Introduction
A derivative describes the instantaneous rate of change of a function. For the natural exponential, the result is especially simple: at every input, its derivative equals its value there. But this fact needs a proof from the definition of derivative; it is not enough to recognize a familiar pattern.
The proof turns on a limit that measures the change in \(e^h\) relative to a small increment \(h\). We will establish that limit from the compound-growth construction of the exponential, then use it in the difference quotient at an arbitrary point.
The difference quotient
For a function \(f\), its derivative at \(a\), when the limit exists, is defined by
Here \(h\) is a nonzero increment that approaches 0. The numerator is the change in the function's value, and dividing by \(h\) gives the average rate of change over that increment. The limit, if it exists, is the instantaneous rate of change at \(a\).
For \(f(x)=e^x\), the quotient becomes
The exponential law \(e^{a+h}=e^a e^h\) lets us factor this expression:
For fixed \(a\), the factor \(e^a\) does not depend on \(h\). Therefore the central question is whether the limit
exists, and what its value is.
Bounds from the compound-growth construction
The natural exponential can be constructed using the compound-growth limit: for each fixed real \(x\),
where \(n\) is taken large enough that the base is positive. In particular, \(e=e^1\). We use this construction to bound \(e^x\) when \(x\) is positive and small.
Suppose \(0<x<1\). For each positive integer \(n\), the binomial theorem gives
The term for \(k=0\) is 1, and the term for \(k=1\) is \(x\). Every remaining term is nonnegative, so
For an upper bound, when \(0\le k\le n\),
Consequently,
Taking the limit as \(n\to\infty\) preserves these inequalities. Thus, for \(0<x<1\),
These bounds are useful because both sides are close to 1 when \(x\) is close to 0. We now use them to prove the key limit separately for positive and negative increments.
Proof of the key limit
Let \(0<h<1\). Subtract 1 from the bounds \(1+h\le e^h\le 1/(1-h)\), then divide by the positive number \(h\). This gives
Write \(h=-t\), where \(0<t<1\). Use the positive-increment bounds for \(e^t\), and take reciprocals, using \(e^{-t}=1/e^t\).
In both cases, the quotient is trapped between expressions that approach 1. The squeeze theorem then gives the two-sided limit.
For \(0<h<1\), the positive-increment bounds give
Subtracting 1 and dividing by \(h>0\) yields
As \(h\to 0^+\), the lower bound approaches 1 and the upper bound also approaches 1. The squeeze theorem therefore gives
For negative increments, put \(h=-t\), where \(0<t<1\). The positive bounds for \(e^t\) are
All quantities here are positive, so taking reciprocals reverses the inequalities:
Since \(e^{-t}=1/e^t\), subtracting from 1 and dividing by \(t>0\) gives
As \(t\to 0^+\), both bounds approach 1. Also,
It follows that the quotient approaches 1 from the negative side as well. Since its one-sided limits agree, the two-sided limit exists and equals 1:
Deriving the derivative at any point
We can now return to the difference quotient at an arbitrary real number \(a\). Using the exponential law and factoring out the constant \(e^a\), we obtain
As \(h\to 0\), the second factor tends to 1 by the limit just proved. The constant multiple rule for limits then gives
The proof establishes the derivative directly from its definition. It does not assume a general rule for differentiating exponentials with more complicated exponents.
Worked examples
Worked Example: Derivative at \(x=0\)
Find the derivative of \(e^x\) at 0 from the difference quotient. Begin by substituting \(a=0\) into the definition:
The last equality is the key limit proved above. The general result gives \(e^0=1\), which agrees with this calculation. As a check, the difference quotient at 0 is exactly the expression whose limit was established, so its value is indeed 1.
Worked Example: Slope at \(x=1\)
Find the slope of the graph \(y=e^x\) at \(x=1\) by evaluating the difference quotient:
Thus the slope at the point \((1,e)\) is \(e\). Checking against the theorem, the derivative at \(x=1\) is \(e^1=e\), the same value obtained from the quotient.
Worked Example: Slope at \(x=-2\)
Evaluate the difference quotient at \(a=-2\):
The slope is \(e^{-2}=1/e^2\). Substituting \(a=-2\) into the theorem gives \(e^a=e^{-2}\), confirming the quotient calculation.
A common mistake: losing the factor \(e^a\)
A tempting but incorrect simplification is to treat \(e^{a+h}-e^a\) as though it were just \(e^h-1\). For instance, at \(a=1\), this mistake would give
The equality is wrong because \(e^{1+h}=e\cdot e^h\), so the correct factorization is
The factor \(e^a\) is constant with respect to \(h\), but it is not generally equal to 1 and cannot be dropped. Keeping it visible in the factorization prevents this error.
Why this result is distinctive
The natural exponential is distinguished by the fact that its instantaneous rate of change at each input equals its value at that input. For example, at \(x=0\), both the value and the slope are 1; at \(x=1\), both are \(e\). This property follows here from the compound-growth limit and the derivative definition, rather than from a memorized differentiation rule.
The argument also illustrates a useful proof strategy: rewrite a difference quotient so that its dependence on the increment is isolated in one standard limit. In this case the exponential law separates the fixed input \(a\) from the changing increment \(h\). When differentiating functions with more complicated exponents, additional reasoning will be needed; the present result concerns \(e^x\) itself.
Check Your Understanding
Use the definition of derivative and the key limit to answer these questions.
- What difference quotient defines the derivative of \(e^x\) at a real number \(a\)?
- Why is it necessary to establish the limit of \((e^h-1)/h\) from both sides?
- What bounds for \(e^h\) hold when \(0<h<1\), and how do they bound the quotient?
- How does the identity \(e^{-t}=1/e^t\) help handle negative increments?
- At which step in the derivative proof does the factor \(e^a\) arise, and why must it remain?