TheoremBase

Four preliminary steps (bounds from covers, an epsilon principle, the lengths ε/2^m, and combining a double sequence of covers through a bijection of ℕ onto ℕ×ℕ) give each clause; the interval lengths use the covering lemma for closed intervals, the split along a ray modifies one cover by maxima and minima, and countable subadditivity chooses covers by countable choice.

Proof

For sequences a=(am)a=(a_{m}) and b=(bm)b=(b_{m}) in R\mathbb{R} with am≤bma_{m}\le b_{m} for every m∈Nm\in\mathbb{N} write ℓ(a,b)=∑m∈N(bm−am)∈[0,∞]\ell(a,b)=\sum_{m\in\mathbb{N}}(b_{m}-a_{m})\in[0,\infty]. For E⊆RE\subseteq\mathbb{R}, covers (a,b)(a,b) of EE, their total lengths ℓ(a,b)\ell(a,b) and the set C(E)\mathcal{C}(E) of these are as in Lebesgue Outer Measure on the Real Line (i), by which (a,b)(a,b) is a cover of EE exactly when am≤bma_{m}\le b_{m} for every m∈Nm\in\mathbb{N} and every x∈Ex\in E lies in (am,bm)(a_{m},b_{m}) for some m∈Nm\in\mathbb{N}; and λ∗(E)=inf⁡C(E)\lambda^{*}(E)=\inf\mathcal{C}(E) by Lebesgue Outer Measure on the Real Line (ii). A real number u≥0u\ge0 is an element of [0,∞][0,\infty]; sums and products in [0,∞][0,\infty] of such numbers are their real sums and products (Addition and Multiplication on [0,∞] (i) and Addition and Multiplication on [0,∞] (ii)), and on elements of R\mathbb{R} (The Real Numbers, with the Natural Numbers, Integers and Rationals Identified with Subsets of the Reals, and Completeness (ii)) the order of [−∞,∞][-\infty,\infty] is that of R\mathbb{R}, a total order on [−∞,∞][-\infty,\infty] (Order and Suprema in the Extended Real Line (i)). These identifications are used throughout.

Step 1: bounds from covers. Let E⊆RE\subseteq\mathbb{R}. As λ∗(E)\lambda^{*}(E) is the greatest lower bound of C(E)\mathcal{C}(E) in [−∞,∞][-\infty,\infty] (Bounds, Least and Greatest Elements, Suprema and Infima for a Partial Order (iv)): (a) λ∗(E)≤ℓ(a,b)\lambda^{*}(E)\le\ell(a,b) for every cover (a,b)(a,b) of EE; (b) if t∈[−∞,∞]t\in[-\infty,\infty] and t≤ℓ(a,b)t\le\ell(a,b) for every cover (a,b)(a,b) of EE, then tt is a lower bound of C(E)\mathcal{C}(E), so t≤λ∗(E)t\le\lambda^{*}(E). (c) If t∈[−∞,∞]t\in[-\infty,\infty] and λ∗(E)<t\lambda^{*}(E)<t, then there is a cover (a,b)(a,b) of EE with ℓ(a,b)<t\ell(a,b)<t, by Order and Suprema in the Extended Real Line (x).

Step 2: an ε\varepsilon-principle. Let x∈[0,∞]x\in[0,\infty] and y∈Ry\in\mathbb{R} with 0≤y0\le y, and suppose x≤y+εx\le y+\varepsilon for every ε∈R\varepsilon\in\mathbb{R} with ε>0\varepsilon>0. Then x≤yx\le y. Indeed, x≤y+1<∞x\le y+1<\infty, so x∈Rx\in\mathbb{R} by Small Facts about [0,∞]: Zero, Finiteness and Sums of Zeros (iii). If y<xy<x, then ε=(x−y)/2>0\varepsilon=(x-y)/2>0 and x≤y+(x−y)/2<y+(x−y)=xx\le y+(x-y)/2<y+(x-y)=x by Inequalities in an Ordered Field: Mixed Transitivity, Strict Sums, Signs, Products, Natural Numbers, Halving, Reciprocals, Absolute Values and Squares (vii), which is impossible; so x≤yx\le y, the order being total.

Step 3: the lengths ε/2m\varepsilon/2^{m}. Let ε∈R\varepsilon\in\mathbb{R} with ε>0\varepsilon>0, let r=1/2r=1/2 and δm=εrm\delta_{m}=\varepsilon r^{m} for m∈Nm\in\mathbb{N}. Then δm∈R\delta_{m}\in\mathbb{R} and δm>0\delta_{m}>0 for every mm, and ∑m∈Nδm=ε\sum_{m\in\mathbb{N}}\delta_{m}=\varepsilon. Indeed, 0<r<10<r<1 by Inequalities in an Ordered Field: Mixed Transitivity, Strict Sums, Signs, Products, Natural Numbers, Halving, Reciprocals, Absolute Values and Squares (vii), so ∣r∣<1|r|<1 and rm>0r^{m}>0 (Powers in a Commutative Ring, a Field and an Ordered Field: Exponent Laws, Factorisation, Geometric Sums, Monotonicity and Bernoulli's Inequality (vi)), whence δm>0\delta_{m}>0 by Inequalities in an Ordered Field: Mixed Transitivity, Strict Sums, Signs, Products, Natural Numbers, Halving, Reciprocals, Absolute Values and Squares (iv). The real series ∑k=1∞rk\sum_{k=1}^{\infty}r^{k} converges with sum r/(1−r)=1r/(1-r)=1, as 1−r=r1-r=r, by Series of Real Numbers with Nonnegative Terms, Comparison, Absolute Convergence and the Geometric Series (v); so ∑k∈Nrk=1\sum_{k\in\mathbb{N}}r^{k}=1 by Sums of Sequences in [0,∞]: Real Series, Linearity, Comparison, Rearrangement, Double Series and Monotone Convergence (vi), and ∑m∈Nδm=ε∑m∈Nrm=ε\sum_{m\in\mathbb{N}}\delta_{m}=\varepsilon\sum_{m\in\mathbb{N}}r^{m}=\varepsilon by Sums of Sequences in [0,∞]: Real Series, Linearity, Comparison, Rearrangement, Double Series and Monotone Convergence (ix).

