Alphabeta Math
TheoremStatement: AI-adaptedProof: AI-adaptedSession-authored (Fable 5 assisted)precheck passjudge pass (z-ai/glm-5.2)audited 2026-07-28
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.

Converse to Froda: for every at most countable ERE \subseteq \mathbb{R} there is a bounded nondecreasing f:RRf : \mathbb{R} \to \mathbb{R} whose set of discontinuities is exactly EE, every one of them a jump

Statement

Let ERE \subseteq \mathbb{R} be at most countable (Finite, countably infinite, countable, uncountable). Then there is a function f:RRf : \mathbb{R} \to \mathbb{R} such that

  1. ff is nondecreasing (Nondecreasing, increasing (strictly increasing), nonincreasing, decreasing, monotone and strictly monotone real functions on a subset of R\mathbb{R}, with the dictionary to monotone sequences) and 0f(x)10 \le f(x) \le 1 for every real xx, so ff is bounded (Lower bound, bounded below, bounded set);
  2. ff is continuous at every xEx \notin E and discontinuous at every xEx \in E (Continuity of f:ARf : A \to \mathbb{R} at a point of AA and on AA: the ε\varepsilon-δ\delta condition, its agreement with limxcf(x)=f(c)\lim_{x \to c} f(x) = f(c) at a limit point, and continuity at an isolated point), so the discontinuity set of ff is exactly EE;
  3. every discontinuity of ff is a jump (Discontinuity of ff at a point of its domain, and its classification: removable discontinuity, jump discontinuity and essential discontinuity, equivalently Rudin's discontinuities of the first and of the second kind), with limxcf(x)=f(c)<limxc+f(x)\lim_{x \to c^{-}} f(x) = f(c) < \lim_{x \to c^{+}} f(x) at every cEc \in E.

Together with Froda's theorem: the set of discontinuities of a monotone function on an interval is at most countable, the injection into N\mathbb{N} being built from one fixed enumeration of the rationals by least index, so no choice principle is used this settles the question completely: the sets that occur as discontinuity sets of monotone functions on R\mathbb{R} are exactly the at most countable ones.

The construction. For E=E = \varnothing take f:=0f := 0. Otherwise fix a surjection s:NEs : \mathbb{N} \to E (A nonempty set is at most countable iff it is a surjective image of N\mathbb{N}) and set

f(x)  :=  k=0ak(x),ak(x):={1/2k+1if s(k)<x,0otherwise,f(x) \;:=\; \sum_{k=0}^{\infty} a_{k}(x), \qquad a_{k}(x) := \begin{cases} 1/2^{\,k+1} & \text{if } s(k) < x,\\ 0 & \text{otherwise,}\end{cases}

(Series, partial sums, convergence and the sum, divergence, and the tail series, Integer powers ama^m): the mass 1/2k+11/2^{\,k+1} is placed at the point s(k)s(k) and is collected by ff strictly to the right of it. Repetitions in the enumeration are harmless; they only make the jump at a point larger.

Facts & Assumptions

Given: An at most countable ERE \subseteq \mathbb{R}.

[L1]

A nonempty at most countable set is the image of a surjection s:NEs : \mathbb{N} \to E (A nonempty set is at most countable iff it is a surjective image of N\mathbb{N}, Finite, countably infinite, countable, uncountable).

[L2]

A series of nonnegative terms converges if and only if its partial sums are bounded above, and its sum is then the supremum of its partial sums; in particular every partial sum is at most the sum (A series of nonnegative terms converges iff its partial sums are bounded, and then the sum is their supremum, Series, partial sums, convergence and the sum, divergence, and the tail series, Lower bound, bounded below, bounded set).

[L3]

Finite sums: k<n\sum_{k<n} is monotone in the terms, splits as k<n=k<m+k=mn1\sum_{k<n} = \sum_{k<m} + \sum_{k=m}^{n-1} for mnm \le n, scales, and telescopes as k<n(ck+1ck)=cnc0\sum_{k<n}(c_{k+1} - c_{k}) = c_{n} - c_{0} (Laws of finite sums and finite products, Finite sums and finite products, by recursion).

