igneum/docs/analysis/cryptanalysis/logs/adv-mixer/q1-fold-integral-merged.log
igneum-labs 996d0ad353 adv-mixer: finalize report (Q1 BOUND), durable logs, run-box.sh log path
Internal adversarial pass, not an independent review. Q1 (this lane, the
algebraic structure of M_r): the fold probe and the integral cube-sum degree
test on the frozen-identical binary show no composition cheaper than 8x. No
affine fold, no word separability, no key-order commutation; algebraic degree
>= 16 after one application and saturated after two. The 9,360 ops per item
stand. Q2 margin (full diffusion at 2 applications) and Q3 census (best day a
1.17x FPGA multiply datapath, 1 in 2^24, zero wall-time gain, ROT-all-equal
never seen) kept as courtesy runs attributed to adv-mixer-3 and adv-mixer-2.

igneum-pow merged to build/master and re-proven byte-identical to c3d32437;
pre-merge runs re-run on the merged tree and matched. Logs copied into the
branch so the report paths survive; run-box.sh writes outside the mirror.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
2026-10-07 18:40:05 +00:00

70 lines
5.4 KiB
Text

FROZEN_HEAD d8eea5f7 igneum-pow IDENTICAL to 017e7037
bin 179b77a5bb1e2158004d4044de6d6622cf649ce6e8cc5e69623053e849651274
UTC_START 2026-10-07T18:32:17Z
### fold confirm
self-check: spec 1.8.1 and 1.8.4 genesis vectors OK; 72 application keys are 1..72 times 0x9E3779B9
fold day=20729 trials=1000000
(a) GF(2) affinity violations of the 2-application map: 1000000 of 1000000 (0 would be a BREAK: the map is affine)
(b) dead (in_word -> out_word) pairs over the full 8-application block: 0 of 256 (any would be a broken dependency)
(c) key-order agreements M(M(.,rk1),rk2) == M(M(.,rk2),rk1): 0 of 1000000 (any would let keys fold)
VERDICT: BOUND: no fold on these probes
### integral apps=1
self-check: spec 1.8.1 and 1.8.4 genesis vectors OK; 72 application keys are 1..72 times 0x9E3779B9
integral day=20729 apps=1 placements_per_d=4000 threads=48
d = cube dimension; 'nonzero' = placements whose cube-sum is nonzero on >=1 output bit (degree >= d reached)
d= 1 nonzero 4000/4000 = 1.0000 mean out-bits set 236.1/512 [degree >= d reached]
d= 2 nonzero 4000/4000 = 1.0000 mean out-bits set 185.4/512 [degree >= d reached]
d= 3 nonzero 4000/4000 = 1.0000 mean out-bits set 174.6/512 [degree >= d reached]
d= 4 nonzero 4000/4000 = 1.0000 mean out-bits set 172.3/512 [degree >= d reached]
d= 5 nonzero 4000/4000 = 1.0000 mean out-bits set 172.5/512 [degree >= d reached]
d= 6 nonzero 4000/4000 = 1.0000 mean out-bits set 172.3/512 [degree >= d reached]
d= 7 nonzero 4000/4000 = 1.0000 mean out-bits set 174.6/512 [degree >= d reached]
d= 8 nonzero 4000/4000 = 1.0000 mean out-bits set 174.7/512 [degree >= d reached]
d= 9 nonzero 4000/4000 = 1.0000 mean out-bits set 176.1/512 [degree >= d reached]
d=10 nonzero 4000/4000 = 1.0000 mean out-bits set 177.9/512 [degree >= d reached]
d=11 nonzero 4000/4000 = 1.0000 mean out-bits set 181.2/512 [degree >= d reached]
d=12 nonzero 4000/4000 = 1.0000 mean out-bits set 183.6/512 [degree >= d reached]
d=13 nonzero 4000/4000 = 1.0000 mean out-bits set 187.3/512 [degree >= d reached]
d=14 nonzero 4000/4000 = 1.0000 mean out-bits set 191.5/512 [degree >= d reached]