Step 4: combining covers. (W) Let w:N×N→[0,∞]w:\mathbb{N}\times\mathbb{N}\to[0,\infty] be a map. For m∈Nm\in\mathbb{N}, Values, Restriction and Equality of a Map, and the Map u↦t(u)u\mapsto t(u) as a Defined Set Symbol (v), with N\mathbb{N} and [0,∞][0,\infty] for its AA and BB, mm for its v1v_{1}, the guard m∈Nm\in\mathbb{N} for its γ\gamma, nn for its uu and the value w(m,n)w(m,n) of ww at (m,n)(m,n), a defined set symbol used properly and lying in [0,∞][0,\infty], for its t(u,v1)t(u,v_{1}), gives a predicative formula θ(h,m)\theta(h,m) whose only solution hh is the map n↦w(m,n)n\mapsto w(m,n) from N\mathbb{N} to [0,∞][0,\infty]. So this map is a defined set symbol with argument mm, used properly at every m∈Nm\in\mathbb{N}, and an element of the set [0,∞]N[0,\infty]^{\mathbb{N}} of sequences in [0,∞][0,\infty] (The Elements of the Class of Functions from a Set to a Set, and This Class Is a Set (i) and The Elements of the Class of Functions from a Set to a Set, and This Class Is a Set (ii)); and Maps and Relations Given by Formulas (i), applied to N\mathbb{N}, [0,∞]N[0,\infty]^{\mathbb{N}} and this expression, gives the map m↦(n↦w(m,n))m\mapsto(n\mapsto w(m,n)) from N\mathbb{N} to [0,∞]N[0,\infty]^{\mathbb{N}}. By Sums of Sequences in [0,∞] as Typed Terms and as a Map (ii), applied to D=ND=\mathbb{N} and this map, the inner sums form a sequence m↦∑n∈Nw(m,n)m\mapsto\sum_{n\in\mathbb{N}}w(m,n) in [0,∞][0,\infty].

Let E⊆RE\subseteq\mathbb{R} and let α,β:N×N→R\alpha,\beta:\mathbb{N}\times\mathbb{N}\to\mathbb{R} be maps with α(m,n)≤β(m,n)\alpha(m,n)\le\beta(m,n) for all m,n∈Nm,n\in\mathbb{N}, such that for every x∈Ex\in E there are m,n∈Nm,n\in\mathbb{N} with α(m,n)<x<β(m,n)\alpha(m,n)<x<\beta(m,n). Then

λ∗(E)≤∑m∈N∑n∈N(β(m,n)−α(m,n)).\lambda^{*}(E)\le\sum_{m\in\mathbb{N}}\sum_{n\in\mathbb{N}}\bigl(\beta(m,n)-\alpha(m,n)\bigr).

Here w(m,n)=β(m,n)−α(m,n)w(m,n)=\beta(m,n)-\alpha(m,n) is a real number with 0≤w(m,n)0\le w(m,n), so it lies in [0,∞][0,\infty], and w:N×N→[0,∞]w:\mathbb{N}\times\mathbb{N}\to[0,\infty] is a map by Maps and Relations Given by Formulas (ii); the inner sums form the sequence m↦∑n∈Nw(m,n)m\mapsto\sum_{n\in\mathbb{N}}w(m,n) in [0,∞][0,\infty] of (W). To prove the inequality, let τ:N→N×N\tau:\mathbb{N}\to\mathbb{N}\times\mathbb{N} be a bijection, which exists by Countable Sets: a Bijection from ℕ onto ℕ×ℕ, Finite Sets, Subsets, Surjective Images, Images, Products, and the Integers and Rationals (ii). The sequences ek=α(τ(k))e_{k}=\alpha(\tau(k)) and fk=β(τ(k))f_{k}=\beta(\tau(k)) satisfy ek≤fke_{k}\le f_{k}, and (e,f)(e,f) is a cover of EE: for x∈Ex\in E take m,nm,n as above and k∈Nk\in\mathbb{N} with τ(k)=(m,n)\tau(k)=(m,n); then ek<x<fke_{k}<x<f_{k}. By Step 1(a) and Sums of Sequences in [0,∞]: Real Series, Linearity, Comparison, Rearrangement, Double Series and Monotone Convergence (xiii),

λ∗(E)≤ℓ(e,f)=∑k∈Nw(τ(k))=∑m∈N∑n∈Nw(m,n).\lambda^{*}(E)\le\ell(e,f)=\sum_{k\in\mathbb{N}}w(\tau(k))=\sum_{m\in\mathbb{N}}\sum_{n\in\mathbb{N}}w(m,n).

