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.

✓ 13 results · all verified · 7 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.

Properties of the Integral and the Working FTC: Examples and Counterexamples

1 · Prerequisites

2 · Summary

3 · Logical flowchart

4 · Definitions, theorems and proofs

None yet.

5 · Examples, counterexamples and false statements

ExampleConstruction: Literature-sourcedVerification: AI-generatedprecheck passjudge pass (z-ai/glm-5.2)audited 2026-07-29Open item page →

∫01xm=1/ι(m+1), computed by the fundamental theorem and checked against the definition

Example

Let m∈N and let pm:[0,1]→R be pm(x)=xm (Integer powers am). Then pm is integrable on [0,1] and

∫01xm dx  =  1ι(m+1),

where ι(m+1) is the canonical natural of m+1 in R (The canonical natural ι(n)=n⋅1F of a field) and is positive because m+1≥1 (Canonical naturals are positive and strictly increasing).

The ι is not decoration. A natural number is a von Neumann natural, that is a set, so m+1 is not an element of R and 1/(m+1) is not an expression of the field; what the display says is 1/ι(m+1), and that is why the reader meets ι here at all.

Two independent checks are carried out below: the value at m=0 against the published formula for the integral of a constant (If m≤f≤M on [a,b] then m(b−a)≤L(f,P)≤∫ab‾f≤∫ab‾f≤U(f,P)≤M(b−a) for every partition P; in particular every constant function is integrable, with ∫abc=c(b−a)), and the monotonicity of the answer in m against the pointwise inequality xm+1≤xm on [0,1].

Facts & Assumptions

Given: A natural number m and the function pm(x)=xm on [0,1].

[L5]

Powers: 1n=1 for every n, 0n=0 for n≥1, 00=1, and 0≤x≤1 gives 0≤xn+1≤xn≤1 (Integer powers am, Monotonicity of x↦xn and of n↦an, claims 1 and 3).

[L6]

ι(0)=0, ι(n+1)=ι(n)+1, ι(n)>0 for n≥1, and ι is increasing on the naturals ≥1 (The canonical natural ι(n)=n⋅1F of a field, Canonical naturals are positive and strictly increasing).

Verification

technique · direct
1.1

pm is continuous on [0,1], hence integrable there, by [L3].

L3
1.2

Define H:[0,1]→R by H(x):=xm+1/ι(m+1); this is legitimate because ι(m+1)>0 by [L6].

L6L8construct
2.1

By [L1] with n:=m+1≥1 and [L2], H is differentiable at every point of [0,1] with H′(x)=ι(m+1) xm/ι(m+1)=xm=pm(x).

step 1.2L1L2L6
3.1

By [L4] applied to H on [0,1], whose derivative pm is integrable by step 1.1, ∫01xm=H(1)−H(0).

step 1.1step 2.1L4
4.1

By [L5], H(1)=1m+1/ι(m+1)=1/ι(m+1) and H(0)=0m+1/ι(m+1)=0, since m+1≥1; so ∫01xm=1/ι(m+1).

step 1.2step 3.1L5L6
5.1

First check, at m=0. There p0 is the constant function 1 by [L5], so ∫01p0=1⋅(1−0)=1 by [L7], while the formula gives 1/ι(1)=1/1=1. The two agree.

step 4.1L5L6L7
6.1

Second check, monotonicity in m. By [L5], xm+1≤xm for every x∈[0,1], so ∫01xm+1≤∫01xm by [L7]; the formula gives 1/ι(m+2)≤1/ι(m+1), which holds by [L6] and [L8]. The two agree.

step 4.1L5L6L7L8∎

Remarks

ExampleConstruction: AI-generatedVerification: AI-generatedprecheck passjudge pass (z-ai/glm-5.2)audited 2026-07-29Open item page →

A step function integrated by additivity over subintervals, and the same value from the definition

Example

Let f:[0,2]→R be

f(x)  :=  {10≤x<1,31≤x≤2.

Then f is integrable on [0,2] and

∫02f  =  1+3  =  4.

Two routes are carried out below. The first splits at the jump with For a<c<b: f is integrable on [a,b] if and only if it is integrable on [a,c] and on [c,b], and then ∫abf=∫acf+∫cbf; with the oriented form for arbitrary a,b,c and evaluates each piece with Changing an integrable function at finitely many points changes neither its integrability nor its integral and the constant case of If m≤f≤M on [a,b] then m(b−a)≤L(f,P)≤∫ab‾f≤∫ab‾f≤U(f,P)≤M(b−a) for every partition P; in particular every constant function is integrable, with ∫abc=c(b−a). The second computes lower and upper Darboux sums directly on a three-point partition and squeezes.

Why this is the smallest interesting example. The jump sits at the interior point 1, which is a partition point of both routes, and a partition point lies in two subintervals (Partition of [a,b] as a finite strictly increasing list a=t0<t1<⋯<tn=b, its subintervals and their lengths, its mesh, refinement, and the common refinement of two partitions). That is exactly the case singled out in the proof of Changing an integrable function at finitely many points changes neither its integrability nor its integral, and it is why the value of f at the single point 1 is irrelevant to both the integrability and the value while the upper sums still see it.

Facts & Assumptions

Given: The function f above on [0,2], and a real η with 0<η<1.

[L3]

Changing an integrable function at finitely many points changes neither its integrability nor its integral (Changing an integrable function at finitely many points changes neither its integrability nor its integral).

[L6]

Ordered-field arithmetic: the order is total and transitive, and a real that is ≤4+2η for every η in (0,1) is ≤4 (Ordered field, Complete ordered field (least-upper-bound property), Intervals of R: the nine order-convex forms, nondegeneracy, and length).

Verification

technique · direct
1.1

f is nondecreasing on [0,2]: for x≤y the only possibility not giving f(x)=f(y) is x<1≤y, where f(x)=1<3=f(y). Hence f is bounded and integrable on [0,2], and likewise on [0,1] and on [1,2], by [L1].

givenL1
1.2

Second route. Let Pη be the partition of [0,2] with point set {0, 1−η, 1, 2}, so n=3 with Δ0=1−η, Δ1=η and Δ2=1.

givenL5construct
2.1

First route, the piece [0,1]. The constant function 1 on [0,1] agrees with f at every point of [0,1] except x=1, a single point; so by [L3] and [L4], ∫01f=∫011=1⋅(1−0)=1.

step 1.1L3L4
2.2

The piece [1,2]. There f is the constant 3, so ∫12f=3⋅(2−1)=3 by [L4].

step 1.1L4
2.3

On I0=[0,1−η], f is constantly 1, so m0=M0=1. On I1=[1−η,1], f takes the value 1 at every point below 1 and 3 at 1, so m1=1 and M1=3 by [L5]. On I2=[1,2], f is constantly 3, so m2=M2=3.

step 1.2givenL5
3.1

By [L2] with c:=1, ∫02f=∫01f+∫12f=1+3=4.

step 1.1step 2.1step 2.2L2
3.2

Hence L(f,Pη)=1(1−η)+1⋅η+3⋅1=4 and U(f,Pη)=1(1−η)+3η+3⋅1=4+2η, by [L5].

step 2.3L5
4.1

By [L4] the number ∫02f lies between L(f,Pη)=4 and U(f,Pη)=4+2η for every η∈(0,1), so 4≤∫02f≤4 by [L6]; the second route agrees with step 3.1.

step 1.1step 3.1step 3.2L4L6∎

Remarks

ExampleConstruction: AI-adaptedVerification: AI-generatedprecheck passverified 2026-08-10 (gpt-5.6-terra-codex-subscription)Open item page →

The integral test applied to ∑1/ι(k+1)p for rational p>0, cross-checked against the published p-series theorem

Example

Let p∈Q with p>0 (Order on the rationals) and define

fp:[0,∞)→R,fp(t)  :=  (t+1)−p,

the rational power of the positive base t+1≥1 (Rational powers ar of a positive base). Then fp is nonnegative and nonincreasing, so The integral test: for f≥0 nonincreasing on [0,∞), ∑kf(k) converges if and only if the sequence (∫0Nf)N is bounded, with ∫0Nf≤∑k<Nf(k)≤f(0)+∫0Nf applies, and its terms are

fp(k)  =  1ι(k+1)p(k∈N).

