Inverse Functions and the Logarithm
Inverse Function
If f is a function, then the function g is the inverse function of f if
g(f(x))=x
for all x in which f(x) can be calculated.
Details
The inverse of a function f is denoted by f−1, i.e.
f−1(f(x))=x
Examples
If f(x)=x2 for x<0 then the function g, defined as g(y)=y for y>0, is not the inverse of f since g(f(x))=x2=∣x∣=−x for x<0.
When the Inverse Exists: The Domain Question
Inverses do not always exist.
For an inverse of f to exist, f must be one-to-one, i.e. for each x, f(x) must be unique.
Figure: The function f(x)=x2 does not have an inverse since f(x)=1
has two possible solutions −1 and 1.
Examples
f(x)=x2 does not have an inverse since f(x)=1 has two possible solutions -1 and 1.
Note that iff f is a function, then the function g is the inverse function of f, if g(f(x))=x for all calculated values of x in f(x).
The inverse function of f is denoted by f−1, i.e. f−1(f(x))=x.
What is the inverse function, f−1, of f if f(x)=5+4x.
The simplest approach is to write y=f(x) and solve for x.
With
f(x)=5+4x
we write
y=5+4x
which we can now rewrite as
y−5=4x
and this implies
4y−5=x
And there we have it, very simple:
f−1(f(x))=4y−5
The Base 10 Logarithm
When x is a positive real number in x=10y, y is referred to as the base 10 logarithm of x and is written as:
y=log10(x)
or
y=log(x)
Details
If log(x)=a and log(y)=b, then x=10a and y=10b, and
x⋅y=10a⋅10b=10a+b
so that
log(xy)=a+b
Examples
log(100)log(1000)=2=3 If
log(2)≈0.3
then
10y=2
210=1024≈1000=103
therefore
2≈103/10
so
log(2)≈0.3
The Natural Logarithm
A logarithm with e as a base is referred to as the natural logarithm and is denoted as ln:
y=ln(x)
if
x=ey=exp(y)
Note that ln is the inverse of exp.
Figure: The curve depicts the function y=ln(x) and shows that ln is the inverse of exp.
Note that ln(1)=0 and when y=0 then e0=1.
Properties of Logarithm(s)
Logarithms transform multiplicative models into additive models, i.e.
ln(a⋅b)=lna+lnb
Details
This implies that any statistical model, which is multiplicative becomes additive on a log scale, e.g.
y=a⋅wb⋅xc
lny=(lna)+ln(wb)+ln(xc)
Next, note that
ln(x2)=ln(x⋅x)=lnx+lnx=2⋅lnx and similarly ln(xn)=n⋅lnx for any integer n.
In general ln(xc)=c⋅lnx for any real number c (for x>0).
Thus the multiplicative model (from above)
y=a⋅wb⋅xc
becomes
y=(lna)+b⋅lnw+c⋅lnx
which is a linear model with parameters (lna), b and c.
In addition, the log-transform is often variance-stabilizing.
The Exponential Function and the Logarithm
The exponential function and the logarithms are inverses of each other
x=ey↔y=lnx
Details
Note the properties:
ln(x⋅y)=ln(x)+ln(y)
and
ea⋅eb=ea+b
Examples
Solve the equation
10e1/3x+3=24
for x.
First, get the 3 out of the way:
10e1/3x=21
Then the 10:
e1/3x=2.1
Next, we can take the natural log of 2.1.
Since ln is an inverse function of e this would result in
31x=ln(2.1)
This yields
x=ln(2.1)⋅3
which is
≈2.23