Why Eligibility Comes Before Differentiation
A quotient can look difficult to evaluate, but that alone does not justify using L'Hôpital's Rule. The rule has specific hypotheses: the original quotient must have one of two indeterminate forms, and the derivatives must satisfy conditions near the limit point. Checking those requirements first prevents a derivative calculation from producing an answer to a different problem.
In this tutorial, eligible means that the quotient has the required form and the functions meet the rule's relevant hypotheses. Eligibility does not guarantee that differentiating will make a limit easier, but it tells us when the rule is justified.
The Two Required Forms
For a quotient \(f(x)/g(x)\), the two forms for which L'Hôpital's Rule is stated directly are zero over zero and infinity over infinity. These labels describe the limits of numerator and denominator separately, not the value of the quotient.
For example, if the numerator tends to \(0\) while the denominator tends to \(5\), the quotient is not in either required form. If both tend to infinity in magnitude, their quotient might tend to \(0\), a finite nonzero number, or infinity; the separate numerator and denominator limits do not determine the quotient.
The rule also needs derivative information. In its standard form, if the derivative quotient has a limit \(L\), then the original quotient has that same limit, subject to the conditions below. The derivative quotient's limit may be finite or an infinite limit.
The condition \(g'(x)\ne 0\) is required near the limit, not merely at some points. It ensures that the derivative quotient is defined there and that the mean-value argument used to justify the rule can use \(g'\) in its denominator. The quotient \(f(x)/g(x)\) must also be defined on the interval where the limit is being considered.
A Derivation of the Finite-Endpoint Case
Here is why the rule works for the zero-over-zero case at a finite endpoint. It also explains why the point at which the derivative quotient is evaluated must approach the endpoint.
Consider a one-sided limit as \(x\to a^+\), with \(f(x)\to0\), \(g(x)\to0\), and \(g'(x)\ne0\) near \(a\). Define \(f(a)=g(a)=0\); these definitions make the functions continuous at \(a\), since their limits there are zero. For any nearby \(x>a\) with \(g(x)\ne0\), the Cauchy Mean Value Theorem on \([a,x]\) gives a point \(c\in(a,x)\) such that
Since \(f(a)=g(a)=0\), and \(g'(c)\ne0\), division gives
As \(x\to a^+\), the point \(c\), which lies between \(a\) and \(x\), must also approach \(a\). Therefore, if \(f'(x)/g'(x)\to L\), then \(f'(c)/g'(c)\to L\), giving \(f(x)/g(x)\to L\). For a left-hand limit, use the interval with \(x<a\); the same reasoning applies.
The Cauchy Mean Value Theorem itself gives a cross-multiplied equation. A ratio conclusion is justified here because the hypotheses ensure that the relevant denominator derivative is nonzero. This distinction matters: the theorem does not generally guarantee that a ratio of derivatives is defined at its supplied point.
Limits at Infinity and Finite-Endpoint Infinity Forms
For an infinity-over-infinity limit as \(x\to+\infty\), a fixed-endpoint argument works as follows. Choose a fixed \(b\) far enough along the interval that \(f'(t)/g'(t)\) is close to \(L\) whenever \(t>b\). Apply the Cauchy Mean Value Theorem on \([b,x]\). It gives \(c\in(b,x)\) with
Because \(c>b\), the derivative quotient at \(c\) is in the region where it is close to \(L\). Also, \(|g(x)|\to\infty\), so the fixed terms \(f(b)/g(x)\) and \(g(b)/g(x)\) tend to zero. These facts make \(f(x)/g(x)\) approach \(L\). For \(x\to-\infty\), the argument uses a fixed endpoint sufficiently far to the left.
At a finite endpoint where both functions diverge in magnitude, a change of variable connects the problem to the infinite-endpoint case. For instance, for \(x\to a^+\), set \(t=1/(x-a)\). Then \(x\to a^+\) corresponds to \(t\to+\infty\). If \(F(t)=f(a+1/t)\) and \(G(t)=g(a+1/t)\), the chain rule gives
Thus the derivative quotient has the same limiting behavior after the substitution, and the infinite-endpoint result applies. For \(x\to a^-\), use \(t=1/(a-x)\); the common chain-rule factor again cancels.
Worked Example: Checking a Zero-Over-Zero Quotient
Worked Example: Checking a Zero-Over-Zero Quotient
Decide whether L'Hôpital's Rule is eligible for \(\lim_{x\to0}(e^x-1)/x\), and find the limit of the derivative quotient.
First substitute the limiting value into the numerator and denominator:
The form is \(0/0\). Let \(f(x)=e^x-1\) and \(g(x)=x\). Both are differentiable near \(0\), and \(g'(x)=1\ne0\). The derivative quotient is
All the hypotheses are satisfied, so the rule is eligible and gives the original limit as \(1\). As a check on the derivative calculation, differentiating \(e^x-1\) gives \(e^x\), and differentiating \(x\) gives \(1\); substituting \(x=0\) in their quotient gives \(1/1=1\), consistent with the stated derivative-quotient limit.
