AFTD Auto-Formalizing Theoretical Domains

exp_neg_le_quadratic

theoremverifiedimported Lean

Quadratic upper bound on $e^{-t}$. For every real number $t \ge 0$, the value $e^{-t}$ is at most the second-order Taylor polynomial $1 - t + t^2/2$; that is, \[ e^{-t} \;\le\; 1 - t + \frac{t^2}{2}. \]

Statement

set_option relaxedAutoImplicit false in
set_option autoImplicit false in
open Real Finset BigOperators in
lemma exp_neg_le_quadratic {t : ℝ} (ht : 0 ≤ t) :
    Real.exp (-t) ≤ 1 - t + t ^ 2 / 2
Source
TCSlib, `exp_neg_le_quadratic`
Lean proof
By Karim Abdel Sadek, Mark Bedaywi, from github.com/Shilun-Allan-Li/tcslib at c8a02591b1e5 (Copyright (c) 2026 Karim Abdel Sadek and Mark Bedaywi. 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.
Verified
07 Oct 2026
Axioms
Classical.choiceQuot.soundpropext
Used by 1
  • one_sub_exp_neg_geLower bound on $1 - e^{-\eta}$. For every real number $\eta > 0$, the quantity $1 - e^{-\eta}$ is bounded below by its second-order Taylor…

Lean source view module on GitHub

set_option relaxedAutoImplicit false inset_option autoImplicit false inopen Real Finset BigOperators in/-- For `t ≥ 0`, `exp(-t) ≤ 1 - t + t²/2` (the second-order upper Taylor bound for `exp(-t)`). Proof idea: multiply by `exp t > 0`; from the lower Taylor bound `1 + t + t²/2 ≤ exp t` we get `exp(t) (1 - t + t²/2) ≥ (1 + t + t²/2)(1 - t + t²/2) = 1 + t⁴/4 ≥ 1`. -/lemma exp_neg_le_quadratic {t : ℝ} (ht : 0 ≤ t) :    Real.exp (-t) ≤ 1 - t + t ^ 2 / 2 := by  -- We prove the bound by multiplying both sides by `exp t > 0` and using the  -- standard lower Taylor bound for `exp t`.  have he : (0 : ℝ) < Real.exp t := exp_pos t  have hq : (0 : ℝ) < 1 - t + t ^ 2 / 2 := by nlinarith [sq_nonneg t]  -- exp(t) ≥ 1 + t + t²/2  have hquad := quadratic_le_exp_of_nonneg ht  -- (1 + t + t²/2)(1 - t + t²/2) = 1 + t⁴/4 ≥ 1  -- So exp(t) * (1 - t + t²/2) ≥ (1 + t + t²/2)(1 - t + t²/2) ≥ 1  have key : 1 ≤ Real.exp t * (1 - t + t ^ 2 / 2) := by    have : (1 + t + t ^ 2 / 2) * (1 - t + t ^ 2 / 2) = 1 + t ^ 4 / 4 := by ring    nlinarith [sq_nonneg (t ^ 2), mul_le_mul_of_nonneg_right hquad hq.le]  -- exp(-t) = (exp t)⁻¹ ≤ 1 - t + t²/2  rw [exp_neg]  exact le_of_mul_le_mul_left (by nlinarith [mul_inv_cancel₀ he.ne']) he
See it in the graphDiscuss this resultChallenge it