Alphabeta Math
DefinitionDefinition: AI-adaptedProof: Not applicableSession-authored (Fable 5 assisted)judge 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.

Partition of [a,b][a,b] as a finite strictly increasing list a=t0<t1<<tn=ba = t_0 < t_1 < \dots < t_n = b, its subintervals and their lengths, its mesh, refinement, and the common refinement of two partitions

Definition

Standing hypothesis for this page. Throughout, R\mathbb{R} is the complete ordered field (Complete ordered field (least-upper-bound property), Ordered field), N\mathbb{N} is the set of natural numbers and contains 00 (The natural numbers N\mathbb{N} (von Neumann), Order on the natural numbers), ι:NR\iota : \mathbb{N} \to \mathbb{R} is the canonical natural (The canonical natural ι(n)=n1F\iota(n) = n \cdot 1_F of a field), and a,ba, b are reals with

a  <  b.a \;<\; b .

Intervals and their lengths are those of Intervals of R\mathbb{R}: the nine order-convex forms, nondegeneracy, and length; finite sums are those of Finite sums and finite products, by recursion, indexed as i<n\sum_{i<n} over iNi \in \mathbb{N}.

Partitions

A partition of [a,b][a,b] is a pair P=(n,t)P = (n, t) consisting of a natural number n1n \ge 1 and a sequence t:NRt : \mathbb{N} \to \mathbb{R} (Sequences of reals: bounded, eventually, frequently, tails, subsequences) with

t0=a,ti<ti+1  for every i<n,tk=b  for every kn.t_0 = a, \qquad t_i < t_{i+1} \ \text{ for every } i < n, \qquad t_k = b \ \text{ for every } k \ge n .

The tail convention on the third clause is bookkeeping only: it makes tt a genuine sequence, so that the finite-sum laws of Laws of finite sums and finite products apply to it verbatim, and it costs nothing because no index above nn is ever read. The first two clauses say exactly that

a  =  t0  <  t1  <    <  tn  =  b,a \;=\; t_0 \;<\; t_1 \;<\; \dots \;<\; t_n \;=\; b ,

the last equality because tn=bt_n = b by the third clause. In particular itii \mapsto t_i is strictly increasing, hence injective, on {iN:in}\{\, i \in \mathbb{N} : i \le n \,\} (Injection, surjection, bijection), and atiba \le t_i \le b for every ini \le n.

The point set of PP is the finite set

pts(P)  :=  {ti : in}    [a,b],a,bpts(P).\operatorname{pts}(P) \;:=\; \{\, t_i \ : \ i \le n \,\} \;\subseteq\; [a,b], \qquad a, b \in \operatorname{pts}(P) .

The subintervals of PP are

Ii  :=  [ti, ti+1](i<n),I_i \;:=\; [\,t_i,\ t_{i+1}\,] \qquad (i < n),

and their lengths are Δi:=ti+1ti\Delta_i := t_{i+1} - t_i. Each Δi>0\Delta_i > 0, so each IiI_i is a nondegenerate closed bounded interval (Intervals of R\mathbb{R}: the nine order-convex forms, nondegeneracy, and length). There are nn of them and they are indexed from i=0i = 0, not from i=1i = 1: the first subinterval is [t0,t1]=[a,t1][t_0, t_1] = [a, t_1].

The lengths sum to bab - a. By the telescoping law, clause 5 of Laws of finite sums and finite products,

i<nΔi  =  i<n(ti+1ti)  =  tnt0  =  ba.\sum_{i<n} \Delta_i \;=\; \sum_{i<n} (t_{i+1} - t_i) \;=\; t_n - t_0 \;=\; b - a .

The mesh. The set {Δi:i<n}\{\, \Delta_i : i < n \,\} is a nonempty finite set of reals, nonempty because n1n \ge 1, so it has a maximum (Every nonempty finite set of reals has a maximum and a minimum, Maximum and minimum of a set). The mesh of PP is

P  :=  max{Δi : i<n}  >  0,\|P\| \;:=\; \max\{\, \Delta_i \ : \ i < n \,\} \;>\; 0 ,

and ΔiP\Delta_i \le \|P\| for every i<ni < n.

