Alphabeta Math
LemmaStatement: AI-adaptedProof: AI-adaptedPipeline-generatedprecheck passjudge pass (gpt-6.1-sol)
How statement and proof provenance work

The first chip identifies the source of the statement or construction; the second identifies the source of its local proof or verification.

  • Literature-sourced: the exact statement appears in a cited source; only wording and notation differ.
  • AI-adapted: a semantically identical restatement of literature-sourced material, modulo indexing, notation, and boundary cases adopted by the library.
  • AI-generated: a genuinely novel statement formulated by AI, with no source for the claim itself.

These labels describe origin, not correctness: citations and verification chips remain separate evidence.

Unique solution of a symmetric coercive variational problem on a Hilbert space

Statement

Assume the Axiom of Countable Choice ACω (The Axiom of Countable Choice (ACω)); in particular the hypothesis is met under the Axiom of Choice (The Axiom of Choice). Let H be a real Hilbert space (Hilbert space, Complete metric space: every Cauchy sequence converges in the space), let a:H×H→R be a symmetric bounded bilinear form (A bounded bilinear map between normed spaces) with a(v,v)≥α∥v∥2 for every v∈H and some α>0, and let ℓ:H→R be a bounded linear functional with operator norm ∥ℓ∥ (The operator norm as the least bound and as the unit-sphere or unit-ball supremum). Then there is exactly one u∈H with

a(u,v)=ℓ(v)for every v∈H,

and it satisfies ∥u∥≤∥ℓ∥/α. The argument minimises the explicit functional J(v)=12a(v,v)−ℓ(v); no Lax-Milgram theorem, Zorn-type argument or unproved minimiser existence is invoked, and the only choice use is the selection of one point from each of the countably many nonempty sublevel sets {v∈H:J(v)<m+1/(n+1)}.

Facts & Assumptions

Given: A real Hilbert space H with norm ∥v∥=⟨v,v⟩, a symmetric bounded bilinear form a:H×H→R, a constant α>0 with a(v,v)≥α∥v∥2 for every v∈H, a bounded linear functional ℓ:H→R, and the functional J(v)=12a(v,v)−ℓ(v).

[F1]

Hilbert norm and completeness. ∥v∥=⟨v,v⟩≥0 with ∥v∥=0 exactly for v=0, and the induced metric (v,w)↦∥v−w∥ is complete: every Cauchy sequence in H converges in norm to a point of H (Hilbert space, Complete metric space: every Cauchy sequence converges in the space).

[F2]

Parallelogram law. ∥x+y∥2+∥x−y∥2=2∥x∥2+2∥y∥2 for all x,y∈H (The parallelogram law).

[F3]

Bounded symmetric bilinear form. a is linear in each argument, a(v,w)=a(w,v), and there is a real C≥0 with ∣a(v,w)∣≤C∥v∥ ∥w∥ for all v,w∈H (A bounded bilinear map between normed spaces).

[F4]

Operator norm. ∥ℓ∥=sup⁡{∣ℓ(v)∣:∥v∥≤1} is finite and ∣ℓ(v)∣≤∥ℓ∥ ∥v∥ for every v∈H (The operator norm as the least bound and as the unit-sphere or unit-ball supremum).

[F5]

Infima. Every nonempty subset of R bounded below has an infimum, and m=inf⁡S is characterised by: m≤s for all s∈S, and for every ε>0 there is s∈S with s<m+ε (Greatest lower bound (infimum), Every nonempty set bounded below has an infimum, Epsilon characterisation of the infimum).

[F6]

Countable choice. ACω provides, for every family (Xn)n∈N of nonempty sets, a function f with domain N and f(n)∈Xn for all n (The Axiom of Countable Choice (ACω)).

Proof

technique · direct
1.1givenF3F4algebra

For every v∈H one has 12a(v,v)≥α2∥v∥2 by the coercivity hypothesis and −ℓ(v)≥−∣ℓ(v)∣≥−∥ℓ∥ ∥v∥ by [F4], hence J(v)≥α2∥v∥2−∥ℓ∥ ∥v∥.

