Differential Calculus
L'Hopital's Rule and Indeterminate Forms
Every derivative you have ever computed from the definition was a limit, so it is no accident that derivatives can be used to evaluate limits. L'Hopital's Rule is that tool: under precise hypotheses it replaces the limit of a quotient by the limit of the quotient of the derivatives. It sits immediately after the differentiation rules and feeds forward into curve sketching and optimisation, where you constantly need the end behaviour of expressions like and . It is also the most over-applied theorem in first-year calculus, so half of this topic is learning when not to reach for it.
What "indeterminate" actually means
A form is indeterminate when the limiting behaviour of the pieces does not by itself determine the limit of the whole. All four limits below are of the form as , and they do four different things:
So "" is not a number and never an answer. It is a report that the limit laws gave no information and that more work is required.
The seven indeterminate forms: , , , , , , .
Everything else is determinate: the answer is forced by the pieces, and l'Hopital's Rule must not be used. Here denotes a finite nonzero constant.
| Form as | Forced value | Note |
|---|---|---|
| just substitute | ||
| infinite | the size blows up; the sign, and whether a two-sided limit exists at all, needs a one-sided look at the denominator — is neither nor | |
| infinite | ||
| only is indeterminate | ||
| only is indeterminate | ||
| (base ) | ||
| with | with the value is |
Notice how narrow the exponential indeterminates are: only base with an infinite exponent, base with exponent , and base with exponent .
L'Hopital's Rule, stated with its hypotheses
Theorem. Let be a real number or , and suppose all four of the following hold.
- Differentiability. and are differentiable on an open interval containing , except possibly at itself. (If , read this as: differentiable on some interval .) The values and are irrelevant and need not exist.
- Nonvanishing derivative. for every in with .
- The form is indeterminate. Either and , or and (signs may differ).
- The new limit exists. , where is a real number or .
Conclusion. .
The theorem holds verbatim for one-sided limits and (with a one-sided interval) and for . A refinement worth knowing: in the case it is enough that ; what does is not needed.
Why it is true (the case at a finite , with and defined at by , which makes them continuous there). The engine is the Cauchy Mean Value Theorem, and it carries hypotheses of its own.
Cauchy Mean Value Theorem. If and are continuous on the closed interval , differentiable on the open interval , and for every in , then and there is at least one in with
Hypotheses 1 and 2 of l'Hopital deliver exactly those conditions on the interval between and , so for each such there is a strictly between and with
(the middle step is where is needed, and hypothesis 2 is what supplies it). As the point is trapped between and , so too and the quotient inherits the limit .
Every hypothesis earns its place.
- Drop 3. by substitution, but . Differentiating a quotient that was never indeterminate simply gives the wrong number.
- Drop 4. (divide by ), yet oscillates and has no limit. When step 4 fails the rule is silent — it does not say the original limit fails to exist.
- Drop 2. Take and as . Both tend to , and since and , but oscillates between and forever. The culprit is that at every , so no interval is free of zeros: hypothesis 2 fails, and the conclusion fails with it. (Strictly, is not even defined at those points, so the displayed limit is taken along the set where it does make sense — precisely the loophole hypothesis 2 is written to close.)
Repeated application
If is again of type or , and and themselves satisfy the hypotheses, apply the rule again. Re-verify the form at every single stage. The moment the quotient stops being indeterminate, stop and evaluate.
Growth hierarchy
For every and every , as the logarithm grows more slowly than any positive power of , which in turn grows more slowly than any exponential:
Both are proved by l'Hopital (the second needs one application for every unit of , rounded up). Memorising the hierarchy saves you from grinding out four applications of the rule on .
Converting the other five forms into a quotient
| Form | Recognise it as | Convert by | You land on |
|---|---|---|---|
| with , | or | or | |
| difference of two blow-ups | common denominator; multiply by the conjugate; factor out the dominant term; or | usually | |
| , , | take logarithms | ||
| , , | take logarithms | ||
| , , | take logarithms |
For you have a genuine choice, so pick the arrangement whose derivatives are simpler. Pushing a logarithm into the denominator is almost always the wrong choice: is worse than what you started with, whereas is a huge simplification. Rule of thumb: differentiate the logarithm, invert the power.
The exponential recipe. To evaluate where for all near (this positivity is a genuine requirement — otherwise need not even be defined):
- Set , so that .
- Compute by turning the form into a quotient.
- Since is continuous, . If then the limit is ; if the limit is .
Step 3 is the step students forget.
When the rule does not apply, or loops
- The form is not indeterminate. Check first, every time. This is the single most common error.
- The derivative quotient has no limit. Then the rule tells you nothing; fall back on algebra or the Squeeze Theorem.
