Alphabeta Math
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.

✓ 8 results · all verified · 2 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. The 6 not AI-judged were verified by owner audit (typically over a confirmed judge false positive), not failures.

Regular Continued Fractions and Diophantine Approximation — Examples

1 · Prerequisites

2 · Summary

These worked examples keep the abstract statements concrete. They show the two finite expansions of a rational, the Euclidean-algorithm origin of the digits, the periodic expansions of familiar quadratic irrationals, and the way the approximation theorems control explicit rational approximants.

3 · Logical flowchart

4 · Definitions, theorems and proofs

None yet.

5 · Examples, counterexamples and false statements

ExampleConstruction: AI-generatedVerification: AI-generatedprecheck passjudge pass (gpt-5.6-terra)audited 2026-08-27Open item page →

A rational number has exactly two finite regular continued-fraction expansions

Example

The rational number 8/5 has the two finite regular continued-fraction expansions 85=[1;1,1,2]=[1;1,1,1,1], and the normalized one is [1;1,1,2] because its last digit is at least 2.

Facts & Assumptions

Given: The rational number 8/5.

[F1]

Every rational number has a unique normalized finite regular continued fraction, and exactly one other finite expansion obtained by splitting the last digit an≥2 into an−1,1 (Normalized finite regular continued fractions are unique).

Verification

technique · direct
1.1givenalgebra

Direct calculation gives. [given, algebra] [1;1,1,2]=1+11+11+12=1+11+23=1+35=85.

2.1F1step 1.1algebra∎

Likewise. [F1, step 1.1, algebra] [1;1,1,1,1]=1+11+11+11+11=1+35=85, so the same rational has two finite expansions. The last digit of [1;1,1,2] is 2, so [F1] identifies it as the normalized one and shows there are no further finite expansions.

ExampleConstruction: AI-generatedVerification: AI-generatedprecheck passjudge pass (gpt-5.6-terra)audited 2026-08-27Open item page →

The continued fraction of 37/11 matches Euclid and Bezout

Example

For 37/11, the Euclidean divisions 37=3⋅11+4,11=2⋅4+3,4=1⋅3+1,3=3⋅1+0 give the continued fraction 3711=[3;2,1,3], and the penultimate convergents encode the Bezout relation 37⋅3−11⋅10=1.

Facts & Assumptions

Given: The integers 37 and 11.

[F1]

For a rational number, the continued-fraction algorithm terminates and its digits are exactly the Euclidean quotient digits (The continued-fraction algorithm terminates exactly on rational numbers).

Verification

technique · direct
1.1F1givenalgebra

The Euclidean quotient digits are 3,2,1,3, so [F1] gives. [F1, given, algebra] 3711=[3;2,1,3]. Its convergents are 3,3+12=72,3+12+11=103,3+12+11+13=3711.

2.1F2step 1.1algebra∎

The penultimate convergent is 10/3, and. [F2, step 1.1, algebra] 37⋅3−11⋅10=111−110=1. So 1 is an explicit integer linear combination of 37 and 11, which is exactly the Bezout identity for gcd⁡(37,11)=1 in the sense of [F2].

ExampleConstruction: AI-generatedVerification: AI-generatedprecheck passaudited 2026-08-27Open item page →

The continued fraction [1; overline 2] for sqrt(2)

Example

The continued fraction of 2 is 2=[1;2‾], with convergents 1,32,75,1712,….

Facts & Assumptions

Given: The real number 2.

[F1]

Consecutive convergents satisfy pnqn−1−pn−1qn=(−1)n−1 (Determinant identity for consecutive convergents).

[F2]

For an irrational number α, the convergents alternate around α and satisfy ∣α−pn/qn∣<1/(qnqn+1) (Convergent error bound).

[F3]

The complete-quotient algorithm chooses the unique integer part and then takes the reciprocal of the positive fractional part (Complete quotients in the continued-fraction algorithm).

[F4]

If the complete-quotient algorithm does not terminate, its resulting infinite regular continued fraction converges to the original real number (The continued-fraction algorithm for real numbers).

[F5]

For digits an, the convergent numerators and denominators start from p−2=0,p−1=1,q−2=1,q−1=0 and satisfy pn=anpn−1+pn−2 and qn=anqn−1+qn−2 (Convergents of a regular continued fraction).

