TheoremBase

Joining partitions of [a,c] and [c,b] splits Darboux sums, and adding c to a partition of [a,b] refines it; so the lower and the upper integrals over [a,b] are the sums of those over [a,c] and [c,b], which gives the splitting clause. Subintervals and the oriented rule follow by applying it to restrictions.

Proof

Notation. For u,v∈[a,b]u,v\in[a,b] with u<vu<v we have [u,v]⊆[a,b][u,v]\subseteq[a,b] by Basic Facts about Intervals of the Real Line (v) and Basic Facts about Intervals of the Real Line (ii); let fuv=f∣[u,v]:[u,v]→Rf_{uv}=f|_{[u,v]}:[u,v]\to\mathbb{R} be the restriction, with fuv(t)=f(t)f_{uv}(t)=f(t) for t∈[u,v]t\in[u,v] (Domains and Restrictions of Maps (iii)), so ∣fuv(t)∣≤B|f_{uv}(t)|\le B and fuvf_{uv} is bounded. Let I‾(u,v)\underline{I}(u,v) and I‾(u,v)\overline{I}(u,v) be the lower and the upper integral of fuvf_{uv} over [u,v][u,v] (The Riemann Integral on a Closed Interval via Lower and Upper Integrals (iii)); I‾(u,v)≤I‾(u,v)\underline{I}(u,v)\le\overline{I}(u,v) by Bounds for Darboux Sums, Monotonicity under Refinement, the Supremum of the Lower Sums and the Infimum of the Upper Sums, and Concatenation of Partitions (iv). As fab=ff_{ab}=f (Values, Restriction and Equality of a Map, and the Map u↦t(u)u\mapsto t(u) as a Defined Set Symbol (iii)), by The Riemann Integral on a Closed Interval via Lower and Upper Integrals (iv), The Riemann Integral on a Closed Interval via Lower and Upper Integrals (v) and The Riemann Integral on a Closed Interval via Lower and Upper Integrals (vi): ff is integrable on [u,v][u,v] if and only if I‾(u,v)=I‾(u,v)\underline{I}(u,v)=\overline{I}(u,v), and then ∫uvf=I‾(u,v)\int_{u}^{v}f=\underline{I}(u,v). For a bounded map hh on a closed interval and a partition RR of it, the sums U(h,R)U(h,R) and L(h,R)L(h,R) are those of Upper and Lower Darboux Sums of a Bounded Function on a Partition (ii) and Upper and Lower Darboux Sums of a Bounded Function on a Partition (iii), formed with the increasing listing of RR, which is unique by The Increasing Listing of a Partition: Endpoints, Lengths, Covering, Unions, Inserting a Point, Concatenation and Uniform Partitions (i).

Proof of (iv). Let u,v∈Ku,v\in K with u<vu<v and [u,v]⊆K[u,v]\subseteq K, so [u,v]⊆E[u,v]\subseteq E. By Basic Properties of Functions: Equality, Composition, Identity, Inverse and Restriction (ix), as [u,v]∩K=[u,v]=[u,v]∩E[u,v]\cap K=[u,v]=[u,v]\cap E, both (g∣K)∣[u,v](g|_{K})|_{[u,v]} and g∣[u,v]g|_{[u,v]} are maps from [u,v][u,v] to R\mathbb{R} with value g(t)g(t) at every t∈[u,v]t\in[u,v]; so they are equal, by Basic Properties of Functions: Equality, Composition, Identity, Inverse and Restriction (i). Hence, by The Riemann Integral on a Closed Interval via Lower and Upper Integrals (vi), g∣Kg|_{K} is integrable on [u,v][u,v] if and only if this common map is bounded and integrable on [u,v][u,v], that is, if and only if gg is integrable on [u,v][u,v]; and in that case ∫uvg∣K\int_{u}^{v}g|_{K} and ∫uvg\int_{u}^{v}g are both its integral over [u,v][u,v], so they are equal, and ∫vug∣K=−∫uvg∣K=−∫uvg=∫vug\int_{v}^{u}g|_{K}=-\int_{u}^{v}g|_{K}=-\int_{u}^{v}g=\int_{v}^{u}g by The Riemann Integral on a Closed Interval via Lower and Upper Integrals (vii). For u∈Ku\in K, u∈Eu\in E, and ∫uug∣K=0=∫uug\int_{u}^{u}g|_{K}=0=\int_{u}^{u}g by The Riemann Integral on a Closed Interval via Lower and Upper Integrals (vii).

