Alphabeta Math
TheoremStatement: AI-adaptedProof: AI-adaptedSession-authored (Fable 5 assisted)precheck passjudge pass (z-ai/glm-5.2)verified 2026-07-26 (claude-opus-5)
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.

Algebra of limits: sums, scalar multiples, products and quotients

Statement

Let (xk)(x_k) and (yk)(y_k) be sequences of reals (Sequences of reals: bounded, eventually, frequently, tails, subsequences) converging to xx and yy respectively (Limits and Cauchy sequences of reals), and let cRc \in \mathbb{R}. Then

xk+ykx+y,cxkcx,xkykxy,xkykxy,x_k + y_k \to x + y, \qquad c\,x_k \to c\,x, \qquad x_k - y_k \to x - y, \qquad x_k y_k \to x y,

and if in addition y0y \ne 0 and yk0y_k \ne 0 for every kNk \in \mathbb{N}, then

1yk1yandxkykxy.\frac{1}{y_k} \to \frac{1}{y} \qquad \text{and} \qquad \frac{x_k}{y_k} \to \frac{x}{y}.

The quotient case rests on an eventual lower bound for yk|y_k|, proved below rather than assumed: yk>y/2|y_k| > |y|/2 for all sufficiently large kk.

Facts & Assumptions

Given: Sequences (xk)(x_k), (yk)(y_k) of reals with (xk)(x_k) converging to xx and (yk)(y_k) converging to yy, and a real cc (Sequences of reals: bounded, eventually, frequently, tails, subsequences, Limits and Cauchy sequences of reals). For the last two claims we assume in addition y0y \ne 0 and yk0y_k \ne 0 for every kk.

[L1]

Convergence, quantified over rational ε>0\varepsilon > 0 (Limits and Cauchy sequences of reals).

[L2]

Absolute value and the triangle inequality: uv=uv|uv| = |u|\,|v|, u0|u| \ge 0, u=0|u| = 0 if and only if u=0u = 0, u=u|-u| = |u|, and a+ba+b|a + b| \le |a| + |b| (Basic properties of the absolute value, The triangle inequality).

[L3]

Real versus rational ε\varepsilon: for every real η>0\eta > 0 there is a rational ε\varepsilon with 0<ε<η0 < \varepsilon < \eta, by density (The rationals embed densely in the reals) or by the Archimedean property (Every complete ordered field is Archimedean) applied to 1/η1/\eta (Inverses of positives are positive, and reciprocation reverses order); consequently the convergence test of Limits and Cauchy sequences of reals may equivalently be run with real ε>0\varepsilon > 0 (Sequences of reals: bounded, eventually, frequently, tails, subsequences).

[L4]

Convergent sequences are bounded (Every convergent sequence is bounded), and a constant sequence (c)(c) is bounded by c|c| (Sequences of reals: bounded, eventually, frequently, tails, subsequences).

[L5]

A null sequence times a bounded sequence is null (A null sequence times a bounded sequence is null).

[L6]

Reverse triangle inequality: uvuv\bigl||u| - |v|\bigr| \le |u - v|, hence uvvu|u| \ge |v| - |v - u| (The reverse triangle inequality).

[L7]

Inverses and order: a>0a > 0 implies a1>0a^{-1} > 0; 0<a<b0 < a < b implies 0<b1<a10 < b^{-1} < a^{-1}; (u1)1=u(u^{-1})^{-1} = u for u0u \ne 0 (Inverses of positives are positive, and reciprocation reverses order, Field).

[L8]

Order arithmetic in R\mathbb{R}: adding a constant and adding inequalities preserve the order, multiplying a strict inequality by a positive factor preserves it, and \le and << compose transitively; trichotomy holds, and since u0u \ge 0 means u>0u > 0 or u=0u = 0, an element with u0u \ge 0 and u0u \ne 0 satisfies u>0u > 0 (Order is preserved by adding a constant and by adding inequalities, Sign rules for products and monotonicity of multiplication, Complete ordered field (least-upper-bound property), Ordered field). Moreover 2=1+1>02 = 1 + 1 > 0 and 22 is invertible: 0<10 < 1 in any ordered field (The multiplicative identity is positive) and the positives are closed under addition, so 2>02 > 0 and in particular 202 \ne 0 (Ordered field), whence 212^{-1} exists (Field).