The uniform partition. For a natural N1N \ge 1, the uniform partition of [a,b][a,b] into NN parts is UN:=(N,t)U_N := (N, t) with

ti  :=  a+ι(i)baι(N)(iN),tk:=b(kN).t_i \;:=\; a + \iota(i)\,\frac{b-a}{\iota(N)} \quad (i \le N), \qquad t_k := b \quad (k \ge N).

This is a partition: t0=at_0 = a; tN=a+(ba)=bt_N = a + (b-a) = b; and ti+1ti=(ba)/ι(N)>0t_{i+1} - t_i = (b-a)/\iota(N) > 0 for i<Ni < N, because ι(N)>0\iota(N) > 0 and ι(i+1)=ι(i)+1\iota(i+1) = \iota(i) + 1 (Canonical naturals are positive and strictly increasing, The canonical natural ι(n)=n1F\iota(n) = n \cdot 1_F of a field). Its subinterval lengths are all equal to (ba)/ι(N)(b-a)/\iota(N), so

UN  =  baι(N).\|U_N\| \;=\; \frac{b-a}{\iota(N)} .

A partition is determined by its point set

Claim. If P=(n,t)P = (n,t) and P=(n,t)P' = (n', t') are partitions of [a,b][a,b] with pts(P)=pts(P)\operatorname{pts}(P) = \operatorname{pts}(P'), then n=nn = n' and ti=tit_i = t'_i for every ini \le n.

Proof. First, ti=tit_i = t'_i for every imin{n,n}i \le \min\{n,n'\}, by induction on ii (The principle of mathematical induction). For i=0i = 0 both equal aa. Suppose tj=tjt_j = t'_j for all jij \le i and i+1min{n,n}i + 1 \le \min\{n,n'\}. The set S:={xpts(P):x>ti}S := \{\, x \in \operatorname{pts}(P) : x > t_i \,\} has ti+1t_{i+1} as its least element: ti+1St_{i+1} \in S, and any xSx \in S is tjt_j for some jnj \le n with tj>tit_j > t_i, which forces j>ij > i because tt is increasing on indices n\le n, hence ji+1j \ge i+1 and x=tjti+1x = t_j \ge t_{i+1}. The same argument in PP' makes ti+1t'_{i+1} the least element of {xpts(P):x>ti}\{\, x \in \operatorname{pts}(P') : x > t'_i \,\}, which is the same set SS, since the point sets agree and ti=tit'_i = t_i. A set has at most one least element (Maximum and minimum of a set), so ti+1=ti+1t_{i+1} = t'_{i+1}.

Second, n=nn = n'. If n<nn < n' then tn=tn=bt'_n = t_n = b by the previous paragraph, while tn<tn=bt'_n < t'_{n'} = b because tt' is increasing on indices n\le n' and n<nn < n'; that is impossible. Exchanging the roles of PP and PP' rules out n<nn' < n. \square

So the map Ppts(P)P \mapsto \operatorname{pts}(P) is injective, and a partition may be named by its point set whenever one is exhibited.

Inserting a point

Let P=(n,t)P = (n,t) be a partition of [a,b][a,b] and let c[a,b]c \in [a,b]. Define a partition P+cP + c of [a,b][a,b] as follows.