Step 1: joining partitions. Let c∈Rc\in\mathbb{R} with a<c<ba<c<b, let P1P_{1} be a partition of [a,c][a,c] and P2P_{2} a partition of [c,b][c,b]. By The Increasing Listing of a Partition: Endpoints, Lengths, Covering, Unions, Inserting a Point, Concatenation and Uniform Partitions (vii), P1∪P2P_{1}\cup P_{2} is a partition of [a,b][a,b], and by Bounds for Darboux Sums, Monotonicity under Refinement, the Supremum of the Lower Sums and the Infimum of the Upper Sums, and Concatenation of Partitions (v),

U(f,P1∪P2)=U(fac,P1)+U(fcb,P2),L(f,P1∪P2)=L(fac,P1)+L(fcb,P2).U(f,P_{1}\cup P_{2})=U(f_{ac},P_{1})+U(f_{cb},P_{2}),\qquad L(f,P_{1}\cup P_{2})=L(f_{ac},P_{1})+L(f_{cb},P_{2}).

Step 2: cutting a partition at cc. Let cc be as in Step 1 and let PP be a partition of [a,b][a,b]. By The Increasing Listing of a Partition: Endpoints, Lengths, Covering, Unions, Inserting a Point, Concatenation and Uniform Partitions (vi), P′=P∪{c}P'=P\cup\{c\} is a partition of [a,b][a,b] and a refinement of PP; and by The Increasing Listing of a Partition: Endpoints, Lengths, Covering, Unions, Inserting a Point, Concatenation and Uniform Partitions (viii), P1=P′∩[a,c]P_{1}=P'\cap[a,c] and P2=P′∩[c,b]P_{2}=P'\cap[c,b] are partitions of [a,c][a,c] and of [c,b][c,b] with P1∪P2=P′P_{1}\cup P_{2}=P'. By Step 1 and Bounds for Darboux Sums, Monotonicity under Refinement, the Supremum of the Lower Sums and the Infimum of the Upper Sums, and Concatenation of Partitions (ii),

L(f,P)≤L(f,P′)=L(fac,P1)+L(fcb,P2),U(f,P)≥U(f,P′)=U(fac,P1)+U(fcb,P2).L(f,P)\le L(f,P')=L(f_{ac},P_{1})+L(f_{cb},P_{2}),\qquad U(f,P)\ge U(f,P')=U(f_{ac},P_{1})+U(f_{cb},P_{2}).

Step 3: lower and upper integrals add. Let cc be as in Step 1. We show I‾(a,b)=I‾(a,c)+I‾(c,b)\underline{I}(a,b)=\underline{I}(a,c)+\underline{I}(c,b) and I‾(a,b)=I‾(a,c)+I‾(c,b)\overline{I}(a,b)=\overline{I}(a,c)+\overline{I}(c,b). By The Riemann Integral on a Closed Interval via Lower and Upper Integrals (iii), I‾(u,v)\underline{I}(u,v) is the supremum of the set of lower sums of fuvf_{uv} and I‾(u,v)\overline{I}(u,v) the infimum of its set of upper sums, whose elements are the L(fuv,R)L(f_{uv},R), respectively the U(fuv,R)U(f_{uv},R), for the partitions RR of [u,v][u,v], by The Riemann Integral on a Closed Interval via Lower and Upper Integrals (i) and The Riemann Integral on a Closed Interval via Lower and Upper Integrals (ii). A supremum is an upper bound of the set and lies below each of its upper bounds, and an infimum is a lower bound and lies above each lower bound, by Bounds, Least and Greatest Elements, Suprema and Infima for a Partial Order (iv), Bounds, Least and Greatest Elements, Suprema and Infima for a Partial Order (iii) and Bounds, Least and Greatest Elements, Suprema and Infima for a Partial Order (i).

