golden_bpe_bias_in_open_unit_interval
The golden bias p_* lies strictly in the open unit interval (0, 1).
Statement
theorem golden_bpe_bias_in_open_unit_interval :
0 < golden_bpe_bias ∧ golden_bpe_bias < 1- Related work
- knowledgebase
- Verified
- 12 Nov 2026
- Axioms
Classical.choiceQuot.soundpropext- Built on 1
- golden_bpe_biasThe golden critical bias p_* is defined as (3 - √5) / 2.
Read back from the Lean
The constant golden_bpe_bias lies strictly between 0 and 1; that is, 0 < golden_bpe_bias and golden_bpe_bias < 1.
Written by a model that saw only the Lean, never the English above. If the two disagree, that is worth a challenge.
The proof, explained
By definition, the golden bias is p_* = (3 - √5) / 2. Showing that p_* lies in (0, 1) reduces to bounding √5 strictly between 1 and 3, which is done by comparing 5 to 1² = 1 and 3² = 9 via the strict monotonicity of the square root function. Once these bounds 1 < √5 < 3 are established, linear arithmetic directly concludes that 0 < p_* and p_* < 1.
The goal reduces to 0 < (3 - √5) / 2 and (3 - √5) / 2 < 1.
Unfolding the definition of golden_bpe_bias expresses p_* explicitly as (3 - √5) / 2.
unfold golden_bpe_bias1 < √5
Rewriting 1 as √1 using Real.sqrt_one, the inequality √1 < √5 follows from 1 < 5 by the strict monotonicity of the real square root function (Real.sqrt_lt_sqrt).
uses
Real.sqrt_one,Real.sqrt_lt_sqrthave h1 : (1 : ℝ) < Real.sqrt 5 := by rw [← Real.sqrt_one] exact Real.sqrt_lt_sqrt (by norm_num) (by norm_num)
√5 < 3
Writing 3 as √9, the inequality √5 < √9 follows from 5 < 9 by the strict monotonicity of the real square root function (Real.sqrt_lt_sqrt).
uses
Real.sqrt_lt_sqrthave h2 : Real.sqrt 5 < 3 := by have : (3 : ℝ) = Real.sqrt 9 := by norm_num rw [this] exact Real.sqrt_lt_sqrt (by norm_num) (by norm_num)
0 < (3 - √5) / 2 and (3 - √5) / 2 < 1
Splitting the conjunction into two separate inequalities, each is established by linear arithmetic (linarith): 0 < (3 - √5) / 2 follows from √5 < 3, and (3 - √5) / 2 < 1 follows from 1 < √5.
constructor · linarith · linarith
Real.sqrt_one- The real square root of 1 equals 1.
Real.sqrt_lt_sqrt- The square root function on real numbers is strictly increasing on non-negative values: if 0 ≤ x and x < y, then √x < √y.
Written by a model (gemini-3.8-flash) from the Lean proof and the goal states Lean reported at each step. Lean verified the proof; this explanation is a reading aid and is not itself machine-checked.
Lean source view module on GitHub
Choose a tactic — a marked line — to see the goals Lean had before and after it, as in Lean’s infoview. Step with ← → or j k.
/-- The golden bias p_* satisfies 0 < p_* and p_* < 1. -/theorem golden_bpe_bias_in_open_unit_interval : 0 < golden_bpe_bias ∧ golden_bpe_bias < 1 := by unfold golden_bpe_bias have h1 : (1 : ℝ) < Real.sqrt 5 := by rw [← Real.sqrt_one] exact Real.sqrt_lt_sqrt (by norm_num) (by norm_num) have h2 : Real.sqrt 5 < 3 := by have : (3 : ℝ) = Real.sqrt 9 := by norm_num rw [this] exact Real.sqrt_lt_sqrt (by norm_num) (by norm_num) constructor · linarith · linarith