Files
sim-numerics/results_analyze_flat.txt
vvsandClaude Opus 5 e7e74d71e5 Flat regime (PREREG_flat_regime): the unkept-promise flag - after a model step that did not cross the root, r = f_new/f_old >= 1/4, the tangent-versus-halving break-even at m = 2 - RECOGNISES the regime (stage 1: 53.6% of model steps flagged on the flat families against 21.6% on APS and 17.0% on fun1; recall ten times the m>=2 estimator's 5.8%) and cuts the wide-bracket set by 350-490 wherever the budget does not already bind: safe 3390 -> 3037, step 3524 -> 3037, unguarded 4104 -> 3198.
NOT accepted into the proposed rule by the stop rule: under the projection the flat families are already budget-bound (CHA 3220 -> 3198 only) while the false flags on APS cost 2806 -> 3029 (Z2 fails); with G2 both sets fail. The per-problem edge over Chandrupatla on APS turns into 26/72/55. Guarantee intact in every cell. Next candidate, not pre-registered: require the promise to fail twice in a row.

Regression on the branch: 918 pass (twice); the one red test is the paper-number check awaiting the paper update.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-09-21 13:25:56 +01:00

33 lines
2.1 KiB
Plaintext

sweep: 140 configurations x 199 problems (FULL)
wrote sweep_flat.json (7s total)
tent, 3-point, pairs (no gate -> flat gate) APS CHA both flat fun3-6
none 2806->3029 4104->3198 6910->6227 2707->1758 solved 193 over 0
none +G2 2782->3089 3863->3214 6645->6303 2474->1749 solved 193 over 0
step 3036->3086 3524->3037 6560->6123 2111->1612 solved 193 over 0
step +G2 3089->3098 3529->3077 6618->6175 2106->1609 solved 193 over 0
bracket 7260->7144 3604->3604 10864->10748 1730->1730 solved 193 over 0
bracket +G2 6837->6801 3650->3650 10487->10451 1717->1717 solved 193 over 0
both 7239->7118 3515->3515 10754->10633 1529->1529 solved 193 over 0
both +G2 7130->7109 3463->3463 10593->10572 1504->1504 solved 193 over 0
project 2806->3029 3220->3198 6026->6227 1823->1758 solved 193 over 0
project +G2 2782->3089 3169->3214 5951->6303 1780->1749 solved 193 over 0
safe 3036->3086 3390->3037 6426->6123 1977->1612 solved 193 over 0
safe +G2 3089->3098 3392->3077 6481->6175 1969->1609 solved 193 over 0
credit 3568->4208 3020->3038 6588->7246 1044->1064 solved 193 over 0
credit +G2 3799->4403 3064->3053 6863->7456 1044->1064 solved 193 over 0
VERDICTS
proposed rule APS 2806->3029 (Z2 ✗) CHA 3220->3198 (Z1 ✓) Z3 ✓ (0 viol, 0 cells <193) Z4 ✓ both 6227 vs Chandrupatla APS 26/72/55 p=0.0017 not accepted
proposed rule + G2 APS 2782->3089 (Z2 ✗) CHA 3169->3214 (Z1 ✗) Z3 ✓ (0 viol, 0 cells <193) Z4 ✓ both 6303 vs Chandrupatla APS 27/70/56 p=0.0019 not accepted
nine functions, projection, no gate -> flat gate, both sets:
cbrt 18003 -> 18270
cosine 19741 -> 15910
exp_abs 6652 -> 7104
exp_sq 19786 -> 14298
rat_sq 19793 -> 14565
ratio 6026 -> 6227
ratio_soft 9352 -> 9690
rational 6823 -> 7777
sqrt 13388 -> 13608