Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Theorem 3.45 — Rellich–Kondrachov: bounded sets in W^{1,p}_0(Ω) are precompact in L^q

Proved
HunterPDE.Compactness.rellich_kondrachov

by mikedeng1 · Sep 28, 2026 · Mathlib 0df444a (Lean v4.33.1)

compact-embeddingp2o-batch-books5p2o-gran-per-chapterp2o-plan-bookp2o-v1pderellich-kondrachovsobolev-spaces

Let Ω\OmegaΩ be a bounded open set in Rn\mathbb{R}^nRn, 1≤p<n1 \le p < n1≤p<n, and 1≤q<p∗=npn−p1 \le q < p^* = \dfrac{np}{n-p}1≤q<p∗=n−pnp​. If F\mathcal{F}F is a bounded set in W01,p(Ω)W^{1,p}_0(\Omega)W01,p​(Ω), that is, ∥f∥W1,p(Ω)≤M\|f\|_{W^{1,p}(\Omega)} \le M∥f∥W1,p(Ω)​≤M for all f∈Ff \in \mathcal{F}f∈F, then F\mathcal{F}F is precompact in Lq(Rn)L^q(\mathbb{R}^n)Lq(Rn):

{ f1Ω:f∈F }‾  is compact in Lq(Rn).\overline{\{\, f \mathbf{1}_\Omega : f \in \mathcal{F} \,\}} \ \text{ is compact in } L^q(\mathbb{R}^n).{f1Ω​:f∈F}​  is compact in Lq(Rn).

Equivalently, every bounded sequence in W01,p(Ω)W^{1,p}_0(\Omega)W01,p​(Ω) has a subsequence converging in LqL^qLq. This compact embedding underlies the Fredholm theory and spectral theory of elliptic operators on bounded domains. Both boundedness of Ω\OmegaΩ and q<p∗q < p^*q<p∗ are necessary.

Formalization Note. ppp is real with 1≤p<n1 \le p < n1≤p<n, and q∈[1,∞]q \in [1, \infty]q∈[1,∞] with q<p∗q < p^*q<p∗ (so qqq is finite). Elements of F\mathcal{F}F are functions Rn→R\mathbb{R}^n \to \mathbb{R}Rn→R in MemW0 1 p Ω with a common bound on sobolevNorm 1 p Ω. They are placed in Lq(Rn)L^q(\mathbb{R}^n)Lq(Rn) by extension by zero, Ω.indicator f. The conclusion states that every extension is in LqL^qLq, and that the set of their classes in Mathlib's Lp ℝ q volume has compact closure.

Preamble
import Mathlib
import Definitions.Def_HunterPDE_Compactness_Sobolev

open MeasureTheory
open scoped ENNReal NNReal
Formal statement
namespace HunterPDE.Compactness

/-- Theorem 3.45 of Hunter, *Notes on PDEs* (revised 6/18/2014), p. 73 (Rellich–Kondrachov): let
`Ω` be a bounded open set in `ℝⁿ`, `1 ≤ p < n` and `1 ≤ q < p* = np/(n − p)`. If `F` is a bounded
set in `W^{1,p}_0(Ω)`, then `F` is precompact in `L^q(ℝⁿ)`.
Elements of `F` are functions `ℝⁿ → ℝ` in `W^{1,p}_0(Ω)` (closure of `C_c^∞(Ω)` in the
`W^{1,p}(Ω)` norm) with a common bound on `‖f‖_{W^{1,p}(Ω)}`; they are placed in `L^q(ℝⁿ)` by
extension by zero outside `Ω` (`Ω.indicator f`). The conclusion says every extension lies in
`L^q(ℝⁿ)` and the set of their classes has compact closure in `L^q(ℝⁿ)`. -/
theorem rellich_kondrachov {n : ℕ} (Ω : Set (EuclideanSpace ℝ (Fin n))) (hΩ : IsOpen Ω)
    (hΩb : Bornology.IsBounded Ω) (p : ℝ) (hp : 1 ≤ p) (hpn : p < n)
    (q : ℝ≥0∞) [Fact (1 ≤ q)] (hq : q < ENNReal.ofReal (sobolevConjugate n p))
    (F : Set (EuclideanSpace ℝ (Fin n) → ℝ))
    (hF : ∀ f ∈ F, MemW0 1 (ENNReal.ofReal p) Ω f)
    (hFb : ∃ M : ℝ≥0, ∀ f ∈ F, sobolevNorm 1 (ENNReal.ofReal p) Ω f ≤ (M : ℝ≥0∞)) :
    (∀ f ∈ F, MemLp (Ω.indicator f) q volume) ∧
      IsCompact (closure {u : Lp ℝ q (volume : Measure (EuclideanSpace ℝ (Fin n))) |
        ∃ f ∈ F, (u : EuclideanSpace ℝ (Fin n) → ℝ) =ᵐ[volume] Ω.indicator f}) := by sorry

end HunterPDE.Compactness
Source
Hunter, Notes on Partial Differential Equations (revised 6/18/2014), p. 73, Theorem 3.45
Read-back

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

Let nnn be a natural number and let Rn\mathbb{R}^nRn be nnn-dimensional Euclidean space, with coordinates indexed by {0,…,n−1}\{0,\dots,n-1\}{0,…,n−1}. It carries its Euclidean norm and Lebesgue measure λ\lambdaλ. The statement takes the following data and hypotheses:

  • a set Ω⊆Rn\Omega \subseteq \mathbb{R}^nΩ⊆Rn that is open and bounded (bounded in the metric sense);
  • a real number ppp with 1≤p1 \le p1≤p and p<np < np<n, where nnn is regarded as a real number;
  • an extended nonnegative real exponent q∈[0,∞]q \in [0,\infty]q∈[0,∞] with 1≤q1 \le q1≤q (a standing typeclass assumption) and
q<ofReal⁡(p∗),p∗:=sobolevConjugate(n,p).q < \operatorname{ofReal}\big(p^*\big), \qquad p^* := \mathrm{sobolevConjugate}(n,p).q<ofReal(p∗),p∗:=sobolevConjugate(n,p).

