Curvature-Corrected Delta/Alpha Extraction
Experiment V2.69: Curvature-Corrected Delta/Alpha Extraction
Status: COMPLETE
Executive Summary
V2.69 takes a perturbative approach to the de Sitter self-consistency problem that V2.68’s direct d3S method could not solve. By computing Delta_S(n,H) = S_dS(n,H) - S_flat(n,0) and analyzing the curvature correction as a function of n, we establish three key results:
-
Perturbative regime confirmed: Delta_S/H^2 is H-independent to <1% for H = 0.0005 to 0.003 — curvature corrections scale as H^2 (linear response).
-
alpha(H) ≈ alpha(0) to <1%: The area-law coefficient is essentially curvature-independent. The self-consistency gap is NOT from background geometry.
-
R(0) = 0.0413 vs 1/(8*pi) = 0.0398: The self-consistency ratio is suspiciously close to the rational number 1/(8*pi), differing by only 3.9%.
Key conclusion: The factor-of-25 self-consistency gap must come from species content or the relationship between lattice and continuum regularization, not from de Sitter curvature corrections.
1. Motivation
V2.68 showed that the d3S extraction method catastrophically fails on de Sitter because curvature introduces H^2*n^4 polynomial corrections that overwhelm the logarithmic signal. V2.69 takes a different approach: instead of trying to extract delta(H) directly, we study how the entropy changes with curvature.
2. Method
2.1 Perturbative Approach
For each sphere radius n and Hubble parameter H:
Delta_S(n, H) = S_dS(n, H) - S_flat(n, 0)If the curvature correction is perturbative:
Delta_S(n, H) = H^2 * g(n) + O(H^4)where g(n) encodes the n-dependence of the leading curvature correction.
2.2 Functional Form Determination
Fit Delta_S/H^2 as a function of n to determine g(n):
g(n) = c4*n^4 + c3*n^3 + c2*n^2 + c1*n + c_ln*ln(n) + c02.3 Multi-Term Fit with Curvature Corrections
At each H > 0, fit S(n) to:
S(n) = a1*n^2 + a2*n + a3*ln(n) + a4 + a5/n + a6*H^2*n^4 + a7*H^2*n^3The curvature correction terms (a6, a7) absorb the polynomial contamination that destroyed d3S in V2.68.
3. Results
3.1 Phase 1: Perturbative Regime
| H pair | max |Delta_S/H^2 relative difference| | Perturbative? | |--------|---------------------------------------------|---------------| | 0.0005 vs 0.001 | 0.6% | Yes | | 0.001 vs 0.002 | 0.2% | Yes | | 0.002 vs 0.003 | 0.6% | Yes |
Delta_S/H^2 is H-independent to <1%, confirming linear response.
3.2 Phase 2: Functional Form
Delta_S/H^2 = -0.278*n^4 - 0.540*n^3 + 1.667*n^2 + 9.232*n - 39.35*ln(n) + 35.55R^2 = 1.000000000. The dominant curvature correction is the n^4 term, exactly as expected from the H^2*r^4 scaling of the metric perturbation integrated over the sphere.
The ln(n) coefficient in the curvature correction (-39.35) is a “curvature-induced logarithmic shift” — but this is NOT the trace anomaly delta. It represents how the trace anomaly coefficient changes on a curved background.
3.3 Phase 3: Multi-Term Fits at Multiple H
Standard fit (no curvature correction terms):
| H | alpha | delta | R^2 |
|---|---|---|---|
| 0 | 0.02278 | -0.01115 | 1.0000 |
| 0.001 | 0.01616 | -1549 | 0.99998 |
| 0.005 | 0.01151 | -303 | 0.99998 |
| 0.01 | 0.00599 | -200 | 0.99999 |
Standard fits at H > 0 give garbage delta (curvature terms absorbed into delta).
Curvature-corrected fit (with H^2n^4 and H^2n^3 terms):
| H | alpha | delta | curv_n4 | R^2 |
|---|---|---|---|---|
| 0 | 0.02278 | -0.01115 | — | 1.0000 |
| 0.001 | 0.02274 | -1.438 | -0.282 | 1.0000 |
| 0.002 | 0.02273 | -0.860 | -0.283 | 1.0000 |
| 0.003 | 0.02271 | -0.702 | -0.284 | 1.0000 |
| 0.005 | 0.02267 | -0.615 | -0.285 | 1.0000 |
| 0.008 | 0.02260 | -0.596 | -0.286 | 1.0000 |
| 0.01 | 0.02253 | -0.669 | -0.288 | 1.0000 |
alpha recovery is excellent — within 1% of the flat-space value at all H. But delta remains contaminated (order 1, not order 0.01), indicating that higher-order curvature correction terms (H^2n^2ln(n), etc.) still leak into the delta fit.
3.4 Phase 4: alpha(H) Independence
| H | alpha | Relative change from H=0 |
|---|---|---|
| 0 | 0.02278 | — |
| 0.001 | 0.02274 | -0.15% |
| 0.002 | 0.02273 | -0.22% |
| 0.003 | 0.02271 | -0.29% |
| 0.005 | 0.02267 | -0.46% |
| 0.008 | 0.02260 | -0.76% |
| 0.01 | 0.02253 | -1.08% |
The area-law coefficient is curvature-independent to <1%. This definitively rules out background geometry as the source of the self-consistency gap.
3.5 Phase 5: High-Precision R(0)
R(0) = |delta(0)| / (12 * alpha(0)) = 0.04134
1/(8*pi) = 0.03979
Difference: 3.9%The match to 1/(8pi) is suggestive but not conclusive. If R = 1/(8pi) exactly, then the self-consistency condition R = 1 requires a multiplicative factor of 8*pi ≈ 25.1, which must come from species content or the relationship between lattice and continuum α.
4. Conclusions
-
The self-consistency gap is NOT from geometry. alpha(H) ≈ alpha(0) to <1%, and delta cannot be reliably extracted on curved backgrounds at all. The gap R = 0.040 vs R = 1 must have a different origin.
-
The curvature correction to S(n) is perturbative and well-characterized: dominated by an n^4 term with coefficient -0.278*H^2.
-
R(0) ≈ 1/(8*pi) to 3.9%. If this is exact, it points to a missing factor of 8*pi from species counting or regularization scheme.
-
V2.70 should investigate species content as the resolution. Different field types (scalar, fermion, vector) have different trace anomaly ratios and area-law coefficients.
5. Files
| File | Purpose |
|---|---|
run_experiment.py | 5-phase experiment: perturbative check, functional form, multi-term fits, alpha(H), R(0) |
results/v2_69_results.json | Complete numerical results |
6. Runtime
Total: 1200 seconds (~20 minutes). Dominated by Phase 3 entropy table computations at multiple H values.