[L4]

k0rk\sum_{k \ge 0} r^{k} converges to 1/(1r)1/(1-r) for r<1|r| < 1, the first term being r0=1r^{0} = 1 (For r<1|r| < 1, k0rk=1/(1r)\sum_{k \ge 0} r^k = 1/(1-r), and for r1|r| \ge 1 the series diverges, Integer powers ama^m); a series converges if and only if each of its tails does, and k0uk=k<Nuk+kNuk\sum_{k \ge 0} u_{k} = \sum_{k<N} u_{k} + \sum_{k \ge N} u_{k} (A series converges iff each of its tail series converges, and the sum splits as sNs_N plus the NN-th tail); a convergent sequence of reals comes within every positive ε\varepsilon of its limit from some index on (Limits and Cauchy sequences of reals).

[L5]

A nonempty finite set of reals, presented as {c0,,cm}\{c_{0}, \dots, c_{m}\}, has a maximum and a minimum (Every nonempty finite set of reals has a maximum and a minimum, Maximum and minimum of a set); and strictly between any two distinct reals there lies a real (The rationals embed densely in the reals).

Proof

technique · constructive
1.1

If E=E = \varnothing, the constant function 00 is nondecreasing, takes values in [0,1][0,1], is continuous at every real, and has empty discontinuity set; all three claims hold vacuously for claim 3. Assume from here on that EE \ne \varnothing and fix a surjection s:NEs : \mathbb{N} \to E.

L1construct
1.2

For every nNn \in \mathbb{N}, k<n1/2k+1=11/2n\sum_{k<n} 1/2^{\,k+1} = 1 - 1/2^{\,n}: each term is 1/2k+1=1/2k1/2k+11/2^{\,k+1} = 1/2^{\,k} - 1/2^{\,k+1}, so the sum telescopes to 1/201/2n=11/2n1/2^{\,0} - 1/2^{\,n} = 1 - 1/2^{\,n}.

L3
2.1

Define ak(x):=1/2k+1a_{k}(x) := 1/2^{\,k+1} when s(k)<xs(k) < x and ak(x):=0a_{k}(x) := 0 otherwise, and note 0ak(x)1/2k+10 \le a_{k}(x) \le 1/2^{\,k+1} for every kk and every real xx.

step 1.1construct
2.2

For every real ε>0\varepsilon > 0 there is nNn \in \mathbb{N} with 1/2n<ε1/2^{\,n} < \varepsilon: the partial sums tn:=k<n1/2kt_{n} := \sum_{k<n} 1/2^{\,k} converge to 22, and tn=22/2nt_{n} = 2 - 2/2^{\,n} by the same telescoping as in step 1.2, so tn2=2/2n<ε|t_{n} - 2| = 2/2^{\,n} < \varepsilon for all large nn, whence 1/2n<ε/2<ε1/2^{\,n} < \varepsilon/2 < \varepsilon for those nn. Consequently the partial sums 11/2n1 - 1/2^{\,n} of k1/2k+1\sum_{k} 1/2^{\,k+1} have supremum 11, so that series converges with sum 11.

step 1.2L2L3L4
3.1

For every real xx the series kak(x)\sum_{k} a_{k}(x) converges and 0f(x)10 \le f(x) \le 1: its terms are nonnegative and its partial sums satisfy k<nak(x)k<n1/2k+1=11/2n1\sum_{k<n} a_{k}(x) \le \sum_{k<n} 1/2^{\,k+1} = 1 - 1/2^{\,n} \le 1, so they are bounded above by 11 and the sum, being their supremum, lies in [0,1][0,1].

step 2.1step 1.2L2L3
3.2

Left continuity holds at every real cc: given real ε>0\varepsilon > 0 take nn with 1/2n<ε1/2^{\,n} < \varepsilon; let F:={k<n:s(k)<c}F := \{\, k < n : s(k) < c \,\}; if F=F = \varnothing put x0:=c1x_{0} := c - 1, and otherwise put x0x_{0} to be a real with max{s(k):kF}<x0<c\max\{s(k) : k \in F\} < x_{0} < c, which exists because the maximum of the nonempty finite set {s(k):kF}\{s(k) : k \in F\} is a real strictly below cc.

