Alphabeta Math
Pipeline-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.

✓ 3 results · all verified · 3 also independently AI-judged
Every result on this page is machine-checked by a proof checker and read in full and owner-audited; the judge is an additional, independent cross-model AI review of the proofs; all 3 also cleared it.

Method of Types and Finite Large Deviations — Examples

1 · Prerequisites

2 · Summary

This companion computes binary type-class sizes with their polynomial factors visible, then works a finite Bernoulli upper-tail event and its asymptotic relative-entropy rate. The Bernoulli calculation includes the threshold ceiling, the all-ones endpoint, and a one-sided Hoeffding comparison under its stated Axiom of Choice assumption. The final counterexample shows why an entropy exponential alone is not an exact finite type-class count.

3 · Logical flowchart

4 · Definitions, theorems and proofs

None yet.

5 · Examples, counterexamples and false statements

ExampleConstruction: AI-adaptedVerification: AI-adaptedprecheck passjudge pass (gpt-6-luna)Open item page →

Binary type classes at length ten

Example

For A={0,1} and n=10, the types with q(1)=1/5 and q(1)=1/2 have class sizes 45 and 252. Their entropy exponentials are respectively 510/48 and 210=1024, and the polynomially corrected bounds use the factor (n+1)−1=1/11.

Facts & Assumptions

Given: The binary alphabet A={0,1}, block length n=10, and the two stated attainable types.

[F1]

Empirical types, n-types, and type classes: a type class consists of the words with prescribed symbol counts.

[F2]

Finite distributions, entropy, divergence, and stochastic matrices: H(q)=−∑a:q(a)>0q(a)ln⁡q(a) in nats.

[F3]

Polynomially sharp bounds for type-class cardinality: for an alphabet of size k, (n+1)−(k−1)enH(q)≤∣Tq∣≤enH(q).

Verification

Given: The data above.

1.1F1F2givenalgebra

For q(1)=1/5, the counts are (8,2), so ∣Tq∣=10!/(8!2!)=45 and e10H(q)=(1/5)−2(4/5)−8=510/48.

1.2F1F2F3givenalgebra

For q(1)=1/2, the counts are (5,5), so ∣Tq∣=10!/(5!5!)=252 and e10H(q)=210=1024. Since 1024≤252⋅11=2,772 and 252≤1,024, we have 1024/11≤252≤1024, retaining the polynomial factor in [F3].

2.1F3step 1.1algebra

Since 510=9,765,625, 48=65,536, 45⋅11⋅48=32,440,320, and 45⋅48=2,949,120, cross-multiplication in step 1.1 gives e10H(q)/11≤45≤e10H(q), as [F3] states.

3.1F1F2F3algebra∎

At endpoint types q=(1,0) and q=(0,1), each type class is a singleton and H(q)=0, so the lower cardinality bound is 1/11, while the class size and upper bound equal 1; thus [F3] also covers zero-coordinate boundary cases.

ExampleConstruction: AI-adaptedVerification: AI-adaptedprecheck passjudge pass (gpt-6-luna)Open item page →

A finite Bernoulli large-deviation tail

Example

For p∈[0,1), let bp be the binary probability vector bp(1)=p, bp(0)=1−p. On {0,1}n use the product law bp⊗n from Finite product mass on words, and let Xi(xn)=xi be its coordinate maps; by The finite product law has iid coordinate maps these are independent Bernoulli(p) variables. Let a∈(p,1], define p^n=n−1∑i=1nXi, qn=⌈na⌉/n, and d(q∥p):=qln⁡qp+(1−q)ln⁡1−q1−p, with zero-mass summands omitted, d(0∥0):=0, and d(q∥0)=+∞ for q>0. Then (n+1)−1e−nd(qn∥p)≤bp⊗n(p^n≥a)≤(n+1)e−nd(a∥p). Consequently lim⁡n→∞1nln⁡bp⊗n(p^n≥a)=−d(a∥p), including the value −∞ when p=0. For p=1/2, a=3/4, and n=4, the exact tail is 5/16; under AC, Hoeffding also gives the bound e−1/2 for this same event.

Facts & Assumptions

Given: A finite Bernoulli product model with parameter p∈[0,1), threshold a∈(p,1], and integer n≥1. The Hoeffding comparison assumes AC.

[A1]

The Axiom of Choice: every family of nonempty sets has a choice function. The example invokes this axiom only to apply the Hoeffding supplier in step 4.1.

[F1]

Empirical types, n-types, and type classes: the empirical type of a binary word is its fraction of ones, and its type class consists of words with that fraction.

[F2]

Finite distributions, entropy, divergence, and stochastic matrices: finite relative entropy uses natural logarithms, 0ln⁡0=0, and equals +∞ if positive mass is placed where the reference law is zero.

[F3]

Finite probability bounds for an empirical type: for each binary n-type q, (n+1)−1e−nd(q∥p)≤P(p^n=q)≤e−nd(q∥p).

[F4]

Finite relative entropy is continuous on a simplex with positive reference law: when 0<p<1, q↦d(q∥p) is continuous on [0,1].

