Alphabeta Math
LemmaStatement: Literature-sourcedProof: AI-generatedPipeline-generatedjudge pass (gpt-6-luna)
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.

Trigonometric identities, Machin's formula and reduced Taylor tails

Statement

The power-series sine and cosine satisfy the addition identities and sin⁡2x+cos⁡2x=1. For A(t)=arctan⁡t from Real elementary functions and their domains,

sin⁡(A(t))=tcos⁡(A(t)),cos⁡(A(t))>0,

so A(t) is the inverse tangent with values in (−π/2,π/2) when π=4A(1). It is odd; for u,v≥0 with uv<1, A(u)+A(v)=A((u+v)/(1−uv)); and for t>0, A(t)=π/2−A(1/t). In particular sin⁡(π/2)=1 and cos⁡(π/2)=0. Also,

π=16arctan⁡(1/5)−4arctan⁡(1/239).

For every real x there is an integer k and r=x−kπ/2 with ∣r∣≤π/4; the sine and cosine of x are obtained from those of r by the quarter-period identities. For every real r with ∣r∣≤1 and every N≥0, truncating sine through degree 2N+1 and cosine through degree 2N gives absolute tail bounds

Ts=∣r∣2N+3(2N+3)!(1−∣r∣2/((2N+4)(2N+5)))

and

Tc=∣r∣2N+2(2N+2)!(1−∣r∣2/((2N+3)(2N+4))).

Facts & Assumptions

[L1]

Real elementary functions and their domains: sine and cosine are the factorial-denominator power series; A(t)=∫0tdu/(1+u2); and π=4A(1). The integral has the local Darboux constant-comparison and average-value derivative properties.

[L2]

Range-reduced exponential, logarithm and arctangent series have certified tails: the arctangent series on [−1/2,1/2] has a remainder tending to zero with its truncation order.

[L3]

Mean value theorem on a closed real interval: for u<v a continuous h differentiable on (u,v) has w∈(u,v) with h(v)−h(u)=h′(w)(v−u).

[L4]

Sum, product, quotient, and chain rules from difference quotients: differentiability implies continuity, and sum, product, quotient and composition derivatives follow from the difference-quotient definition.

[L5]

Power series may be differentiated term by term inside their radius: inside its radius of convergence, a power series with absolutely convergent derivative series has the termwise derivative.

[L6]

Dedekind completeness implies the Archimedean property: the natural numbers are unbounded above.

[L7]

Real and complex modulus product and triangle inequalities: real absolute value is multiplicative, satisfies the triangle inequality, and is nonnegative.

[L8]

Elementary factorial and geometric series converge uniformly: the factorial series and their derivative series converge absolutely and uniformly on bounded intervals, and geometric tails have finite bounds.

[L9]

Cauchy products of absolutely convergent series multiply their sums: the product of two absolutely convergent series equals their Cauchy product, grouped by total degree.

Proof

1.1givenL1L7L8L9algebra

On each bounded square, the factorial series and their tails converge uniformly by [L8]. Their Cauchy products are absolutely convergent and can be grouped by total degree by [L9]. For sine, the coefficient of xrys in sin⁡(x+y) vanishes when r+s is even and otherwise equals (−1)(r+s−1)/2/(r!s!), by the binomial theorem; the same coefficient appears in sin⁡xcos⁡y+cos⁡xsin⁡y. For cosine, the coefficient in cos⁡(x+y) vanishes when r+s is odd and otherwise equals (−1)(r+s)/2/(r!s!), matching cos⁡xcos⁡y−sin⁡xsin⁡y. Thus both addition identities hold.

1.2givenL1L3L4L5L7L8algebra

On every bounded interval, [L8] gives absolute convergence of the factorial coefficient series and their derivative series. Applying [L5] to the two power series gives sin⁡′=cos⁡ and cos⁡′=−sin⁡. By the product rule in [L4], the derivative of sin⁡2x+cos⁡2x is zero; [L3] and its value 1 at zero give sin⁡2x+cos⁡2x=1.