step 2.2L5
3.3

Right continuity holds at every cEc \notin E: given real ε>0\varepsilon > 0 take nn with 1/2n<ε1/2^{\,n} < \varepsilon; since cEc \notin E and ss has image EE, no kk has s(k)=cs(k) = c, so every k<nk < n has s(k)<cs(k) < c or s(k)>cs(k) > c. Let G:={k<n:s(k)>c}G := \{\, k < n : s(k) > c \,\}; if G=G = \varnothing put y0:=c+1y_{0} := c + 1, and otherwise put y0y_{0} to be a real with c<y0<min{s(k):kG}c < y_{0} < \min\{s(k) : k \in G\}.

step 1.1step 2.2L5
4.1

ff is nondecreasing: if xyx \le y then s(k)<xs(k) < x implies s(k)<ys(k) < y, so ak(x)ak(y)a_{k}(x) \le a_{k}(y) for every kk, hence k<nak(x)k<nak(y)\sum_{k<n} a_{k}(x) \le \sum_{k<n} a_{k}(y) for every nn, and taking suprema gives f(x)f(y)f(x) \le f(y).

step 2.1step 3.1L2L3
4.2

For all reals xyx \le y and every nNn \in \mathbb{N} with ak(x)=ak(y)a_{k}(x) = a_{k}(y) for every k<nk < n, one has f(y)f(x)1/2nf(y) - f(x) \le 1/2^{\,n}: for NnN \ge n the splitting k<Nak(y)=k<nak(y)+k=nN1ak(y)k<nak(x)+k=nN11/2k+1\sum_{k<N} a_{k}(y) = \sum_{k<n} a_{k}(y) + \sum_{k=n}^{N-1} a_{k}(y) \le \sum_{k<n} a_{k}(x) + \sum_{k=n}^{N-1} 1/2^{\,k+1} holds, the last sum being at most kn1/2k+1=1(11/2n)=1/2n\sum_{k \ge n} 1/2^{\,k+1} = 1 - (1 - 1/2^{\,n}) = 1/2^{\,n}; so every partial sum of kak(y)\sum_{k} a_{k}(y) is at most f(x)+1/2nf(x) + 1/2^{\,n}, and so is their supremum f(y)f(y).

step 2.1step 1.2step 3.1L2L3L4
4.3

Let cEc \in E and fix k0k_{0} with s(k0)=cs(k_{0}) = c. For every y>cy > c and every N>k0N > k_{0} the finite sum k<Nak(y)\sum_{k<N} a_{k}(y) exceeds k<Nak(c)\sum_{k<N} a_{k}(c) by at least 1/2k0+11/2^{\,k_{0}+1}, because the list kak(y)ak(c)k \mapsto a_{k}(y) - a_{k}(c) has nonnegative entries, so the finite sum of its first NN entries is at least its entry at the index k0k_{0}, which is ak0(y)ak0(c)=1/2k0+10a_{k_{0}}(y) - a_{k_{0}}(c) = 1/2^{\,k_{0}+1} - 0. Hence f(y)1/2k0+1k<Nak(c)f(y) - 1/2^{\,k_{0}+1} \ge \sum_{k<N} a_{k}(c) for every NN, the case Nk0N \le k_{0} holding because the partial sums of a nonnegative series are nondecreasing; so f(y)1/2k0+1f(y) - 1/2^{\,k_{0}+1} is an upper bound of those partial sums and therefore at least their supremum f(c)f(c).

step 1.1step 2.1step 3.1L2L3
5.1

With x0x_{0} as in step 3.2 and any xx with x0<xcx_{0} < x \le c: for k<nk < n with s(k)<cs(k) < c we have s(k)max{s(j):jF}<x0<xs(k) \le \max\{s(j) : j \in F\} < x_{0} < x, so ak(x)=1/2k+1=ak(c)a_{k}(x) = 1/2^{\,k+1} = a_{k}(c); and for k<nk < n with s(k)cxs(k) \ge c \ge x we have ak(x)=0=ak(c)a_{k}(x) = 0 = a_{k}(c). So ak(x)=ak(c)a_{k}(x) = a_{k}(c) for every k<nk < n, and step 4.2 applied to the pair xcx \le c gives 0f(c)f(x)1/2n<ε0 \le f(c) - f(x) \le 1/2^{\,n} < \varepsilon.

