closed_network_bound_of_product_bounds
For any integer n ≥ 1, real correlator values I, J, and non-negative tuples x, y : Fin n → ℝ satisfying x i + y i ≤ 1 for all i, if |I| ≤ ∏ i, x i and |J| ≤ ∏ i, y i, then |I| ^ (1 / n) + |J| ^ (1 / n) ≤ 1.
Statement
theorem closed_network_bound_of_product_bounds
(n : ℕ) (hn : 1 ≤ n) (I J : ℝ) (x y : Fin n → ℝ)
(hx : ∀ i, 0 ≤ x i) (hy : ∀ i, 0 ≤ y i)
(hxy : ∀ i, x i + y i ≤ 1)
(hI : |I| ≤ ∏ i : Fin n, x i) (hJ : |J| ≤ ∏ i : Fin n, y i) :
|I| ^ (1 / (n : ℝ)) + |J| ^ (1 / (n : ℝ)) ≤ 1- Source
- arXiv:2610.35838 Eq. 32
- Verified
- 08 Nov 2026
- Axioms
Classical.choiceQuot.soundpropext- Built on 1
- closed_network_am_gm_product_boundFor any integer n ≥ 1 and any non-negative tuples x, y : Fin n → ℝ satisfying x i + y i ≤ 1 for all i, (∏ i, x i) ^ (1 / n) + (∏ i, y i) ^…
Read back from the Lean
For any natural number $n \ge 1$, any real numbers $I$ and $J$, and any families of real numbers $x, y : \text{Fin } n \to \mathbb{R}$, if: 1. $x(i) \ge 0$ for all $i : \text{Fin } n$, 2. $y(i) \ge 0$ for all $i : \text{Fin } n$, 3. $x(i) + y(i) \le 1$ for all $i : \text{Fin } n$, 4. $|I| \le \prod_{i \in \text{Fin } n} x(i)$, and 5. $|J| \le \prod_{i \in \text{Fin } n} y(i)$, then $|I|^{1/n} + |J|^{1/n} \le 1$ (where the exponents are the real number $1/n$).
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
The theorem is proved by taking the n-th root of both sides in the product bounds for |I| and |J|, then applying an established AM-GM product bound. Since n ≥ 1, the exponent 1/n is non-negative, and the monotonicity of real powers preserves the bounds, giving |I| ^ (1 / n) ≤ (∏ i, x i) ^ (1 / n) and |J| ^ (1 / n) ≤ (∏ i, y i) ^ (1 / n). The lemma `closed_network_am_gm_product_bound` guarantees that (∏ i, x i) ^ (1 / n) + (∏ i, y i) ^ (1 / n) ≤ 1. Combining these inequalities by linear arithmetic yields the desired bound.
The exponent 1/n is non-negative.
From 0 < 1 (`Nat.zero_lt_one`) and 1 ≤ n (hypothesis `hn`), transitivity (`Nat.lt_of_lt_of_le`) gives 0 < n in ℕ, which casts to 0 < n in ℝ. Since 0 ≤ 1 (`zero_le_one`) and 0 ≤ n, the division rule `div_nonneg` yields 0 ≤ 1/n.
uses
Nat.lt_of_lt_of_le,Nat.zero_lt_one,div_nonneg,zero_le_onehave hnpos : (0 : ℝ) < n := by exact_mod_cast (Nat.lt_of_lt_of_le Nat.zero_lt_one hn) have hz : 0 ≤ 1 / (n : ℝ) := div_nonneg zero_le_one hnpos.le
|I| ^ (1 / n) ≤ (∏ i, x i) ^ (1 / n) and |J| ^ (1 / n) ≤ (∏ i, y i) ^ (1 / n).
