The Exponential Functions article showed you that exp:R→(0,∞) is strictly increasing and surjective onto (0,∞). A strictly increasing function is automatically injective, so exp is a bijection — it has a unique inverse. That inverse is one of the most useful functions in all of analysis.
The natural logarithm
Definition. The natural logarithmln:(0,∞)→R is the inverse of exp. That is:
ln(exp(x))=xfor all x∈R,(1)exp(ln(y))=yfor all y>0.(2)
Intuitively, ln(y) answers the question: “To what power must e be raised to obtain y?” Because exp maps all of R bijectively onto (0,∞), this question always has a unique answer.
Immediate values
From the definition and what you already know about exp:
ln(1)=0, because exp(0)=1.
ln(e)=1, because exp(1)=e.
ln(en)=n for every n∈Z, because exp(n)=en.
These three anchor points are worth memorizing; they let you reason about ln quickly without a calculator.
The product rule for logarithms
Theorem (product rule). For all x,y>0:
ln(xy)=lnx+lny.(3)
Proof. Let a:=lnx and b:=lny, so exp(a)=x and exp(b)=y. The functional equation of the exponential — exp(a+b)=exp(a)exp(b) — gives:
xy=exp(a)⋅exp(b)=exp(a+b).
Applying ln to both sides and using identity (1):
ln(xy)=ln(exp(a+b))=a+b=lnx+lny.□
Intuitively, ln converts multiplication into addition — precisely the property that made logarithm tables indispensable for arithmetic long before calculators.
Two immediate corollaries follow from (3). Setting y=1/x gives ln(1/x)=−lnx. Setting y=x gives ln(x2)=2lnx. In fact, a full power rule holds for all real exponents.
The power rule for logarithms
Theorem (power rule). For all x>0 and r∈R:
ln(xr)=rlnx.(4)
Proof sketch. For r=n∈N, apply the product rule (3) repeatedly:
For rational r=p/q with q∈N+, note that q⋅ln(xp/q)=ln((xp/q)q)=ln(xp)=plnx, so ln(xp/q)=qplnx. For arbitrary real r, the equality extends by continuity of ln. □
The power rule (4) turns exponentiation into multiplication, just as (3) turned multiplication into addition.
The derivative of ln
Theorem. For all x>0:
dxdlnx=x1.(5)
Proof. Let y=lnx, so that x=exp(y). Differentiate both sides with respect to x using the inverse function theorem:
1=dxdexp(y)=exp(y)⋅dxdy=x⋅dxdy.
Solving for dxdy:
dxdy=x1.□
This is a striking result: the slope of the graph of ln at the point x is simply 1/x, with no multiplicative constant or extra function.
Monotonicity and behavior at the boundary
Because (lnx)′=1/x>0 for all x>0, the function ln is strictly increasing on its entire domain (0,∞).
The behavior at the two boundary points follows directly from the relationship with exp:
x→∞limlnx=+∞,x→0+limlnx=−∞.
The first limit holds because exp(t)→+∞ as t→+∞, so its inverse must also grow without bound. The second holds because exp(t)→0 as t→−∞. Together these confirm that ln is a bijection from (0,∞) onto all of R, the mirror image of the fact that exp maps R bijectively onto (0,∞).
Natural logarithm as an integral
There is a second way to arrive at ln that makes the derivative formula (5) transparent from the outset. Define:
L(x):=∫1xt1dt(x>0).(6)
By the Fundamental Theorem of Calculus, L′(x)=1/x and L(1)=0. One can verify that L satisfies the same product rule as ln (changing variables in the integral), and since both are continuous functions that agree at x=1 and have the same derivative, they are identical:
lnx=∫1xt1dt.
This integral representation is sometimes taken as the definition of ln, with Euler’s number then recovered as the unique e>0 satisfying ∫1et1dt=1.
General exponentials via the logarithm
Now that ln is available, you can define bx for any base b>0 and any real exponent x∈R — including irrationals.
Definition. For b>0:
bx:=exp(xlnb).(7)
When x is a rational number p/q, this agrees with the usual arithmetic meaning of bp/q (you can verify this using the power rule (4)). Definition (7) extends it seamlessly to all real exponents.
Derivative. Differentiating (7) by the chain rule:
dxdbx=exp(xlnb)⋅lnb=(lnb)bx.
The extra factor lnb is what distinguishes bx from ex: when b=e, lne=1 and the factor disappears, which is precisely why e is the “natural” base.
Logarithms to an arbitrary base
Definition. For b>0 with b=1, the logarithm to base b is:
logb(x):=lnblnx(x>0).(8)
By this definition, logb(x)=y if and only if by=x (substitute (7) to verify), which matches the familiar meaning from school algebra.
Change-of-base formula
From definition (8), for any two valid bases a and b:
In practice, scientific calculators provide log10 (common logarithm) and ln; the change-of-base formula lets you reach any other base through either one.
Summary
The natural logarithmln:(0,∞)→R is the inverse of exp: ln(exp(x))=x and exp(lny)=y.
Key values: ln1=0, lne=1, ln(en)=n for all n∈Z.
Product rule:ln(xy)=lnx+lny, derived from the functional equation exp(a+b)=exp(a)exp(b).
Power rule:ln(xr)=rlnx for all r∈R and x>0.
Derivative:(lnx)′=1/x, proved via the inverse function theorem.
ln is strictly increasing on (0,∞); lnx→+∞ as x→∞ and lnx→−∞ as x→0+.
Integral representation:lnx=∫1xt1dt, an alternative definition that makes the derivative formula immediate.
General exponential:bx:=exp(xlnb), with derivative (lnb)bx.
Logarithm base b:logb(x):=lnblnx; the change-of-base formula is loga(x)=logb(a)logb(x).