Alphabeta Math
CounterexampleConstruction: Literature-sourcedVerification: AI-adaptedPipeline-generatedprecheck pass
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.

The convex-flux Riemann formula fails for a nonconvex flux

Statement

Assume Countable Choice (The Axiom of Countable Choice (ACω)) for the analytic prerequisites used below.

Let f(u)=u3 and take the Riemann data uL=1 for x<0, uR=−1 for x>0. The single jump u(t,x)={1,x<t,−1,x>t has Rankine--Hugoniot speed s=1 and is a distributional weak solution with these data: integration by parts on the two sides leaves only the interface coefficient s[u]−[f]=1(−2)−(−2)=0. Its strong local L1 trace is the stated datum because the discrepancy is supported on 0<x<t and has amplitude 2. It is not a Kruzhkov entropy solution. For k=−12, the Kruzhkov pair has ηk(1)=32, ηk(−1)=12, qk(1)=98, and qk(−1)=78, hence [qk]−s[ηk]=−14−(−1)=34>0, violating the required nonpositive entropy production. Equivalently, the chord from (−1,−1) to (1,1) is z↦z, while z3−z>0 on (−1,0) and z3−z<0 on (0,1), so the graph fails the required one-sided condition in The convex entropy condition for a single shock is the chord condition. The strictly convex Riemann solver theorem does not apply: f′′(u)=6u changes sign and f′ is not monotone on [−1,1]. Thus its formula does not extend to this nonconvex flux; the concave-hull construction gives the corresponding composite entropy wave (The self-similar Riemann problem, Kruzhkov entropy solutions).

Facts & Assumptions

Given: Countable Choice, the flux f(u)=u3, the states uL=1>uR=−1, the Riemann datum u0(x)=1 for x<0, u0(x)=−1 for x>0, the single-jump profile u above, and a test function φ∈Cc∞(ΠT).

[F1]

For any constant-state jump A→B across x=st, write u=A+(B−A)1x>st. Fubini and one-dimensional FTC, as in [F3], give ∂t1x>st=−sδx=st and ∂x1x>st=δx=st, where δx=st pairs with φ as ∫φ(t,st)dt. Thus the weak residual is ([f]−s[u])δx=st. Its vanishing is the Rankine--Hugoniot relation (The Rankine--Hugoniot jump condition in space--time normal form, Distributional weak solutions of the Cauchy problem). The same computation applies to smooth regions separated by rays, with the regionwise classical residual and the trace-jump terms added.

[F2]

Entropy production at a jump: the distribution ∂tη(u)+∂xq(u) is the measure ([q]−s[η])δΓ with δΓ=δx=st as defined in [F1], and the entropy inequality holds at the jump if and only if [q]−s[η]≤0; for the Kruzhkov pairs ηk(s)=∣s−k∣, qk(s)=sgn⁡(s−k)(f(s)−f(k)) this condition is necessary for u to be a Kruzhkov entropy solution (The convex entropy condition for a single shock is the chord condition, Kruzhkov entropy solutions).

[F4]

Bounded Kruzhkov entropy solutions with identical initial data are unique (Uniqueness, comparison and order preservation of entropy solutions).

Proof

technique · direct
1.1F1F3

Speed, weak solvability and the initial trace. With uL=1, uR=−1, f(u)=u3: [u]=uR−uL=−2 and [f]=f(uR)−f(uL)=−1−1=−2, so the Rankine--Hugoniot speed is s=[f]/[u]=1. By [F1] the weak residual of the jump profile is ([f]−s[u])δx=st, which vanishes because s[u]=1⋅(−2)=−2=[f]; hence u is a distributional weak solution. For t>0 the set where u(t,x)≠u0(x) is contained in the interval (0,t) (the region swept by the moving discontinuity compared with the initial step at 0), of length t and amplitude at most 2, so ∫K∣u(t,x)−u0(x)∣ dx≤2t→0 for every compact K: the strong local L1 trace is u0.