- The rule loops. In (form ) one application gives ; applying the rule to that gives , exactly where you started. Divide top and bottom by instead: .
- The rule makes things worse. Roots, and quotients of polynomials of high degree, usually yield to algebra faster.
- The rule is circular as a proof. Applying it to does return the correct value , and once is on the books that computation is a sound deduction — but you may never cite the rule to establish the limit, because the derivative formula is itself proved from it. The same objection applies to any limit of the form : that quotient is the definition of , so differentiating top and bottom assumes what you were asked to find.
Worked example A — an needing two applications
Evaluate .
As we get and : the form is , which is indeterminate. Combine over a common denominator (legitimate for , where ):
Now and , so the form is . Both functions are differentiable everywhere, and for , so hypothesis 2 holds there:
Check the new form before continuing: and , still . The next denominator derivative is , which tends to and so is nonzero near :
This last form is , not indeterminate, so we stop. Answer: . (Sanity check: and , so the quotient behaves like .)
Worked example B — a
Evaluate .
The base and the exponent , so the form is : indeterminate. For we have , so both the power and its logarithm are defined. Put :
The numerator tends to and the denominator to : form . Differentiate top and bottom separately, using the chain rule on the top and noting for :
\lim_{x\to 0}\frac{3\ln\left(\cos 2x\right)}{x^{2}} &= \lim_{x\to 0}\frac{3\cdot\dfrac{-2\sin 2x}{\cos 2x}}{2x}\\ &= \lim_{x\to 0}\frac{-6\tan 2x}{2x}\\ &= \lim_{x\to 0}\left(-3\cdot\frac{\tan 2x}{x}\right)\\ &= -3\cdot 2=-6 \end{aligned}$$ (The last step uses the standard limit $\lim_{x\to 0}\frac{\tan 2x}{x}=2$; one more application of l'Hopital gives the same thing, since $\frac{-12\sec^{2}2x}{2}\to-6$.) So $\ln y\to-6$, and because $\exp$ is continuous, $\lim_{x\to 0} y=e^{-6}$. **Answer: $e^{-6}\approx 0.00248$.** (Sanity check: $\ln\cos 2x\approx-\frac{\left(2x\right)^{2}}{2}=-2x^{2}$, so $\ln y\approx\frac{3\left(-2x^{2}\right)}{x^{2}}=-6$.) ### Common mistakes | Wrong | Right | |---|---| | $\lim_{x\to 0}\frac{\sin x}{x+1}=\lim\frac{\cos x}{1}=1$ | the form is $\frac{0}{1}$, not indeterminate — substitute, and the limit is $0$ | | Differentiating with the Quotient Rule: $\lim\frac{f'g-fg'}{g^{2}}$ | l'Hopital uses $\frac{f'}{g'}$: numerator and denominator are differentiated **separately** | | $\lim_{x\to 0}\frac{e^{x}-1}{x^{2}}=\lim\frac{e^{x}}{2x}=\lim\frac{e^{x}}{2}=\frac{1}{2}$ | after one application the form is $\frac{1}{0}$, which is *not* indeterminate, so you must stop: $\lim_{x\to 0^{+}}=+\infty$, $\lim_{x\to 0^{-}}=-\infty$, and the two-sided limit does not exist | | "$\frac{f'}{g'}$ has no limit, therefore $\frac{f}{g}$ has no limit" | the rule is silent; try algebra instead. $\frac{x+\sin x}{x}\to 1$ even though $\frac{1+\cos x}{1}$ has no limit | | Answering $-6$ for $\lim_{x\to 0}\left(\cos 2x\right)^{3/x^{2}}$ | $-6$ is $\lim\ln y$; the answer is $e^{-6}$ | | Citing the rule as a **proof** that $\lim_{x\to 0}\frac{\sin x}{x}=1$ | circular: $(\sin x)'=\cos x$ is itself proved from this limit, so the rule cannot establish it — the value $1$ it returns is nevertheless correct | | Treating $\frac{1}{0}$, $0\cdot 0$ or $0^{\infty}$ as indeterminate | all determinate: infinite, $0$, and $0$ respectively | | $\lim_{x\to\infty}\left(1+\frac{1}{x}\right)^{x}=1^{\infty}=1$ | $1^{\infty}$ is indeterminate; the true value is $e$ | | Handling $f^{g}$ with $\left(f^{g}\right)'=gf^{g-1}$ or $f^{g}\ln f$ | neither is valid when both are functions of $x$: take logarithms, or write $f^{g}=e^{g\ln f}$ and use the chain rule | | Applying the rule directly to $\lim_{x\to 0^{+}}x^{x}$ | the theorem is about **quotients**; convert with logarithms first (the answer is $1$) | | Writing $\ln f$ without checking $f>0$ | $\left(\sin x\right)^{x}$ only makes sense where $\sin x>0$, so only $x\to 0^{+}$ is meaningful | | Chasing $\lim_{x\to\infty}\frac{\sqrt{x^{2}+1}}{x}$ with the rule | it loops back to itself; divide by $x$ to get $\sqrt{1+\frac{1}{x^{2}}}\to 1$ | | Quoting the two-sided rule when $f$ or $g$ exists only on one side of $a$ | $\frac{\ln\left(1+x\right)}{\sqrt{x}}$ is defined only for $x>0$, so only $\lim_{x\to 0^{+}}$ can be asked (its value is $0$); use the one-sided version of the theorem |Key terms
