Context
The lower bound counts distinct complete seam functions, not accepted traces. Recognition may merge prefixes whose full routing actions differ.
Definitions
- For , .
- Two exact-profile states may merge only when these functions agree on every suffix guard.
Hypotheses and scope
- .
- The interface exposes the complete horizon map on hypothetical suffix guards; it is not merely an accept/reject state.
Proof or evidence
The multiplicative order yields the distinct slopes. The guide proves that order by the self-contained induction . Recording slope and intercept modulo gives the matching linear-bit upper bound.
Verification notes
This run checked the valuation induction, the state-count-to-bit conversion, and the trace-budget caveat. The source correctly limits the displayed family to horizons satisfying the stated exponential prefix-length budget.
Limitations
- This is not a lower bound for paradoxical-trace recognition, finite automata for the accepted language, or Collatz computation.
- Within a trace budget , this family alone yields only .
Notes
Distinctness uses ord_(2^ell)(3)=2^(ell-2), proved in the guide by a self-contained 2-adic valuation induction. This is a lower bound for exact routing profiles, not for paradoxical-trace recognition, finite automata for the accepted language, or general Collatz computation. Acceptance may merge prefixes whose complete seam actions differ. The displayed calibration uses prefix lengths up to 2^(ell-2)-1, so within a trace budget K it yields only ell<=floor(log_2(K+1))+2; no linear-in-K memory lower bound is claimed. This budgets the prefix only because the interface acts on hypothetical suffix guards; a convention requiring the suffix to occur inside the same trace would also charge ell symbols.