Proof of (i). The constant sequences am=bm=0a_{m}=b_{m}=0 form a cover of ∅\emptyset, and ℓ(a,b)=∑m∈N0=0\ell(a,b)=\sum_{m\in\mathbb{N}}0=0 by Small Facts about [0,∞]: Zero, Finiteness and Sums of Zeros (iv). So λ∗(∅)≤0\lambda^{*}(\emptyset)\le0 by Step 1(a), and λ∗(∅)=0\lambda^{*}(\emptyset)=0 by Small Facts about [0,∞]: Zero, Finiteness and Sums of Zeros (i).

Proof of (ii). Let A⊆BA\subseteq B. Every cover of BB is a cover of AA, so λ∗(A)≤ℓ(a,b)\lambda^{*}(A)\le\ell(a,b) for every cover (a,b)(a,b) of BB by Step 1(a), and λ∗(A)≤λ∗(B)\lambda^{*}(A)\le\lambda^{*}(B) by Step 1(b).

Proof of (iii). If λ∗(A)=∞\lambda^{*}(A)=\infty or λ∗(B)=∞\lambda^{*}(B)=\infty, then λ∗(A)+λ∗(B)=∞\lambda^{*}(A)+\lambda^{*}(B)=\infty (Addition and Multiplication on [0,∞] (i)), and the claim holds as ∞\infty is the greatest element of [−∞,∞][-\infty,\infty] (Order and Suprema in the Extended Real Line (ii)). Otherwise λ∗(A)\lambda^{*}(A) and λ∗(B)\lambda^{*}(B) are real by Small Facts about [0,∞]: Zero, Finiteness and Sums of Zeros (ii). Fix first ε∈R\varepsilon\in\mathbb{R} with ε>0\varepsilon>0; then ε/2>0\varepsilon/2>0 and ε/2+ε/2=ε\varepsilon/2+\varepsilon/2=\varepsilon (Inequalities in an Ordered Field: Mixed Transitivity, Strict Sums, Signs, Products, Natural Numbers, Halving, Reciprocals, Absolute Values and Squares (vii)). As λ∗(A)<λ∗(A)+ε/2\lambda^{*}(A)<\lambda^{*}(A)+\varepsilon/2, Step 1(c) gives a cover (a,b)(a,b) of AA with ℓ(a,b)<λ∗(A)+ε/2\ell(a,b)<\lambda^{*}(A)+\varepsilon/2; likewise fix a cover (c,d)(c,d) of BB with ℓ(c,d)<λ∗(B)+ε/2\ell(c,d)<\lambda^{*}(B)+\varepsilon/2. Only these two covers are taken. Let α,β:N×N→R\alpha,\beta:\mathbb{N}\times\mathbb{N}\to\mathbb{R} be the maps with α(1,n)=an\alpha(1,n)=a_{n}, α(2,n)=cn\alpha(2,n)=c_{n}, β(1,n)=bn\beta(1,n)=b_{n}, β(2,n)=dn\beta(2,n)=d_{n}, and α(m,n)=β(m,n)=0\alpha(m,n)=\beta(m,n)=0 for m∉{1,2}m\notin\{1,2\} (Maps Defined by Cases (i), applied twice for each map). Then α≤β\alpha\le\beta, and every x∈A∪Bx\in A\cup B lies in (an,bn)(a_{n},b_{n}) or in (cn,dn)(c_{n},d_{n}) for some nn, so Step 4 applies to E=A∪BE=A\cup B. For um=∑n∈N(β(m,n)−α(m,n))u_{m}=\sum_{n\in\mathbb{N}}(\beta(m,n)-\alpha(m,n)) we have u1=ℓ(a,b)u_{1}=\ell(a,b), u2=ℓ(c,d)u_{2}=\ell(c,d), and um=0u_{m}=0 for m>2m>2 by Small Facts about [0,∞]: Zero, Finiteness and Sums of Zeros (iv); so ∑m∈Num=u1+u2\sum_{m\in\mathbb{N}}u_{m}=u_{1}+u_{2} by Sums of Sequences in [0,∞]: Real Series, Linearity, Comparison, Rearrangement, Double Series and Monotone Convergence (vii), Iterated Operations over Finite Sets: Singletons, Disjoint Unions, Reindexing, Products of Sets, Termwise Combination, Homomorphisms and Intervals (vii) and Iterated Operations over Finite Sets: Singletons, Disjoint Unions, Reindexing, Products of Sets, Termwise Combination, Homomorphisms and Intervals (i), which apply as ++ on [0,∞][0,\infty] is associative and commutative with neutral element 00 (Arithmetic in [0,∞]: Monoid Laws, Distributivity, Monotonicity and Cancellation (i)); here m↦umm\mapsto u_{m} is the sequence in [0,∞][0,\infty] of (W) for the map ww of Step 4, w(m,n)=β(m,n)−α(m,n)w(m,n)=\beta(m,n)-\alpha(m,n). Hence, by Step 4 and Arithmetic in [0,∞]: Monoid Laws, Distributivity, Monotonicity and Cancellation (iv),

λ∗(A∪B)≤ℓ(a,b)+ℓ(c,d)≤(λ∗(A)+λ∗(B))+ε,\lambda^{*}(A\cup B)\le\ell(a,b)+\ell(c,d)\le\bigl(\lambda^{*}(A)+\lambda^{*}(B)\bigr)+\varepsilon,

all terms on the right being real. As ε>0\varepsilon>0 was arbitrary, Step 2 gives the claim.

