Reading guide · Proof index

The logarithm and the exponential

Jiří Lebl, Basic Analysis I–II, version 6.3. Free author edition of this section. Selection and attribution · Notation.

L5.4.1: Parts (i)–(iv) only: logarithm integral, derivative, strict increase, full range, endpoint limits, product law and uniqueness. Rational powers in (v) excluded.

Proof.

To prove existence, we define a candidate and show it satisfies all the properties. Let
L(x)≔∫1x1t dt.\begin{equation*} L(x) \coloneqq \int_1^x \frac{1}{t}\,dt . \end{equation*}
Obviously, i holds. Property ii holds via the second form of the fundamental theorem of calculus (Theorem 5.3.3).
To prove property iv, we change variables u=ytu=yt to obtain
L(x)=∫1x1t dt=∫yxy1u du=∫1xy1u du−∫1y1u du=L(xy)−L(y).\begin{equation*} L(x) = \int_1^{x} \frac{1}{t}\,dt = \int_y^{xy} \frac{1}{u}\,du = \int_1^{xy} \frac{1}{u}\,du - \int_1^{y} \frac{1}{u}\,du = L(xy)-L(y) . \end{equation*}
Let us prove iii. Property ii together with the fact that L′(x)=1 ⁣/ ⁣x>0L'(x) = \nicefrac{1}{x} > 0 for x>0,x > 0\text{,} implies that LL is strictly increasing and hence one-to-one. Let us show LL is onto. As 1 ⁣/ ⁣t≥1 ⁣/ ⁣2\nicefrac{1}{t} \geq \nicefrac{1}{2} when t∈[1,2],t \in [1,2]\text{,}
L(2)=∫121t dt≥1 ⁣/ ⁣2.\begin{equation*} L(2) = \int_1^2 \frac{1}{t} \,dt \geq \nicefrac{1}{2} . \end{equation*}
By induction, iv implies that for n∈N,n \in \N\text{,}
L(2n)=L(2)+L(2)+⋯+L(2)=nL(2).\begin{equation*} L(2^n) = L(2) + L(2) + \cdots + L(2) = n L(2) . \end{equation*}
Given y>0,y > 0\text{,} by the Archimedean property of the real numbers (notice L(2)>0L(2) > 0), there is an n∈Nn \in \N such that L(2n)>y.L(2^n) > y\text{.} The intermediate value theorem gives an x1∈(1,2n)x_1 \in (1,2^n) such that L(x1)=y.L(x_1) = y\text{.} Thus (0,∞)(0,\infty) is in the image of L.L\text{.} As LL is increasing, L(x)>yL(x) > y for all x>2n,x > 2^n\text{,} and so
lim⁡x→∞L(x)=∞.\begin{equation*} \lim_{x\to\infty} L(x) = \infty . \end{equation*}
Next 0=L(x ⁣/ ⁣x)=L(x)+L(1 ⁣/ ⁣x),0 = L(\nicefrac{x}{x}) = L(x) + L(\nicefrac{1}{x})\text{,} and so L(x)=−L(1 ⁣/ ⁣x).L(x) = - L(\nicefrac{1}{x})\text{.} Using x=2−n,x=2^{-n}\text{,} we obtain as above that LL achieves all negative numbers. And
lim⁡x→0L(x)=lim⁡x→0−L(1 ⁣/ ⁣x)=lim⁡x→∞−L(x)=−∞.\begin{equation*} \lim_{x \to 0} L(x) = \lim_{x \to 0} -L(\nicefrac{1}{x}) = \lim_{x \to \infty} -L(x) = - \infty . \end{equation*}
In the limits, note that only x>0x > 0 are in the domain of L.L\text{.}
Let us prove v. Fix x>0.x > 0\text{.} As above, iv implies L(xn)=nL(x)L(x^n) = n L(x) for all n∈N.n \in \N\text{.} We already found that L(x)=−L(1 ⁣/ ⁣x),L(x) = - L(\nicefrac{1}{x})\text{,} so L(x−n)=−L(xn)=−nL(x).L(x^{-n}) = - L(x^n) = -n L(x)\text{.} Then for m∈Nm \in \N
L(x)=L((x1/m)m)=mL(x1/m).\begin{equation*} L(x) = L\Bigl({(x^{1/m})}^m\Bigr) = m L\bigl(x^{1/m}\bigr) . \end{equation*}
Putting everything together for n∈Zn \in \Z and m∈N,m \in \N\text{,} we have L(xn/m)=nL(x1/m)=(n ⁣/ ⁣m)L(x).L(x^{n/m}) = n L(x^{1/m}) = (\nicefrac{n}{m}) L(x)\text{.}
Uniqueness follows using properties i and ii. Via the first form of the fundamental theorem of calculus (Theorem 5.3.1),
L(x)=∫1x1t dt\begin{equation*} L(x) = \int_1^x \frac{1}{t}\,dt \end{equation*}
is the unique function such that L(1)=0L(1) = 0 and L′(x)=1 ⁣/ ⁣x.L'(x) = \nicefrac{1}{x}\text{.}

L5.4.2: Parts (i)–(iv) and full uniqueness proof, with smooth inverse supplied by P3. Rational-power part (v) excluded.