The series ∑kfp(k) is exactly the p-series ∑k≥11/ι(k)p in the sense of Series, partial sums, convergence and the sum, divergence, and the tail series, which converges if and only if p>1 (For rational p>0, ∑1/kp converges iff p>1). The integral test therefore delivers, with no primitive computed anywhere:

(∫0Nfp)N∈N  is bounded above⟺p>1.

The cross-check. At p=2 the integral can also be computed directly: the primitive G(t)=−(t+1)−1 gives

∫0N(t+1)−2 dt  =  1−1ι(N+1)  <  1,

so the sequence is bounded by 1, in agreement with the verdict above at p=2>1. At p=1 the verdict is that (∫0N(t+1)−1)N is unbounded, since the harmonic series diverges. No named logarithmic primitive is available from the current dependency vocabulary, and none is needed for this conclusion.

The exponent must be rational. Real exponents do not exist in this library at this point in the reading order (Why real exponents are deferred on the rational-powers page), so "for p∈[1,∞)" is not a statement that can be made here.

Facts & Assumptions

Given: A rational p>0, the function fp(t)=(t+1)−p on [0,∞), and a natural number N.

[L1]

For a>0 and rationals r,s: ar>0, ar+s=aras, a−r=1/ar, and a0=1 (Laws of rational exponents, Rational powers ar of a positive base).

[L2]

For rational r>0 and 0<a<b: ar<br (Monotonicity of r↦ar and of a↦ar, claim 2); the nonstrict form follows by adjoining equality.

[L3]

ι(k+1)≥1>0 for k∈N, ι(k+1)=ι(k)+1, and ι is nondecreasing (The canonical natural ι(n)=n⋅1F of a field, Canonical naturals are positive and strictly increasing).

[L5]

∑k≥11/ι(k)p converges if and only if p>1, and by Series, partial sums, convergence and the sum, divergence, and the tail series that series is by definition the series of the sequence j↦1/ι(j+1)p on N (For rational p>0, ∑1/kp converges iff p>1).

[L6]

a1/1=a, so for a negative integer exponent the rational power of Rational powers ar of a positive base is the integer power of Integer powers am (Existence and uniqueness of n-th roots: a unique a1/n≥0 with (a1/n)n=a).

[L10]

A quotient of continuous functions is continuous where the denominator does not vanish, and every polynomial function is continuous (Sums, scalar multiples, products, absolute values, maxima, minima and quotients with nonvanishing denominator of continuous functions are continuous, as are constants, the identity and every polynomial function, claims 4 and 5).

[L9]

Ordered-field arithmetic: a positive real has a positive inverse, 0<s≤t gives 1/t≤1/s, and the order is total and transitive (Ordered field, Complete ordered field (least-upper-bound property), Intervals of R: the nine order-convex forms, nondegeneracy, and length).

Verification

technique · direct
1.1

For t≥0 the base t+1 is ≥1>0, so fp(t)=(t+1)−p is defined and positive by [L1].

givenL1
1.2

The cross-check at p=2. By [L6], f2(t)=(t+1)−2 is the integer power, and by [L7] the function G(t):=−(t+1)−1 is differentiable at every t≥0 with G′(t)=−(−(t+1)−2)⋅1=(t+1)−2=f2(t).

givenL6L7
2.1

fp is nonincreasing: for 0≤t≤u one has 0<t+1≤u+1, so (t+1)p≤(u+1)p by [L2], and taking reciprocals reverses the inequality by [L9], giving fp(u)≤fp(t) by [L1].

step 1.1L1L2L9
2.2

fp(k)=(ι(k)+1)−p=1/ι(k+1)p by [L1] and [L3], so the sequence k↦fp(k) is the one named in [L5].

step 1.1L1L3
2.3

f2(t)=1/(t+1)2 is a quotient of polynomial functions whose denominator does not vanish on [0,N], hence continuous there by [L10], hence integrable there by [L8]; so [L8] applied to G gives ∫0Nf2=G(N)−G(0)=−1/ι(N+1)+1.

step 1.2L3L8
3.1

By [L4], ∑kfp(k) converges if and only if (∫0Nfp)N is bounded above.

step 1.1step 2.1L4
4.1

Hence, by [L5] and step 3.1, (∫0Nfp)N is bounded above if and only if p>1.

step 3.1step 2.2L5
5.1

Since ι(N+1)≥1>0, 0<1/ι(N+1)≤1, so 0≤∫0Nf2<1 for every N: the sequence is bounded above by 1, which agrees with step 4.1 at p=2>1.

step 2.3L3L9
6.1

The verdict at p=1. By [L5] the series ∑k≥11/ι(k) diverges, so by step 4.1 the sequence (∫0N(t+1)−1 dt)N is not bounded above. No primitive of (t+1)−1 is exhibited, and none is needed for this conclusion.

step 4.1L5∎

Remarks

ExampleConstruction: AI-adaptedVerification: AI-generatedprecheck passverified 2026-08-10 (gpt-5.6-terra-codex-subscription)Open item page →

H(x)=2x on [0,1]: H is continuous, H′ is unbounded on (0,1], and H′ is therefore not Riemann integrable

Example

Write x:=x1/2 for the unique nonnegative square root of x≥0 (Existence and uniqueness of n-th roots: a unique a1/n≥0 with (a1/n)n=a, Rational powers ar of a positive base) and put

H:[0,1]→R,H(x)  :=  2x.

Then:

  1. H is continuous on [0,1];
  2. H is differentiable at every x∈(0,1], with H′(x)=1/x, and it is not differentiable at 0;
  3. H′ is unbounded on (0,1]: H′(1/ι(n+1)2)=ι(n+1) for every n∈N;
  4. consequently no function on [0,1] agreeing with H′ on (0,1] is Riemann integrable on [0,1], because Darboux sums are defined only for bounded functions (For bounded f on [a,b] and a partition P: the infimum mi and supremum Mi of f on the i-th subinterval, and the lower and upper Darboux sums L(f,P)=∑imiΔi and U(f,P)=∑iMiΔi).

So the second fundamental theorem does not apply on [0,1], even though H is continuous there and differentiable on (0,1]: both hypotheses of The second fundamental theorem: if G is differentiable on [a,b] with G′=f and f is integrable, then ∫abf=G(b)−G(a) fail, differentiability at 0 and integrability of the derivative.

What is available, and what is not. On [η,1] with 0<η<1 everything works: H′ is continuous there, hence integrable, and

∫η1H′  =  H(1)−H(η)  =  2−2η.

The value that the right-hand side approaches as η shrinks is not computed here and is not called an integral: ∫01H′ is undefined, and the object that repairs it is the improper integral, which belongs to a later page.

Facts & Assumptions

Given: The function H(x)=2x1/2 on [0,1], a real η with 0<η<1, and a natural number n.

[L1]

For a≥0 there is a unique s≥0 with s2=a, written a1/2; a1/2>0 when a>0, and 01/2=0 (Existence and uniqueness of n-th roots: a unique a1/n≥0 with (a1/n)n=a, Rational powers ar of a positive base, Laws of rational exponents).

[L10]

Ordered-field arithmetic: a positive real has a positive inverse, the order is total and transitive, and multiplying an inequality by a positive real preserves it (Ordered field, Complete ordered field (least-upper-bound property), The ε-δ limit lim⁡x→cf(x)=L of f:A→R at a limit point c of A).

Verification

technique · direct
1.1

By [L2] the map q(x)=x2 is continuous and injective on the order-convex [0,∞), and its image is exactly [0,∞): every value is ≥0, and every a≥0 is q(a1/2) by [L1].

