power_divergence_half_root_bounds
For x ≥ 0 and b ≥ 0, if a^2 ≤ x ≤ b^2 then a ≤ x^(1/2) ≤ b.
Statement
theorem power_divergence_half_root_bounds (x a b : ℝ) (hx : 0 ≤ x) (hb : 0 ≤ b)
(h1 : a ^ 2 ≤ x) (h2 : x ≤ b ^ 2) : a ≤ x ^ ((1 : ℝ) / 2) ∧ x ^ ((1 : ℝ) / 2) ≤ b- Source
- Helper for the refutations of subadditivity of mutual f_β-information, arXiv:2610.21231 (On the subadditivity of mutual f-information): rational bounds on real roots.
- Verified
- 16 Oct 2026
- Axioms
Classical.choiceQuot.soundpropext- Used by 1
- mutual_hellinger_information_not_subadditiveSubadditivity fails in the Hellinger case β = 1/2: there are binary random variables X, Y, Z with Y ← X → Z and Ĩ_{f_{1/2}}(X; YZ) >…
The proof, explained
The proof first establishes that x^(1/2) is non-negative and that (x^(1/2))^2 = x. It then splits the goal into the lower bound a ≤ x^(1/2) and the upper bound x^(1/2) ≤ b. Each bound is deduced from the corresponding assumption (a^2 ≤ x and x ≤ b^2) using the monotonicity of the squaring operation on non-negative real numbers.
0 ≤ x^(1/2)
Since x ≥ 0, any real power of x is non-negative by Real.rpow_nonneg, so x^(1/2) ≥ 0.
uses
Real.rpow_nonneghave ht0 : 0 ≤ x ^ ((1 : ℝ) / 2) := Real.rpow_nonneg hx _(x^(1/2))^2 = x
Rewriting the exponent 1/2 as 2⁻¹, the identity follows from Real.rpow_inv_natCast_pow because x ≥ 0 and 2 ≠ 0.
uses
Real.rpow_inv_natCast_powhave ht : (x ^ ((1 : ℝ) / 2)) ^ 2 = x := by rw [show (1 : ℝ) / 2 = ((2 : ℕ) : ℝ)⁻¹ by norm_num] exact Real.rpow_inv_natCast_pow hx (by norm_num)
It suffices to show a ≤ x^(1/2) and x^(1/2) ≤ b separately.
The target is a conjunction; the constructor tactic splits it into the lower bound and upper bound subgoals.
constructora ≤ x^(1/2)
By le_of_pow_le_pow_left₀, since 2 ≠ 0 and x^(1/2) ≥ 0, it suffices to show a^2 ≤ (x^(1/2))^2. Since (x^(1/2))^2 = x, this is precisely the assumption a^2 ≤ x (h1).
uses
le_of_pow_le_pow_left₀· exact le_of_pow_le_pow_left₀ (by norm_num : (2 : ℕ) ≠ 0) ht0 (by rw [ht]; exact h1)
x^(1/2) ≤ b
Similarly, by le_of_pow_le_pow_left₀, since 2 ≠ 0 and b ≥ 0 (hb), it suffices to show (x^(1/2))^2 ≤ b^2. Since (x^(1/2))^2 = x, this is precisely the assumption x ≤ b^2 (h2).
uses
le_of_pow_le_pow_left₀· exact le_of_pow_le_pow_left₀ (by norm_num : (2 : ℕ) ≠ 0) hb (by rw [ht]; exact h2)
Real.rpow_nonneg- For any non-negative real number x and any real exponent y, x^y ≥ 0.
Real.rpow_inv_natCast_pow- For any non-negative real number x and non-zero natural number n, (x^(1/n))^n = x.
le_of_pow_le_pow_left₀- If n ≠ 0, b ≥ 0, and a^n ≤ b^n in a linearly ordered monoid with zero, then a ≤ b.
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.
theorem power_divergence_half_root_bounds (x a b : ℝ) (hx : 0 ≤ x) (hb : 0 ≤ b) (h1 : a ^ 2 ≤ x) (h2 : x ≤ b ^ 2) : a ≤ x ^ ((1 : ℝ) / 2) ∧ x ^ ((1 : ℝ) / 2) ≤ b := by have ht0 : 0 ≤ x ^ ((1 : ℝ) / 2) := Real.rpow_nonneg hx _ have ht : (x ^ ((1 : ℝ) / 2)) ^ 2 = x := by rw [show (1 : ℝ) / 2 = ((2 : ℕ) : ℝ)⁻¹ by norm_num] exact Real.rpow_inv_natCast_pow hx (by norm_num) constructor · exact le_of_pow_le_pow_left₀ (by norm_num : (2 : ℕ) ≠ 0) ht0 (by rw [ht]; exact h1) · exact le_of_pow_le_pow_left₀ (by norm_num : (2 : ℕ) ≠ 0) hb (by rw [ht]; exact h2)