Alphabeta Math
TheoremStatement: Literature-sourcedProof: AI-adaptedPipeline-generatedverified 2026-10-15 (gpt-6-luna, gpt-6.1-sol, and delegated Codex review/repair agents; see completed run evidence)
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.

Finite-dimensional Fenchel-Moreau biconjugation

Statement

Let d≥1 and let f:Rd→R‾ be closed, proper, and convex in the sense of Proper, convex, and lower-semicontinuous extended-real functions on Euclidean space. With the Fenchel conjugate and biconjugate of Fenchel conjugate, biconjugate, and subdifferential, one has

f∗∗(x)=f(x)(x∈Rd).

Facts & Assumptions

Given: The closed proper convex function and conjugates in the Statement.

[F1]

Proper, convex, and lower-semicontinuous extended-real functions on Euclidean space: proper means dom⁡f≠∅ and f>−∞; convexity is the extended chord inequality; lower semicontinuity means closed weak sublevel sets.

[F2]

Extended-real convexity and lower semicontinuity are characterized by the epigraph: under [F1], epi⁡f is nonempty, closed, and convex.

[F3]

Strictly separate a point from a nonempty closed convex Euclidean set: a point outside a nonempty closed convex subset of Rd+1 has a strict linear separator.

[F4]

Fenchel conjugate, biconjugate, and subdifferential: f∗(p) and f∗∗(x) are the stated extended suprema over the effective domain and finite conjugate domain.

[L1]

Every complete ordered field is Archimedean: the canonical natural numbers are cofinal in R; for every q∈R there is a positive integer k with q<k.

Proof

technique · direct
1.1F4givenalgebra

For any x with f(x)<+∞ and any p with f∗(p)<+∞, the defining supremum gives f∗(p)≥⟨p,x⟩−f(x), hence ⟨p,x⟩−f∗(p)≤f(x); taking the supremum over p yields f∗∗(x)≤f(x). If f(x)=+∞, the same upper bound is automatic.

1.2F2F3givenalgebra

Fix x∈dom⁡f and a real t<f(x). Apply [F3] to (x,t) and the nonempty closed convex set epi⁡f, and negate the resulting separator to write ⟨a,y⟩+bs≥c>⟨a,x⟩+bt for all (y,s)∈epi⁡f. Upward closure forces b≥0. If b=0, the inequality at (x,f(x))∈epi⁡f would give ⟨a,x⟩≥c>⟨a,x⟩, impossible. Thus b>0.

1.3F2F3F4algebra

Now fix x∉dom⁡f and t∈R. Apply [F3] to (x,t) and negate the separator to obtain ⟨a,y⟩+bs≥c>⟨a,x⟩+bt for all (y,s)∈epi⁡f. If b>0, then f(y)≥−⟨a/b,y⟩+c/b for every y∈dom⁡f, so with p=−a/b one has f∗(p)≤−c/b<+∞ and f∗∗(x)≥⟨p,x⟩−f∗(p)≥(c−⟨a,x⟩)/b>t.

2.1F4step 1.2algebra

Substituting (y,f(y)) for every y∈dom⁡f gives the affine minorant ℓ(y)=−⟨a/b,y⟩+c/b≤f(y). With p=−a/b, [F4] gives f∗(p)≤−c/b, hence f∗∗(x)≥⟨p,x⟩−f∗(p)≥ℓ(x)>t. Since this holds for every t<f(x), f∗∗(x)≥f(x) on dom⁡f.

2.2F1F2F3F4L1step 1.3choosealgebra

In the b=0 case of step 1.3, the separator gives ⟨a,y⟩≥c>⟨a,x⟩ for every y∈dom⁡f. Choose y0∈dom⁡f and a finite t0<f(y0). Apply [F3] to (y0,t0) and epi⁡f, reversing the separator orientation to obtain ⟨a0,y⟩+b0s≥c0>⟨a0,y0⟩+b0t0 for every (y,s)∈epi⁡f. Upward closure gives b0≥0, and b0=0 contradicts the inequality at (y0,f(y0)), so b0>0. Set p0=−a0/b0 and r0=c0/b0; substituting (y,f(y)) gives the affine minorant ℓ0(y)=⟨p0,y⟩+r0≤f(y) on dom⁡f, so [F4] gives f∗(p0)≤−r0. For each positive integer k, define ℓk(y)=ℓ0(y)+k(c−⟨a,y⟩)=⟨p0−ka,y⟩+(r0+kc). Since c−⟨a,y⟩≤0 on dom⁡f, this is still a minorant; [F4] gives f∗(p0−ka)≤−(r0+kc)<+∞ and hence f∗∗(x)≥ℓk(x)=ℓ0(x)+kδ, where δ:=c−⟨a,x⟩>0 by step 1.3. For any prescribed B∈R, apply [L1] to q=(B−ℓ0(x))/δ to choose a positive integer k with q<k; multiplying by δ>0 gives ℓ0(x)+kδ>B. Hence f∗∗(x) exceeds every real bound, so f∗∗(x)=+∞=f(x).

3.1step 1.1step 2.1step 2.2∎

The upper bound from step 1.1 and the lower bounds from steps 2.1 and 2.2 cover respectively finite and infinite values of f(x), proving equality everywhere.

Depends on

Dependency tree · 2 levels

Sources