Proof of (iv). The intervals [p,q][p,q]. Let p,q∈Rp,q\in\mathbb{R} with p≤qp\le q; we show λ∗([p,q])=q−p\lambda^{*}([p,q])=q-p. For every cover (a,b)(a,b) of [p,q][p,q], q−p≤ℓ(a,b)q-p\le\ell(a,b) by A Closed Interval Covered by Open Intervals Is No Longer than Their Total Length (ii); so q−p≤λ∗([p,q])q-p\le\lambda^{*}([p,q]) by Step 1(b). Conversely, let ε∈R\varepsilon\in\mathbb{R} with ε>0\varepsilon>0, and let (am)(a_{m}) and (bm)(b_{m}) be the sequences with a1=p−ε/2a_{1}=p-\varepsilon/2, b1=q+ε/2b_{1}=q+\varepsilon/2 and am=bm=0a_{m}=b_{m}=0 for m≠1m\neq1 (Maps Defined by Cases (i)). Then am≤bma_{m}\le b_{m} for every mm, every x∈[p,q]x\in[p,q] satisfies a1<x<b1a_{1}<x<b_{1}, and bm−am=0b_{m}-a_{m}=0 for m>1m>1; so (a,b)(a,b) is a cover of [p,q][p,q] with ℓ(a,b)=b1−a1=(q−p)+ε\ell(a,b)=b_{1}-a_{1}=(q-p)+\varepsilon by Sums of Sequences in [0,∞]: Real Series, Linearity, Comparison, Rearrangement, Double Series and Monotone Convergence (vii) and Iterated Operations over Finite Sets: Singletons, Disjoint Unions, Reindexing, Products of Sets, Termwise Combination, Homomorphisms and Intervals (i), which apply as ++ on [0,∞][0,\infty] is associative and commutative with neutral element 00 (Arithmetic in [0,∞]: Monoid Laws, Distributivity, Monotonicity and Cancellation (i)). By Step 1(a), λ∗([p,q])≤(q−p)+ε\lambda^{*}([p,q])\le(q-p)+\varepsilon, and by Step 2, λ∗([p,q])≤q−p\lambda^{*}([p,q])\le q-p. So λ∗([p,q])=q−p\lambda^{*}([p,q])=q-p, the order being antisymmetric. In particular λ∗([p,p])=0\lambda^{*}([p,p])=0 for every p∈Rp\in\mathbb{R}.

The four intervals. Let a≤ba\le b. Every x∈[a,b]x\in[a,b] satisfies x=ax=a, x=bx=b or a<x<ba<x<b, so [a,b]⊆((a,b)∪[a,a])∪[b,b][a,b]\subseteq((a,b)\cup[a,a])\cup[b,b]. By (ii), (iii) applied twice, Arithmetic in [0,∞]: Monoid Laws, Distributivity, Monotonicity and Cancellation (iv) and the closed case,

λ∗([a,b])≤λ∗((a,b)∪[a,a])+λ∗([b,b])≤(λ∗((a,b))+0)+0=λ∗((a,b)),\lambda^{*}([a,b])\le\lambda^{*}((a,b)\cup[a,a])+\lambda^{*}([b,b])\le\bigl(\lambda^{*}((a,b))+0\bigr)+0=\lambda^{*}((a,b)),

00 being neutral (Arithmetic in [0,∞]: Monoid Laws, Distributivity, Monotonicity and Cancellation (i)). Let II be one of (a,b)(a,b), [a,b][a,b], [a,b)[a,b) and (a,b](a,b]. Then (a,b)⊆I⊆[a,b](a,b)\subseteq I\subseteq[a,b], so by (ii) and the closed case

b−a=λ∗([a,b])≤λ∗((a,b))≤λ∗(I)≤λ∗([a,b])=b−a,b-a=\lambda^{*}([a,b])\le\lambda^{*}((a,b))\le\lambda^{*}(I)\le\lambda^{*}([a,b])=b-a,

and λ∗(I)=b−a\lambda^{*}(I)=b-a.

Proof of (v). Let c∈Rc\in\mathbb{R}; then A∩(c,∞)={x∈A:c<x}A\cap(c,\infty)=\{x\in A:c<x\} and A∖(c,∞)={x∈A:x≤c}A\setminus(c,\infty)=\{x\in A:x\le c\}. Maxima and minima of two real numbers are as in The Maximum and Minimum of Two Elements of a Total Order (i), for the total order of R\mathbb{R}.

First let (a,b)(a,b) be a cover of AA and ε∈R\varepsilon\in\mathbb{R} with ε>0\varepsilon>0, and let (δm)(\delta_{m}) be as in Step 3 for this ε\varepsilon. Define, for m∈Nm\in\mathbb{N},

am′=max⁡{am,c},bm′=max⁡{bm,c},am′′=min⁡{am,c},bm′′=min⁡{bm,c}+δm.a'_{m}=\max\{a_{m},c\},\quad b'_{m}=\max\{b_{m},c\},\quad a''_{m}=\min\{a_{m},c\},\quad b''_{m}=\min\{b_{m},c\}+\delta_{m}.