[F5]

Finite-alphabet Sanov bounds on the support simplex: on the support simplex, the Sanov liminf uses the infimum over relative interior and the limsup uses the infimum over relative closure; equality of these infima gives the limit.

[F6]

Hoeffding inequality for independent bounded sums: under AC, for independent integrable Xi∈[ai,bi] and t>0, the one-sided bound is P(∑i(Xi−EXi)≥t)≤exp⁡(−2t2/∑i(bi−ai)2) when the denominator is positive.

[F8]

The mean value theorem, as the case g(x)=x of Cauchy's: for f continuous on [a,b] with a<b and differentiable on (a,b) there is c∈(a,b) with f(b)−f(a)=f′(c)(b−a): if a real function is continuous on a closed interval and differentiable on its interior, its increment equals its derivative at some interior point times the interval length.

[F9]

The type classes partition the word space and have polynomial count: on a binary alphabet there are at most n+1 attainable n-types.

[F10]

The finite product law has iid coordinate maps: the sample law on {0,1}n is a probability law and its coordinate variables are independent with Bernoulli(p) marginal law.

Verification

Given: The stated Bernoulli model and thresholds.

1.1F1F2F3F10givenalgebra

If p=0, every coordinate equals 0 almost surely by [F10], so the tail is zero and d(qn∥p)=d(a∥p)=+∞; both finite bounds have right-hand sides zero. If 0<p<1, the tail is the union of binary type classes with q≥a, and the attainable type qn=⌈na⌉/n belongs to the event, so [F3] gives the lower bound.

1.2F4F7F8givenalgebra

Suppose 0<p<1 and 0<a<1. For every x∈(p,1), [F7] gives ddxd(x∥p)=ln⁡x(1−p)p(1−x)>0. For any a≤u<v<1, the finite logarithmic formula is continuous on [u,v] and differentiable on (u,v), so [F8] gives d(v∥p)>d(u∥p). Thus d(⋅∥p) is increasing on [a,1), and continuity [F4] extends d(q∥p)≥d(a∥p) to q=1.

2.1F3F9step 1.2algebra

For 0<p<1 and 0<a<1, there are at most n+1 binary types by [F9]. Each tail type has probability at most e−nd(q∥p) by [F3], and step 1.2 gives d(q∥p)≥d(a∥p), so summing the types proves the finite upper bound.

2.2F4F5step 1.1step 1.2algebra

If 0<p<a<1, then for Γ=[a,1] the relative interior in [0,1] is (a,1] and the closure is [a,1]. By step 1.2 and continuity [F4], both divergence infima equal d(a∥p), so [F5] gives the stated limit. If a=1 and 0<p<1, the event is exactly all ones and has probability pn=e−nd(1∥p), giving the limit directly. The case p=0 is in step 1.1.

3.1step 1.1step 2.1step 2.2algebra

For 0<p<1 and 0<a<1, summing the type upper bounds from step 2.1 yields the displayed finite upper bound; when a=1 and 0<p<1, step 2.2 gives the exact value pn≤(n+1)pn, while p=0 is covered in step 1.1.

4.1A1F1F6F10givenalgebra∎

For p=1/2, a=3/4, and n=4, the event contains four words with three ones and one with four, so its probability is 5/16. By [F10], the four coordinates are independent and take values in [0,1], hence are integrable. Thus [F6] applies with ∑i(Xi−EXi)≥1 on the event and ∑i(bi−ai)2=4 for bounds ai=0,bi=1. Its exponent is −2(1)2/4=−1/2, giving b1/2⊗4(p^4≥3/4)≤e−1/2. This comparison uses [A1]; the finite type and Sanov calculations in steps 1.1–3.1 are choice-free.

CounterexampleConstruction: AI-adaptedVerification: AI-adaptedprecheck passjudge pass (gpt-6-luna)Open item page →

A balanced binary type class is not exactly exponential in entropy

Statement refuted

For every finite alphabet, every n≥1, and every n-type q, the type-class size is exactly ∣Tq∣=enH(q).

Facts & Assumptions

Given: The binary alphabet A={0,1} and block length n=2.

[F1]

Empirical types, n-types, and type classes: Tq consists of the words whose empirical vector is q.

[F2]

Finite distributions, entropy, divergence, and stochastic matrices: H(q)=−∑a:q(a)>0q(a)ln⁡q(a).

[F3]

Polynomially sharp bounds for type-class cardinality: (n+1)−(k−1)enH(q)≤∣Tq∣≤enH(q) for an alphabet of size k.

Counterexample

Given: n=2 and the binary type q(0)=q(1)=1/2.

1.1F1givenconstruct

The words with type q are 01 and 10, so ∣Tq∣=2.

2.1F2step 1.1algebra

By [F2], H(q)=ln⁡2, and therefore e2H(q)=4≠2=∣Tq∣. This single attainable type refutes the universal exact-equality claim.

3.1F1F2F3algebra∎

At the boundary type q=(1,0), the only word is 00, and both ∣Tq∣ and e2H(q) equal 1; this boundary equality does not rescue the refuted universal claim and is consistent with [F3].

Sources