Verification

technique · direct
1.1F3F4givenalgebra

Since 1<2<2, [F3] gives a0=1. Then [F3, F4, given, algebra] α1=12−1=2+1, so 2<α1<3 and a1=2. Moreover 1α1−2=12−1=α1, so every later complete quotient is again α1. Thus the algorithm never terminates and produces the digits 1,2,2,…; by [F4] its continued fraction converges to the original number. Hence 2=[1;2‾].

2.1F1F5step 1.1algebra

Applying [F5] to the digits from step 1.1 gives [F5, step 1.1, algebra] (p0,q0)=(1,1),(p1,q1)=(3,2),(p2,q2)=(7,5),(p3,q3)=(17,12), so the convergents begin 1,32,75,1712. The same recurrence gives q4=29. For the displayed pairs one checks 3⋅1−1⋅2=1,7⋅2−3⋅5=−1,17⋅5−7⋅12=1, exactly as [F1] predicts.

3.1F2step 2.1algebra∎

The error formula [F2] now gives [F2, step 2.1, algebra] ∣2−32∣=3−222<110,∣2−75∣=52−75<160, and similarly ∣2−1712∣=17−12212<1348. So the concrete convergents alternate around 2 with the expected quality of approximation.

ExampleConstruction: AI-generatedVerification: AI-generatedprecheck passaudited 2026-08-27Open item page →

The continued fraction [3; overline 1,2,1,6] for sqrt(14)

Example

The complete quotients of 14 cycle through the states 14+35,14+22,14+25,14+3, so 14=[3;1,2,1,6‾].

Facts & Assumptions

Given: The real number 14.

[F1]

The continued-fraction algorithm is deterministic: each complete quotient αn determines the digit an and, when αn≠an, the next complete quotient αn+1 (Complete quotients in the continued-fraction algorithm).

Verification

technique · direct
1.1givenalgebra

Since 3<14<4, the first digit is a0=3. Then [given, algebra] α1=114−3=14+35,α2=1α1−1=14+22, α3=1α2−2=14+25,α4=1α3−1=14+3, and finally α5=1α4−6=114−3=α1. So the digits from α1 onward are 1,2,1,6 and then repeat.

2.1F1step 1.1∎

Step 1.1 shows that α5=α1. By the determinism in [F1], the same four digits 1,2,1,6 therefore repeat from α1 onward. Hence 14=[3;1,2,1,6‾].

ExampleConstruction: AI-generatedVerification: AI-generatedprecheck passaudited 2026-08-27Open item page →

The continued fraction [1; overline 1] for the golden ratio

Example

If x=[1;1‾], then x=1+1x, so x is the positive root of x2−x−1=0, namely the golden ratio φ=1+52.

Facts & Assumptions

Given: The purely periodic continued fraction x=[1;1‾].

[F1]

Every eventually periodic regular continued fraction has quadratic- irrational value (Eventually periodic regular continued fractions are quadratic irrationals).

[F2]

The value of an infinite regular continued fraction is the common limit of its convergents; for [1;1,1,…] the increasing even subsequence starts at 1 (Every infinite regular continued fraction converges to a unique real number).

[F3]

Finite regular continued fractions are evaluated by the recursion [a0;a1,…,an]=a0+1/[a1;…,an] (Finite and infinite regular continued fractions).

Verification

technique · direct
1.1F2F3givenalgebra

Let xm=[1;1,…,1⏟m further digits]. By [F2], xm→x and x≥1, while [F3] gives xm+1=1+1/xm and xm≥1. Hence ∣1xm−1x∣=∣xm−x∣xmx≤∣xm−x∣⟶0, so taking limits in the recursion gives x=1+1x. Multiplication by x now gives x2−x−1=0, whose positive solution is x=1+52.

2.1F1step 1.1∎

The continued fraction is purely periodic, so [F1] says its value is a quadratic irrational. Step 1.1 exhibits the quadratic equation explicitly, and its positive root is the golden ratio φ.

ExampleConstruction: AI-generatedVerification: AI-generatedprecheck passaudited 2026-08-27Open item page →

The fractions 22/7, 333/106, and 355/113 as approximations to pi

Example

Among the classical fractions 227,333106,355113, the best approximation to π is 355/113, and it is already forced by Legendre's criterion.