As am≤bm≤bm′a_{m}\le b_{m}\le b'_{m} and c≤bm′c\le b'_{m}, The Maximum and Minimum of Two Elements of a Total Order (iii) and The Maximum and Minimum of Two Elements of a Total Order (iv) give am′≤bm′a'_{m}\le b'_{m}; as am′′≤am≤bma''_{m}\le a_{m}\le b_{m} and am′′≤ca''_{m}\le c by The Maximum and Minimum of Two Elements of a Total Order (iii), The Maximum and Minimum of Two Elements of a Total Order (v) gives am′′≤min⁡{bm,c}<bm′′a''_{m}\le\min\{b_{m},c\}<b''_{m}.

(a′,b′)(a',b') is a cover of A∩(c,∞)A\cap(c,\infty): let x∈Ax\in A with c<xc<x, and mm with am<x<bma_{m}<x<b_{m}. Then am′a'_{m} is ama_{m} or cc (The Maximum and Minimum of Two Elements of a Total Order (ii)), so am′<xa'_{m}<x; and x<bm≤bm′x<b_{m}\le b'_{m}.

(a′′,b′′)(a'',b'') is a cover of A∖(c,∞)A\setminus(c,\infty): let x∈Ax\in A with x≤cx\le c, and mm with am<x<bma_{m}<x<b_{m}. Then am′′≤am<xa''_{m}\le a_{m}<x, and x≤min⁡{bm,c}<bm′′x\le\min\{b_{m},c\}<b''_{m} by The Maximum and Minimum of Two Elements of a Total Order (v).

By The Maximum and Minimum of Two Elements of an Ordered Field: Sum, Difference and Absolute Value (i), bm′+min⁡{bm,c}=bm+cb'_{m}+\min\{b_{m},c\}=b_{m}+c and am′+am′′=am+ca'_{m}+a''_{m}=a_{m}+c, so

(bm′−am′)+(bm′′−am′′)=(bm+c)−(am+c)+δm=(bm−am)+δm.(b'_{m}-a'_{m})+(b''_{m}-a''_{m})=(b_{m}+c)-(a_{m}+c)+\delta_{m}=(b_{m}-a_{m})+\delta_{m}.

Hence, by Sums of Sequences in [0,∞]: Real Series, Linearity, Comparison, Rearrangement, Double Series and Monotone Convergence (viii) (twice) and Step 3,

ℓ(a′,b′)+ℓ(a′′,b′′)=∑m∈N((bm−am)+δm)=ℓ(a,b)+ε,\ell(a',b')+\ell(a'',b'')=\sum_{m\in\mathbb{N}}\bigl((b_{m}-a_{m})+\delta_{m}\bigr)=\ell(a,b)+\varepsilon,

and by Step 1(a) and Arithmetic in [0,∞]: Monoid Laws, Distributivity, Monotonicity and Cancellation (iv), applied twice,

λ∗(A∩(c,∞))+λ∗(A∖(c,∞))≤ℓ(a,b)+ε.(∗)\lambda^{*}(A\cap(c,\infty))+\lambda^{*}(A\setminus(c,\infty))\le\ell(a,b)+\varepsilon.\tag{$\ast$}

Now, if λ∗(A)=∞\lambda^{*}(A)=\infty the claim holds by Order and Suprema in the Extended Real Line (ii). Otherwise λ∗(A)∈R\lambda^{*}(A)\in\mathbb{R} by Small Facts about [0,∞]: Zero, Finiteness and Sums of Zeros (ii). Fix first ε∈R\varepsilon\in\mathbb{R} with ε>0\varepsilon>0; then Step 1(c), as λ∗(A)<λ∗(A)+ε/2\lambda^{*}(A)<\lambda^{*}(A)+\varepsilon/2, gives one cover (a,b)(a,b) of AA with ℓ(a,b)<λ∗(A)+ε/2\ell(a,b)<\lambda^{*}(A)+\varepsilon/2, and (∗)(\ast) for this cover and ε/2\varepsilon/2 gives

λ∗(A∩(c,∞))+λ∗(A∖(c,∞))≤λ∗(A)+ε.\lambda^{*}(A\cap(c,\infty))+\lambda^{*}(A\setminus(c,\infty))\le\lambda^{*}(A)+\varepsilon.

No choice is made beyond this one cover: the covers (a′,b′)(a',b') and (a′′,b′′)(a'',b'') are given by expressions in it. As ε>0\varepsilon>0 was arbitrary, Step 2 gives the claim.

Proof of (vi). Let (Am)(A_{m}) be a sequence in P(R)\mathcal{P}(\mathbb{R}), let E=⋃m∈NAmE=\bigcup_{m\in\mathbb{N}}A_{m}, so that x∈Ex\in E if and only if x∈Amx\in A_{m} for some m∈Nm\in\mathbb{N} (Indexed Families of Sets and Their Union, Intersection and Product (ii)), and let σ=∑m∈Nλ∗(Am)\sigma=\sum_{m\in\mathbb{N}}\lambda^{*}(A_{m}). If σ=∞\sigma=\infty, the claim holds by Order and Suprema in the Extended Real Line (ii). Otherwise, for every mm, λ∗(Am)≤σ<∞\lambda^{*}(A_{m})\le\sigma<\infty by Sums of Sequences in [0,∞]: Real Series, Linearity, Comparison, Rearrangement, Double Series and Monotone Convergence (iii), so λ∗(Am)∈R\lambda^{*}(A_{m})\in\mathbb{R} by Small Facts about [0,∞]: Zero, Finiteness and Sums of Zeros (iii).