L1L2
2.1

By [L3] the inverse g:[0,∞)→[0,∞) of q is continuous, and g(a)=a1/2 by the uniqueness in [L1]. Hence H=2g restricted to [0,1] is continuous, which is claim 1.

step 1.1L1L3
2.2

Let x∈(0,1] and put c:=x1/2>0 by [L1]. By [L5], q is differentiable at c with q′(c)=ι(2)c=2c≠0; so [L4] gives g differentiable at x=q(c) with g′(x)=1/(2c)=1/(2x).

step 1.1L1L4L5L10
3.1

Hence H=2g is differentiable at every x∈(0,1] with H′(x)=2/(2x)=1/x, by [L5].

step 2.2L5
4.1

Claim 3. For n∈N put xn:=1/ι(n+1)2, a real in (0,1] by [L9]. Then xn=1/ι(n+1), since that number is positive with square xn and [L1] gives uniqueness; so H′(xn)=ι(n+1) by step 3.1.

step 3.1L1L9L10
5.1

H′ is unbounded on (0,1]: given a real M≥0, [L9] supplies n with M<ι(n+1)=H′(xn).

step 4.1L9
5.2

H is not differentiable at 0. The difference quotient of H at 0 is x↦2x/x=2/x for x∈(0,1], by [L1] and [L10]; at xn it takes the value 2ι(n+1), which exceeds every real by [L9]. So no real L can satisfy the ε-δ condition with ε=1: any δ>0 admits some xn<δ, again by [L9], at which the quotient exceeds L+1.

step 4.1L1L9L10
6.1

Claim 4. Let u:[0,1]→R agree with H′ on (0,1]. By step 5.1, u is unbounded on [0,1], so it has no Darboux sums and is not Riemann integrable there, by [L8].

step 5.1L8
7.1

The hypotheses of the second fundamental theorem both fail on [0,1], by step 5.2 and step 6.1; so [L7] gives nothing there, and ∫01H′ is an undefined symbol.

step 5.2step 6.1L7L8
8.1

On [η,1] everything works. There ⋅ is continuous and does not vanish, so H′=1/⋅ is continuous on [η,1] by [L6] and integrable there; H is differentiable at every point of [η,1] by step 3.1; so [L7] gives ∫η1H′=H(1)−H(η)=2−2η.

step 2.1step 3.1L1L6L7∎

Remarks

CounterexampleConstruction: AI-adaptedVerification: AI-generatedprecheck passjudge pass (z-ai/glm-5.2)audited 2026-08-01Open item page →

The sign function is Riemann integrable on [−1,1] and has no primitive there

Statement refuted

False claim: every Riemann integrable function on a closed bounded interval has a primitive there, that is, is the derivative of some function on that interval.

Let s:[−1,1]→R be the sign function,

s(x)  :=  {−1−1≤x<0,0x=0,10<x≤1.

Then s is Riemann integrable on [−1,1] with ∫−11s=0, and there is no G:[−1,1]→R differentiable at every point of [−1,1] with G′=s.

The integral function of s is F(x)=∣x∣−1, which is differentiable at every point of [−1,1] except x=0 — exactly the one point where s is discontinuous. That is consistent with The first fundamental theorem: if f is integrable on [a,b] and continuous at c, then F′(c)=f(c); in particular a continuous f has F as a primitive, which claims F′(c)=s(c) only at points of continuity of s, and it is what the refutation below turns into a contradiction.

Facts & Assumptions

Given: The sign function s on [−1,1] and its integral function F(x)=∫−1xs.

[A1]

There is G:[−1,1]→R, differentiable at every point of [−1,1] as a function on [−1,1], with G′(x)=s(x) for every x∈[−1,1].

[L7]

Absolute value: ∣x∣=x for x≥0 and ∣x∣=−x for x≤0; ∣x∣/x=1 for x>0 and =−1 for x<0 (Absolute value in an ordered field, Basic properties of the absolute value, Ordered field, Complete ordered field (least-upper-bound property)).

Counterexample

technique · contradiction
1.1

s is bounded, with ∣s∣≤1, and continuous at every point of [−1,1] other than 0: near a point x≠0 it is locally constant. So s is integrable on [−1,1], and likewise on [−1,0] and on [0,1], by [L1].

givenL1
1.2

Assume [A1]. Then G′=s everywhere on [−1,1].

assume-contraA1
2.1

On [−1,0] the function s agrees with the constant −1 except at the single point 0, so ∫−10s=−1⋅(0−(−1))=−1 by [L2]; on [0,1] it agrees with the constant 1 except at 0, so ∫01s=1.

step 1.1L2
2.2

On I+:=[0,1] the function u+(x):=G(x)−x is continuous, by [L5], and differentiable at every interior point x∈(0,1) with u+′(x)=s(x)−1=0; so u+ is constant on [0,1] by [L4], say G(x)=x+c1 there.

step 1.2L4L5choose
2.3

On I−:=[−1,0] the function u−(x):=G(x)+x is continuous and differentiable at every x∈(−1,0) with u−′(x)=s(x)+1=0; so G(x)=−x+c2 on [−1,0].

step 1.2L4L5choose
3.1

By [L3], ∫−11s=∫−10s+∫01s=−1+1=0.

step 2.1L3
3.2

By [L3] again, F(x)=∫−1xs equals −1+x for x∈[0,1] and equals −(x−(−1))=−x−1 for x∈[−1,0], by [L2] applied on the relevant piece; in both cases F(x)=∣x∣−1 by [L7].

step 2.1L2L3L7
3.3

Evaluating both formulas at x=0 gives c1=G(0)=c2; write c for the common value, so G(x)=∣x∣+c for every x∈[−1,1] by [L7].

step 2.2step 2.3L7
4.1

The difference quotient of G at 0 is x↦(G(x)−G(0))/x=∣x∣/x, which equals 1 for x>0 and −1 for x<0 by [L7]; so its right-hand limit at 0 is 1 and its left-hand limit is −1.

step 3.3L7
5.1

By [L6] the two-sided limit of that quotient at 0 does not exist, so G is not differentiable at 0; this contradicts [A1].

step 4.1A1L6
6.1

Hence no such G exists: s is integrable on [−1,1] by step 1.1 and has no primitive there, so the claim is false.

step 1.1step 3.1step 5.1discharge-contradiction∎

Remarks

CounterexampleConstruction: AI-adaptedVerification: AI-generatedprecheck passverified 2026-08-10 (gpt-5.6-terra-codex-subscription)Open item page →

A function differentiable on [0,1] whose derivative is unbounded, hence not Riemann integrable

Statement refuted

False claim: if G:[0,1]→R is differentiable at every point of [0,1] (The derivative f′(c)=lim⁡x→cf(x)−f(c)x−c of f:A→R at a point c∈A that is a limit point of A, and differentiability on a set), then G′ is Riemann integrable on [0,1] (so that ∫01G′=G(1)−G(0) makes sense).

The claim is false. Put

ϕ(t)  :=  t2(1−t)2(t∈R),

a polynomial with ϕ(0)=ϕ(1)=0 and ϕ′(0)=ϕ′(1)=0, and for n∈N set

αn:=1ι(n+2),hn:=1ι(n+2)4,cn:=1ι(n+2)2,βn:=αn+hn,In:=[αn,βn]

(The canonical natural ι(n)=n⋅1F of a field, Integer powers am). The intervals In are pairwise disjoint and lie in (0,1], and

G:[0,1]→R,G(x)  :=  {cn ϕ ⁣(x−αnhn)x∈In for some n,0otherwise

is differentiable at every point of [0,1], while

G′ ⁣(αn+14hn)  =  316 ι(n+2)2,

so G′ is unbounded on [0,1] and therefore has no Darboux sums at all (For bounded f on [a,b] and a partition P: the infimum mi and supremum Mi of f on the i-th subinterval, and the lower and upper Darboux sums L(f,P)=∑imiΔi and U(f,P)=∑iMiΔi) and is not Riemann integrable.

The construction is entirely polynomial, and deliberately so. The classical witness is x2sin⁡(1/x2); the trigonometric functions are built on a later page of this library, so a bump glued from a single quartic is used instead. Only one bump is nonzero near any point of (0,1], so no series converges anywhere in the argument and no limit function is formed.

Facts & Assumptions

Given: The polynomial ϕ(t)=t2(1−t)2, the numbers αn,hn,cn,βn and intervals In above, the function G above, and a real ε>0.

[L3]

ι(n+2)≥2>1, ι is increasing on N, ι(m)<ι(n) for m<n, and for every real w there is a natural n with w<ι(n+2); also ι(n)k≥ι(n) for k≥1 when ι(n)≥1 (The canonical natural ι(n)=n⋅1F of a field, Canonical naturals are positive and strictly increasing, Every complete ordered field is Archimedean, For every ε>0 in a complete ordered field there is a natural n≥1 with 1/n<ε, Monotonicity of x↦xn and of n↦an, claims 3 and 4).

[L8]
[L9]

Ordered-field arithmetic: t(1−t)≤1/4 for every real t, since (t−1/2)2≥0; a positive real has a positive inverse; multiplying an inequality by a positive real preserves it; the order is total and transitive (Ordered field, Complete ordered field (least-upper-bound property), Monotonicity of x↦xn and of n↦an, claim 1).

