Under the axiom of dependent choice, ([a,b],B[a,b],λ[a,b]) is a measure space (Measurable Functions and the Lebesgue Integral: Standing Notation (vii)).
Integrals over [a,t] as integrals over [a,b]. Let t∈[a,b] and f∈L1([a,b]). As [a,t]∈B[a,b] (Integrals over Subintervals of a Closed Interval: Notation (iii)), Integrals over Measurable Subsets (vi), applied to μ=λ[a,b] and A=[a,t], shows that f is integrable on [a,t]. Since λ[a,t] is the restriction of λ[a,b] to [a,t] (Integrals over Subintervals of a Closed Interval: Notation (iii)), this means f∣[a,t]∈L1(λ[a,t]) (Integrals of Functions over a Measurable Subset (iii)), and the integral of f over [a,t] with respect to λ[a,b] is ∫f∣[a,t]dλ[a,t]=∫[a,t]fdλ (Integrals of Functions over a Measurable Subset (iv)). By Integrals over Measurable Subsets (ii), the function f1[a,t], which equals f on [a,t] and 0 on [a,b]∖[a,t] (Indicator Function of a Subset (i)), lies in L1([a,b]), and
∫[a,t]fdλ=∫f1[a,t]dλ[a,b].(1)
Thus f∣[a,t]∈L1(λ[a,t]) and f1[a,t]∈L1([a,b]), so both sides of (1) are defined.
Let g∈L1([a,b];Rn), and let gt:[a,b]→Rn be the function with gt(r)=g(r) for r≤t and gt(r)=0 for r>t (Maps Defined by Cases (i)). As the origin of Rn has all components 0 (Euclidean Space Rn: n-Tuples of Real Numbers with Their Componentwise Operations (v)), the components of gt are (gt)k=gk1[a,t] for k∈[n]. Since gk∈L1([a,b]) (Integrable Vector- and Matrix-Valued Functions and Their Integrals (i)), the scalar case gives gk∣[a,t]∈L1(λ[a,t]) and gk1[a,t]∈L1([a,b]). The restriction g∣[a,t] is measurable (Measurable Maps: Generators, Composition, Continuous Maps, Restrictions, and Criteria for Real- and [0,∞]-Valued Maps (viii)) and has the components gk∣[a,t], and every component of gt is measurable, so gt is measurable (Vector- and Matrix-Valued Measurable Maps: Components, Entries and Continuous Images (i)); hence g∣[a,t]∈L1(λ[a,t];Rn) and gt∈L1([a,b];Rn) (Integrable Vector- and Matrix-Valued Functions and Their Integrals (i)). By Integrable Vector- and Matrix-Valued Functions and Their Integrals (ii), taken in both measure spaces, and (1), the k-th component of ∫gtdλ[a,b] is
∫gk1[a,t]dλ[a,b]=∫gk∣[a,t]dλ[a,t],
the k-th component of ∫g∣[a,t]dλ[a,t]=∫[a,t]gdλ; hence (Euclidean Space Rn: n-Tuples of Real Numbers with Their Componentwise Operations (iii))
∫[a,t]gdλ=∫gtdλ[a,b].(2)
Thus g∣[a,t]∈L1(λ[a,t];Rn) and gt∈L1([a,b];Rn), so both sides of (2) are defined.
A predicative description. Let I:L1([a,b])→R and J:L1([a,b];Rn)→Rn be the integrals with respect to λ[a,b] (Integrable Functions and Their Lebesgue Integral (iii) and Integrable Vector- and Matrix-Valued Functions and Their Integrals (ii)). With [a,b], the order of R, the origin 0 of Rn, L1([a,b];Rn) and J as parameters, let θ(y,g,t) be the formula
∃h(h∈L1([a,b];Rn)∧∀r(r∈[a,b]⇒((r≤t⇒(r,g(r))∈h)∧(t<r⇒(r,0)∈h)))∧(h,y)∈J),
predicative under the guard "g∈L1([a,b];Rn) and t∈[a,b]". Under this guard, the only h satisfying the first two conjuncts is gt, which lies in L1([a,b];Rn); so, by (2), θ(y,g,t) holds exactly for y=J(gt)=∫[a,t]gdλ. The same formula with L1([a,b]), I and 0∈R in place of L1([a,b];Rn), J and the origin holds, for f∈L1([a,b]) and t∈[a,b], exactly for y=I(f1[a,t])=∫[a,t]fdλ, by (1). By Predicativity of Expressions with Abbreviations, and Symbol-Free Forms of the Basic Defined Set Symbols (ii), ∫[a,t]fdλ and ∫[a,t]gdλ may therefore be written in expressions as defined set symbols in f,t and in g,t, used properly for f∈L1([a,b]), g∈L1([a,b];Rn) and t∈[a,b], with values in R and in Rn.
Proof of (i) and (ii). The expressions ∫[a,t]fdλ∈R and ∫[a,t]gdλ∈Rn in t∈[a,b] give exactly one such function X each, by Maps and Relations Given by Formulas (i).
Proof of (iii). By Maps and Relations Given by Formulas (ii) there is a function Φ:L1([a,b];Rn)×[a,b]→Rn with Φ(g,t)=∫[a,t]gdλ, and by Projection, Evaluation and Currying Maps (iii) there is exactly one function J:L1([a,b];Rn)→(Rn)[a,b] with J(g)(t)=Φ(g,t) for all g and t.
Proof of (iv). For z∈(Rn)[a,b] and r∈[a,b] we have (r,z(r))∈[a,b]×Rn, so F(r,z(r))∈Rn, and Maps and Relations Given by Formulas (ii) gives a function G:(Rn)[a,b]×[a,b]→Rn with G(z,r)=F(r,z(r)). By Maps Formed Pointwise: Constant Maps, Pairs of Values, Maps Applied to Values, and Families of Maps (iv) there is exactly one function Θ:(Rn)[a,b]→(Rn)[a,b] with Θ(z)(r)=G(z,r) for all z and r.
Proof of (v). By Maps Formed Pointwise: Constant Maps, Pairs of Values, Maps Applied to Values, and Families of Maps (iv) there is a function H:[a,b]→(Rn)[a,b] with H(t)=h(t,⋅) for every t∈[a,b], and ∫[a,t]h(t,s)ds=∫[a,t]H(t)dλ by Integrals over Subintervals of a Closed Interval: Notation (iii). As H(t)∈L1([a,b];Rn) for every t∈[a,b] by hypothesis, the expression ∫[a,t]H(t)dλ∈Rn in t∈[a,b] gives exactly one such function Y, by Maps and Relations Given by Formulas (i).