Fix first ε∈R\varepsilon\in\mathbb{R} with ε>0\varepsilon>0, and let (δm)(\delta_{m}) be as in Step 3. The set RN\mathbb{R}^{\mathbb{N}} of sequences in R\mathbb{R} is a set by The Elements of the Class of Functions from a Set to a Set, and This Class Is a Set (ii), and so is Ω=RN×RN\Omega=\mathbb{R}^{\mathbb{N}}\times\mathbb{R}^{\mathbb{N}}, by Membership in a Cartesian Product, and the Cartesian Product of Two Sets Is a Set (ii). For m∈Nm\in\mathbb{N} let

Km={(a,b)∈Ω:(a,b) is a cover of Am and ℓ(a,b)<λ∗(Am)+δm},K_{m}=\{(a,b)\in\Omega:(a,b)\text{ is a cover of }A_{m}\text{ and }\ell(a,b)<\lambda^{*}(A_{m})+\delta_{m}\},

a subset of Ω\Omega, hence an element of the set P(Ω)\mathcal{P}(\Omega) (The Union Set and the Power Set of a Set (ii)); its condition, on mm and an element pp of Ω\Omega, is expressed by a predicative formula, with the sequences (Am)(A_{m}) and (δm)(\delta_{m}), the map λ∗\lambda^{*}, Ω\Omega, the strict relation << of [−∞,∞][-\infty,\infty] (Partial and Total Orders on a Set and the Associated Strict Relation (iii)) and N\mathbb{N}, R\mathbb{R}, RN\mathbb{R}^{\mathbb{N}}, the order, the operations ++ and ⋅\cdot and the elements 00 and 11 of R\mathbb{R}, P(R)\mathcal{P}(\mathbb{R}), [−∞,∞][-\infty,\infty] and its order, [0,∞][0,\infty], [0,∞]N[0,\infty]^{\mathbb{N}}, the sum on [0,∞][0,\infty], and the order and the addition of N0\mathbb{N}_{0} as parameters. The class variables for these parameters are chosen pairwise distinct whenever their values differ: in particular the order of R\mathbb{R} and the order of [−∞,∞][-\infty,\infty] are different variables, although Intervals of the Real Line as Defined Set Symbols, Typed Terms and Values of Maps (ii) and Sums of Sequences in [0,∞] as Typed Terms and as a Map (i) both write LL for theirs, and likewise ++ of R\mathbb{R} and the sum ++ on [0,∞][0,\infty].