1.2F2

Failure of the Kruzhkov inequality at k=−12. For k=−12, ηk(1)=∣1+12∣=32, ηk(−1)=∣−1+12∣=12, ηk(1)−ηk(−1)=1; and qk(s)=sgn⁡(s+12)(s3+18) gives qk(1)=1⋅98=98, qk(−1)=(−1)⋅(−78)=78, so [qk]=qk(−1)−qk(1)=−14 and [ηk]=ηk(−1)−ηk(1)=−1. By [F2] the entropy production measure is ([qk]−s[ηk])δΓ=(−14+1)δΓ=34 δΓ>0; testing against a nonnegative test function concentrated near the interface produces a strictly positive entropy production, so the Kruzhkov entropy inequality fails and u is not a Kruzhkov entropy solution.

1.3F1F3

The composite weak solution. Put c=3/4 and define v(t,x)=1 for x<ct, v(t,x)=−x/(3t) for ct<x<3t, and v(t,x)=−1 for x≥3t. At the shock the traces are 1 and −1/2, with [v]=−3/2 and [f]=−9/8=c[v]. On the fan, ψ(ξ)=−ξ/3 satisfies f′(ψ(ξ))=ξ, so vt+f′(v)vx=t−1ψ′(ξ)(−ξ+f′(ψ(ξ)))=0. At x=3t the traces match. Regionwise integration using [F1, F3] therefore gives zero weak residual. The discrepancy with the initial datum is confined to (0,3t) and has amplitude at most 2, so its local L1 norm is at most 6t; this supplies the strong trace and the Cauchy boundary term.

2.1F3step 1.2

Chord condition and nonconvexity. With u−=1, u+=−1 and s=1, the chord residual is F(z)=f(z)−f(1)−s(z−1)=z3−1−(z−1)=z3−z, and the criterion of The convex entropy condition for a single shock is the chord condition requires F(z)(u+−u−)≥0, that is, F(z)≤0 on [−1,1]. But F(z)=z(z−1)(z+1)>0 for z∈(−1,0), so the chord condition fails, independently confirming the entropy failure of step 1.2. Moreover f′′(u)=6u changes sign on [−1,1], so f′ is not increasing and the hypotheses of the strictly convex Riemann solver The Riemann solver for a strictly convex flux are not satisfied.

2.2F2F3F4step 1.3

Entropy admissibility of the composite. At its descending shock the chord residual is z3−1−c(z−1)=(z−1)(z+1/2)2≤0 for −1/2≤z≤1. Thus the chord criterion of [F2] gives [q]−c[η]≤0 for every convex C2 pair. On the smooth fan the entropy residual is η′(v)(vt+f′(v)vx)=0, and it vanishes on both constant regions; matching traces at x=3t give no interface measure. Hence every smooth convex entropy inequality holds. For each k, take ηδ(z)=(z−k)2+δ2−δ and qδ(z)=∫kzηδ′(r)f′(r)dr. On the bounded range, ηδ→∣z−k∣ uniformly. The fluxes converge uniformly to sgn⁡(z−k)(f(z)−f(k)): outside an arbitrarily small interval about k, the derivatives converge uniformly to the sign, and inside it the integral error is bounded by twice its length times a bound for ∣f′∣. Passing against compact tests gives every Kruzhkov inequality. With step 1.3 and [F4], v is the unique entropy solution.

3.1step 1.1step 1.2step 1.3step 2.2∎

The concave hull and conclusion. On [−1,−1/2] the hull follows z3; on [−1/2,1] it is ℓ(z)=1+c(z−1). The residual in step 2.2 shows ℓ≥z3 on the latter interval. The arc is concave, and its derivative decreases to 3/4 at −1/2, matching the slope of ℓ, so the joined function is a concave majorant. Any concave majorant lies above the cubic on the arc and above the line joining the values at −1/2 and 1 on the chord interval; it therefore lies above this function. This proves it is the least concave majorant. Its arc and chord yield exactly the fan and shock verified in steps 1.3--2.2. The single shock of step 1.1 is weak but non-entropic, whereas this composite is entropic, proving the claimed failure of the convex-flux formula and its replacement here.

Depends on

Used by

Dependency tree · two levels

58 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