconstructed data implies a higher matter density, suggesting a matter-dominated present universe relative to standard expectations, which the authors attribute to parameter degeneracies between matter, curvature, and DE evolution.

• Model Comparison:

  • The proposed model yields lower $\chi^2_{min}$ and AIC values compared to both the flat $\Lambda$CDM and the non-flat $\Lambda$CDM+$\Omega_k$ models for the original dataset, indicating a better goodness of fit without excessive parameter penalization.
  • For the reconstructed dataset, the proposed model continues to outperform the baseline scenarios, though the absolute $\chi^2$ values increase slightly due to the reconstruction process.
  • In contrast, the standard $\Lambda$CDM models yield significantly higher $H_0$ values ($\approx 73$ km s$^{-1}$ Mpc$^{-1}$), consistent with local distance-ladder measurements but distinct from the lower $H_0$ found in the authors' curved, dynamical DE framework.

Significance and Claims
The paper claims that allowing for both spatial curvature and dynamical DE evolution reveals a complex degeneracy structure that is sensitive to data processing methods. Specifically:

1. Sensitivity to Reconstruction: The shift from an open to a closed universe preference when moving from original to ANN-reconstructed data highlights that curvature constraints are highly sensitive to how observational noise and sparsity are handled.
2. Parameter Interplay: The results demonstrate a correlated response between the DE evolution parameter ($\alpha$), matter density ($\Omega_{m0}$), and curvature ($\Omega_{k0}$). The introduction of curvature freedom in standard models primarily adjusts the expansion rate via $H_0$, whereas the authors' model distributes this adjustment between curvature and the evolving DE sector.
3. Robustness: The ANN-based reconstruction serves as a complementary consistency check, validating that the inferred cosmological trends (such as the deviation from $w=-1$) are not merely artifacts of measurement scatter.
4. Statistical Preference: The proposed parametrization provides a competitive, and in some cases superior, fit to current low-redshift datasets compared to standard flat or non-flat $\Lambda$CDM models, particularly when accounting for the dynamical nature of dark energy.

The authors conclude that non-flat models should not be ruled out and that future surveys (DESI, Euclid, LSST, CMB-S4) will be crucial for further testing these curvature-DE degeneracies.