Let P1P_{1} and P2P_{2} be partitions of [a,c][a,c] and [c,b][c,b]. By Step 1, L(fac,P1)+L(fcb,P2)=L(f,P1∪P2)≤I‾(a,b)L(f_{ac},P_{1})+L(f_{cb},P_{2})=L(f,P_{1}\cup P_{2})\le\underline{I}(a,b), so L(fac,P1)≤I‾(a,b)−L(fcb,P2)L(f_{ac},P_{1})\le\underline{I}(a,b)-L(f_{cb},P_{2}) by Rules of Arithmetic and Order in an Ordered Field (iv). Fixing P2P_{2}, I‾(a,b)−L(fcb,P2)\underline{I}(a,b)-L(f_{cb},P_{2}) is thus an upper bound of the lower sums of facf_{ac}, so I‾(a,c)≤I‾(a,b)−L(fcb,P2)\underline{I}(a,c)\le\underline{I}(a,b)-L(f_{cb},P_{2}), that is, L(fcb,P2)≤I‾(a,b)−I‾(a,c)L(f_{cb},P_{2})\le\underline{I}(a,b)-\underline{I}(a,c) (Rules of Arithmetic and Order in an Ordered Field (iv)). As P2P_{2} was arbitrary, I‾(a,b)−I‾(a,c)\underline{I}(a,b)-\underline{I}(a,c) is an upper bound of the lower sums of fcbf_{cb}, so I‾(c,b)≤I‾(a,b)−I‾(a,c)\underline{I}(c,b)\le\underline{I}(a,b)-\underline{I}(a,c), that is, I‾(a,c)+I‾(c,b)≤I‾(a,b)\underline{I}(a,c)+\underline{I}(c,b)\le\underline{I}(a,b). Conversely, let PP be a partition of [a,b][a,b], and let P1P_{1} and P2P_{2} be the partitions of Step 2; then L(f,P)≤L(fac,P1)+L(fcb,P2)≤I‾(a,c)+I‾(c,b)L(f,P)\le L(f_{ac},P_{1})+L(f_{cb},P_{2})\le\underline{I}(a,c)+\underline{I}(c,b) by Step 2 and Rules of Arithmetic and Order in an Ordered Field (iv). So I‾(a,c)+I‾(c,b)\underline{I}(a,c)+\underline{I}(c,b) is an upper bound of the lower sums of fab=ff_{ab}=f, and I‾(a,b)≤I‾(a,c)+I‾(c,b)\underline{I}(a,b)\le\underline{I}(a,c)+\underline{I}(c,b). The two inequalities give the equality for the lower integrals. The argument for the upper integrals is the same, with all inequalities reversed, upper sums in place of lower sums, lower bounds in place of upper bounds and infima in place of suprema.

Proof of (i). For u<vu<v in [a,b][a,b] let D(u,v)=I‾(u,v)−I‾(u,v)≥0D(u,v)=\overline{I}(u,v)-\underline{I}(u,v)\ge0. By Step 3, D(a,b)=D(a,c)+D(c,b)D(a,b)=D(a,c)+D(c,b). If ff is integrable on [a,c][a,c] and on [c,b][c,b], then D(a,c)=D(c,b)=0D(a,c)=D(c,b)=0, so D(a,b)=0D(a,b)=0 and ff is integrable on [a,b][a,b]. If ff is integrable on [a,b][a,b], then 0≤D(a,c)=−D(c,b)≤00\le D(a,c)=-D(c,b)\le0, so D(a,c)=D(c,b)=0D(a,c)=D(c,b)=0 and ff is integrable on [a,c][a,c] and on [c,b][c,b]. In that case, by the notation paragraph and Step 3, ∫abf=I‾(a,b)=I‾(a,c)+I‾(c,b)=∫acf+∫cbf\int_{a}^{b}f=\underline{I}(a,b)=\underline{I}(a,c)+\underline{I}(c,b)=\int_{a}^{c}f+\int_{c}^{b}f.

