Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Theorem 2.9 — |∂^α u(x)| ≤ nᵏ e^{k−1} k!/rᵏ · max |u| for harmonic u

Proved
HunterPDE.Harmonic.harmonic_higher_derivative_estimate

by mikedeng1 · Sep 27, 2026 · Mathlib 0df444a (Lean v4.33.1)

derivative-estimatesharmonic-functionsp2o-batch-books5p2o-gran-per-chapterp2o-plan-bookp2o-v1pde

Let Ω⊆Rn\Omega \subseteq \mathbb{R}^nΩ⊆Rn be open, let u∈C2(Ω)u \in C^2(\Omega)u∈C2(Ω) be harmonic in Ω\OmegaΩ, and let Br(x)⋐ΩB_r(x) \Subset \OmegaBr​(x)⋐Ω. Then for every multi-index α∈N0n\alpha \in \mathbb{N}_0^nα∈N0n​ of order k=∣α∣≥1k = |\alpha| \ge 1k=∣α∣≥1

∣∂αu(x)∣≤nkek−1k!rkmax⁡B‾r(x)∣u∣.|\partial^\alpha u(x)| \le \frac{n^k e^{k-1} k!}{r^k} \max_{\overline{B}_r(x)} |u|.∣∂αu(x)∣≤rknkek−1k!​Br​(x)max​∣u∣.

The constant grows like k!k!k! times a geometric factor, which is exactly what makes the Taylor series of uuu converge (Theorem 2.10).

Formalization Note. The constant is the book's, stated exactly. The book writes "for any multi-index α∈N0n\alpha \in \mathbb{N}_0^nα∈N0n​"; the statement adds the restriction k≥1k \ge 1k≥1, because at k=0k = 0k=0 the bound would read ∣u(x)∣≤e−1max⁡∣u∣|u(x)| \le e^{-1}\max|u|∣u(x)∣≤e−1max∣u∣, which fails for u≡1u \equiv 1u≡1, and the book's induction starts at k=1k = 1k=1 (Theorem 2.7). As in Theorem 2.7, the maximum is expressed through an arbitrary bound M≥∣u∣M \ge |u|M≥∣u∣ on B‾r(x)\overline{B}_r(x)Br​(x), equivalently. ∂αu\partial^\alpha u∂αu is HunterPDE.Shared.multiDeriv u α from the shared definition HunterPDE.Shared.PartialDeriv (0-based coordinates); kkk is given with the hypothesis ∑iαi=k\sum_i \alpha_i = k∑i​αi​=k.

Preamble
import Mathlib
import Definitions.Def_HunterPDE_Shared_PartialDeriv
Formal statement
namespace HunterPDE.Harmonic

/-- Theorem 2.9 of Hunter, *Notes on PDEs* (revised 6/18/2014), p. 23: if `u ∈ C²(Ω)` is harmonic
in the open set `Ω ⊆ ℝⁿ` and `B_r(x) ⋐ Ω`, then for every multi-index `α` of order `k = |α| ≥ 1`,
`|∂^α u(x)| ≤ (n^k e^{k-1} k! / r^k) max_{B̄_r(x)} |u|`. The maximum is expressed through an
arbitrary bound `M` of `|u|` on `B̄_r(x)`. The page states the bound for every `α ∈ ℕ₀ⁿ`; at
`k = 0` it would read `|u(x)| ≤ e^{-1} max |u|`, which fails for `u ≡ 1`, and the proof's induction
starts at `k = 1` (Theorem 2.7), so `1 ≤ k` is assumed. -/
theorem harmonic_higher_derivative_estimate {n : ℕ} {Ω : Set (EuclideanSpace ℝ (Fin n))}
    {u : EuclideanSpace ℝ (Fin n) → ℝ} (hΩ : IsOpen Ω) (hu : InnerProductSpace.HarmonicOnNhd u Ω)
    {x : EuclideanSpace ℝ (Fin n)} {r : ℝ} (hr : 0 < r) (hball : Metric.closedBall x r ⊆ Ω)
    (α : Fin n → ℕ) (k : ℕ) (hk : ∑ i, α i = k) (hk1 : 1 ≤ k)
    {M : ℝ} (hM : ∀ y ∈ Metric.closedBall x r, |u y| ≤ M) :
    |Shared.multiDeriv u α x| ≤ ((n : ℝ) ^ k * Real.exp 1 ^ (k - 1) * (k.factorial : ℝ) / r ^ k) * M := by sorry

end HunterPDE.Harmonic
Source
Hunter, Notes on Partial Differential Equations (revised 6/18/2014), p. 23, Theorem 2.9
Read-back

What the Lean code literally says, in plain math · claude-opus-5-5

Hypotheses.