- Indeterminate form
- 0/0 form
- Infinity over infinity form
- Determinate form
- L'Hopital's Rule
- Hypotheses of l'Hopital's Rule
- Cauchy Mean Value Theorem
- Repeated application
- 0 times infinity
- Infinity minus infinity
- 1 to the infinity form
- 0 to the 0 form
- Infinity to the 0 form
- Logarithmic conversion
- Continuity of the exponential function
- Looping application
- Growth hierarchy
- Circular reasoning
Practice Problems
Evaluate . State the form of the limit and check the hypotheses of l'Hopital's Rule before you use it.
Show hint
Substitute into the numerator and the denominator separately first — you are not allowed to differentiate until you know what form you are looking at.
Show answer
Step 1 — identify the form. As ,
so the form is : indeterminate, and l'Hopital's Rule is a candidate.
Step 2 — check the remaining hypotheses. Write and . Both are differentiable at every real number, so hypothesis 1 holds on any interval around . Also , and whenever , so hypothesis 2 holds on the interval .
Step 3 — differentiate top and bottom separately.
Step 4 — evaluate the new limit. The quotient is continuous at and its denominator there is , so direct substitution applies:
This limit exists, so hypothesis 4 holds and the rule may be applied:
Answer: .
Independent check without l'Hopital. Using the standard limits and ,
Numerically at the quotient is , consistent with .
For each limit below, say whether the form is indeterminate, name the form, and evaluate the limit. Use l'Hopital's Rule only where it is legitimate.
(a)
(b)
(c)
(d)
(e)
Show hint
For each one, work out separately what the numerator/base does and what the denominator/exponent does, then compare with the list of seven indeterminate forms.
Show answer
(a) Numerator ; denominator . The form is , which is not indeterminate. The quotient is continuous at (denominator nonzero), so substitute:
Applying l'Hopital here would give , which is simply wrong.
(b) Both and : form , indeterminate. On both functions are differentiable and , so the rule applies:
(c) Numerator ; denominator . The form is , which is not indeterminate — it is determinate and infinite:
(At the quotient is about .) L'Hopital would give : badly wrong.
(d) Base , exponent : the form is , which is not indeterminate. To see the value, note that for
which is a determinate product. Since is continuous, .
(e) Base , exponent : form , indeterminate. For the base is positive, so put and take logarithms:
This is . Substitute , so :
Numerator and denominator : form . Both functions are differentiable near and the denominator's derivative is , so
Hence and .
Summary: (a) not indeterminate, ; (b) , ; (c) not indeterminate, ; (d) not indeterminate (), ; (e) , .
Evaluate using l'Hopital's Rule, checking the form at every stage. Then confirm your answer by an algebraic method that does not use the rule.
Show hint
One application will not finish the job — but before you apply it a second time, look again at what the new numerator and new denominator are doing.
Show answer
Stage 0 — the form. As , and : form , indeterminate. Both functions are polynomials, hence differentiable everywhere.
Stage 1. With and we get and , and for all , so hypothesis 2 holds on :
Stage 2 — re-check the form. Now and : still , so we may go again. The new derivatives are and , and :
This limit exists, which validates stage 2, which in turn validates stage 1. Therefore
Answer: .
Independent check by algebra. Divide numerator and denominator by (legitimate for ):
Numerically at : , consistent with . Note the shortcut this illustrates: for a rational function, the limit at infinity is the ratio of the leading coefficients when the degrees match — much faster than the rule.
Evaluate . There are two ways to write this product as a quotient; explain which one you choose and why the other is a bad idea.
Show hint
A product can be written as or as . Ask yourself which of and you would rather differentiate, and which you would rather push into a denominator.
Show answer
Step 1 — identify the form. As : and . The form is , which is indeterminate. The rule applies only to quotients, so the product must be rewritten.
