integrable_exp_mul_sq_gaussianReal_zero
theoremverifiedimported Lean
Integrability of a Gaussian moment generating integrand. Let $v$ be a nonnegative real number with $v \ne 0$, and let $N(0,v)$ denote the centred Gaussian probability measure on $\bbr$ with variance $v$. If $t \in \bbr$ satisfies $2tv < 1$, then the function $y \mapsto e^{t y^2}$ is integrable with respect to $N(0,v)$.
Statement
set_option relaxedAutoImplicit false in set_option autoImplicit false in open MeasureTheory ProbabilityTheory Real NNReal Matrix Finset in variable {d k : ℕ} in lemma integrable_exp_mul_sq_gaussianReal_zero (v : ℝ≥0) (hv : v ≠ 0) (t : ℝ) (ht : 2 * t * (v : ℝ) < 1) : Integrable (fun y => Real.exp (t * y ^ 2)) (gaussianReal 0 v)
- Source
- TCSlib, `integrable_exp_mul_sq_gaussianReal_zero`
- Lean proof
- By Ganesh Sankar, from github.com/Shilun-Allan-Li/tcslib at
c8a02591b1e5(Copyright (c) 2026 Ganesh Sankar. All rights reserved, Apache-2.0). Compiled here against leanprover/lean4:v4.33.1, Mathlib v4.33.1; written for leanprover/lean4:v4.25.0.integrable_exp_mul_sq_gaussianReal_zero← integrable_exp_mul_sq_gaussianReal_zero (TCSlib/LearningTheory/JohnsonLindenstrauss/ChiSquaredMGF.lean:319): verbatim
- Verified
- 07 Oct 2026
- Axioms
Classical.choiceQuot.soundpropext- Used by 1
- centered_chi_squared_stepCentered chi-squared MGF bound. Let $(\Omega,\mu)$ be a probability space and let $k$ be a positive natural number. Suppose…
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.
set_option relaxedAutoImplicit false inset_option autoImplicit false inopen MeasureTheory ProbabilityTheory Real NNReal Matrix Finset invariable {d k : ℕ} in/-- **Integrability of `exp(t · y²)` under `gaussianReal 0 v`.** For `v ≠ 0` and every real `t` with `2tv < 1`, the function `y ↦ exp(t y²)` is integrable with respect to the law `N(0, v)`. This is the integrability implicit in the finiteness of `E exp(sX²) = (1 − 2s)^{−1/2}` in [DG03, proof of Lemma 2.2]. **Proof sketch.** Step 1: write `N(0, v)` as Lebesgue measure with the Gaussian density, so integrability against it is integrability of the density times the integrand against Lebesgue measure. Step 2: rewrite the product pointwise as `(√(2πv))⁻¹ exp(−(1/(2v) − t) y²)`, where `b = 1/(2v) − t > 0` by the hypothesis. Step 3: `exp(−b y²)` is Lebesgue-integrable for `b > 0` (Mathlib's `integrable_exp_neg_mul_sq`); transfer along the pointwise identity. -/lemma integrable_exp_mul_sq_gaussianReal_zero (v : ℝ≥0) (hv : v ≠ 0) (t : ℝ) (ht : 2 * t * (v : ℝ) < 1) : Integrable (fun y => Real.exp (t * y ^ 2)) (gaussianReal 0 v) := by -- Step 1: convert `gaussianReal 0 v` to `volume.withDensity (gaussianPDF 0 v)`. rw [gaussianReal_of_var_ne_zero _ hv] rw [integrable_withDensity_iff_integrable_smul' (measurable_gaussianPDF _ _) (ae_of_all _ fun _ => gaussianPDF_lt_top)] -- Goal: Integrable (fun y => (gaussianPDF 0 v y).toReal • exp(t y²)) volume. -- Rewrite the integrand as `(√(2πv))⁻¹ * exp(-(1/(2v) - t) y²)`, then use -- `integrable_exp_neg_mul_sq` with `b = 1/(2v) - t > 0`. have hv_pos : (0 : ℝ) < (v : ℝ) := NNReal.coe_pos.mpr (pos_iff_ne_zero.mpr hv) have h2v_pos : 0 < 2 * (v : ℝ) := by linarith have hb_pos : 0 < 1 / (2 * (v : ℝ)) - t := by rw [sub_pos, lt_div_iff₀ h2v_pos]; linarith -- Step 2: pointwise rewrite of integrand. have h_eq : ∀ y : ℝ, (gaussianPDF 0 v y).toReal • Real.exp (t * y ^ 2) = (Real.sqrt (2 * Real.pi * (v : ℝ)))⁻¹ * Real.exp (-(1 / (2 * (v : ℝ)) - t) * y ^ 2) := by intro y rw [toReal_gaussianPDF, gaussianPDFReal, smul_eq_mul, sub_zero, mul_assoc, ← Real.exp_add] congr 2 field_simp ring -- Step 3: integrability of the rewritten form. have h_int_rewritten : Integrable (fun y => (Real.sqrt (2 * Real.pi * (v : ℝ)))⁻¹ * Real.exp (-(1 / (2 * (v : ℝ)) - t) * y ^ 2)) volume := Integrable.const_mul (integrable_exp_neg_mul_sq hb_pos) _ -- Transfer to original integrand via AE-equality. exact h_int_rewritten.congr (ae_of_all _ (fun y => (h_eq y).symm))