[論文レビュー] Do we need wavelets in the late Universe?
この論文は、ΛCDMに基づくハッブル半径のウェーブレット拡張を導入し、エルミートウェーブレットを用いて後期に局所的な振動を生み出し、それをベイズ推論とモデル比較で宇宙データに対して検証する。ウェーブレットはBAOデータの適合を改善し、データセットに依存して赤方偏移中心が選好される。
We parameterize the Hubble function by adding Hermitian wavelets to the Hubble radius of $Λ$CDM. This allows us to build Hubble functions that oscillate around $Λ$CDM at late times without modifying its angular diameter distance to last scattering. We perform parameter inference and model selection procedures on these new Hubble functions at the background level. In our analyses consisting of a wide variety of cosmological observations, we find that baryon acoustic oscillations (BAO) data play a crucial role in determining the constraints on the wavelet parameters. In particular, we focus on the differences between SDSS- and DESI-BAO datasets and find that wavelets provide a better fit to the data when either of the BAO datasets is present. However, DESI-BAO has a preference for the center of the wavelets to be around $z \sim 0.7$, while SDSS-BAO prefers higher redshifts of $z > 1$. This difference appears to be driven by the discrepancies between these two datasets in their $D_H / r_{ m d}$ measurements at $z = 0.51$ and $z \sim 2.3$. Finally, we also derive the consequences of the wavelets on a dark energy component. We find that the dark energy density oscillates by construction and also attains negative values at large redshifts ($z\gtrsim2$) as a consequence of the SDSS-BAO data. We conclude that while the early universe and the constraints on the matter density and the Hubble constant remain unchanged, wavelets are favored in the late universe by the BAO data. Specifically, there is a significant improvement at more than $3σ$ in the fit when new DESI-BAO data are included in the analysis.
研究の動機と目的
- Motivate and explore a wavelet extension of the Hubble radius to capture potential late-time deviations from ΛCDM without altering the angular diameter distance to last scattering.
- Develop a parametric family of H(z) deviations using Hermitian wavelets derived from Gaussian derivatives.
- Constrain the wavelet parameters using a broad set of cosmological data and compare models via Bayesian evidence.
- Investigate the influence of different BAO datasets (SDSS-BAO vs DESI-BAO) on wavelet constraints and the implied dark energy behavior.
提案手法
- Define Hubble function deviations as 1/H(z) = 1/𝓗(z) + ψ(z), where 𝓗(z) is ΛCDM and ψ(z) is a Hermitian wavelet.
- Construct wavelets ψ_n(z) as nth derivatives of a Gaussian-like base G(z) to control oscillation nodes.
- Use the first four Hermitian wavelets (ψ1–ψ4) to generate model variants with three wavelet parameters (α_h, β_h, z†).
- Perform Bayesian parameter estimation and model selection with an adapted SimpleMC MCMC framework and dynesty for Bayesian evidence.
- Assemble data from CMB-related BAO (Pl), SN Ia (Pantheon+), cosmic chronometers, and two BAO datasets (SB: SDSS-BAO, DB: DESI-BAO) plus H0 priors as needed.
- Compute χ2 from data with the full covariance and sum contributions across Pl, SN, BAO, and H0 where applicable.
実験結果
リサーチクエスチョン
- RQ1Can a wavelet-based deviation to the Hubble radius reproduce late-time oscillatory features while preserving the angular diameter distance to last scattering?
- RQ2How do Hermitian wavelets modify the inferred expansion history and dark energy density compared to ΛCDM?
- RQ3What is the impact of using SDSS-BAO versus DESI-BAO on the preferred wavelet parameters and the goodness of fit?
- RQ4Do the BAO-driven wavelet fits require dynamical dark energy and, if so, what is the inferred evolution of ρ_DE(z) and w_DE(z)?