d=15 nonzero 4000/4000 = 1.0000 mean out-bits set 196.4/512 [degree >= d reached]
d=16 nonzero 4000/4000 = 1.0000 mean out-bits set 203.1/512 [degree >= d reached]
READING: the smallest d at which nonzero is ~1.0 is the degree floor; high degree at small d means no low-degree algebraic shortcut across the applications
### integral apps=2
self-check: spec 1.8.1 and 1.8.4 genesis vectors OK; 72 application keys are 1..72 times 0x9E3779B9
integral day=20729 apps=2 placements_per_d=4000 threads=48
d = cube dimension; 'nonzero' = placements whose cube-sum is nonzero on >=1 output bit (degree >= d reached)
d= 1 nonzero 4000/4000 = 1.0000 mean out-bits set 256.1/512 [degree >= d reached]
d= 2 nonzero 4000/4000 = 1.0000 mean out-bits set 255.8/512 [degree >= d reached]
d= 3 nonzero 4000/4000 = 1.0000 mean out-bits set 255.7/512 [degree >= d reached]
d= 4 nonzero 4000/4000 = 1.0000 mean out-bits set 255.9/512 [degree >= d reached]
d= 5 nonzero 4000/4000 = 1.0000 mean out-bits set 256.2/512 [degree >= d reached]
d= 6 nonzero 4000/4000 = 1.0000 mean out-bits set 255.7/512 [degree >= d reached]
d= 7 nonzero 4000/4000 = 1.0000 mean out-bits set 255.8/512 [degree >= d reached]
d= 8 nonzero 4000/4000 = 1.0000 mean out-bits set 255.7/512 [degree >= d reached]
d= 9 nonzero 4000/4000 = 1.0000 mean out-bits set 256.0/512 [degree >= d reached]
d=10 nonzero 4000/4000 = 1.0000 mean out-bits set 256.4/512 [degree >= d reached]
d=11 nonzero 4000/4000 = 1.0000 mean out-bits set 256.4/512 [degree >= d reached]
d=12 nonzero 4000/4000 = 1.0000 mean out-bits set 256.1/512 [degree >= d reached]
d=13 nonzero 4000/4000 = 1.0000 mean out-bits set 256.1/512 [degree >= d reached]
d=14 nonzero 4000/4000 = 1.0000 mean out-bits set 255.8/512 [degree >= d reached]
d=15 nonzero 4000/4000 = 1.0000 mean out-bits set 255.8/512 [degree >= d reached]
d=16 nonzero 4000/4000 = 1.0000 mean out-bits set 255.9/512 [degree >= d reached]
READING: the smallest d at which nonzero is ~1.0 is the degree floor; high degree at small d means no low-degree algebraic shortcut across the applications
### diffusion spot K=1,2
self-check: spec 1.8.1 and 1.8.4 genesis vectors OK; 72 application keys are 1..72 times 0x9E3779B9
diffusion day=20729 plant=none apps=1 start=0 states=2000000 threads=48
mean output-flip probability over all 262144 cells: 0.461179 (ideal 0.5)
census band: 8 sigma = 0.002828 (2000000 states)
dependency holes (p == 0 or p == 1 exactly): 551 of 262144
strong-bias cells (|p-0.5| > band, not a hole): 112032 of 262144
worst cell: in_bit 29 -> out_bit 76 p = 1.000000 dev = 0.500000 (1414.2 sigma)
VERDICT: FINDING (holes or strong bias at this K)
self-check: spec 1.8.1 and 1.8.4 genesis vectors OK; 72 application keys are 1..72 times 0x9E3779B9
diffusion day=20729 plant=none apps=2 start=0 states=2000000 threads=48
mean output-flip probability over all 262144 cells: 0.500001 (ideal 0.5)
census band: 8 sigma = 0.002828 (2000000 states)
dependency holes (p == 0 or p == 1 exactly): 0 of 262144
strong-bias cells (|p-0.5| > band, not a hole): 0 of 262144
worst cell: in_bit 203 -> out_bit 42 p = 0.501711 dev = 0.001711 (4.8 sigma)
VERDICT: no distinguisher at this K
UTC_END 2026-10-07T18:34:01Z