Alphabeta Math
TheoremStatement: Literature-sourcedProof: AI-adaptedPipeline-generated
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.

Log and Brier losses are strictly proper

Statement

For every K≥2 and p,q∈ΔK, the logarithmic and Brier losses in the definition of finite-label score loss satisfy Sˉ(p,p)≤Sˉ(p,q), with equality only when q=p. The expected logarithmic loss at the truthful forecast is finite, including when some coordinates of p vanish.

Statement references

Strictly proper scoring rules on a finite label set: defines the simplex, expected loss, logarithmic loss, Brier loss, and propriety conventions.

The natural logarithm as the inverse of the exponential function: fixes the natural logarithm used in the logarithmic score.

Facts & Assumptions

Given: An integer K≥2 and p,q∈ΔK.

[F1]

For pj=0, the expected-score term is zero even if the corresponding score is infinite; the logarithmic and Brier losses are the two functions in the definition. Strictly proper scoring rules on a finite label set

[F2]

For t>0, log⁡t=∫1tdu/u and log⁡1=0. The natural logarithm has derivative 1/x and equals the integral from 1 to x of 1/t

[F3]

The expected score is the finite sum Sˉ(p,q)=∑jpjS(q,j). Strictly proper scoring rules on a finite label set

Proof

1.1F2algebra

If t>1, then 1/u<1 for 1<u≤t, so [F2] gives log⁡t<t−1. If 0<t<1, then 1/u>1 for t≤u<1, so log⁡t=−∫t1du/u<−(1−t)=t−1. At t=1 equality holds. Therefore −log⁡t≥1−t for every t>0, with equality exactly at t=1.

1.2F1F3algebra

If pj>0 and qj=0 for some j, then [F1] gives Sˉlog⁡(p,q)=+∞, while Sˉlog⁡(p,p)=−∑j:pj>0pjlog⁡pj<∞ because this is a finite sum of finite terms. Thus propriety is strict in this case.

2.1F1F3step 1.1algebra

Otherwise qj>0 for every j with pj>0. The truthful risk is finite, and cancellation on the support of p gives Sˉlog⁡(p,q)−Sˉlog⁡(p,p)=∑j:pj>0pjlog⁡pjqj. Applying step 1.1 to t=qj/pj yields ∑j:pj>0pjlog⁡pjqj≥1−∑j:pj>0qj≥0. If the displayed difference is zero, both nonnegative gaps in this chain are zero. Equality in step 1.1 then forces qj/pj=1 on the support of p, and the second gap forces q to have no mass outside that support. Hence q=p. This proves strict propriety for the logarithmic loss.

3.1F1F3algebra∎

Expanding the Brier loss and using ∑jpj=1 gives SˉB(p,q)=∑iqi2−2∑ipiqi+1, whereas SˉB(p,p)=1−∑ipi2. Subtraction yields SˉB(p,q)−SˉB(p,p)=∑i(qi−pi)2≥0, with equality exactly when every qi=pi. Thus the Brier loss is strictly proper as well.

Source notes

The distinct encyclopedia treatment gives the scoring-rule conventions and categorical formulas (with the logarithmic reward negated). Strictness is proved locally from the declared logarithm supplier, rather than imported from a representation theorem. The displayed finite-sum arguments above separately verify the boundary cases pj=0 and qj=0 under this library's nonnegative-loss convention.

Depends on

Used by

Dependency tree · two levels

12 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