Facts & Assumptions

Given: The real number π and the three rational numbers above.

[L1]

The number π is irrational (Ivan Niven, "A simple proof that pi is irrational", Bulletin of the AMS 53 (1947), 509).

[F1]

If α is irrational and a reduced rational number r/s satisfies ∣α−r/s∣<1/(2s2), then it is a convergent of α (Legendre's criterion for convergents).

[F2]

A convergent of an irrational number is the best approximation among all rationals with smaller next denominator (Convergents are best rational approximations of the first kind).

Verification

technique · direct
1.1givenalgebra

Direct decimal comparison gives. [given, algebra] ∣π−227∣≈1.264×10−3,∣π−333106∣≈8.322×10−5, ∣π−355113∣≈2.668×10−7<12⋅1132≈3.916×10−5. Moreover 355=3⋅113+16 and 113=7⋅16+1, so gcd⁡(355,113)=1. Thus 355/113 is reduced and satisfies Legendre's criterion.

2.1F1F2step 1.1

By [L1] and [F1], the fraction 355/113 is a convergent of π. Then [F2] says no. [L1, F1, F2, step 1.1] rational with denominator at most 113 approximates π more closely. Since 22/7 and 333/106 have denominators 7 and 106, neither can beat 355/113.

2.2step 1.1algebra∎

The direct errors from step 1.1 also show that 333/106 improves on. [step 1.1, algebra] 22/7, but that both are far worse than 355/113.

CounterexampleConstruction: AI-generatedVerification: AI-generatedprecheck passaudited 2026-08-27Open item page →

The constant 1/2 in Legendre's criterion cannot be replaced by 3/4

Statement refuted

For every irrational real number α and every reduced rational number r/s with s>0, ∣α−rs∣<34s2 implies that r/s is a convergent of α.

Facts & Assumptions

Given: The irrational number 2 and the rational number 10/7.

[F1]

Legendre's criterion with the sharp constant 1/2 says that ∣α−r/s∣<1/(2s2) forces r/s to be a convergent (Legendre's criterion for convergents).

[F2]

Convergents of an irrational satisfy the standard error bound ∣α−pn/qn∣<1/(qnqn+1) (Convergent error bound).

Counterexample

technique · direct
1.1givenalgebra

The first few convergents of [1;2‾]=2 are. [given, algebra] 1,32,75,1712, so 10/7 is not a convergent of 2.

1.2givenalgebra

Nevertheless. [given, algebra] 107−2=100−987(10+72)=27(10+72)<34⋅72. Thus 10/7 satisfies the displayed 3/(4s2) bound.

2.1F1F2step 1.1step 1.2∎

Step 1.1 and step 1.2 together refute the statement. In the light of [F1],. [F1, F2, step 1.1, step 1.2] this shows that the sharp constant 1/2 in Legendre's criterion cannot simply be replaced by the larger constant 3/4.

ExampleConstruction: AI-generatedVerification: AI-generatedprecheck passaudited 2026-08-27Open item page →

A negative irrational has a regular continued fraction with positive later digits

Example

For the negative irrational −2, the continued-fraction algorithm gives −2=[−2;1,1,2‾]. The only negative digit is the initial one; every later digit is positive.

Facts & Assumptions

Given: The real number −2.

[F1]

The complete-quotient algorithm chooses the unique integer part an with an≤αn<an+1, and whenever the next complete quotient exists it is αn+1=1/(αn−an) (Complete quotients in the continued-fraction algorithm).

[F2]

The continued-fraction algorithm reconstructs the original real number from its digits (The continued-fraction algorithm for real numbers).

Verification

technique · direct
1.1F1givenalgebra

Since −2<−2<−1, the first digit is a0=−2. Then. [F1, given, algebra] α1=1−2+2=2+22,α2=1α1−1=2, so a1=1 and a2=1.

2.1F1F2step 1.1algebra∎

One more step gives. [F1, F2, step 1.1, algebra] α3=12−1=2+1, so every later digit is 2. Hence the digit string is [−2;1,1,2‾], and [F2] identifies its value with −2. In particular the negative sign is absorbed entirely into the first digit, while every later digit stays positive as required by [F1].

Sources