2.1givenL1L3L4L7step 1.2algebra

Write A(t)=arctan⁡t, S(t)=sin⁡A(t) and C(t)=cos⁡A(t). From [L1], the difference quotient for A is the average of the continuous integrand 1/(1+u2) over the interval from t to t+h, so A′(t)=1/(1+t2). By the preceding sine and cosine identity and the chain and product rules [L4], g(t)=S(t)−tC(t) satisfies g′(t)=tg(t)/(1+t2) and g(0)=0. The quotient rule [L4] gives (g(t)2/(1+t2))′=0. By [L3], this quotient is constant on every interval between 0 and t, hence zero; therefore g(t)=0. Now C(t)2=1/(1+t2) by the preceding sine and cosine identity, and C′(t)=−tC(t)/(1+t2). Put q(t)=1+t2. Since q(t)≥0 and q(t)2≥1, it follows that q(t)≥1: otherwise (1−q(t))(1+q(t))>0 would imply q(t)2<1. The same holds at t+h. For h near zero, rationalization gives ∣q(t+h)−q(t)∣=∣h∣ ∣2t+h∣q(t+h)+q(t)≤∣h∣(2∣t∣+∣h∣)2⟶0. Thus q is continuous, and the rationalized difference quotient (q(t+h)−q(t))/h=(2t+h)/(q(t+h)+q(t)) tends to t/q(t); hence q′(t)=t/q(t). The product rule [L4] gives (q(t)C(t))′=0, so [L3] implies q(t)C(t) is constant on every segment between 0 and t. Its value at zero is 1, hence C(t)=1/q(t)>0. For t>0 and 0≤u≤t, the integrand satisfies 1/(1+u2)≥1/(1+t2). Constant comparison for the integral [L1] gives A(t)≥t/(1+t2)>0. Also A′(t)=1/(1+t2), so H(t)=A(t)+A(1/t) has derivative H′(t)=11+t2−1t211+1/t2=0. By [L3], H is constant on the positive segment between 1 and t; H(1)=2A(1). Therefore A(t)=2A(1)−A(1/t) for every t>0, and the positivity just proved shows 0<A(t)<2A(1). Similarly, the derivative of A(−t)+A(t) is zero and its value at zero is zero, so [L3] gives A(−t)=−A(t) for every real t. Since π/2=2A(1), these bounds yield A(t)∈(−π/2,π/2) for all real t and tan⁡A(t)=t. At t=1, S(1)=C(1)>0 and step 1.2 gives both equal 1/2; the addition identities yield sin⁡(π/2)=1 and cos⁡(π/2)=0.

3.1step 2.1L3L4L7algebra

For u,v≥0 with uv<1, define H(v)=A(u)+A(v)−A((u+v)/(1−uv)). Differentiating with the inverse-tangent derivative formula and the chain rule [L4], the derivative of the last term is (1+u2)/(1−uv)21+((u+v)/(1−uv))2=11+v2, which is also A′(v). Thus H′(v)=0 and H(0)=0; [L3] gives A(u)+A(v)=A((u+v)/(1−uv)). Set α=A(1/5) and β=A(1/239). This identity gives 2α=A(5/12), then 4α=A(120/119), and also β+A(1)=A(120/119). Hence 4α=β+A(1) and 16α−4β=4A(1)=π.

3.2L1L6L7step 1.1step 2.1choosealgebra

Put q=2x/π. By [L6], choose an integer n>∣q∣+1. Among the finite ordered list −n,−n+1,…,n, take its greatest member k0≤q; then k0≤q<k0+1. At least one of k0,k0+1 is within 1/2 of q; call it k. Since 0<1/(1+u2)≤1 on [0,1], integral comparison [L1] and step 2.1 give 0<A(1)≤1, hence π/4=A(1)≤1. Therefore the residual r=x−kπ/2 satisfies ∣r∣≤π/4≤1. Repeated application of the addition identities in step 1.1 and the values proved in step 2.1 gives the quarter-period sign and swap formulas.

4.1step 3.1L2construct

Since 1/5 and 1/239 lie in [0,1/2], [L2] gives rational intervals for α and β with the displayed tail widths. Multiplying those intervals by 16 and 4 and using the exact identity in step 3.1 gives a certified rational interval for π; increasing the truncation count decreases both tail bounds.

5.1givenL7L8algebra∎

Fix any ∣r∣≤1 and N≥0. The sine terms after degree 2N+1 have first omitted magnitude ∣r∣2N+3/(2N+3)! and each later ratio is at most ∣r∣2/((2N+4)(2N+5))<1. Summing this geometric majorant gives Ts. The cosine terms after degree 2N have first omitted magnitude ∣r∣2N+2/(2N+2)! and ratio at most ∣r∣2/((2N+3)(2N+4))<1, giving Tc.

Depends on

Used by

Dependency tree · two levels

15 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