Parameterizing Dark Energy at the density level: A two-parameter alternative to CPL

1. Problem Statement

Recent measurements from the Dark Energy Spectroscopic Instrument (DESI), combined with Cosmic Microwave Background (CMB) and Type Ia Supernovae (SNe) data, have suggested a $\sim 3\sigma$ preference for evolving Dark Energy (DE) over a cosmological constant ($\Lambda$). This preference is typically assessed using the Chevallier-Polarski-Linder (CPL) parameterization, which describes the DE equation of state $w(z)$ as a linear function of the scale factor: $w(z) = w_0 + w_a \frac{z}{1+z}$.

However, the authors identify critical limitations in the CPL framework:

• Non-linear Mapping: The DE density $f_{DE}(z)$ is derived from $w(z)$ via a non-linear integral. This mapping can amplify parameter uncertainties and introduce approximate degeneracies.
• Weak Sensitivity to $w_a$: The expansion history $H(z)$ is only weakly sensitive to the slope parameter $w_a$. Consequently, the preference for evolving DE is often driven by a poorly constrained $w_a$, while the pivot value $w_p$ remains consistent with $-1$ (the cosmological constant value).
• Cosmic Coincidence: Reparameterizing CPL to $(w_p, w_a)$ reveals that the data are consistent with $w_p = -1$ at the pivot redshift (where constraints are tightest), yet the model requires a large $w_a$ to fit the data. This creates a "coincidence" where the DE equation of state equals the cosmological constant value precisely at the epoch where the data are most sensitive.
• Degeneracy: The $(w_0, w_a)$ basis suffers from elongated degeneracy contours, making it difficult to isolate the physical properties of DE at specific epochs.

2. Methodology

The authors propose a minimal two-parameter formulation that parameterizes the DE density evolution directly, rather than the equation of state.

• Polynomial Expansion: They define the normalized DE density $f_{DE}(z) \equiv \rho_{DE}(z)/\rho_{DE,0}$ as a second-order polynomial expansion in terms of $(1-a)$, where $a$ is the scale factor:
  $$f_{DE}(a) = 1 + f_a(1-a) + f_b(1-a)^2$$
  This form naturally recovers the cosmological constant ($\Lambda$) when coefficients are zero and ensures $f_{DE} \to 1$ at $a=1$.

• Pivot Reparameterization: To improve physical interpretability and reduce correlations, the coefficients are re-expressed in terms of the DE density fraction ($f_p$) and the equation of state ($w_p$) evaluated at a pivot redshift $z_p$:

  • $f_p \equiv f_{DE}(z_p)$: Quantifies the DE density relative to today at the pivot epoch.
  • $w_p \equiv w(z_p)$: Captures the rate of evolution at the pivot epoch.
  • The pivot $z_p$ is chosen (set to $0.5$ in this analysis) to minimize the correlation between $f_p$ and $w_p$.

• Validation:

  • Quintessence Benchmarks: The model is tested against thawing quintessence scenarios (exponential and hilltop potentials). The authors minimize the maximum relative error in the Hubble rate ($E_H$) between the parameterization and the true model.
  • Mock Data: Analyses are performed on mock DESI, CMB, and SNe data generated from both $\Lambda$CDM and exponential quintessence cosmologies to test for unbiased parameter recovery.

• Data Analysis:

  • Datasets: DESI DR2 BAO, a compressed CMB likelihood (QCMB) from Planck, and two SNe samples (Pantheon+ and DESY5).
  • Framework: Markov Chain Monte Carlo (MCMC) analysis using MontePython and CLASS. The analysis is performed at the background level (ignoring perturbations) to avoid singularities in $w(z)$ that might arise if $f_{DE}$ becomes negative at high redshifts.
  • Comparison: Results are compared directly with the standard CPL ($w_0, w_a$) parameterization.