[L10]

If H is differentiable at every point of [p,q] with H′ integrable there, then ∫pqH′=H(q)−H(p) (The second fundamental theorem: if G is differentiable on [a,b] with G′=f and f is integrable, then ∫abf=G(b)−G(a)).

Counterexample

technique · direct
1.1

By [L1], ϕ is differentiable everywhere with ϕ′(t)=2t(1−t)2−2t2(1−t)=2t(1−t)(1−2t); in particular ϕ(0)=ϕ(1)=0 and ϕ′(0)=ϕ′(1)=0, and ϕ′(1/4)=2⋅14⋅34⋅12=316.

L1L9
1.2

For t∈[0,1], 0≤ϕ(t)=(t(1−t))2≤(1/4)2=1/16, by [L9].

L9
1.3

Each αn,hn,cn is a positive real, hn≤αn and cn≤αn because ι(n+2)>1, and αn≤1/2; so In⊆(0,1], since βn=αn+hn≤2αn≤1.

givenL3L9
2.1

The In are pairwise disjoint. For n≥1, αn−1−αn=1/(ι(n+1)ι(n+2)) while hn=1/ι(n+2)4, and ι(n+2)4>ι(n+1)ι(n+2) because ι(n+2)3≥2 ι(n+2)>ι(n+1) by [L3]; so βn<αn−1. Since (αn) is strictly decreasing, m<n gives βn<αn−1≤αm, and In lies strictly below Im.

step 1.3L3L9
2.2

For each n define the polynomial Φn(y):=cn ϕ((y−αn)/hn) on R. By [L1] and [L2] it is differentiable everywhere with Φn′(y)=(cn/hn) ϕ′((y−αn)/hn), and Φn(αn)=Φn(βn)=0, Φn′(αn)=Φn′(βn)=0 by step 1.1.

step 1.1L1L2construct
2.3

Differentiability at a point of (0,1] outside every In. Let x∈(0,1] with x∉Im for every m. By [L3] fix N with 2/x<ι(N+2), so 2/ι(N+2)<x; for m≥N, βm≤2αm=2/ι(m+2)≤2/ι(N+2)<x by step 1.3 and [L3].

step 1.3L3L9choose
3.1

So G is a well-defined function on [0,1], no x lying in two of the In.

step 2.1given
3.2

Gaps around the endpoints. For each n, (βn+1,αn) meets no Im: for m≤n one has αm≥αn by step 2.1, and for m≥n+1 one has βm≤βn+1, again by step 2.1. Likewise (βn,αn−1) meets no Im for n≥1, and (β0,1] meets none.

step 2.1L3
3.3

The finitely many closed intervals I0,…,IN−1 do not contain x, so each of the positive reals ∣x−αm∣ and ∣x−βm∣ for m<N, together with x−2/ι(N+2), forms a nonempty finite set of positive reals; let ρ be its least element, which is positive by [L8].

step 2.3L8L9choose
4.1

G agrees with Φn on In and with the zero function off ⋃mIm; in particular G(αn)=G(βn)=0 and G≥0 everywhere.

step 3.1step 2.2L9
5.1

Differentiability at an interior point of a bump. Let x∈(αn,βn). The difference quotients of G and of Φn at x agree on the punctured neighbourhood 0<∣y−x∣<min⁡{x−αn, βn−x} inside [0,1], so by [L5] and step 2.2, G is differentiable at x with G′(x)=Φn′(x).

step 2.2step 4.1L5L6L8
5.2

Differentiability at a left endpoint αn. On the right, G agrees with Φn on [αn,βn] and G(αn)=Φn(αn), so the right-hand limit of the difference quotient is Φn′(αn)=0 by [L5] and step 2.2. On the left, G vanishes on (βn+1,αn] by step 3.2 and step 4.1, so the quotient is identically 0 there and its left-hand limit is 0. By [L4], G′(αn)=0.

step 2.2step 4.1step 3.2L4L5L6
5.3

Differentiability at a right endpoint βn. Symmetrically: on the left G agrees with Φn, giving limit Φn′(βn)=0; on the right G vanishes on [βn,αn−1) when n≥1 and on [β0,1] when n=0, by step 3.2, giving limit 0. By [L4], G′(βn)=0.

step 2.2step 4.1step 3.2L4L5L6
5.4

Then G vanishes on (x−ρ,x+ρ)∩[0,1], so its difference quotient at x is identically 0 there and G′(x)=0 by [L5] and [L6].

step 2.3step 3.3step 4.1L5L6
5.5

Differentiability at 0. For y∈(0,1]: if y∉⋃mIm then G(y)/y=0; and if y∈Im then 0≤G(y)≤cm/16 by step 1.2 and step 4.1 and y≥αm, so 0≤G(y)/y≤cm/(16 αm)=1/(16 ι(m+2)).

step 1.2step 4.1L9
6.1

Given ε>0, fix by [L3] a natural N with 1/ε<ι(N+2) and put δ:=αN>0. If 0<y<δ and y∈Im then αm≤y<αN, so ι(N+2)<ι(m+2) by [L3], and step 5.5 gives ∣G(y)/y∣≤1/(16 ι(m+2))<ε; otherwise G(y)/y=0.

step 5.5L3L9choose
6.2

G′ is unbounded. Put un:=αn+hn/4, an interior point of In; by step 5.1 and step 2.2, G′(un)=(cn/hn)ϕ′(1/4)=ι(n+2)2⋅316, using cn/hn=ι(n+2)4/ι(n+2)2=ι(n+2)2. Given a real M≥0, [L3] supplies n with 163M<ι(n+2)≤ι(n+2)2, so G′(un)>M.

step 1.1step 2.2step 5.1L3L9
7.1

So the difference quotient of G at 0, which is y↦G(y)/y on (0,1], tends to 0; hence G is differentiable at 0 with G′(0)=0 by [L6].

step 6.1L6
8.1

By steps 5.1, 5.2, 5.3, 5.4 and 7.1, G is differentiable at every point of [0,1]: every x is either 0, or an interior point of some In, or an endpoint of some In, or a point of (0,1] outside every In.

step 5.1step 5.2step 5.3step 5.4step 7.1L9
9.1

Hence G′ is a function on [0,1] that is not bounded, so it has no Darboux sums and is not Riemann integrable on [0,1] by [L7]; the claim is false, and ∫01G′ is an undefined symbol, so [L10] gives nothing here.

step 8.1step 6.2L7L10∎

Remarks

  • Only one bump is active near any point of (0,1], and that is what makes every step finite. The intervals In accumulate only at 0, so a point of (0,1] has a neighbourhood meeting at most one of them (steps 3.2 and 3.3), and the only place where infinitely many bumps are seen at once is the origin, where step 5.5 controls all of them by a single estimate. No series is summed anywhere.

  • The two exponents are what the construction turns on. Differentiability at 0 needs cn/αn→0, and unboundedness of G′ needs cn/hn→∞; with αn=1/ι(n+2) the choices cn=αn2 and hn=αn4 give cn/αn=αn→0 and cn/hn=αn−2→∞. Any pair of exponents with the same two properties would do; these are verified explicitly in steps 6.1 and 6.2 because the construction is only as good as those two inequalities.

  • What this refutes and what it does not. It refutes the claim that every derivative is Riemann integrable, hence the naive reading of The second fundamental theorem: if G is differentiable on [a,b] with G′=f and f is integrable, then ∫abf=G(b)−G(a) with its integrability hypothesis deleted. It says nothing about whether G′ has a primitive — it does, namely G — and nothing about the sharp class of functions for which ∫abG′=G(b)−G(a) holds, which this library records but does not prove (Conventions of this page, and which sharpenings of the integral are taken up later in the reading order).

CounterexampleConstruction: Literature-sourcedVerification: AI-generatedprecheck passjudge pass (z-ai/glm-5.2)audited 2026-07-29Open item page →

Integrable φ and integrable f with φ∘f not integrable: the order of the hypotheses in the composition theorem cannot be reversed

Statement refuted

False claim: if f:[a,b]→[m,M] is Riemann integrable and φ:[m,M]→R is Riemann integrable, then φ∘f is Riemann integrable on [a,b].