Worked Example: Checking an Infinity-Over-Infinity Quotient
Worked Example: Checking an Infinity-Over-Infinity Quotient
Check eligibility and evaluate the derivative quotient for \(\lim_{x\to+\infty}(3x^2+1)/(x^2-4x)\).
As \(x\to+\infty\), the numerator \(3x^2+1\) and denominator \(x^2-4x\) both tend to \(+\infty\), so the original form is \(\infty/\infty\). Both functions are differentiable, and the denominator derivative is
For \(x>2\), \(2x-4>0\), so this derivative is nonzero throughout a sufficiently distant interval. Differentiating the numerator and forming the derivative quotient gives
The original quotient is eligible, and its limit is \(3\). As a check, divide numerator and denominator by \(x^2\): the quotient becomes \((3+1/x^2)/(1-4/x)\), which tends to \(3/1=3\). This independent check agrees with the derivative-quotient result.
Worked Example: The Initial Form Is Not Enough
Worked Example: The Initial Form Is Not Enough
Examine \(\lim_{x\to0}\sin x/x^2\). Does seeing \(0/0\) let us conclude that the limit exists by L'Hôpital's Rule?
Substitution gives \(\sin 0=0\) and \(0^2=0\), so the starting form is \(0/0\). The functions are differentiable, and the denominator derivative \(2x\) is nonzero for \(x\ne0\). But the derivative quotient is
As \(x\to0^+\), \(\cos x\to1\) and \(2x\to0^+\), so this quotient tends to \(+\infty\). As \(x\to0^-\), \(\cos x\to1\) and \(2x\to0^-\), so it tends to \(-\infty\). Thus the two-sided derivative-quotient limit does not exist. The two-sided L'Hôpital conclusion is unavailable; the rule cannot be used to claim a two-sided limit. In fact, the original quotient also tends to \(+\infty\) from the right and \(-\infty\) from the left, so its two-sided limit does not exist. This check confirms the need to examine the derivative quotient rather than stopping at \(0/0\).
Forms That Need More Than Direct Application
A quotient is not automatically eligible just because it is written as a fraction. For example, if the numerator and denominator approach nonzero finite values, the quotient is not \(0/0\) or \(\infty/\infty\). Applying L'Hôpital's Rule in that situation can give an incorrect answer.
Worked Example: A Quotient That Is Not Indeterminate
Consider \(\lim_{x\to0}(1+x)/(2+x)\). Substitution gives numerator \(1+0=1\) and denominator \(2+0=2\), so the form is \(1/2\), not \(0/0\) or \(\infty/\infty\). Direct substitution gives \(1/2\).
A wrong calculation would differentiate and report
That answer is incorrect: the original quotient tends to \(1/2\), not \(1\). The correction is to check the form first and use direct substitution when the functions are continuous and the denominator's limiting value is nonzero. Here the denominator tends to \(2\ne0\), so continuity verifies the substitution.
Other expressions may be called indeterminate forms, but they are not directly one of the two quotient forms in the theorem. Examples include a product with one factor tending to zero and another growing without bound, a difference of two expressions that both diverge, and expressions involving powers such as \(0^0\). They require an appropriate transformation or another method before the quotient version of L'Hôpital's Rule can be considered. Recognizing that a form needs transformation is different from assuming the rule applies to it as written.
A Practical Eligibility Check
Before differentiating, check the conditions in order. If any required condition is missing, do not claim that L'Hôpital's Rule justifies the limit.
Confirm that the limit is a quotient \(f(x)/g(x)\), defined near the point except possibly at the point itself.
Check that both functions tend to zero or that both tend to infinity in magnitude. A different form is not directly eligible.
Verify differentiability on the relevant punctured interval and that \(g'(x)\ne0\) throughout a sufficiently near or distant interval.
Determine whether \(f'(x)/g'(x)\) has a limit in the direction being considered. Without that limit, the rule does not establish the original quotient's limit.
Why Careful Eligibility Checks Matter
L'Hôpital's Rule relates two limits under stated assumptions; it is not a general instruction to differentiate the top and bottom of every quotient. A quotient may be directly evaluable, may require a transformation, or may meet the starting form while failing to produce a usable derivative-quotient limit. Separating those cases is an important part of sound limit reasoning.
Keep the hypotheses visible as you work: the initial form, differentiability, a nonzero denominator derivative near the limit, and the limiting behavior of the derivative quotient. If the point is finite, use the correct one-sided interval when the domain requires it. If it is infinite, check conditions sufficiently far along the relevant direction. Only after these checks should you use the rule.
Check Your Understanding
For each question, decide what the form and hypotheses tell you before attempting a limit calculation.
- What two forms permit direct use of L'Hôpital's Rule?
- For \(f(x)/g(x)\), why must \(g'(x)\) be nonzero near the limit point?
- For \(\lim_{x\to0}(1-\cos x)/x^2\), identify the initial form and write the derivative quotient. Does that quotient have a two-sided limit?
- Why does the form \(0\cdot\infty\) not permit direct application of the quotient version of the rule?
- What additional check is needed after confirming that a quotient has form \(0/0\) or \(\infty/\infty\)?