1.2givenF2algebra

Expanding with bilinearity and symmetry of a (the same expansion as the parallelogram identity [F2], applied to the quadratic form v↦a(v,v)) gives a(v+w2,v+w2)=12a(v,v)+12a(w,w)−14a(v−w,v−w), hence J(v+w2)=J(v)+J(w)2−18a(v−w,v−w) for all v,w∈H.

2.1step 1.1F5algebra

Completing the square gives α2∥v∥2−∥ℓ∥ ∥v∥=α2(∥v∥−∥ℓ∥α)2−∥ℓ∥22α≥−∥ℓ∥22α, so the image S={J(v):v∈H} is nonempty and bounded below in R, and its infimum m=inf⁡S∈R exists by [F5].

3.1step 2.1F5F6choose

For each n∈N the set Xn:={v∈H:J(v)<m+1n+1} is nonempty: otherwise m+1n+1 would be a lower bound of S larger than m, contradicting [F5] (equivalently, [F5] with ε=1n+1 produces v with J(v)<m+1n+1). Applying ACω [F6] to the family (Xn)n∈N gives a sequence (vn) in H with J(vn)<m+1n+1 for all n; since m is a lower bound of S, also J(vn)≥m, so J(vn)→m.

4.1step 3.1step 1.2givenalgebra

Since m is a lower bound of S one has J(vn+vk2)≥m, so step 1.2 and the coercivity bound a(vn−vk,vn−vk)≥α∥vn−vk∥2 give α8∥vn−vk∥2≤J(vn)+J(vk)2−m≤12(1n+1+1k+1); the right-hand side tends to 0 as n,k→∞, so (vn) is a Cauchy sequence in H.

5.1step 4.1F1

By completeness [F1] there is u∈H with ∥vn−u∥→0.

6.1step 5.1step 3.1F3F4algebra

The functional J is continuous in the norm: for v,w∈H one has ∣J(v)−J(w)∣≤12∣a(v,v)−a(w,w)∣+∣ℓ(v−w)∣=12∣a(v+w,v−w)∣+∣ℓ(v−w)∣≤(C2∥v+w∥+∥ℓ∥)∥v−w∥ by [F3] and [F4], which tends to 0 as v→w; hence J(u)=lim⁡nJ(vn)=m.

7.1step 6.1givenalgebra

Fix v∈H and consider g(s)=J(u+sv); expanding with bilinearity and symmetry gives g(s)=J(u)+s(a(u,v)−ℓ(v))+s22a(v,v) for all s∈R. Since J(u)=m is the minimum of J, g(s)≥g(0) for all s: for s>0 division by s gives a(u,v)−ℓ(v)+s2a(v,v)≥0, and letting s→0+ gives a(u,v)−ℓ(v)≥0, while for s<0 the same division gives a(u,v)−ℓ(v)+s2a(v,v)≤0, and letting s→0− gives a(u,v)−ℓ(v)≤0. Therefore a(u,v)=ℓ(v), and since v∈H was arbitrary, u solves the variational equation.

8.1step 7.1givenF1algebra

Uniqueness: if u1,u2∈H both satisfy the equation, then testing the difference at v=u1−u2 gives a(u1−u2,u1−u2)=0, and coercivity gives α∥u1−u2∥2≤a(u1−u2,u1−u2)=0, so ∥u1−u2∥=0 and u1=u2 by [F1].

9.1step 7.1step 8.1givenF1F4algebra∎

Estimate: testing the equation at v=u gives a(u,u)=ℓ(u), and coercivity together with [F4] gives α∥u∥2≤a(u,u)=ℓ(u)≤∥ℓ∥ ∥u∥; if ∥u∥=0 then 0≤∥ℓ∥/α holds, and otherwise division by α∥u∥>0 gives ∥u∥≤∥ℓ∥/α. This proves existence, uniqueness and the estimate, and identifies the sole use of ACω as the selection in step 3.1.

Depends on

Dependency tree · 2 levels

Sources