[L9]

Rational arithmetic: ε/2\varepsilon/2 is a rational >0> 0 whenever ε\varepsilon is, and ε/2+ε/2=ε\varepsilon/2 + \varepsilon/2 = \varepsilon (The rationals form a totally ordered field); the order on N\mathbb{N} is total, so finitely many thresholds admit a common index (\le is a linear order on N\mathbb{N}).

Proof

technique · direct
1.1

Reduction to null sequences: for any sequence (zk)(z_k) of reals and any real zz, the statements "(zk)(z_k) converges to zz" and "(zkz)(z_k - z) converges to 00" are literally the same condition, because (zkz)0=zkz|(z_k - z) - 0| = |z_k - z| for every kk.

L1L2
1.2

Sum rule, in general form. Let (uk)u(u_k) \to u and (vk)v(v_k) \to v be any convergent sequences of reals and let ε>0\varepsilon > 0 be rational; take K1K_1 with uku<ε/2|u_k - u| < \varepsilon/2 for kK1k \ge K_1 and K2K_2 with vkv<ε/2|v_k - v| < \varepsilon/2 for kK2k \ge K_2, and let KK be an index at least as large as both. For kKk \ge K, (uk+vk)(u+v)=(uku)+(vkv)uku+vkv<ε/2+ε/2=ε|(u_k + v_k) - (u + v)| = |(u_k - u) + (v_k - v)| \le |u_k - u| + |v_k - v| < \varepsilon/2 + \varepsilon/2 = \varepsilon; hence uk+vku+vu_k + v_k \to u + v, and in particular xk+ykx+yx_k + y_k \to x + y.

L1L2L8L9algebra
1.3

Boundedness: every convergent sequence of reals is bounded, and every constant sequence (d)(d) is bounded by d|d|.

L4
1.4

Quotient preparation. Assume y0y \ne 0 and yk0y_k \ne 0 for every kk. Then y>0|y| > 0 by [L2], so y/2>0|y|/2 > 0; running the convergence test of (yk)(y_k) with the real number y/2|y|/2 as tolerance, which [L3] licenses, produces K0NK_0 \in \mathbb{N} with yky<y/2|y_k - y| < |y|/2 for all kK0k \ge K_0.

L1L2L3L7L8choose
2.1

Scalar rule, in general form. Let (uk)u(u_k) \to u and let dRd \in \mathbb{R}. By step 1.1 the sequence (uku)(u_k - u) is null and by step 1.3 the constant sequence (d)(d) is bounded, so (d(uku))=(dukdu)(d(u_k - u)) = (d u_k - d u) is null by [L5]; by step 1.1 again, dukdud u_k \to d u, and in particular cxkcxc x_k \to c x.

step 1.1step 1.3L5algebra
2.2

Product rule, in general form. Let (uk)u(u_k) \to u and (vk)v(v_k) \to v, and write ukvkuv=uk(vkv)+v(uku)u_k v_k - u v = u_k(v_k - v) + v(u_k - u). By step 1.1 both (vkv)(v_k - v) and (uku)(u_k - u) are null; by step 1.3 both (uk)(u_k) and the constant sequence (v)(v) are bounded; so both (uk(vkv))(u_k(v_k - v)) and (v(uku))(v(u_k - u)) are null by [L5], and their sum is null by step 1.2 applied with both limits equal to 00. By step 1.1, ukvkuvu_k v_k \to u v, and in particular xkykxyx_k y_k \to x y.

step 1.1step 1.2step 1.3L5algebra
2.3

Eventual lower bound. For every kK0k \ge K_0, the reverse triangle inequality gives ykyyyk=yyky>yy/2=y/2>0|y_k| \ge |y| - |y - y_k| = |y| - |y_k - y| > |y| - |y|/2 = |y|/2 > 0; so yk>y/2|y_k| > |y|/2 for all kK0k \ge K_0, and in particular yk0|y_k| \ne 0 there.

step 1.4L2L6L8
3.1

Difference rule. Applying step 2.1 to the sequence (yk)(y_k) with d=1d = -1 gives yky-y_k \to -y; the sum rule of step 1.2 applied to (xk)(x_k) and (yk)(-y_k) then gives xkyk=xk+(yk)x+(y)=xyx_k - y_k = x_k + (-y_k) \to x + (-y) = x - y.

