TheoremBase

The difference F(q)-F(p) is the integral of f over [p,q], so bounds on f over [p,q] bound it; this gives the Lipschitz estimate and continuity, and, with the bounds f(x0)-eps/2 and f(x0)+eps/2 near a point of continuity, the derivative.

Proof

Proof of (i). [a,b][a,b] is an interval containing the two distinct points aa and bb, by Basic Facts about Intervals of the Real Line (v), a∈[a,b]a\in[a,b], and ff is integrable on [u,v][u,v] for all u,v∈[a,b]u,v\in[a,b] with u<vu<v, by Additivity of the Riemann Integral over Adjacent Intervals, Subintervals and Oriented Integrals, and Integrals of Restrictions (ii). So A Predicative Formula for the Oriented Riemann Integral, and Maps Given by Integrals (ii), applied with K=[a,b]K=[a,b], c=ac=a and g=fg=f, gives exactly one map F:[a,b]β†’RF:[a,b]\to\mathbb{R} with F(x)=∫axfF(x)=\int_{a}^{x}f for every x∈[a,b]x\in[a,b].

Step 1: increments of FF. Let p,q∈[a,b]p,q\in[a,b] with p<qp<q. Then [p,q]βŠ†[a,b][p,q]\subseteq[a,b] by Basic Facts about Intervals of the Real Line (v) and Basic Facts about Intervals of the Real Line (ii), and ff is integrable on [p,q][p,q] by Additivity of the Riemann Integral over Adjacent Intervals, Subintervals and Oriented Integrals, and Integrals of Restrictions (ii). By Additivity of the Riemann Integral over Adjacent Intervals, Subintervals and Oriented Integrals, and Integrals of Restrictions (iii), applied with u=au=a, v=pv=p and w=qw=q, we have ∫aqf=∫apf+∫pqf\int_{a}^{q}f=\int_{a}^{p}f+\int_{p}^{q}f, that is, by (i),

F(q)βˆ’F(p)=∫pqf.(1)F(q)-F(p)=\int_{p}^{q}f.\qquad\text{(1)}

Suppose moreover that c,C∈Rc,C\in\mathbb{R} satisfy c≀f(s)≀Cc\le f(s)\le C for every s∈[p,q]s\in[p,q]. The restriction f∣[p,q]:[p,q]β†’Rf|_{[p,q]}:[p,q]\to\mathbb{R} has the values f(s)f(s) for s∈[p,q]s\in[p,q], so it satisfies the same bounds, and it is bounded and integrable on [p,q][p,q]. By Rules for the Riemann Integral: Sums, Scalar Multiples, Monotonicity, Bounds and Constants (iv), applied to f∣[p,q]f|_{[p,q]} on [p,q][p,q], and as ∫pqf=∫pqf∣[p,q]\int_{p}^{q}f=\int_{p}^{q}f|_{[p,q]} by The Riemann Integral on a Closed Interval via Lower and Upper Integrals (vi), (1) gives

c (qβˆ’p)≀F(q)βˆ’F(p)≀C (qβˆ’p).(2)c\,(q-p)\le F(q)-F(p)\le C\,(q-p).\qquad\text{(2)}

Proof of (ii). Let B∈RB\in\mathbb{R} with ∣f(t)βˆ£β‰€B|f(t)|\le B for every t∈[a,b]t\in[a,b], and let x,y∈[a,b]x,y\in[a,b]. If x=yx=y, both sides of the claimed inequality are 00. Otherwise let pp be the smaller and qq the larger of xx and yy, so p<qp<q, ∣F(y)βˆ’F(x)∣=∣F(q)βˆ’F(p)∣|F(y)-F(x)|=|F(q)-F(p)| and ∣yβˆ’x∣=qβˆ’p|y-x|=q-p (Difference and Quotient Identities in ℝ (i)). Every s∈[p,q]s\in[p,q] lies in [a,b][a,b] by Step 1, so βˆ’B≀f(s)≀B-B\le f(s)\le B. By (2) with c=βˆ’Bc=-B and C=BC=B, βˆ’B (qβˆ’p)≀F(q)βˆ’F(p)≀B (qβˆ’p)-B\,(q-p)\le F(q)-F(p)\le B\,(q-p), that is, ∣F(y)βˆ’F(x)∣=∣F(q)βˆ’F(p)βˆ£β‰€B (qβˆ’p)=Bβ€‰βˆ£yβˆ’x∣|F(y)-F(x)|=|F(q)-F(p)|\le B\,(q-p)=B\,|y-x|.