Proof.

Again, we prove existence of such a function by defining a candidate and proving that it satisfies all the properties. The L=ln⁡L = \ln defined above is invertible. Let EE be the inverse function of L.L\text{.} Property i is immediate.
Property ii follows via the inverse function theorem, in particular via Lemma 4.4.1: LL satisfies all the hypotheses of the lemma, and hence
E′(x)=1L′(E(x))=E(x).\begin{equation*} E'(x) = \frac{1}{L'\bigl(E(x)\bigr)} = E(x) . \end{equation*}
Let us look at property iii. The function EE is strictly increasing since E′(x)=E(x)>0.E'(x) = E(x) > 0\text{.} As EE is the inverse of L,L\text{,} it must also be bijective. To find the limits, we use that EE is strictly increasing and onto (0,∞).(0,\infty)\text{.} For every M>0,M > 0\text{,} there is an x0x_0 such that E(x0)=ME(x_0) = M and E(x)≥ME(x) \geq M for all x≥x0.x \geq x_0\text{.} Similarly, for every ϵ>0,\epsilon > 0\text{,} there is an x0x_0 such that E(x0)=ϵE(x_0) = \epsilon and E(x)<ϵE(x) < \epsilon for all x<x0.x < x_0\text{.} Therefore,
lim⁡x→−∞E(x)=0andlim⁡x→∞E(x)=∞.\begin{equation*} \lim_{x\to -\infty} E(x) = 0 \qquad \text{and} \qquad \lim_{x\to \infty} E(x) = \infty . \end{equation*}
To prove property iv, we use the corresponding property for the logarithm. Take x,y∈R.x, y \in \R\text{.} As LL is bijective, find aa and bb such that x=L(a)x = L(a) and y=L(b).y = L(b)\text{.} Then
E(x+y)=E(L(a)+L(b))=E(L(ab))=ab=E(x)E(y).\begin{equation*} E(x+y) = E\bigl(L(a)+L(b)\bigr) = E\bigl(L(ab)\bigr) = ab = E(x)E(y) . \end{equation*}
Property v also follows from the corresponding property of L.L\text{.} Given x∈R,x \in \R\text{,} let aa be such that x=L(a)x = L(a) and
E(qx)=E(qL(a))=E(L(aq))=aq=E(x)q.\begin{equation*} E(qx) = E\bigl(qL(a)\bigr) = E\bigl(L(a^q)\bigr) = a^q = {E(x)}^q . \end{equation*}
Uniqueness follows from i and ii. Let EE and FF be two functions satisfying i and ii.
ddx(F(x)E(−x))=F′(x)E(−x)−E′(−x)F(x)=F(x)E(−x)−E(−x)F(x)=0.\begin{equation*} \frac{d}{dx} \Bigl( F(x)E(-x) \Bigr) = F'(x)E(-x) - E'(-x)F(x) = F(x)E(-x) - E(-x)F(x) = 0 . \end{equation*}
Therefore, by Proposition 4.2.6, F(x)E(−x)=F(0)E(−0)=1F(x)E(-x) = F(0)E(-0) = 1 for all x∈R.x \in \R\text{.} Doing the computation with F=E,F = E\text{,} we obtain E(x)E(−x)=1.E(x)E(-x) = 1\text{.} Then
0=1−1=F(x)E(−x)−E(x)E(−x)=(F(x)−E(x))E(−x).\begin{equation*} 0 = 1-1 = F(x)E(-x) - E(x)E(-x) = \bigl(F(x)-E(x)\bigr) E(-x) . \end{equation*}
Finally, E(−x)≠0E(-x) \neq 0
 1 
EE is a function into (0,∞)(0,\infty) after all. However, E(−x)≠0E(-x) \neq 0 also follows from E(x)E(−x)=1.E(x)E(-x) = 1\text{.} Therefore, we can prove uniqueness of EE given i and ii, even for functions E ⁣:R→R.E \colon \R \to \R\text{.}
for all x∈R.x \in \R\text{.} So F(x)−E(x)=0F(x)-E(x) = 0 for all x,x\text{,} and we are done.