Namely, the condition is ∃a ∃b (p=(a,b)∧κ)\exists a\,\exists b\,(p=(a,b)\wedge\kappa), with p=(a,b)p=(a,b) in the predicative form of Predicativity of Expressions with Abbreviations, and Symbol-Free Forms of the Basic Defined Set Symbols (v), and κ\kappa formed as follows under one guard GG on aa, bb and mm throughout: GG is γ∧m∈N\gamma\wedge m\in\mathbb{N}, where γ\gamma is a∈RN∧b∈RN∧∀n (n∈N⇒∃u ∃v ((n,u)∈a∧(n,v)∈b∧u≤v))a\in\mathbb{R}^{\mathbb{N}}\wedge b\in\mathbb{R}^{\mathbb{N}}\wedge\forall n\,(n\in\mathbb{N}\Rightarrow\exists u\,\exists v\,((n,u)\in a\wedge(n,v)\in b\wedge u\le v)); γ\gamma and GG are predicative by Predicativity of Expressions with Abbreviations, and Symbol-Free Forms of the Basic Defined Set Symbols (i). As (n,z)∈a(n,z)\in a if and only if z=anz=a_{n} for a map aa with domain N\mathbb{N} and n∈Nn\in\mathbb{N} (Predicativity of Expressions with Abbreviations, and Symbol-Free Forms of the Basic Defined Set Symbols (x)), and likewise for bb, GG states that aa and bb are sequences in R\mathbb{R} with an≤bna_{n}\le b_{n} for every n∈Nn\in\mathbb{N}, and that m∈Nm\in\mathbb{N}. For the index terms we use the guard G∧n∈NG\wedge n\in\mathbb{N} on aa, bb, mm and nn: under it ana_{n} and bnb_{n} are terms of type R\mathbb{R} (Terms and Typed Terms under a Guard (ii)), bn−anb_{n}-a_{n} is one of type [0,∞][0,\infty], its equation being expressed with the formula of Predicative Substitution of Values of Functions and Binary Operations, and Predicative Formulas for Negatives, Differences, Absolute Values, Distances, Reciprocals and Quotients in an Ordered Field (iv), and (an,bn)(a_{n},b_{n}) is one of type P(R)\mathcal{P}(\mathbb{R}) by Intervals of the Real Line as Defined Set Symbols, Typed Terms and Values of Maps (ii). Thus bn−anb_{n}-a_{n} is an expression built from the defined set symbols ana_{n}, bnb_{n} and u−vu-v, used properly and lying in [0,∞][0,\infty] for all aa, bb and mm satisfying GG and every n∈Nn\in\mathbb{N}. So Values, Restriction and Equality of a Map, and the Map u↦t(u)u\mapsto t(u) as a Defined Set Symbol (v), with N\mathbb{N} and [0,∞][0,\infty] for its AA and BB, aa, bb and mm for its v1v_{1}, v2v_{2} and v3v_{3}, GG for its γ\gamma, nn for its uu and bn−anb_{n}-a_{n} for its t(u,v1,v2,v3)t(u,v_{1},v_{2},v_{3}), gives a predicative formula θ(h,a,b,m)\theta(h,a,b,m) such that, for all aa, bb and mm satisfying GG, the only set hh with θ(h,a,b,m)\theta(h,a,b,m) is the map n↦bn−ann\mapsto b_{n}-a_{n} from N\mathbb{N} to [0,∞][0,\infty], an element of [0,∞]N[0,\infty]^{\mathbb{N}} by The Elements of the Class of Functions from a Set to a Set, and This Class Is a Set (i). Hence G∧θ(h,a,b,m)G\wedge\theta(h,a,b,m) expresses G∧h=(n↦bn−an)G\wedge h=(n\mapsto b_{n}-a_{n}), and n↦bn−ann\mapsto b_{n}-a_{n} is a term of type [0,∞]N[0,\infty]^{\mathbb{N}} under GG (Terms and Typed Terms under a Guard (ii)); and Sums of Sequences in [0,∞] as Typed Terms and as a Map (i), with GG for its γ\gamma and this term for its aa, makes ℓ(a,b)\ell(a,b) a term of type [0,∞][0,\infty] under GG. "Every x∈Amx\in A_{m} lies in (an,bn)(a_{n},b_{n}) for some n∈Nn\in\mathbb{N}" is expressed by ∀x (∃y (y=Am∧x∈y)⇒∃n (n∈N∧∃z (z=(an,bn)∧x∈z)))\forall x\,(\exists y\,(y=A_{m}\wedge x\in y)\Rightarrow\exists n\,(n\in\mathbb{N}\wedge\exists z\,(z=(a_{n},b_{n})\wedge x\in z))), the equations y=Amy=A_{m} and z=(an,bn)z=(a_{n},b_{n}) having predicative forms, under GG and under G∧n∈NG\wedge n\in\mathbb{N} respectively, by Terms and Typed Terms under a Guard (iv). Here mm is a term of type N\mathbb{N} under GG by Typed Terms under a Predicative Guard: Variables, Parameters, Values of Maps and Ordered Pairs, and Predicative Equations and Memberships between Them (i), so AmA_{m}, the value of (Am)(A_{m}) at mm, is a term of type P(R)\mathcal{P}(\mathbb{R}) under GG and λ∗(Am)\lambda^{*}(A_{m}) one of type [0,∞][0,\infty], by Typed Terms under a Predicative Guard: Variables, Parameters, Values of Maps and Ordered Pairs, and Predicative Equations and Memberships between Them (iii), λ∗\lambda^{*} being a map from P(R)\mathcal{P}(\mathbb{R}) to [0,∞][0,\infty] (Lebesgue Outer Measure on the Real Line (ii)). As every δn\delta_{n} is a positive real number, (δn)(\delta_{n}) is also a map from N\mathbb{N} to [0,∞][0,\infty] (Functions, Values of a Function, and Functions from One Class to Another (iii)), so δm\delta_{m} is a term of type [0,∞][0,\infty] under GG by Typed Terms under a Predicative Guard: Variables, Parameters, Values of Maps and Ordered Pairs, and Predicative Equations and Memberships between Them (iii). All terms below are under GG. Then (λ∗(Am),δm)(\lambda^{*}(A_{m}),\delta_{m}) is a term of type [0,∞]×[0,∞][0,\infty]\times[0,\infty] by Typed Terms under a Predicative Guard: Variables, Parameters, Values of Maps and Ordered Pairs, and Predicative Equations and Memberships between Them (iv), and λ∗(Am)+δm\lambda^{*}(A_{m})+\delta_{m}, the value at it of the sum on [0,∞][0,\infty], a map from [0,∞]×[0,∞][0,\infty]\times[0,\infty] to [0,∞][0,\infty] (Addition and Multiplication on [0,∞] (i) and Binary Operations on a Set (i)), is a term of type [0,∞][0,\infty] by Typed Terms under a Predicative Guard: Variables, Parameters, Values of Maps and Ordered Pairs, and Predicative Equations and Memberships between Them (iii). Finally (ℓ(a,b),λ∗(Am)+δm)(\ell(a,b),\lambda^{*}(A_{m})+\delta_{m}) is a term of type [0,∞]×[0,∞][0,\infty]\times[0,\infty] by Typed Terms under a Predicative Guard: Variables, Parameters, Values of Maps and Ordered Pairs, and Predicative Equations and Memberships between Them (iv), and ℓ(a,b)<λ∗(Am)+δm\ell(a,b)<\lambda^{*}(A_{m})+\delta_{m}, which says that this pair lies in <<, is expressed by a predicative formula by Typed Terms under a Predicative Guard: Variables, Parameters, Values of Maps and Ordered Pairs, and Predicative Equations and Memberships between Them (vi). κ\kappa is the conjunction of the formulas obtained, and ∃a ∃b (p=(a,b)∧κ)\exists a\,\exists b\,(p=(a,b)\wedge\kappa) is predicative by Predicativity of Expressions with Abbreviations, and Symbol-Free Forms of the Basic Defined Set Symbols (i). So KmK_{m} is a defined set symbol used properly at every m∈Nm\in\mathbb{N}, and Maps and Relations Given by Formulas (i), applied to N\mathbb{N}, P(Ω)\mathcal{P}(\Omega) and this expression, gives the map m↦Kmm\mapsto K_{m}, a family indexed by N\mathbb{N}. Each KmK_{m} is nonempty: λ∗(Am)<λ∗(Am)+δm\lambda^{*}(A_{m})<\lambda^{*}(A_{m})+\delta_{m}, so Step 1(c) gives a cover (a,b)(a,b) of AmA_{m} with ℓ(a,b)<λ∗(Am)+δm\ell(a,b)<\lambda^{*}(A_{m})+\delta_{m}, and (a,b)∈Ω(a,b)\in\Omega by The Elements of the Class of Functions from a Set to a Set, and This Class Is a Set (i) and Membership in a Cartesian Product, and the Cartesian Product of Two Sets Is a Set (i). By Countable Choice Follows from Dependent Choice (i), applied to the set Ω\Omega and the family (Km)(K_{m}), there is a map ξ:N→Ω\xi:\mathbb{N}\to\Omega with ξm∈Km\xi_{m}\in K_{m} for every m∈Nm\in\mathbb{N}. This is the only choice in the argument, made after ε\varepsilon is fixed. Write ξm=(a(m),b(m))\xi_{m}=(a^{(m)},b^{(m)}), and let α,β:N×N→R\alpha,\beta:\mathbb{N}\times\mathbb{N}\to\mathbb{R} be the maps α(m,n)=an(m)\alpha(m,n)=a^{(m)}_{n} and β(m,n)=bn(m)\beta(m,n)=b^{(m)}_{n}, which exist by Maps and Relations Given by Formulas (ii), as an(m)a^{(m)}_{n} and bn(m)b^{(m)}_{n} are values at nn of the components of the value ξm\xi_{m}, maps N→R\mathbb{N}\to\mathbb{R} by The Elements of the Class of Functions from a Set to a Set, and This Class Is a Set (i), and so lie in R\mathbb{R}. Then α≤β\alpha\le\beta, as each (a(m),b(m))(a^{(m)},b^{(m)}) is a cover, and every x∈Ex\in E lies in some AmA_{m}, hence in (an(m),bn(m))(a^{(m)}_{n},b^{(m)}_{n}) for some nn, as (a(m),b(m))(a^{(m)},b^{(m)}) is a cover of AmA_{m}. The sequence m↦ℓ(a(m),b(m))m\mapsto\ell(a^{(m)},b^{(m)}) in [0,∞][0,\infty] is given by Sums of Sequences in [0,∞] as Typed Terms and as a Map (ii), as in (W) for the map ww of Step 4: since β(m,n)−α(m,n)=bn(m)−an(m)\beta(m,n)-\alpha(m,n)=b^{(m)}_{n}-a^{(m)}_{n} for all m,n∈Nm,n\in\mathbb{N}, the mm-th inner sum of Step 4 is ∑n∈N(bn(m)−an(m))=ℓ(a(m),b(m))\sum_{n\in\mathbb{N}}(b^{(m)}_{n}-a^{(m)}_{n})=\ell(a^{(m)},b^{(m)}). By Step 4, Sums of Sequences in [0,∞]: Real Series, Linearity, Comparison, Rearrangement, Double Series and Monotone Convergence (x), Sums of Sequences in [0,∞]: Real Series, Linearity, Comparison, Rearrangement, Double Series and Monotone Convergence (viii) and Step 3,

