Goal: Reproduce the geometry constant \( C_\kappa \), Maxwell normalization \( \kappa_B \), bubble ratios, and Ward diagnostics. The recommended baseline is amp=none
which collapses scheme dependence.
Contents
1 · Environment
- Python 3.10+; numpy, scipy, matplotlib, pandas.
- Repo:
RTG/scripts/rtg_eft_sim.py
andrtg_eft_post2.py
.
2 · Baseline runs (PowerShell‑friendly)
# SU(2), μ/Δω* = 0.18, scheme comparison (amp=none)
python rtg_eft_sim.py `
--do scheme `
--window SU2 `
--scheme litim `
--mu_frac 0.18 `
--amp none `
--outdir results_su2_mu018_ampNone
# U(1), μ/Δω* = 0.18 (amp=none)
python rtg_eft_sim.py `
--do ckappa `
--scheme litim `
--window U1 `
--mu_frac 0.18 `
--amp none `
--outdir results_u1_mu018_ampNone
# U(1)^2, μ/Δω* = 0.18 (reweight, 15M links; amp=none)
python rtg_eft_sim.py `
--do ckappa `
--scheme litim `
--window U1xU1 `
--mode reweight `
--nlinks 15000000 `
--mu_frac 0.18 `
--amp none `
--outdir results_u1xu1_mu018_ampNone
3 · Post‑processing
# Reports per summary (κ_B with J_ex, K')
python rtg_eft_post2.py report results_su2_mu018_ampNone/summary_litim_SU2.json --Jex 2.2 --Kp 12.0 --output-csv eft_release.csv
python rtg_eft_post2.py report results_u1_mu018_ampNone/summary_litim_U1.json --Jex 2.2 --Kp 12.0 --output-csv eft_release.csv
python rtg_eft_post2.py report results_u1xu1_mu018_ampNone/summary_litim_U1xU1.json --Jex 2.2 --Kp 12.0 --output-csv eft_release.csv
# Scheme deltas (should be ~0% under amp=none)
python rtg_eft_post2.py scheme results_su2_mu018_ampNone/summary_litim_SU2.json results_su2_mu018_ampNone/summary_sharp_SU2.json
# Window comparison (SU2, U1, U1xU1)
python rtg_eft_post2.py windows results_su2_mu018_ampNone/summary_litim_SU2.json results_u1_mu018_ampNone/summary_litim_U1.json results_u1xu1_mu018_ampNone/summary_litim_U1xU1.json
# Bubble scan with small-ω correction
python rtg_eft_post2.py bubble results_su2_mu018_ampNone/summary_litim_SU2.json --mu-frac 0.18 --omega-over-mu 0.05 0.10 0.20
# Ward fit (with covariance)
python rtg_eft_post2.py ward results_ward_litim/ward_residual.csv
4 · Acceptance gates
- Δ(scheme) < 0.1% (amp=none expected 0%).
- Window robustness \(|\Delta C_\kappa|/C_\kappa < 1\%\).
- \(\max_a |\kappa_B(a)-\kappa_B(a_{\rm ref})|/\kappa_B(a_{\rm ref}) < 1\%\).
- Bubble fraction < 1.5% on \(\omega/\mu \in [0.05,0.20]\); spreads >0.5% → sys; >2% reject.
- Ward \(a_0\) compatible with 0 within errors.
Artifacts for publication: Upload eft_release.csv
, eft_release.zip
, and the generated plots (C(L) fits, Ward, scheme weights) to /wp-content/uploads/2025/09/
, then reference them on the EFT and Release pages.