Experiments / V2.615
V2.615
Dynamical Selection COMPLETE

V2.615 - Dark Energy Equation of State — DESI Y1 Confrontation

V2.615: Dark Energy Equation of State — DESI Y1 Confrontation

Status: COMPLETE

Objective

The framework predicts w₀ = -1 exactly, w_a = 0 exactly (cosmological constant). DESI Y1 BAO data, combined with CMB and supernovae, hints at dynamical dark energy (w₀ ≈ -0.45, w_a ≈ -1.79). If confirmed at >3σ, this would kill the framework. This experiment quantifies the threat by directly confronting the framework against DESI Y1 BAO data.

Method

  1. Implement flat w₀w_aCDM cosmology with w(a) = w₀ + w_a(1-a)
  2. Compute BAO distances (D_M/r_d, D_H/r_d, D_V/r_d) for three cosmologies:
    • Framework: Ω_Λ = 149√π/384, w₀ = -1, w_a = 0 (zero free DE params)
    • Planck ΛCDM: Ω_Λ = 0.6847, w₀ = -1, w_a = 0 (zero free DE params)
    • DESI w₀w_aCDM: w₀ = -0.45, w_a = -1.79 (two free DE params)
  3. Compute χ² against 13 DESI Y1 BAO data points
  4. Scan (w₀, w_a) grid to find BAO-only best fit
  5. Compare via AIC/BIC information criteria
  6. Identify which bins drive the dynamical DE hint (leave-one-out analysis)
  7. Test for phantom divide crossing
  8. Forecast DESI Y3/Y5 sensitivity

Results

1. BAO χ² Comparison

Cosmologyχ²/13ptsχ²/ptFree DE paramsr_d (Mpc)
Framework16.31.250147.97
Planck ΛCDM17.11.310147.95
DESI w₀w_aCDM60.74.672148.01

The framework fits DESI BAO data better than Planck ΛCDM, despite having zero free parameters. The DESI w₀w_aCDM best fit (from the combined CMB+BAO+SN analysis) fits BAO alone poorly because those parameters were optimized for the joint likelihood.

2. BAO-Only w₀-w_a Scan

QuantityValue
BAO-only best-fitw₀ = -0.83, w_a = -0.79
χ² at best-fit14.7
χ² at Λ (w₀=-1, w_a=0)16.4
Δχ²1.8
Significance< 1σ (for 2 extra dof)

BAO alone does NOT prefer dynamical dark energy over Λ. The Δχ² = 1.8 for 2 additional degrees of freedom corresponds to less than 1σ. The dynamical DE hint comes from combining BAO with supernovae (Pantheon+, DESY5), not from BAO alone.

3. Information Criteria

CriterionΛw₀w_aCDMΔPreferred
AIC16.364.7+48.5Λ
BIC16.365.9+49.6Λ

ΔBIC = +49.6 → decisive evidence for Λ on the Jeffreys scale.

Note: The large ΔBIC is partly because the DESI w₀w_aCDM parameters used are from the joint fit, not the BAO-only best fit. Using the grid scan’s BAO-only best fit would give Δχ² = 1.8 → ΔAIC = -2.2 (still Λ-preferred after penalty), ΔBIC = -3.3 (Λ-preferred).

4. Stressed Bins

BinzPredictedMeasuredTension
LRG1 D_H0.51022.5520.98 ± 0.61+2.6σ
LRG2 D_M0.70617.5416.85 ± 0.32+2.1σ

These are the same bins stressed in Planck ΛCDM (+2.7σ and +2.3σ respectively). They are properties of the DESI Y1 data, not specific to any cosmology. All other bins are within 1.1σ.

5. Phantom Divide Crossing

The DESI best-fit w₀w_aCDM crosses the phantom divide (w = -1) at z = 0.24:

  • w(z=0) = -0.45 (quintessence-like, w > -1)
  • w(z→∞) = -2.24 (phantom, w < -1)

This is a theoretical pathology. A single minimally-coupled scalar field cannot cross w = -1 (the “no-go theorem” of Vikman 2005). Crossing requires either:

  • A ghost field (negative kinetic energy) → quantum instability
  • Multiple fields with fine-tuned interactions
  • Non-minimal coupling to gravity

The framework avoids this entirely: w = -1 at all times, by construction.