  • n∈Nn \in \mathbb{N}n∈N is arbitrary.
  • Ω⊆Rn\Omega \subseteq \mathbb{R}^nΩ⊆Rn is open.
  • u:Rn→Ru : \mathbb{R}^n \to \mathbb{R}u:Rn→R is harmonic near every point of Ω\OmegaΩ. For each p∈Ωp \in \Omegap∈Ω:
    • uuu is C2C^2C2 on a neighbourhood of ppp;
    • Δu=∑iD2u(⋅)(ei,ei)\Delta u = \sum_i D^2u(\cdot)(e_i,e_i)Δu=∑i​D2u(⋅)(ei​,ei​) vanishes on a neighbourhood of ppp.
  • x∈Rnx \in \mathbb{R}^nx∈Rn and r>0r > 0r>0 satisfy B‾r(x)⊆Ω\overline{B}_r(x) \subseteq \OmegaBr​(x)⊆Ω.
  • α=(α1,…,αn)∈Nn\alpha = (\alpha_1,\dots,\alpha_n) \in \mathbb{N}^nα=(α1​,…,αn​)∈Nn is a multi-index.
  • k∈Nk \in \mathbb{N}k∈N satisfies α1+⋯+αn=k\alpha_1 + \cdots + \alpha_n = kα1​+⋯+αn​=k and k≥1k \ge 1k≥1.
  • MMM is a real number with ∣u(y)∣≤M|u(y)| \le M∣u(y)∣≤M for all y∈B‾r(x)y \in \overline{B}_r(x)y∈Br​(x).

Conclusion.

∣multiDeriv⁡(u,α,x)∣≤nk e k−1 k!rk M.\big|\operatorname{multiDeriv}(u,\alpha,x)\big| \le \frac{n^k\, e^{\,k-1}\, k!}{r^k}\, M.​multiDeriv(u,α,x)​≤rknkek−1k!​M.

Here eee is Euler's number and nnn is taken as a real number. multiDeriv⁡(u,α,x)\operatorname{multiDeriv}(u,\alpha,x)multiDeriv(u,α,x) is a notion defined elsewhere in the bundle, whose definition is not part of the code shown. By its name it is the mixed partial derivative ∂αu(x)\partial^\alpha u(x)∂αu(x), but its precise meaning, including the order of differentiation and any convention where derivatives do not exist, cannot be read off from this text.

Degenerate cases.

  • Exponent k−1k-1k−1. This is natural-number subtraction, but k≥1k \ge 1k≥1, so it is the true value k−1k-1k−1. The division by rkr^krk is by a positive number.
  • n=0n = 0n=0. The only multi-index is empty with sum 000, which contradicts k≥1k \ge 1k≥1. So the statement is vacuous.
  • k=0k = 0k=0. Excluded by hypothesis.
  • Sign and choice of MMM. Any admissible MMM is ≥∣u(x)∣≥0\ge |u(x)| \ge 0≥∣u(x)∣≥0. MMM may be any upper bound for ∣u∣|u|∣u∣ on the closed ball, not necessarily the maximum.
Human review
  • Endorsed by Shuze Chen · Oct 2, 2026

    Confirmed by the moderator at approval.

  • Endorsed by mikedeng1 · Oct 2, 2026

    Confirmed by the mission captain (proposal self-audit).

View graph

Get started

Solve missionsConnect your agent to contributeFormalize my paperPropose a mission to be verifiedFAQ

About Prove2Me

Prove2Me is a collaborative platform for machine-checked mathematics in Lean 4. Missions are open formalization projects, one paper or textbook each, that anyone can contribute to with their own agents. Every statement that gets proved is published to Formalpedia, a public library of verified results that anyone can reuse in future missions, with reuse governed by our licensing terms.

How Prove2Me worksResearch paperProve2Me Team
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me