Alphabeta Math
ExampleConstruction: AI-adaptedVerification: AI-adaptedPipeline-generatedjudge pass (gpt-5.6-terra)audited 2026-09-13
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.

Derivatives of piecewise smooth functions include jump deltas

Example

Assume Countable Choice. Let A⊆R be locally finite, and let f∈Lloc1(R) be C1 on each component of R∖A. Assume finite one-sided limits f(a±) at every a∈A, and assume that the classical derivative g=f′ off A, assigned arbitrary finite values on A, belongs to Lloc1. Then Duf=ug+∑a∈A(f(a+)−f(a−))δa. The sum is locally finite. These hypotheses hold, in particular, when f is C1 up to each side of every break point.

Facts & Assumptions

[F1]

Locally integrable functions have regular functionals, and under Countable Choice the embedding theorem makes them distributions; distribution derivatives are signed test transposes and Dirac masses evaluate tests (Regular distribution from a locally integrable function, Locally integrable functions embed in distributions, Distributional derivative, Dirac delta and its derivatives).

[F2]

Complex integration by parts on closed intervals holds under Countable Choice (Complex integration by parts on intervals and decaying lines).

[F3]

Dominated convergence passes limits through integrable complex functions (Dominated convergence).

[F4]

Compactwise finite-order bounds characterize distributions (Local finite order characterization of distributions).

[F5]

Assume The Axiom of Countable Choice (ACω) for the Lebesgue integration interfaces.

Proof

Given: A,f,g and the stated assumptions.

1.1givenF1F2F5

By F1, uf and ug are distributions. Fix a test φ and a closed interval [b,c] containing its support in its interior, with endpoints outside A. Local finiteness and compactness imply A∩[b,c] is finite: take a finite subcover of neighborhoods each meeting finitely many points. List these break points in increasing order. On each intervening open interval (s,t) apply F2 to f,φ on [s+ε,t−ε] for sufficiently small positive ε. This gives −∫s+εt−εfφ′=∫s+εt−εgφ+f(s+ε)φ(s+ε)−f(t−ε)φ(t−ε)..

2.1step 1.1givenF1F3

Let ε decrease to zero, for example through the reciprocal integers once the truncated interval is nonempty. F3 applies to the two integrals, with majorants ∣fφ′∣ and ∣gφ∣, integrable on [b,c] by the local integrability assumptions. The boundary values tend to f(s+)φ(s) and f(t−)φ(t) by the finite one-sided limits; at b,c the test vanishes. Sum over the finitely many intervals. At each break point a, the left interval contributes −f(a−)φ(a) and the right contributes f(a+)φ(a). The result is exactly the asserted formula when paired with φ.

3.1step 2.1givenF1F4

On any fixed compact test support K, the delta sum is finite and bounded in modulus by (∑a∈A∩K∣f(a+)−f(a−)∣)p0(φ). It therefore defines a distribution by F4, so the test equality proves the distribution identity. If there are no break points in K the sum is zero, and a zero jump contributes no delta. If f is C1 up to both sides, f and g are bounded on each of the finitely many compact pieces meeting a compact interval, hence locally integrable, verifying the stated sufficient case. Merely being C1 on the open pieces does not supply local integrability of g at the breaks. □.

Depends on

Used by

Nothing in the library uses this result yet.

Dependency tree · two levels

35 results within two dependency steps of this one, each drawn at its shortest distance from it. An arrow runs from a result to what uses it, so the chart reads left to right and ends at this result, which carries a heavier outline. Every node is a link to that result. Click elsewhere on the chart to enlarge it.

Sources