λ∗(E)≤∑m∈Nℓ(a(m),b(m))≤∑m∈N(λ∗(Am)+δm)=σ+ε.\lambda^{*}(E)\le\sum_{m\in\mathbb{N}}\ell\bigl(a^{(m)},b^{(m)}\bigr)\le\sum_{m\in\mathbb{N}}\bigl(\lambda^{*}(A_{m})+\delta_{m}\bigr)=\sigma+\varepsilon.

As ε>0\varepsilon>0 was arbitrary and σ∈R\sigma\in\mathbb{R} by Small Facts about [0,∞]: Zero, Finiteness and Sums of Zeros (ii), Step 2 gives λ∗(E)≤σ\lambda^{*}(E)\le\sigma.

Proof of (vii). λ∗\lambda^{*} is a map from P(R)\mathcal{P}(\mathbb{R}) to [0,∞][0,\infty] by Lebesgue Outer Measure on the Real Line (ii); it satisfies λ∗(∅)=0\lambda^{*}(\emptyset)=0 by (i), is monotone by (ii), and is countably subadditive by (vi). As R\mathbb{R} is a set by The Real Numbers (i), these are the conditions of Outer Measure (i) for X=RX=\mathbb{R}.

Citations

Loading…

Dependencies

Uses0

Loading…

Comments

Log in to comment.

Loading…