Proof of (ii). Let ff be integrable on [a,b][a,b] and u,v∈[a,b]u,v\in[a,b] with u<vu<v. First, ff is integrable on [u,b][u,b]: if u=au=a this is the hypothesis, as fab=ff_{ab}=f; if a<ua<u, then a<u<ba<u<b and (i) with c=uc=u applies. Let g=fubg=f_{ub}, bounded and integrable on [u,b][u,b]. If v=bv=b we are done. If v<bv<b, (i), applied to gg on [u,b][u,b] with c=vc=v, shows that gg is integrable on [u,v][u,v]. As g=f∣[u,b]g=f|_{[u,b]} and [u,v]⊆[u,b]⊆[a,b][u,v]\subseteq[u,b]\subseteq[a,b], (iv), with [a,b][a,b] for EE, [u,b][u,b] for KK and ff for gg, shows that ff is integrable on [u,v][u,v].

Proof of (iii). Let ff be integrable on [a,b][a,b]. For x,y∈[a,b]x,y\in[a,b] let I(x,y)=∫xyfI(x,y)=\int_{x}^{y}f: for x<yx<y this is ∫xyfxy\int_{x}^{y}f_{xy}, defined by (ii) and The Riemann Integral on a Closed Interval via Lower and Upper Integrals (vi); I(x,x)=0I(x,x)=0; and for y<xy<x, I(x,y)=−I(y,x)I(x,y)=-I(y,x), as in The Riemann Integral on a Closed Interval via Lower and Upper Integrals (vii). Hence I(y,x)=−I(x,y)I(y,x)=-I(x,y) for all x,y∈[a,b]x,y\in[a,b], checking the cases x<yx<y, x=yx=y and y<xy<x.

Let x,y,z∈[a,b]x,y,z\in[a,b] with x<y<zx<y<z. By (ii), h=fxzh=f_{xz} is bounded and integrable on [x,z][x,z], and by (i), applied to hh on [x,z][x,z] with c=yc=y, ∫xzh=∫xyh+∫yzh\int_{x}^{z}h=\int_{x}^{y}h+\int_{y}^{z}h. Here ∫xzh=I(x,z)\int_{x}^{z}h=I(x,z); and, as h=f∣[x,z]h=f|_{[x,z]} with [x,y]⊆[x,z][x,y]\subseteq[x,z] and [y,z]⊆[x,z][y,z]\subseteq[x,z], (iv), with [a,b][a,b] for EE, [x,z][x,z] for KK and ff for gg, gives ∫xyh=∫xyf=I(x,y)\int_{x}^{y}h=\int_{x}^{y}f=I(x,y) and ∫yzh=∫yzf=I(y,z)\int_{y}^{z}h=\int_{y}^{z}f=I(y,z). Thus I(x,z)=I(x,y)+I(y,z)I(x,z)=I(x,y)+I(y,z).

For u,v,w∈[a,b]u,v,w\in[a,b] let E(u,v,w)=I(u,v)+I(v,w)+I(w,u)E(u,v,w)=I(u,v)+I(v,w)+I(w,u). Since I(w,u)=−I(u,w)I(w,u)=-I(u,w), the assertion is E(u,v,w)=0E(u,v,w)=0. If two of u,v,wu,v,w coincide, E(u,v,w)=0E(u,v,w)=0 by I(s,s)=0I(s,s)=0 and I(s,t)=−I(t,s)I(s,t)=-I(t,s). Otherwise let x<y<zx<y<z be u,v,wu,v,w in increasing order. Then E(x,y,z)=I(x,y)+I(y,z)−I(x,z)=0E(x,y,z)=I(x,y)+I(y,z)-I(x,z)=0 by the preceding paragraph. EE is unchanged when its arguments are rotated, E(u,v,w)=E(v,w,u)E(u,v,w)=E(v,w,u), and changes sign when the first two are exchanged: E(v,u,w)=I(v,u)+I(u,w)+I(w,v)=−E(u,v,w)E(v,u,w)=I(v,u)+I(u,w)+I(w,v)=-E(u,v,w). Every arrangement of x,y,zx,y,z is a rotation of (x,y,z)(x,y,z) or of (y,x,z)(y,x,z), so E(u,v,w)=±E(x,y,z)=0E(u,v,w)=\pm E(x,y,z)=0.

Citations

Loading…

Dependencies

Uses0

Loading…

Comments

Log in to comment.

Loading…