step 2.1step 4.1step 4.2step 3.2
5.2

With y0y_{0} as in step 3.3 and any yy with cy<y0c \le y < y_{0}: for k<nk < n with s(k)<cys(k) < c \le y we get ak(y)=1/2k+1=ak(c)a_{k}(y) = 1/2^{\,k+1} = a_{k}(c), and for k<nk < n with s(k)>cs(k) > c we have s(k)min{s(j):jG}>y0>ys(k) \ge \min\{s(j) : j \in G\} > y_{0} > y, so ak(y)=0=ak(c)a_{k}(y) = 0 = a_{k}(c). So ak(y)=ak(c)a_{k}(y) = a_{k}(c) for every k<nk < n, and step 4.2 applied to the pair cyc \le y gives 0f(y)f(c)1/2n<ε0 \le f(y) - f(c) \le 1/2^{\,n} < \varepsilon.

step 2.1step 4.1step 4.2step 3.3
5.3

So ff is discontinuous at cc: for ε:=1/2k0+1>0\varepsilon := 1/2^{\,k_{0}+1} > 0 and any real δ>0\delta > 0 the point y:=c+δ/2y := c + \delta/2 satisfies yc<δ|y - c| < \delta and f(y)f(c)ε|f(y) - f(c)| \ge \varepsilon, so no δ\delta witnesses the continuity condition at cc.

step 4.3
6.1

Hence ff is continuous at every cEc \notin E: fix a real ε>0\varepsilon > 0, take x0x_{0} as in step 3.2 and y0y_{0} as in step 3.3 for that same ε\varepsilon, and put δ:=min{cx0,y0c}>0\delta := \min\{c - x_{0}, y_{0} - c\} > 0; then every real xx with xc<δ|x - c| < \delta satisfies x0<x<y0x_{0} < x < y_{0} and therefore f(x)f(c)<ε|f(x) - f(c)| < \varepsilon, by step 5.1 when xcx \le c and by step 5.2 when xcx \ge c.

step 5.1step 5.2L5
6.2

Every point of EE is an interior point of the order-convex set R\mathbb{R}, so both one-sided limits of ff exist there; step 5.1 gives limxcf(x)=f(c)\lim_{x \to c^{-}} f(x) = f(c) and step 4.3 gives limxc+f(x)f(c)+1/2k0+1>f(c)\lim_{x \to c^{+}} f(x) \ge f(c) + 1/2^{\,k_{0}+1} > f(c). The two one-sided limits therefore differ, and the discontinuity at cc is a jump.

step 5.1step 4.3step 5.3L6
7.1

Claims 1, 2 and 3 hold for the function ff constructed in steps 1.1 and 2.1: claim 1 by steps 3.1 and 4.1, claim 2 by steps 6.1 and 5.3, and claim 3 by step 6.2.

step 3.1step 4.1step 6.1step 5.3step 6.2discharge-construct

Remarks

  • Why the mass is collected strictly to the right. The definition uses s(k)<xs(k) < x rather than s(k)xs(k) \le x, and that is what makes ff left continuous everywhere, as steps 3.2 and 5.1 show without any hypothesis on cc. The value f(c)f(c) at a point of EE is therefore the left limit, and the whole jump sits on the right. Using s(k)xs(k) \le x would produce a right continuous function with the same discontinuity set; nothing else would change.

  • Repetitions in the enumeration are harmless. If ss takes the value cc at several indices, the jump at cc is the total mass {1/2k+1:s(k)=c}\sum \{1/2^{\,k+1} : s(k) = c\} rather than a single term. Step 4.3 uses only one index k0k_{0} and so needs no such sum; it establishes a lower bound for the jump, which is all that discontinuity requires.

  • Boundedness is free, and it is worth recording. The total mass available is k01/2k+1=1\sum_{k \ge 0} 1/2^{\,k+1} = 1, so ff maps R\mathbb{R} into [0,1][0,1] however large EE is. A bounded nondecreasing function on R\mathbb{R} can therefore have a dense set of discontinuities; the companion page takes E=QE = \mathbb{Q} and gets exactly that.

Depends on

Used by

Dependency tree · next 3 levels

Direct dependencies and their dependencies through the next three levels: 144 results over 38 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