  • If cpts(P)c \in \operatorname{pts}(P), put P+c:=PP + c := P.
  • Otherwise cac \ne a and cbc \ne b, so a<c<ba < c < b. The set T:={ti:in and ti<c}T := \{\, t_i : i \le n \text{ and } t_i < c \,\} is a nonempty finite set of reals, nonempty because t0=a<ct_0 = a < c, so it has a maximum (Every nonempty finite set of reals has a maximum and a minimum); let i0ni_0 \le n be the unique index with ti0=maxTt_{i_0} = \max T, unique because tt is injective on indices n\le n. Then i0<ni_0 < n, since tn=b>ct_n = b > c puts tnTt_n \notin T; and ti0  <  c  <  ti0+1,t_{i_0} \;<\; c \;<\; t_{i_0+1}, the right inequality because ti0+1ct_{i_0+1} \ne c (as cpts(P)c \notin \operatorname{pts}(P)) and ti0+1<ct_{i_0+1} < c would put ti0+1Tt_{i_0+1} \in T with ti0+1>ti0=maxTt_{i_0+1} > t_{i_0} = \max T. Put P+c:=(n+1,s)P + c := (n+1, s) with si:=ti (ii0),si0+1:=c,si:=ti1 (i0+2in+1),sk:=b (kn+1).s_i := t_i \ (i \le i_0), \qquad s_{i_0+1} := c, \qquad s_i := t_{i-1} \ (i_0 + 2 \le i \le n+1), \qquad s_k := b \ (k \ge n+1).

In both cases P+cP + c is a partition of [a,b][a,b] and

pts(P+c)  =  pts(P){c},P+c    P.\operatorname{pts}(P + c) \;=\; \operatorname{pts}(P) \cup \{c\}, \qquad \|P+c\| \;\le\; \|P\| .

The displayed identity is immediate from the two cases. For the mesh: in the first case nothing changes; in the second the list of subinterval lengths of P+cP + c is that of PP with Δi0\Delta_{i_0} replaced by the two numbers cti0c - t_{i_0} and ti0+1ct_{i_0+1} - c, each of which is smaller than Δi0=(cti0)+(ti0+1c)\Delta_{i_0} = (c - t_{i_0}) + (t_{i_0+1} - c) because the other is positive. So every length of P+cP + c is at most a length of PP, and the maximum cannot increase. Finally the index count grows by exactly 11 in the second case and not at all in the first.

Refinement and the common refinement

PP' refines PP, and is a refinement of PP, when

pts(P)    pts(P).\operatorname{pts}(P) \;\subseteq\; \operatorname{pts}(P') .

Let P=(n,t)P = (n,t) and Q=(m,s)Q = (m,s) be partitions of [a,b][a,b]. Applying the recursion theorem (The recursion theorem) to the set N×P[a,b]\mathbb{N} \times \mathcal{P}_{[a,b]}, where P[a,b]\mathcal{P}_{[a,b]} is the set of partitions of [a,b][a,b], with starting element (0,P)(0, P) and the map (j,R)(j+1, R+sj)(j, R) \mapsto (j+1,\ R + s_j) — legitimate because sj[a,b]s_j \in [a,b] for every jNj \in \mathbb{N} — gives a unique family (Rj)jN(R_j)_{j \in \mathbb{N}} of partitions with R0=PR_0 = P and Rj+1=Rj+sjR_{j+1} = R_j + s_j. The common refinement of PP and QQ is

PQ  :=  Rm+1.P \vee Q \;:=\; R_{m+1} .

Its point set is the union. By induction on jj (The principle of mathematical induction), pts(Rj)=pts(P){sl:l<j}\operatorname{pts}(R_j) = \operatorname{pts}(P) \cup \{\, s_l : l < j \,\}; taking j=m+1j = m+1 gives

pts(PQ)  =  pts(P)pts(Q).\operatorname{pts}(P \vee Q) \;=\; \operatorname{pts}(P) \cup \operatorname{pts}(Q).

Hence PQP \vee Q refines both PP and QQ; by the uniqueness claim above it is the only partition with that point set, so PQ=QPP \vee Q = Q \vee P, and

P refines PPP=P,P' \text{ refines } P \quad \Longrightarrow \quad P \vee P' = P' ,

since then pts(P)pts(P)=pts(P)\operatorname{pts}(P) \cup \operatorname{pts}(P') = \operatorname{pts}(P').

Two size bounds, both used later. Writing nRn_R for the first component of a partition RR:

PQ    P,nPQ    nP+nQ1.\|P \vee Q\| \;\le\; \|P\|, \qquad n_{P \vee Q} \;\le\; n_P + n_Q - 1 .