That is If f is integrable on [a,b] with values in [m,M] and φ is continuous on [m,M], then φ∘f is integrable with "continuous" weakened to "integrable" on the outer function, and it is false. On [0,1] take f:=t, Thomae's function (The Dirichlet function 1Q, and Thomae's function t with t(x)=1/q at a rational x=p/q in lowest terms with q≥1 and t(x)=0 at every irrational x), whose values lie in [0,1], and

φ:[0,1]→R,φ(u)  :=  {0u=0,10<u≤1.

Both are Riemann integrable, t because its discontinuity set is at most countable and φ because it is nondecreasing. But

φ∘t  =  1Qon [0,1],

the Dirichlet function, which is not Riemann integrable.

Exactly one hypothesis of the composition theorem fails, and it is named: φ is not continuous, being discontinuous at 0. The theorem's hypothesis is continuous after integrable; here the outer function is merely integrable, and that is not enough.

Facts & Assumptions

Given: Thomae's function t on [0,1], with t(x)=1/ι(q(x)) at a rational x of least denominator q(x)≥1 and t(x)=0 at an irrational x, and the function φ above.

[L3]

Q is countably infinite and every subset of an at most countable set is at most countable (Q is countably infinite, Every subset of an at most countable set is at most countable, Finite, countably infinite, countable, uncountable).

Counterexample

technique · direct
1.1

t is bounded on [0,1] with values in [0,1], by [L1].

givenL1
1.2

The set of discontinuities of t in [0,1] is Q∩[0,1] by [L2], which is at most countable by [L3].

L2L3
1.3

φ is nondecreasing on [0,1]: for u≤v the only case not giving φ(u)=φ(v) is u=0<v, where φ(u)=0<1=φ(v). So φ is bounded and integrable on [0,1] by [L5].

givenL5L7
2.1

Hence t is Riemann integrable on [0,1] by [L4].

step 1.1step 1.2L4
2.2

For x∈[0,1]: if x is rational then t(x)>0 by [L1], so φ(t(x))=1; if x is irrational then t(x)=0, so φ(t(x))=0. Hence φ∘t agrees with 1Q at every point of [0,1].

step 1.1givenL1
3.1

1Q is bounded on [0,1] and continuous at no point of it by [L2], so its discontinuity set is [0,1], which does not have measure zero by [L6]; therefore φ∘t is not Riemann integrable on [0,1] by [L6].

step 2.2L2L6
4.1

So t and φ are integrable while φ∘t is not, and the claim is false. The hypothesis of If f is integrable on [a,b] with values in [m,M] and φ is continuous on [m,M], then φ∘f is integrable that fails here is the continuity of the outer function: φ is discontinuous at 0, since φ(0)=0 while φ(u)=1 for every u∈(0,1] and every neighbourhood of 0 in [0,1] contains such a u.

step 2.1step 1.3step 3.1L7∎

Remarks

CounterexampleConstruction: AI-adaptedVerification: AI-generatedprecheck passverified 2026-08-10 (gpt-5.6-terra-codex-subscription)Open item page →

A function that is not Riemann integrable although ∣f∣ is

Statement refuted

False claim: if ∣f∣ is Riemann integrable on [a,b] then so is f; that is, the first clause of If f,g are integrable on [a,b] then so are ∣f∣, f2, fg, max⁡(f,g) and min⁡(f,g), and ∣∫abf∣≤∫ab∣f∣ has a converse.

Let 1Q be the Dirichlet function (The Dirichlet function 1Q, and Thomae's function t with t(x)=1/q at a rational x=p/q in lowest terms with q≥1 and t(x)=0 at every irrational x) and put

f:[0,1]→R,f(x)  :=  2 1Q(x)−1  =  {1x rational,−1x irrational.

Then ∣f∣ is the constant function 1, integrable with ∫01∣f∣=1, while f is not Riemann integrable on [0,1]: every lower Darboux sum of f is −1 and every upper Darboux sum is 1, so the lower and upper integrals are −1 and 1.

Facts & Assumptions

Given: The function f=21Q−1 on [0,1], and a partition P=(n,t) of [0,1].

[L2]

Both Q and the irrationals are dense in R, so every nonempty open interval contains a rational and an irrational (Both Q and R∖Q are dense in R, and every nonempty open subset of R is uncountable).

[L4]

L(u,P)=∑i<nmiΔi and U(u,P)=∑i<nMiΔi with mi=inf⁡u[Ii] and Mi=sup⁡u[Ii]; a set with a least element has it as its infimum and with a greatest element has it as its supremum (For bounded f on [a,b] and a partition P: the infimum mi and supremum Mi of f on the i-th subinterval, and the lower and upper Darboux sums L(f,P)=∑imiΔi and U(f,P)=∑iMiΔi, Greatest lower bound (infimum), Maximum and minimum of a set).

[L6]

Finite sums: scaling and ∑i<nλΔi=λ∑i<nΔi (Finite sums and finite products, by recursion, Laws of finite sums and finite products, clause 2).

[L8]

Absolute value and ordered-field arithmetic: ∣1∣=∣−1∣=1, and the order is total (Absolute value in an ordered field, Ordered field, Complete ordered field (least-upper-bound property)).

Counterexample

technique · direct
1.1

∣f∣ is the constant function 1 on [0,1]: at a rational x, f(x)=2⋅1−1=1, and at an irrational x, f(x)=2⋅0−1=−1, and ∣1∣=∣−1∣=1 by [L1] and [L8]. Hence ∣f∣ is integrable with ∫01∣f∣=1 by [L7].

givenL1L7L8
1.2

Let P=(n,t) be any partition of [0,1] and let i<n. The open interval (ti,ti+1) is nonempty by [L3], so it contains a rational and an irrational by [L2]; both lie in Ii, so 1∈f[Ii] and −1∈f[Ii].

givenL2L3
2.1

f is bounded, with values in {−1,1}, so its Darboux sums are defined by [L4] and [L5].

step 1.1givenL4L5
3.1

Since f[Ii]⊆{−1,1} and both values occur, mi=−1 and Mi=1 by [L4].

step 2.1step 1.2L4
4.1

Hence L(f,P)=∑i<n(−1)Δi=−1 and U(f,P)=∑i<n1⋅Δi=1, by [L4], [L6] and [L3].

step 3.1L3L4L6
5.1

That holds for every partition P, so the set of lower sums is {−1} and the set of upper sums is {1}; by [L5], ∫01‾f=−1≠1=∫01‾f and f is not integrable.

step 4.1L5
6.1

So ∣f∣ is integrable on [0,1] while f is not, and the claim is false.

step 1.1step 5.1∎

Remarks

CounterexampleConstruction: AI-adaptedVerification: AI-generatedprecheck passjudge pass (z-ai/glm-5.2)audited 2026-07-29Open item page →

Continuous fn→0 pointwise on [0,1] with ∫01fn=1 for every n

Statement refuted

False claim: if (fn) is a sequence of Riemann integrable functions on [0,1] converging pointwise to f, and f is integrable, then ∫01fn→∫01f.

For n∈N write cn:=ι(n+1)≥1 (The canonical natural ι(n)=n⋅1F of a field, Canonical naturals are positive and strictly increasing) and define the tent fn:[0,1]→R by

fn(x)  :=  {4cn2 x0≤x≤12cn,4cn2(1cn−x)12cn≤x≤1cn,01cn≤x≤1.

Each fn is continuous on [0,1], hence integrable, with

∫01fn  =  1for every n,

while fn(x)→0 for every x∈[0,1]. So the pointwise limit is the zero function, whose integral is 0, and the integrals do not converge to it.

The heights are unbounded: fn attains the value 2cn at x=1/(2cn), and 2cn→∞. That is what the example refutes and what it does not: it refutes the interchange for pointwise convergence, and it says nothing whatever about sequences that are uniformly bounded, for which no theorem is stated on this page in any direction.

Facts & Assumptions

Given: For n∈N, cn=ι(n+1) and the function fn above; a point x∈[0,1] and a real ε>0.