step 1.2step 2.1L8algebra
3.2

Reciprocal estimate. For kK0k \ge K_0 we have yk0y_k \ne 0 and yk>y/2>0|y_k| > |y|/2 > 0, so [L7] applied to 0<y/2<yk0 < |y|/2 < |y_k| gives 0<yk1<2/y0 < |y_k|^{-1} < 2/|y|, and therefore 1yk1y=yykyky2y2yky\left|\dfrac{1}{y_k} - \dfrac{1}{y}\right| = \dfrac{|y - y_k|}{|y_k|\,|y|} \le \dfrac{2}{|y|^2}\,|y_k - y|.

step 2.3L2L7L8algebra
4.1

Reciprocal rule. Let ε>0\varepsilon > 0 be an arbitrary real and put η:=εy2/2\eta := \varepsilon |y|^2/2, a real >0> 0; by [L3] there is K1K_1 with yky<η|y_k - y| < \eta for all kK1k \ge K_1. For every kk at least as large as both K0K_0 and K1K_1, step 3.2 gives 1yk1y2y2yky<2y2η=ε\left|\dfrac{1}{y_k} - \dfrac{1}{y}\right| \le \dfrac{2}{|y|^2}|y_k - y| < \dfrac{2}{|y|^2}\,\eta = \varepsilon; hence 1/yk1/y1/y_k \to 1/y.

step 3.2L1L3L7L8L9algebra
5.1

Quotient rule. By step 4.1 the sequence (1/yk)(1/y_k) converges to 1/y1/y, so the product rule of step 2.2 applied to (xk)(x_k) and (1/yk)(1/y_k) gives xk/yk=xk(1/yk)x(1/y)=x/yx_k/y_k = x_k \cdot (1/y_k) \to x \cdot (1/y) = x/y.

step 2.2step 4.1algebra
6.1

All the claims are established: the sum rule in step 1.2, the scalar rule in step 2.1, the difference rule in step 3.1, the product rule in step 2.2, and the reciprocal and quotient rules in steps 4.1 and 5.1.

step 1.2step 2.1step 2.2step 3.1step 4.1step 5.1

Remarks

  • The quotient case is where proofs usually cheat. The estimate 1/yk1/y=yyk/(yky)|1/y_k - 1/y| = |y - y_k| / (|y_k||y|) is worthless until yk|y_k| is known to stay away from 00: without a lower bound the denominator can be arbitrarily small and the fraction arbitrarily large, even while yyk|y - y_k| shrinks. Step 2.3 supplies that bound, yk>y/2|y_k| > |y|/2 for kK0k \ge K_0, and it is proved from the reverse triangle inequality, not assumed.

  • The hypothesis yk0y_k \ne 0 for every kk is only there so that 1/yk1/y_k is defined for every index. It is not needed for the limit: step 2.3 shows yk>y/2>0|y_k| > |y|/2 > 0 from K0K_0 on, so a sequence with y0y \ne 0 has at most finitely many zero terms, and by Convergence depends only on the tail one may pass to the K0K_0-th tail and read the conclusion there.

  • The hypothesis y0y \ne 0 cannot be dropped. With xk=1x_k = 1 and yk=1/(k+1)y_k = 1/(k+1), both sequences converge: the first is constant (Sequences of reals: bounded, eventually, frequently, tails, subsequences) and the second is null (FALSE: limits preserve strict inequalities), so y=0y = 0. Yet xk/yk=k+1x_k/y_k = k+1, and no real MM bounds every k+1k+1, by the Archimedean property (Every complete ordered field is Archimedean); so the quotient sequence is unbounded, hence not convergent by Every convergent sequence is bounded.

  • Nothing in the proof uses completeness of R\mathbb{R} beyond the Archimedean property invoked in [L3], so the same rules hold verbatim for sequences of rationals.

Depends on

Used by

…and 11 more results.

Dependency tree · next 3 levels

Direct dependencies and their dependencies through the next three levels: 74 results over 29 levels. An arrow runs from a result to what uses it, and this result sits at the bottom with a heavier outline. Click the chart to enlarge it.

Sources