3. Key Contributions

1. Density-Level Parameterization: Introduction of the $(w_p, f_p)$ framework, which avoids the non-linear integral mapping inherent in CPL. This eliminates the structural degeneracy where $w_a$ is weakly constrained.
2. Decorrelation of Parameters: The authors demonstrate that $w_p$ and $f_p$ are largely decorrelated. Crucially, they are constrained by distinct subsets of data:
   • $w_p$ is tightly constrained by BAO and CMB (sensitive to the relative variation of $H(z)$).
   • $f_p$ is constrained by SNe (which provide an independent measurement of $\Omega_m$ via the luminosity distance shape, breaking the degeneracy between DE density and expansion normalization).
3. Physical Interpretability: Unlike the Taylor coefficients of CPL, $w_p$ and $f_p$ have direct physical meanings at the pivot epoch, allowing for a clearer characterization of DE dynamics.
4. Robustness Check: The study confirms that the "coincidence" of DE parameters aligning with $\Lambda$CDM values is not an artifact of the parameterization choice but is reinforced when using a less degenerate basis.

4. Key Results

• Constraints: Using the combined dataset (DESI + QCMB + Pantheon+), the authors obtain:
  • $w_p = -1.04 \pm 0.04$
  • $f_p = 1.07 \pm 0.04$
  • Similar results are found with the DESY5 SNe sample.
• Precision: Both parameters are measured to percent-level precision. This contrasts with CPL, where even after including SNe, the parameters remain correlated, and $w_a$ remains poorly constrained.
• Statistical Preference for Evolving DE:
  • The preference for evolving DE over $\Lambda$CDM remains below $3\sigma$ (ranging from $2.2\sigma$ to $2.9\sigma$ depending on the dataset and parameterization).
  • The $\Delta \chi^2_{MAP}$ values for $(w_p, f_p)$ and $(w_0, w_a)$ are comparable, indicating similar goodness-of-fit.
• Reconstruction: The reconstructed DE density $f_{DE}(z)$ shows a mild rise (a few percent) above unity at intermediate redshifts ($z \sim 0.5$) before decreasing, with an equation of state that remains marginally phantom ($w < -1$) at higher redshifts.
• Coincidence Reinforced: The results confirm that $w_p \approx -1$ and $f_p \approx 1$ (within $2\sigma$) at the pivot redshift. This reinforces the "coincidence" that the data favor a cosmological constant value precisely where the data are most constraining, even when the overall model prefers evolution.
• Single-Parameter Limit: A single-parameter version (linear density evolution) yields even weaker evidence for evolution ($< 2\sigma$), highlighting that the marginal preference for evolution in the two-parameter case is sensitive to the number of free parameters.

5. Significance and Conclusion

The paper provides a robust alternative to the standard CPL parameterization that addresses its intrinsic degeneracies and interpretability issues.

• Cautionary Note: The findings suggest that the current evidence for dynamical Dark Energy is suggestive but not compelling. The fact that the best-fit parameters in this less-degenerate framework still align closely with the cosmological constant ($\Lambda$) at the pivot epoch reinforces the need for caution in interpreting the current $\sim 3\sigma$ hints of evolution.
• Methodological Value: The $(w_p, f_p)$ formulation offers a clearer diagnostic tool. By distributing constraining power across two individually well-measured parameters tied to specific physical observables (distance slopes vs. matter density), it makes the interplay between datasets and the proximity to $\Lambda$CDM more transparent.
• Future Outlook: The authors note that including DE perturbations and the full CMB likelihood (rather than the compressed prior) may slightly increase the significance, but the fundamental conclusion—that the evidence is marginal and sensitive to parameterization choices—remains. The framework is well-suited for future surveys (e.g., LSST, Euclid) to further tighten constraints on the nature of Dark Energy.

In summary, this work argues that while the data show a mild tension with $\Lambda$CDM, the "coincidence" of parameters aligning with $\Lambda$ at the most sensitive redshift, combined with the lack of a decisive statistical breakthrough ($>3\sigma$), warrants a cautious interpretation of evolving Dark Energy claims.