[L6]

For n≥1 the map x↦xn has derivative ι(n)x n−1; sums and scalar multiples of differentiable functions are differentiable with the corresponding derivatives; and ∫pqc=c(q−p) for a constant (For a natural n≥1 the function x↦xn is differentiable everywhere with derivative ι(n) x n−1; for n=0 it is the constant 1, with derivative 0; for a natural n≥1 the function x↦x−n is differentiable at every x≠0 with derivative −ι(n) x−n−1; consequently every polynomial function is differentiable at every real, with the derivative computed term by term, Sums, scalar multiples, products and quotients: (f+g)′(c)=f′(c)+g′(c), (αf)′(c)=αf′(c), (fg)′(c)=f′(c)g(c)+f(c)g′(c), and (f/g)′(c)=(f′(c)g(c)−f(c)g′(c))/g(c)2 when g(c)≠0, If m≤f≤M on [a,b] then m(b−a)≤L(f,P)≤∫ab‾f≤∫ab‾f≤U(f,P)≤M(b−a) for every partition P; in particular every constant function is integrable, with ∫abc=c(b−a), The derivative f′(c)=lim⁡x→cf(x)−f(c)x−c of f:A→R at a point c∈A that is a limit point of A, and differentiability on a set, Integer powers am).

[L7]

A sequence of reals converges to L when for every real ε>0 there is N with ∣an−L∣<ε for all n≥N (Limits and Cauchy sequences of reals, Sequences of reals: bounded, eventually, frequently, tails, subsequences).

Counterexample

technique · direct
1.1

cn≥1 by [L1], so 0<1/(2cn)<1/cn≤1 and the three pieces of the definition subdivide [0,1].

givenL1
2.1

The three formulas agree at the shared endpoints: at x=1/(2cn) both give 4cn2/(2cn)=2cn, and at x=1/cn the second gives 0, which is the third. So fn is a well-defined function and is continuous on [0,1] by [L2], hence integrable by [L3].

step 1.1L2L3
2.2

On [1/cn,1] the function fn is constantly 0, so ∫1/cn1fn=0 by [L6]; when 1/cn=1 this piece is degenerate and the integral is 0 by [L4].

step 1.1L4L6
2.3

Pointwise convergence to 0. At x=0 every fn(0)=0. For x>0, [L1] gives N with 1/ι(N+1)<x, and for n≥N one has cn=ι(n+1)≥ι(N+1), hence 1/cn≤1/ι(N+1)<x, so x lies in the third piece and fn(x)=0.

step 1.1givenL1L8
3.1

On [0,1/(2cn)] the function H(x):=2cn2x2 has H′(x)=4cn2x=fn(x) by [L6], so ∫01/(2cn)fn=H(1/(2cn))−H(0)=2cn2/(4cn2)=1/2 by [L5].

step 2.1L5L6
3.2

On [1/(2cn),1/cn] the function H2(x):=4cn2(x/cn−x2/2) has H2′(x)=4cn2(1/cn−x)=fn(x) by [L6], and H2(1/cn)=4cn2(1/cn2−1/(2cn2))=2 while H2(1/(2cn))=4cn2(1/(2cn2)−1/(8cn2))=3/2; so ∫1/(2cn)1/cnfn=2−3/2=1/2 by [L5].

step 2.1L5L6
3.3

Hence ∣fn(x)−0∣=0<ε for all n≥N, so fn(x)→0 for every x∈[0,1] by [L7].

step 2.3L7
4.1

By [L4] applied twice, ∫01fn=1/2+1/2+0=1 for every n.

step 3.1step 3.2step 2.2L4
5.1

The pointwise limit is the zero function, which is integrable with integral 0 by [L6], while ∫01fn=1 for every n by step 4.1; so the integrals do not converge to the integral of the limit and the claim is false.

step 4.1step 3.3L6L7∎

Remarks

  • What this refutes, stated exactly. It refutes the interchange of a limit with an integral under pointwise convergence alone, even when every fn is continuous and the limit function is as regular as possible. It does not refute, and does not address, any statement about uniformly convergent sequences or about uniformly bounded ones; no such statement is proved on this page, and none is contradicted here.

  • Unboundedness of the heights is essential to the construction and is stated as a feature, not hidden. sup⁡[0,1]fn=2cn grows without bound, and the mass 1 escapes into a spike of shrinking width. A reader who wants a theorem in this direction should note that the hypothesis to look for is a bound on the whole sequence, and that whichever theorem supplies it is not on this page.

  • The integral of the limit exists here. The failure is not that the limit function is non-integrable — it is the zero function — but that the numbers ∫01fn simply do not converge to ∫010. A separate failure, in which the pointwise limit of integrable functions is not integrable at all, is recorded as a false statement on the companion page of The Riemann Integral.

CounterexampleConstruction: AI-generatedVerification: AI-generatedprecheck passjudge pass (z-ai/glm-5.2)audited 2026-07-29Open item page →

Continuous f and integrable sign-changing g with ∫abfg≠f(ξ)∫abg for every ξ

Statement refuted

False claim: if f is continuous on [a,b] and g is integrable on [a,b], then there is ξ∈[a,b] with ∫abfg=f(ξ)∫abg.

That is If f is continuous on [a,b] and g is integrable with g≥0, there is ξ∈[a,b] with ∫abfg=f(ξ)∫abg with the hypothesis g≥0 deleted, and it is false. On [−1,1] take

f(t)  =  t,g(t)  =  t.

Both are continuous, hence integrable, and g changes sign. Then

∫−11fg  =  ∫−11t2 dt  =  2ι(3)  >  0,∫−11g  =  ∫−11t dt  =  0,

so f(ξ)∫−11g=0 for every ξ∈[−1,1], while the left-hand side is positive. No ξ works.

Facts & Assumptions

Given: The functions f(t)=g(t)=t on [−1,1], and ξ∈[−1,1] arbitrary.

[L5]

Powers and canonical naturals: 12=1, (−1)2=1, 13=1, (−1)3=−1, ι(2)=2>0 and ι(3)=3>0 (Integer powers am, The canonical natural ι(n)=n⋅1F of a field, Canonical naturals are positive and strictly increasing, Ordered field).

Counterexample

technique · direct
1.1

f and g are continuous on [−1,1], hence integrable there by [L1], and fg, the function t↦t2, is integrable by [L1] or [L2].

givenL1L2
1.2

The function H1(t):=t3/ι(3) is differentiable at every real with H1′(t)=ι(3)t2/ι(3)=t2, by [L3] and [L5].

L3L5construct
1.3

The function H2(t):=t2/ι(2) is differentiable at every real with H2′(t)=ι(2) t/ι(2)=t, by [L3] and [L5].

L3L5construct
2.1

By [L4] applied to H1 on [−1,1], ∫−11t2 dt=H1(1)−H1(−1)=1/ι(3)−(−1)/ι(3)=2/ι(3), a positive real by [L5].

step 1.1step 1.2L4L5
2.2

By [L4] applied to H2 on [−1,1], ∫−11t dt=H2(1)−H2(−1)=1/ι(2)−1/ι(2)=0.

step 1.1step 1.3L4L5
3.1

For every ξ∈[−1,1], f(ξ)∫−11g=ξ⋅0=0 by step 2.2 and [L6], while ∫−11fg=2/ι(3)>0 by step 2.1.

step 2.1step 2.2L5L6
4.1

Hence ∫−11fg≠f(ξ)∫−11g for every ξ∈[−1,1], and the claim fails at this pair.

step 3.1L5∎

Remarks

False statementConstruction: AI-adaptedVerification: AI-generatedprecheck passaudited 2026-07-29verified 2026-09-09 (gpt-6-astra)Open item page →

FALSE: if u and v are differentiable on [a,b] then ∫abuv′=u(b)v(b)−u(a)v(a)−∫abu′v

Statement

False claim: if u,v:[a,b]→R are differentiable at every point of [a,b] (The derivative f′(c)=lim⁡x→cf(x)−f(c)x−c of f:A→R at a point c∈A that is a limit point of A, and differentiability on a set), then

∫abu v′  =  u(b)v(b)−u(a)v(a)  −  ∫abu′ v.

That is If u,v are differentiable on [a,b] with u′,v′ integrable, then ∫abuv′=u(b)v(b)−u(a)v(a)−∫abu′v with the hypothesis "u′ and v′ are integrable" deleted, and it is false.