主な発見
| Model | Datasets | h | Ω_m,0 | ln B_{ΛCDM,i} | -2ΔlnL_max |
|---|---|---|---|---|---|
| ΛCDM | SB+Pl | 0.679 (0.006) | 0.309 (0.007) | 0 | 0 |
| ψ1 | SB+Pl | 0.684 (0.005) | 0.306 (0.007) | -0.22 (0.35) | -5.22 |
| ψ2 | SB+Pl | 0.686 (0.006) | 0.303 (0.008) | -0.51 (0.34) | -4.72 |
| ψ3 | SB+Pl | 0.684 (0.005) | 0.307 (0.007) | -0.54 (0.34) | -5.29 |
| ψ4 | SB+Pl | 0.685 (0.006) | 0.306 (0.007) | -0.66 (0.34) | -6.58 |
| ΛCDM | SB+SN | 0.686 (0.013) | 0.306 (0.013) | 0 | 0 |
| ψ1 | SB+SN | 0.692 (0.014) | 0.315 (0.015) | -0.61 (0.17) | -7.42 |
| ψ2 | SB+SN | 0.692 (0.014) | 0.318 (0.016) | -1.22 (0.17) | -8.59 |
| ψ3 | SB+SN | 0.693 (0.014) | 0.315 (0.015) | -0.83 (0.18) | -8.07 |
| ψ4 | SB+SN | 0.692 (0.014) | 0.314 (0.014) | -0.71 (0.18) | -8.61 |
| ΛCDM | SB+SN+Pl | 0.676 (0.006) | 0.312 (0.007) | 0 | 0 |
| ψ1 | SB+SN+Pl | 0.684 (0.005) | 0.306 (0.007) | -0.22 (0.35) | -5.22 |
| ψ2 | SB+SN+Pl | 0.685 (0.005) | 0.306 (0.006) | -0.51 (0.34) | -4.72 |
| ψ3 | SB+SN+Pl | 0.684 (0.005) | 0.307 (0.007) | -0.54 (0.34) | -5.29 |
| ψ4 | SB+SN+Pl | 0.685 (0.006) | 0.306 (0.007) | -0.66 (0.34) | -6.58 |
| ΛCDM | SB+SN+H0 | 0.709 (0.014) | 0.311 (0.013) | 0 | 0 |
| ψ1 | SB+SN+H0 | 0.705 (0.012) | 0.315 (0.015) | 0.14 (0.24) | -2.91 |
| ψ2 | SB+SN+H0 | 0.706 (0.012) | 0.318 (0.015) | 0.34 (0.24) | -3.17 |
| ψ3 | SB+SN+H0 | 0.705 (0.012) | 0.314 (0.015) | 0.55 (0.23) | -3.89 |
| ψ4 | SB+SN+H0 | 0.705 (0.011) | 0.314 (0.015) | -0.11 (0.23) | -3.23 |
| ΛCDM | SB+SN+Pl+H0 | 0.679 (0.005) | 0.308 (0.007) | 0 | 0 |
| ψ1 | SB+SN+Pl+H0 | 0.687 (0.005) | 0.303 (0.007) | -1.38 (0.34) | -6.18 |
| ψ2 | SB+SN+Pl+H0 | 0.688 (0.005) | 0.302 (0.007) | -1.52 (0.34) | -6.34 |
| ψ3 | SB+SN+Pl+H0 | 0.688 (0.006) | 0.302 (0.007) | -1.48 (0.35) | -7.16 |
| ψ4 | SB+SN+Pl+H0 | 0.688 (0.005) | 0.302 (0.006) | -1.21 (0.34) | -7.78 |
| ΛCDM | SN+Pl | 0.671 (0.006) | 0.319 (0.008) | 0 | 0 |
| ψ1 | SN+Pl | 0.676 (0.026) | 0.331 (0.017) | 0.58 (0.23) | -3.71 |
| ψ2 | SN+Pl | 0.675 (0.026) | 0.332 (0.018) | 0.12 (0.24) | -3.85 |
| ψ3 | SN+Pl | 0.676 (0.028) | 0.332 (0.018) | 0.13 (0.22) | -5.21 |
| ψ4 | SN+Pl | 0.676 (0.026) | 0.330 (0.017) | 0.42 (0.23) | -5.41 |
| ΛCDM | SN+H0 | 0.711 (0.019) | 0.322 (0.017) | 0 | 0 |
| ψ1 | SN+H0 | 0.711 (0.017) | 0.323 (0.016) | -0.28 (0.22) | -3.88 |
| ψ2 | SN+H0 | 0.712 (0.017) | 0.323 (0.017) | -0.12 (0.22) | -4.18 |
| ψ3 | SN+H0 | 0.712 (0.017) | 0.324 (0.016) | -0.69 (0.22) | -5.04 |
| ψ4 | SN+H0 | 0.712 (0.018) | 0.322 (0.017) | -0.09 (0.24) | -5.66 |
| ΛCDM | SN+Pl+H0 | 0.678 (0.005) | 0.311 (0.007) | 0 | 0 |
| ψ1 | SN+Pl+H0 | 0.686 (0.005) | 0.304 (0.007) | -1.61 (0.35) | -13.25 |
| ψ2 | SN+Pl+H0 | 0.686 (0.005) | 0.304 (0.007) | -1.41 (0.34) | -11.87 |
| ψ3 | SN+Pl+H0 | 0.686 (0.005) | 0.304 (0.006) | -1.99 (0.34) | -12.48 |
| ψ4 | SN+Pl+H0 | 0.686 (0.005) | 0.304 (0.007) | -1.31 (0.34) | -12.01 |
- Wavelet deviations can fit BAO data better than ΛCDM, with negative ΔlnLmax indicating improved fits across many data combinations.
- For DESI-BAO, wavelets are favored at high significance (over 3σ in most cases) when combined with SN and Planck data, though results depend on the BAO dataset used.
- The posterior for the wavelet center z† shifts with data: DESI-BAO prefers z† ~ 0.7, while SDSS-BAO prefers z† > 1, driven by discrepancies in certain D_H/r_d measurements.
- Including Planck data constrains ΛCDM-like baselines, yet wavelets remain competitive, offering a dynamical-deviation interpretation that can reconcile BAO tensions.
- Wavelet-induced oscillations lead to an oscillatory dark energy density ρ_DE(z), which can become negative at high redshift (z ≳ 2) depending on the data combination.
- Overall, early-universe constraints (Ω_m, h) stay consistent with ΛCDM, while late-time BAO data favor wavelet extensions of the expansion history.
- There is a notable improvement in fit when DESI-BAO data are included, with improvements exceeding expectations from the additional wavelet parameters.
より良い研究を、今すぐ始めましょう
論文設計から論文執筆まで、研究時間を劇的に削減しましょう。
クレジットカード登録不要
このレビューはAIが作成し、人間の編集者が確認しました。