Proof of (iii). As ff is bounded, there is B∈RB\in\mathbb{R} with ∣f(t)βˆ£β‰€B|f(t)|\le B for every t∈[a,b]t\in[a,b]; then Bβ‰₯∣f(a)∣β‰₯0B\ge|f(a)|\ge0. Let x∈[a,b]x\in[a,b] and let Ξ΅>0\varepsilon>0 be real, and put Ξ΄=Ξ΅/(B+1)\delta=\varepsilon/(B+1), a real number with Ξ΄>0\delta>0. Every y∈[a,b]y\in[a,b] with ∣yβˆ’x∣<Ξ΄|y-x|<\delta satisfies, by (ii),

∣F(y)βˆ’F(x)βˆ£β‰€Bβ€‰βˆ£yβˆ’xβˆ£β‰€B δ<(B+1) δ=Ξ΅.|F(y)-F(x)|\le B\,|y-x|\le B\,\delta<(B+1)\,\delta=\varepsilon .

By The Real Line: Standing Notation and Background for Calculus on Intervals (vi), FF is therefore continuous at xx; as x∈[a,b]x\in[a,b] was arbitrary, FF is continuous on [a,b][a,b], again by The Real Line: Standing Notation and Background for Calculus on Intervals (vi).

Proof of (iv). Let x0∈[a,b]x_{0}\in[a,b] be a point at which ff is continuous, and let Ξ΅>0\varepsilon>0 be real. By The Real Line: Standing Notation and Background for Calculus on Intervals (vi), applied with Ξ΅/2\varepsilon/2, there is a real Ξ΄>0\delta>0 such that every s∈[a,b]s\in[a,b] with ∣sβˆ’x0∣<Ξ΄|s-x_{0}|<\delta satisfies ∣f(s)βˆ’f(x0)∣<Ξ΅/2|f(s)-f(x_{0})|<\varepsilon/2. Let t∈[a,b]t\in[a,b] with 0<∣tβˆ’x0∣<Ξ΄0<|t-x_{0}|<\delta, and let pp be the smaller and qq the larger of tt and x0x_{0}, so p<qp<q and qβˆ’p=∣tβˆ’x0∣q-p=|t-x_{0}|. Every s∈[p,q]s\in[p,q] satisfies s∈[a,b]s\in[a,b] and ∣sβˆ’x0βˆ£β‰€qβˆ’p<Ξ΄|s-x_{0}|\le q-p<\delta, as x0x_{0} is one of pp and qq; hence f(x0)βˆ’Ξ΅/2≀f(s)≀f(x0)+Ξ΅/2f(x_{0})-\varepsilon/2\le f(s)\le f(x_{0})+\varepsilon/2. By (2) with c=f(x0)βˆ’Ξ΅/2c=f(x_{0})-\varepsilon/2 and C=f(x0)+Ξ΅/2C=f(x_{0})+\varepsilon/2, and dividing by qβˆ’p>0q-p>0,

f(x0)βˆ’Ξ΅2≀F(q)βˆ’F(p)qβˆ’p≀f(x0)+Ξ΅2.f(x_{0})-\frac{\varepsilon}{2}\le\frac{F(q)-F(p)}{q-p}\le f(x_{0})+\frac{\varepsilon}{2}.

If t=qt=q the quotient here is F(t)βˆ’F(x0)tβˆ’x0\frac{F(t)-F(x_{0})}{t-x_{0}}; if t=pt=p, numerator and denominator of F(t)βˆ’F(x0)tβˆ’x0\frac{F(t)-F(x_{0})}{t-x_{0}} are the negatives of those of F(q)βˆ’F(p)qβˆ’p\frac{F(q)-F(p)}{q-p} (Difference and Quotient Identities in ℝ (i)), so the two quotients agree again. Hence

∣F(t)βˆ’F(x0)tβˆ’x0βˆ’f(x0)βˆ£β‰€Ξ΅2<Ξ΅.\Bigl|\frac{F(t)-F(x_{0})}{t-x_{0}}-f(x_{0})\Bigr|\le\frac{\varepsilon}{2}<\varepsilon .

As Ξ΅\varepsilon was arbitrary and Ξ΄\delta was chosen after Ξ΅\varepsilon, f(x0)f(x_{0}) is a derivative of FF at x0x_{0}. So FF is differentiable at x0x_{0} by Derivative of a Real Function on an Interval (ii), and since this derivative is unique by Derivative of a Real Function on an Interval (ii) and Uniqueness of the Derivative of a Real Function on an Interval (i), Fβ€²(x0)=f(x0)F'(x_{0})=f(x_{0}).

Citations

Loading…

Dependencies

Uses0

Loading…

Comments

Log in to comment.

Loading…