Here sobolevConjugate\mathrm{sobolevConjugate}sobolevConjugate is a real-valued function defined in the imported module Definitions.Def_HunterPDE_Compactness_Sobolev. Its formula is not shown in this code and is not unfolded here. ofReal⁡\operatorname{ofReal}ofReal converts a real number to an extended nonnegative real: it sends negative numbers to 000 and is always finite.

  • a set FFF of functions f:Rn→Rf : \mathbb{R}^n \to \mathbb{R}f:Rn→R, each defined on all of Rn\mathbb{R}^nRn, such that:
    • every f∈Ff \in Ff∈F satisfies MemW0(1,ofReal⁡(p),Ω,f)\mathrm{MemW0}\big(1, \operatorname{ofReal}(p), \Omega, f\big)MemW0(1,ofReal(p),Ω,f), a predicate from the same imported module. By its name it is membership in a space "W01,p(Ω)W^{1,p}_0(\Omega)W01,p​(Ω)", but its actual definition is not visible here and is not unfolded.
    • there is a single constant M≥0M \ge 0M≥0 such that for every f∈Ff \in Ff∈F
sobolevNorm(1,ofReal⁡(p),Ω,f)≤M.\mathrm{sobolevNorm}\big(1, \operatorname{ofReal}(p), \Omega, f\big) \le M.sobolevNorm(1,ofReal(p),Ω,f)≤M.
Here $\mathrm{sobolevNorm}$ is another imported definition, whose value lies in $[0,\infty]$. Because of this bound, it is finite on $F$.

Under these hypotheses, the statement concludes both of the following:

  1. For every f∈Ff \in Ff∈F, the zero extension 1Ωf\mathbf{1}_\Omega f1Ω​f (equal to fff on Ω\OmegaΩ and 000 off Ω\OmegaΩ) is in Lq(Rn,λ)L^q(\mathbb{R}^n, \lambda)Lq(Rn,λ). That is, it is almost-everywhere strongly measurable and has finite LqL^qLq seminorm; when q=∞q = \inftyq=∞, this means it is essentially bounded.
  2. Let SSS be the set of elements uuu of the Banach space Lq(Rn,λ;R)L^q(\mathbb{R}^n, \lambda; \mathbb{R})Lq(Rn,λ;R) (functions identified when equal λ\lambdaλ-almost everywhere) for which some f∈Ff \in Ff∈F satisfies u=1Ωfu = \mathbf{1}_\Omega fu=1Ω​f λ\lambdaλ-almost everywhere. Then the closure of SSS in the LqL^qLq-norm topology is compact.

Only the values of each fff on Ω\OmegaΩ enter the conclusion. How the hypotheses constrain fff depends entirely on the imported definitions MemW0\mathrm{MemW0}MemW0 and sobolevNorm\mathrm{sobolevNorm}sobolevNorm. The exponent qqq is tied to ppp only through q<ofReal⁡(p∗)q < \operatorname{ofReal}(p^*)q<ofReal(p∗). Nothing requires q≤pq \le pq≤p or q<pq < pq<p. The Sobolev exponent is fixed at order 111, and ppp enters those definitions only as ofReal⁡(p)\operatorname{ofReal}(p)ofReal(p).

Degenerate cases:

  • n=0n = 0n=0 or n=1n = 1n=1: the hypotheses 1≤p1 \le p1≤p and p<np < np<n cannot both hold, so the statement is vacuously true. It says something only when n≥2n \ge 2n≥2.
  • q=∞q = \inftyq=∞: excluded, because ofReal⁡(p∗)\operatorname{ofReal}(p^*)ofReal(p∗) is finite and qqq must be strictly smaller.
  • ofReal⁡(p∗)≤1\operatorname{ofReal}(p^*) \le 1ofReal(p∗)≤1: if p∗≤1p^* \le 1p∗≤1 (for example, if the imported formula produced a non-positive value, which ofReal⁡\operatorname{ofReal}ofReal turns into 000), no qqq satisfies both 1≤q1 \le q1≤q and q<ofReal⁡(p∗)q < \operatorname{ofReal}(p^*)q<ofReal(p∗), and the statement is vacuous. Whether this can happen for 1≤p<n1 \le p < n1≤p<n depends on the unseen definition of sobolevConjugate\mathrm{sobolevConjugate}sobolevConjugate.
  • FFF empty: both conclusions hold trivially. The first is vacuous, and SSS is empty, so its closure is empty and therefore compact.
  • Ω\OmegaΩ empty: every 1Ωf\mathbf{1}_\Omega f1Ω​f is the zero function. So whenever FFF is nonempty, S={0}S = \{0\}S={0} and the conclusion is automatic. Whether any fff satisfies MemW0\mathrm{MemW0}MemW0 on the empty set depends on the imported definition.
  • Hidden conventions: any default or conventional values inside MemW0\mathrm{MemW0}MemW0, sobolevNorm\mathrm{sobolevNorm}sobolevNorm or sobolevConjugate\mathrm{sobolevConjugate}sobolevConjugate (for example, division by zero in p∗p^*p∗, or how weak derivatives are handled for non-integrable fff) cannot be seen from this code. The statement inherits whatever those definitions assign.
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