The first is the mesh bound above applied m+1m+1 times. For the second, each insertion raises the index count by at most 11, and the two insertions of s0=as_0 = a and sm=bs_m = b raise it by 00, since aa and bb already lie in pts(P)\operatorname{pts}(P) and hence in pts(Rj)\operatorname{pts}(R_j) for every jj; so at most m1m - 1 of the m+1m+1 insertions increase it.

The index map of a refinement

Let P=(n,t)P' = (n',t') refine P=(n,t)P = (n,t). For each ini \le n the point tit_i lies in pts(P)\operatorname{pts}(P'), so there is exactly one φ(i)n\varphi(i) \le n' with tφ(i)=tit'_{\varphi(i)} = t_i, uniqueness because tt' is injective on indices n\le n'. The resulting map φ\varphi satisfies

φ(0)=0,φ(n)=n,φ(i)<φ(i+1)  (i<n),\varphi(0) = 0, \qquad \varphi(n) = n', \qquad \varphi(i) < \varphi(i+1) \ \ (i < n),

the first two because t0=a=t0t'_0 = a = t_0 and tn=b=tnt'_{n'} = b = t_n together with injectivity, and the third because ti<ti+1t_i < t_{i+1} and tt' is increasing on indices n\le n'. In particular nnn \le n'. Moreover, for i<ni < n and φ(i)j<φ(i+1)\varphi(i) \le j < \varphi(i+1),

Ij  =  [tj, tj+1]    [ti, ti+1]  =  Ii,I'_j \;=\; [\,t'_j,\ t'_{j+1}\,] \;\subseteq\; [\,t_i,\ t_{i+1}\,] \;=\; I_i ,

because ti=tφ(i)tjt_i = t'_{\varphi(i)} \le t'_j and tj+1tφ(i+1)=ti+1t'_{j+1} \le t'_{\varphi(i+1)} = t_{i+1}.

The blocks are counted by telescoping. By clause 5 of Laws of finite sums and finite products, i<n(φ(i+1)φ(i))=φ(n)φ(0)=n\sum_{i<n}\big(\varphi(i+1) - \varphi(i)\big) = \varphi(n) - \varphi(0) = n', so, subtracting i<n1=n\sum_{i<n} 1 = n,

i<n(φ(i+1)φ(i)1)  =  nn,\sum_{i<n} \big(\varphi(i+1) - \varphi(i) - 1\big) \;=\; n' - n ,

a sum of nn nonnegative integers, one for each block, which vanishes exactly at the blocks consisting of a single index. This identity is the whole content of the quantitative bound in Refining a partition raises the lower Darboux sum and lowers the upper one, and every lower sum is at most every upper sum: L(f,P)L(f,P)U(f,P)U(f,P)L(f,P) \le L(f,P') \le U(f,P') \le U(f,P) when PP' refines PP, and L(f,P)U(f,Q)L(f,P) \le U(f,Q) for arbitrary partitions PP and QQ; moreover the two changes are at most 2M(nn)P2M(n' - n)\|P\|, and it is also why nnn \le n'.

Finally, the lengths inside a block sum to the length of the block:

j=φ(i)φ(i+1)1Δj  =  tφ(i+1)tφ(i)  =  ti+1ti  =  Δi(i<n),\sum_{j = \varphi(i)}^{\varphi(i+1)-1} \Delta'_j \;=\; t'_{\varphi(i+1)} - t'_{\varphi(i)} \;=\; t_{i+1} - t_i \;=\; \Delta_i \qquad (i < n),

again by telescoping, applied to the sequence ltφ(i)+ll \mapsto t'_{\varphi(i)+l} and read through the index-shift convention of Finite sums and finite products, by recursion.

Remarks

Depends on

Used by

…and 18 more results.

Dependency tree · next 3 levels

Direct dependencies and their dependencies through the next three levels: 58 results over 15 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