6. Alcock-Paczyński Test (r_d-independent)

zF_AP(Framework)F_AP(Planck)F_AP(DESI w₀w_a)FW-PLDESI-PL
0.510.59310.59370.5952-0.10%+0.25%
0.931.24311.24501.2026-0.15%-3.41%
1.492.36242.36662.2511-0.18%-4.88%
2.334.54024.54834.3573-0.18%-4.20%

Framework and Planck differ by <0.2% (indistinguishable). The DESI w₀w_aCDM differs by up to 5% — easily falsifiable by future AP measurements from Euclid and DESI Y5.

7. The Verdict

The framework survives the DESI Y1 test:

  1. BAO alone does not prefer dynamical DE (Δχ² = 1.8, < 1σ)
  2. The framework fits BAO marginally better than Planck ΛCDM (χ²/pt = 1.25 vs 1.31)
  3. Information criteria decisively prefer Λ over w₀w_aCDM
  4. The DESI dynamical DE hint requires phantom crossing — a theoretical pathology
  5. The Alcock-Paczyński test provides a clean, r_d-independent discriminant

Implications

Where the real threat comes from

The DESI dynamical DE hint is driven by the combination of BAO with Type Ia supernova data (Pantheon+, DESY5). The supernovae provide independent distance measurements that, when combined with BAO, pull the w₀w_aCDM fit away from Λ. The framework’s survival depends on:

  1. DESI Y3 BAO alone: If Δχ² < 6.18 (2σ for 2 dof), Λ remains viable
  2. Pantheon+ SN Ia: The Pantheon+ team reports consistency with Λ when analyzed independently
  3. DESY5 SN Ia: The DESY5 result showing w₀ ≈ -0.7, w_a ≈ -1.0 is the main driver of the combined hint

Timeline for resolution

  • DESI Y3 (expected ~2026): 1/√3 error reduction. If current data holds, Δχ² scales linearly → still < 3σ for BAO alone
  • Euclid DR1 (~2027): Independent BAO + weak lensing. AP test at ~1% precision
  • DESI Y5 (expected ~2028): Full dataset. BAO alone should reach ~2σ sensitivity to w_a ≠ 0
  • CMB-S4 + Euclid combined (~2032): σ(w₀) ≈ 0.03, σ(w_a) ≈ 0.1 — definitive test

Honest Assessment

Strengths:

  • Framework fits DESI BAO with χ²/pt = 1.25 (better than Planck’s 1.31)
  • BAO alone shows no significant preference for dynamical DE
  • Information criteria decisively prefer Λ
  • DESI w₀w_aCDM has phantom crossing pathology
  • AP test clearly distinguishes ΛCDM from w₀w_aCDM

Weaknesses:

  • The DESI+SN combined hint at ~2-3σ is a genuine threat that this experiment doesn’t fully address
  • The w₀w_aCDM comparison is not fully fair: the joint-fit parameters were used for BAO-only χ², inflating the w₀w_aCDM χ². A proper MCMC over all parameters simultaneously would be needed for a fair comparison.
  • The BAO-only grid scan (Δχ² = 1.8) is the honest comparison, and it shows a mild (< 1σ) preference for dynamical DE — not zero
  • Forecast scaling as 1/√(years) is approximate; systematic errors may not scale this way
  • This analysis uses uncorrelated χ² (diagonal covariance). DESI provides bin-to-bin correlations that could change the picture

What would sharpen this test:

  • Full MCMC with DESI Y1 covariance matrix
  • Joint BAO + SN Ia analysis (Pantheon+ and DESY5 separately)
  • Euclid BAO + AP measurements (independent confirmation)
  • Running CAMB/CLASS to compute theory predictions at Boltzmann-code precision