The silver exponent is a counting exponent.
ProvedBerggrenZeta.count_ge_silver_rpowaether-catalognovelty
The silver exponent is a counting exponent.
For H ≥ 5 the number of Berggren nodes with hypotenuse at most H is at least
(1/3)·(H/5)^{σ₀}, where σ₀ = log 3 / (2 log(1+√2)) is the silver abscissa.
Combined with nodesBelow_ncard_le (N(H) = O(H)) and silverAbscissa_lt_one, this shows
the silver value is a genuine lower growth exponent but not the true one.
theorem BerggrenZeta.count_ge_silver_rpow{H : ℕ} (hH : 5 ≤ H) :
(1 / 3 : ℝ) * ((H : ℝ) / 5) ^ silverAbscissa ≤ ((nodesBelow H).ncard : ℝ) := by sorry
Formalization Note Transplanted verbatim from the Aether Catalog source Novelty/BerggrenTreeCounting.lean; the statement is byte-identical to the source declaration, elaborated with autoImplicit disabled in the platform environment.
Preamble
-- Thm stub generated from Novelty/BerggrenTreeCounting.lean
import Mathlib
import Definitions.Def_Novelty_BerggrenTreeCounting
import Definitions.Def_Novelty_BerggrenTreeCriticalLine
import Definitions.Def_Novelty_BerggrenTreeSilverGrowth
/-!
# Counting Berggren nodes below a height: the silver exponent as a lower bound
Let `N(H) = #{ w : c(w) ≤ H }` be the number of nodes of the Berggren tree whose hypotenuse
is at most `H`. This file proves the two-sided estimate
* `nodesBelow_ncard_le` : `N(H) ≤ (⌊√H⌋ + 1)²`, so `N(H) = O(H)` — a consequence of the
seed injectivity of `Novelty.BerggrenTreeZetaCore`, since a node with `c(w) = m²+n² ≤ H`
has both Euclid parameters in `[0, √H]`;
* `silver_count_lower` / `count_ge_silver_rpow` : `N(H) ≥ 3^d` whenever `5(3+2√2)^d ≤ H`,
hence `N(H) ≥ (1/3)·(H/5)^{σ₀}` with `σ₀ = log 3 / (2 log(1+√2))` the *silver abscissa*
of `Novelty.BerggrenTreeCriticalLine`.
Together with `silverAbscissa_lt_one` this pins down the role of the silver exponent
exactly: the silver value `σ₀ ≈ 0.6237` is the growth exponent of the **depth-graded**
(purely exponential) model of the tree, and it is a genuine *lower* bound for the counting
function, but it is strictly smaller than the true abscissa of convergence `1` computed in
`Novelty.BerggrenTreeZetaAbscissa`. The gap between `H^{σ₀}` and `H` is produced by the
polynomially growing outer spines (`Lspine_hyp`, `Rspine_hyp`), which contribute nodes far
below the silver speed limit.
## Lab notes (computed with `#eval` on the definitions of this development)
| `H` | nodes with `c(w) ≤ H` | `(⌊√H⌋+1)²` | largest `d` with `5(3+2√2)^d ≤ H` | `3^d` |
|-------|----------------------|--------------|-----------------------------------|-------|
| 5 | 1 | 9 | 0 | 1 |
| 50 | 7 | 64 | 1 | 3 |
| 200 | 32 | 225 | 2 | 9 |
| 1000 | 158 | 1024 | 3 | 27 |
The counts (1, 7, 32, 158) — obtained by enumerating the admissible Euclid seeds `(m,n)`
with `m²+n² ≤ H`, which by `seedEquiv` is exactly the node count — confirm both bounds and
show that the truth sits strictly between `3^d ≍ H^{σ₀}` and `(√H+1)²`, in line with the
abscissa-`1` theorem (the counts grow essentially linearly in `H`).
-/
open BerggrenZeta
-- open removed: section is not a namespace
/-! ## Lower bound: the depth-`d` slice -/
/-! ## The silver exponent as a counting exponent -/Formal statement
theorem BerggrenZeta.count_ge_silver_rpow{H : ℕ} (hH : 5 ≤ H) :
(1 / 3 : ℝ) * ((H : ℝ) / 5) ^ silverAbscissa ≤ ((nodesBelow H).ncard : ℝ) := by sorrySource