Because 0 ≤ |I| (`abs_nonneg`), |I| ≤ ∏ i, x i (hypothesis `hI`), and 0 ≤ 1/n, raising both sides to the power 1/n preserves the inequality via `Real.rpow_le_rpow`. The exact same monotonicity step applied to J and y using hypothesis `hJ` produces |J| ^ (1 / n) ≤ (∏ i, y i) ^ (1 / n).
uses
Real.rpow_le_rpow,abs_nonneghave h1 : |I| ^ (1 / (n : ℝ)) ≤ (∏ i : Fin n, x i) ^ (1 / (n : ℝ)) := Real.rpow_le_rpow (abs_nonneg I) hI hz have h2 : |J| ^ (1 / (n : ℝ)) ≤ (∏ i : Fin n, y i) ^ (1 / (n : ℝ)) := Real.rpow_le_rpow (abs_nonneg J) hJ hz
(∏ i, x i) ^ (1 / n) + (∏ i, y i) ^ (1 / n) ≤ 1.
Applying the AM-GM product bound for closed networks (`closed_network_am_gm_product_bound`) to n ≥ 1, x_i ≥ 0, y_i ≥ 0, and x_i + y_i ≤ 1 yields that the sum of the n-th roots of the two products is at most 1.
uses
closed_network_am_gm_product_boundhave hmain := closed_network_am_gm_product_bound n hn x y hx hy hxy|I| ^ (1 / n) + |J| ^ (1 / n) ≤ 1.
Linear arithmetic (`linarith`) sums the bounds |I| ^ (1 / n) ≤ (∏ i, x i) ^ (1 / n) and |J| ^ (1 / n) ≤ (∏ i, y i) ^ (1 / n) and chains them with (∏ i, x i) ^ (1 / n) + (∏ i, y i) ^ (1 / n) ≤ 1 to establish the target inequality.
linarith
Nat.lt_of_lt_of_le- If n < m and m ≤ k for natural numbers, then n < k.
Nat.zero_lt_one- In the natural numbers, 0 < 1.
div_nonneg- If a ≥ 0 and b ≥ 0, then a / b ≥ 0.
zero_le_one- Zero is less than or equal to one.
Real.rpow_le_rpow- For non-negative reals x ≤ y and exponent z ≥ 0, x^z ≤ y^z.
abs_nonneg- The absolute value of any real number is non-negative: 0 ≤ |a|.
closed_network_am_gm_product_bound- For tuples x, y : Fin n → ℝ with x i ≥ 0, y i ≥ 0, and x i + y i ≤ 1 for 1 ≤ n, (∏ i, x i)^(1/n) + (∏ i, y i)^(1/n) ≤ 1.
- The cast of the strict inequality 0 < n from the natural numbers to the real numbers in line 8 is handled by `exact_mod_cast`.
- In line 15, `linarith` automatically combines the inequalities via linear arithmetic.
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.
/-- Bounding the n-th root sum of correlators by the AM-GM product bound. -/theorem closed_network_bound_of_product_bounds (n : ℕ) (hn : 1 ≤ n) (I J : ℝ) (x y : Fin n → ℝ) (hx : ∀ i, 0 ≤ x i) (hy : ∀ i, 0 ≤ y i) (hxy : ∀ i, x i + y i ≤ 1) (hI : |I| ≤ ∏ i : Fin n, x i) (hJ : |J| ≤ ∏ i : Fin n, y i) : |I| ^ (1 / (n : ℝ)) + |J| ^ (1 / (n : ℝ)) ≤ 1 := by have hnpos : (0 : ℝ) < n := by exact_mod_cast (Nat.lt_of_lt_of_le Nat.zero_lt_one hn) have hz : 0 ≤ 1 / (n : ℝ) := div_nonneg zero_le_one hnpos.le have h1 : |I| ^ (1 / (n : ℝ)) ≤ (∏ i : Fin n, x i) ^ (1 / (n : ℝ)) := Real.rpow_le_rpow (abs_nonneg I) hI hz have h2 : |J| ^ (1 / (n : ℝ)) ≤ (∏ i : Fin n, y i) ^ (1 / (n : ℝ)) := Real.rpow_le_rpow (abs_nonneg J) hJ hz have hmain := closed_network_am_gm_product_bound n hn x y hx hy hxy linarith