The falsity is undefinedness, not a wrong number. Take [a,b]=[0,1], let u:=G be the everywhere-differentiable function of A function differentiable on [0,1] whose derivative is unbounded, hence not Riemann integrable, and let v(x):=x. Then u and v are differentiable at every point of [0,1], and uv′=G is continuous hence integrable, so the left-hand side exists. But u′v is the function x↦x G′(x), which is unbounded on [0,1], hence has no Darboux sums at all (For bounded f on [a,b] and a partition P: the infimum mi and supremum Mi of f on the i-th subinterval, and the lower and upper Darboux sums L(f,P)=∑imiΔi and U(f,P)=∑iMiΔi) and is not Riemann integrable: the symbol ∫01u′v on the right-hand side does not denote. An equation one of whose sides is undefined is not a true equation.

The correct hypothesis, and when it is automatic. If u,v are differentiable on [a,b] with u′,v′ integrable, then ∫abuv′=u(b)v(b)−u(a)v(a)−∫abu′v asks that u′ and v′ be integrable, which is what makes (uv)′=u′v+uv′ integrable and lets the second fundamental theorem be applied to uv. It holds automatically when u and v are continuously differentiable, since a continuous function on [a,b] is integrable (A continuous function on [a,b] is Riemann integrable, by Heine-Cantor and Riemann's criterion).

Facts & Assumptions

Given: The function G:[0,1]→R of A function differentiable on [0,1] whose derivative is unbounded, hence not Riemann integrable, differentiable at every point of [0,1], together with the points un:=αn+14hn of that item, where αn=1/ι(n+2), and v(x):=x on [0,1].

[A1]

The false claim above.

[L1]

G is differentiable at every point of [0,1], G′ is unbounded there, and G′(un)=316 ι(n+2)2 with un>αn=1/ι(n+2) (A function differentiable on [0,1] whose derivative is unbounded, hence not Riemann integrable).

[L5]
[L6]

Ordered-field arithmetic: multiplying an inequality by a positive real preserves it, the order is total and transitive, and a positive real has a positive inverse (Ordered field, Complete ordered field (least-upper-bound property), Intervals of R: the nine order-convex forms, nondegeneracy, and length).

Refutation

technique · direct
1.1

u:=G and v are differentiable at every point of [0,1], by [L1] and [L2]; so the hypothesis of [A1] is satisfied by this pair.

givenA1L1L2
2.1

uv′=G⋅1=G, which is continuous on [0,1] by [L3] and therefore integrable there; so the left-hand side of [A1] exists.

step 1.1L2L3
2.2

u′v is the function x↦x G′(x) on [0,1]. At the point un its value is un G′(un)>αn⋅316ι(n+2)2=316 ι(n+2), using αn=1/ι(n+2)>0 and [L1].

step 1.1L1L5L6
3.1

Given a real M≥0, [L5] supplies n with 163M<ι(n+2), so unG′(un)>M by step 2.2; hence u′v is unbounded on [0,1].

step 2.2L5L6
4.1

By [L4] the function u′v has no Darboux sums and is not Riemann integrable on [0,1], so the symbol ∫01u′v appearing in [A1] does not denote a real number.

step 3.1L4
5.1

So [A1] fails at this pair: its left-hand side is defined by step 2.1 and its right-hand side is not, by step 4.1, and the asserted identity is therefore not a true statement about them.

step 2.1step 4.1A1∎

Remarks

False statementConstruction: Literature-sourcedVerification: AI-generatedprecheck passverified 2026-08-10 (gpt-5.6-terra-codex-subscription)Open item page →

FALSE: in the substitution theorem the continuity of f may be weakened to integrability, f∘φ still being integrable

Statement

False claim: let c<d be reals, let φ:[c,d]→R be differentiable at every point of [c,d] with φ′ integrable, and let f be Riemann integrable on an interval J containing φ[ [c,d] ]. Then f∘φ is Riemann integrable on [c,d] — so the hypothesis "f is continuous" in Substitution: if φ is differentiable on [c,d] with φ′ integrable and f is continuous on an interval containing φ([c,d]), then ∫φ(c)φ(d)f=∫cd(f∘φ) φ′ may be weakened to "f is integrable" without the right-hand side ∫cd(f∘φ)φ′ losing its meaning.

The claim is false. Let S⊆[0,1] be the Smith-Volterra-Cantor set (The Smith-Volterra-Cantor set: the same construction removing, at stage n≥1, an open middle interval of length 4−n from each of the 2n−1 remaining intervals), which is compact, nowhere dense and not null (The Smith-Volterra-Cantor set is compact, perfect and nowhere dense, and does not have measure zero), and let

dS(u)  :=  inf⁡{ ∣u−s∣ : s∈S },φ(x)  :=  ∫0xdS(x∈[0,1]).

Then φ is differentiable at every point of [0,1] with φ′=dS continuous, hence integrable; φ is strictly increasing; and φ[S] has measure zero. Taking

f  :=  1φ[S]on J:=[0, φ(1)]

gives an integrable f, because its discontinuity set is contained in the null closed set φ[S], while

f∘φ  =  1Son [0,1],

whose discontinuity set is exactly S and is not null. So f∘φ is not Riemann integrable.

What this does and does not show. It shows that continuity of f in Substitution: if φ is differentiable on [c,d] with φ′ integrable and f is continuous on an interval containing φ([c,d]), then ∫φ(c)φ(d)f=∫cd(f∘φ) φ′ cannot simply be weakened to integrability: the composite in the right-hand side need not be integrable. It does not exhibit a pair for which both sides of the substitution identity exist and differ, and no such pair is claimed here.

Facts & Assumptions

Given: The Smith-Volterra-Cantor set S⊆[0,1], the function dS and the function φ above, a real ε>0 and a natural number N≥1.

[L6]

A continuous w≥0 on [p,q] with p<q and ∫pqw=0 vanishes identically (A continuous f≥0 on [a,b] with ∫abf=0 is identically 0).

[L8]

The continuous image of a compact set is compact, and a compact subset of R is closed and bounded (The image of a compact subset of R under a continuous real function is compact, A subset of R is compact if and only if it is closed and bounded).

[L10]

Finite sums: monotonicity in the terms and ∑i<Nλ=ι(N)λ (Finite sums and finite products, by recursion, Laws of finite sums and finite products, clauses 2 and 4); ι(N)≥1>0 for N≥1, and for every real η>0 there is N≥1 with 1/ι(N)<η (The canonical natural ι(n)=n⋅1F of a field, Canonical naturals are positive and strictly increasing, For every ε>0 in a complete ordered field there is a natural n≥1 with 1/n<ε, Every complete ordered field is Archimedean).

[L11]

Ordered-field arithmetic and suprema: a nonempty bounded set has a supremum and an infimum; min⁡{s,t} is at most the average of s and t when s+t is fixed; multiplying inequalities by positive reals preserves them; the order is total and transitive (Complete ordered field (least-upper-bound property), Ordered field, Greatest lower bound (infimum), Intervals of R: the nine order-convex forms, nondegeneracy, and length, Injection, surjection, bijection).

Refutation

technique · direct
1.1

dS≥0 everywhere, dS(u)=0 for u∈S, and dS(u)>0 for u∉S: S is closed by [L1], so some ρ>0 has (u−ρ,u+ρ)∩S=∅, whence ∣u−s∣≥ρ for every s∈S and dS(u)≥ρ.

givenL1L3L11
2.1

dS is continuous by [L3], hence integrable on every [0,x] with x>0 by [L4]; so φ is defined on [0,1], and by [L5] it is differentiable at every point of [0,1] with φ′=dS, which is integrable by [L4].

step 1.1L3L4L5
3.1

φ is strictly increasing, hence injective. For 0≤x<y≤1, φ(y)−φ(x)=∫xydS≥0 by [L5] and [L7]; if it were 0 then [L6] would force dS≡0 on [x,y], so [x,y]⊆S by step 1.1, contradicting [L2]. Hence φ(x)<φ(y). In particular φ(0)=0<φ(1).

step 1.1step 2.1L2L5L6L7
3.2

A quadratic contraction on S. Let x<y both lie in S. For u∈[x,y] one has dS(u)≤min⁡{u−x, y−u}≤(y−x)⋅2−1, since x,y∈S; so by [L5] and [L7], 0≤φ(y)−φ(x)=∫xydS≤(y−x)2⋅2−1.

step 2.1L5L7L11
4.1

φ[S] has content zero. Fix N≥1 and for i<N put Ji:=[ι(i)/ι(N), ι(i+1)/ι(N)], so the Ji cover [0,1] and each has length 1/ι(N). If S∩Ji≠∅ let Ei:=φ[S∩Ji], a nonempty bounded set, and put ai:=inf⁡Ei, bi:=sup⁡Ei; otherwise put ai:=bi:=0.

step 3.2L10L11construct
5.1

For z,w∈Ei one has ∣z−w∣≤(1/ι(N))2⋅2−1 by step 3.2, the two preimages lying in S∩Ji; hence bi≤ai+(1/ι(N))2⋅2−1, since every z∈Ei is at most w+(1/ι(N))22−1 for each fixed w, and then w≥bi−(1/ι(N))22−1 for every w.

step 3.2step 4.1L11
6.1

Every point of φ[S] lies in some [ai,bi], because every point of S lies in some Ji; and ∑i<N(bi−ai)≤ι(N)⋅(1/ι(N))2⋅2−1=1/(2 ι(N)) by [L10].

step 4.1step 5.1L10
7.1

Given ε>0, [L10] supplies N≥1 with 1/(2ι(N))≤ε; so φ[S] has content zero and therefore measure zero by [L9].

step 6.1L9L10
8.1

f:=1φ[S] is integrable on J=[0,φ(1)]. It is bounded, with values in {0,1}. S is compact by [L1] and φ is continuous by [L5] and [L3], so φ[S] is compact, hence closed, by [L8]; therefore at every z∈J∖φ[S] some neighbourhood misses φ[S] and f vanishes on it, so f is continuous there. The discontinuity set of f is thus contained in φ[S], which is null by step 7.1, so f is integrable by [L9].

step 2.1step 7.1L1L3L8L9
9.1

f∘φ=1S on [0,1]. For x∈[0,1]: if x∈S then φ(x)∈φ[S] and f(φ(x))=1; if x∉S then φ(x)∉φ[S], since φ is injective by step 3.1, and f(φ(x))=0. Also φ[ [0,1] ]⊆J by step 3.1.

step 3.1step 8.1
10.1

1S is discontinuous at every point of S. Let x∈S and ρ>0; the set (x−ρ,x+ρ)∩(0,1) contains a nonempty open interval, which by [L2] is not contained in S, so some y in it has 1S(y)=0 while 1S(x)=1; no δ works for ε=2−1. At x∉S the function vanishes on a neighbourhood, S being closed, so it is continuous there.

step 9.1L1L2L11
11.1

The discontinuity set of f∘φ on [0,1] is therefore exactly S, which is not null by [L1]; so f∘φ is bounded and not Riemann integrable, by [L9].

step 9.1step 10.1L1L9
12.1

So φ is differentiable on [0,1] with φ′ integrable, f is integrable on an interval containing φ[ [0,1] ], and f∘φ is not integrable: the claim is false, and the continuity hypothesis on f in Substitution: if φ is differentiable on [c,d] with φ′ integrable and f is continuous on an interval containing φ([c,d]), then ∫φ(c)φ(d)f=∫cd(f∘φ) φ′ cannot be weakened to integrability.

step 2.1step 8.1step 9.1step 11.1∎

Remarks

False statementConstruction: AI-adaptedVerification: AI-generatedprecheck passjudge pass (z-ai/glm-5.2)audited 2026-07-29Open item page →

FALSE: for every integrable f on [a,b], the integral function F(x)=∫axf satisfies F′=f on [a,b]

Statement

False claim: let a<b be reals and let f:[a,b]→R be Riemann integrable (The lower and upper Darboux integrals of a bounded f on [a,b] as sup⁡PL(f,P) and inf⁡PU(f,P), Darboux integrability as their equality, and the notation ∫abf). Then its integral function F(x)=∫axf (The integral function F(x):=∫axf of an integrable f) is differentiable at every point of [a,b] with F′(x)=f(x) there.

The claim fails in two independent ways, and both are exhibited below.

  1. F′ may fail to exist. For the sign function s on [−1,1] (The sign function is Riemann integrable on [−1,1] and has no primitive there) one has F(x)=∣x∣−1, which is not differentiable at 0.
  2. F′ may exist and differ from f. On [0,1] let f(x):=0 for x≠1/2 and f(1/2):=1. Then F is the zero function, so F′(1/2)=0 while f(1/2)=1.

The second witness shows the failure is not exotic: any integrable f that differs from a continuous g at a single point has the same integral function as g, by Changing an integrable function at finitely many points changes neither its integrability nor its integral, and therefore has F′=g≠f at that point. Continuity of f at the point is what the true theorem The first fundamental theorem: if f is integrable on [a,b] and continuous at c, then F′(c)=f(c); in particular a continuous f has F as a primitive asks for, and it asks for nothing more. It is not claimed here to be necessary: what the conclusion needs is the equality F′(c)=f(c), and the last Remark below exhibits an f discontinuous at a point where that equality nevertheless holds.

Facts & Assumptions

Given: The sign function s on [−1,1] of The sign function is Riemann integrable on [−1,1] and has no primitive there, with s(x)=−1 for x<0, s(0)=0 and s(x)=1 for x>0; and the function f on [0,1] with f(1/2)=1 and f(x)=0 otherwise.

[A1]

The false claim: for every integrable u on [p,q], the integral function of u is differentiable everywhere on [p,q] with derivative u.

[L5]

Absolute value: ∣x∣=x for x≥0, ∣x∣=−x for x≤0, and ∣x∣/x is 1 for x>0 and −1 for x<0 (Absolute value in an ordered field, Basic properties of the absolute value).

Refutation

technique · direct
1.1

First witness. s is integrable on [−1,1] by [L1], and its integral function is F(x)=∫−1xs.

givenL1
1.2

Second witness. The function f on [0,1] is bounded and agrees with the constant 0 off the single point 1/2, so it is integrable with ∫0xf=∫0x0=0 for every x∈[0,1] by [L2] and [L3]; hence its integral function is the zero function.

givenL2L3
2.1

For x∈[0,1]: by [L3], F(x)=∫−10s+∫0xs, and s agrees with the constant −1 on [−1,0] off the single point 0 and with the constant 1 on [0,x] off the single point 0, so [L2] gives F(x)=−1+x. For x∈[−1,0]: s agrees with the constant −1 on [x,0] off 0, so ∫−1xs=−(x−(−1))=−x−1 by [L2] and [L3]. In both cases F(x)=∣x∣−1 by [L5].

step 1.1L2L3L5
2.2

The zero function is differentiable everywhere with derivative 0, so its derivative at 1/2 is 0, while f(1/2)=1≠0. Here F′ exists at the point and differs from f there, so [A1] fails again, in a different way.

step 1.2givenL2L7
3.1

The difference quotient of F at 0 is x↦(F(x)−F(0))/x=∣x∣/x, which is 1 for x>0 and −1 for x<0 by [L5]; so its one-sided limits at 0 are 1 and −1.

step 2.1L5
4.1

By [L4] the limit of that quotient at 0 does not exist, so F is not differentiable at 0 and [A1] fails at s: the claim is false.

step 3.1A1L4
5.1

Both failures occur exactly at a discontinuity of the integrand: s is discontinuous at 0 and f at 1/2. Off those points [L6] applies and gives F′=f, so the correct statement is The first fundamental theorem: if f is integrable on [a,b] and continuous at c, then F′(c)=f(c); in particular a continuous f has F as a primitive, whose hypothesis is continuity of the integrand at the point in question.

step 4.1step 2.2L6∎

Remarks

  • The two witnesses are genuinely different failures. In the first, F has no derivative at the bad point at all; in the second, F is as smooth as could be wished and simply computes a different number. A repair attempting to weaken the conclusion to "F is differentiable wherever it can be" would still be refuted by the second witness.

  • What is always true of F is one dimension weaker. For every integrable f the integral function is Lipschitz, hence uniformly continuous (The integral function of a bounded integrable f is Lipschitz, hence uniformly continuous); differentiability is exactly what continuity of the integrand buys, and nothing more is available.

  • The failure set can be much larger than a point. For Thomae's function t on [0,1] the integral function is identically 0, so F′=0 while t is positive at every rational: the claim above then fails at every point of an infinite set, not merely at finitely many. No general statement is made here about an arbitrary integrable f — at a discontinuity where f happens to take the value F′ does, the two agree, and f vanishing off {1/ι(n+1)} is such a case at the point 0.

Sources