Step 2 — choose the rearrangement. The two options are
Choose the first. Differentiating gives the very simple , and differentiating the power keeps a power. The second option is a bad idea because
so the derivative quotient becomes , which is more complicated than the original — the same indeterminate form with the logarithm now squared.
Step 3 — check the hypotheses. On both and are differentiable, and there. The form of is , which is the case.
Step 4 — apply the rule and simplify.
\lim_{x\to 0^{+}}\frac{\ln x}{x^{-2}} &= \lim_{x\to 0^{+}}\frac{\dfrac{1}{x}}{-2x^{-3}}\\ &= \lim_{x\to 0^{+}}\frac{1}{x}\cdot\frac{x^{3}}{-2}\\ &= \lim_{x\to 0^{+}}\left(-\frac{x^{2}}{2}\right)\\ &= 0 \end{aligned}$$ The final expression is a polynomial, so no indeterminacy remains and we stop. **Answer: $\displaystyle\lim_{x\to 0^{+}}x^{2}\ln x=0$.** *Independent check.* At $x=10^{-3}$: $x^{2}\ln x=10^{-6}\times\left(-6.9078\right)\approx-6.9\times 10^{-6}$, and at $x=10^{-6}$ it is about $-1.4\times 10^{-11}$ — collapsing to $0$. This is the general fact that a positive power of $x$ beats $\ln x$ as $x\to 0^{+}$, the mirror image of the growth hierarchy at infinity, where $x^{p}$ outgrows $\ln x$.Evaluate , justifying the form at every stage and stating where you could stop early.
Show hint
Each application removes only one power of from the denominator, so expect to go round more than once — and check the form each time before you go again.
Show answer
Stage 0 — the form. and : form . Both functions are differentiable everywhere, and for every , so hypotheses 1 and 2 hold on any interval around .
Stage 1.
Re-check: and , still ; the new denominator derivative is nonzero for .
Stage 2.
You may stop here, because is a standard limit:
Stage 3 (if you prefer to continue). The form of is again , and the denominator derivative is , so a third application is permitted:
Read this stage as bookkeeping, not as a proof: it differentiates , and is itself derived from , so stage 3 may never be used to establish that standard limit. Stage 2 is the honest place to stop. Either way the limit exists, and working backwards validates stage 2, then stage 1. Therefore
Answer: .
Independent check. The Maclaurin series gives , so
Numerically at : , and .
Evaluate .
Show hint
Each piece blows up, so the limit laws cannot be used termwise. Put the two fractions over a single denominator first and see what form appears.
Show answer
Step 1 — identify the form. As : so , and so . The form is , indeterminate (from the left both pieces tend to , giving , equally indeterminate). You may not split the limit across the subtraction, because neither piece has a limit.
Step 2 — make it a single quotient. For with both denominators are nonzero, so
Set and . Then and , and both are continuous at , so the form is .
Step 3 — first application. Both and are differentiable on :
Is near ? We have for , so is strictly increasing; since , it follows that for every . Hypothesis 2 holds. Hence
Step 4 — re-check the form. Numerator ; denominator . Still , so apply again. The new derivatives are
and the second is nonzero for all .
Step 5 — finish.
The last expression is continuous at with nonzero denominator, so no indeterminacy remains and we stop. Working backwards, both applications were valid, so
Answer: .
Independent numerical check. At : and , difference . At : and , difference . Both sides approach .
Evaluate .
Show hint
Identify what the base and the exponent each tend to, then take the natural logarithm of the expression before you do anything else. Remember to undo the logarithm at the end.
Show answer
Step 1 — identify the form. For the base is positive, and
so the form is : indeterminate. (Do not write .)
Step 2 — take logarithms. Put , valid for where the base is positive. Then
This has form , still indeterminate, so it must be made into a quotient.
Step 3 — make a quotient. Move the factor downstairs:
Numerator and denominator , so this is . Both functions are differentiable on , and the denominator's derivative is nonzero there, so hypotheses 1 and 2 hold.
Step 4 — apply the rule.
\frac{d}{dx}\left[\ln\left(x+2\right)-\ln\left(x-1\right)\right] &= \frac{1}{x+2}-\frac{1}{x-1}\\ &= \frac{\left(x-1\right)-\left(x+2\right)}{\left(x+2\right)\left(x-1\right)}\\ &= \frac{-3}{\left(x+2\right)\left(x-1\right)} \end{aligned}$$ and $\dfrac{d}{dx}\left(\dfrac{1}{3x}\right)=-\dfrac{1}{3x^{2}}$. Therefore $$\begin{aligned} \lim_{x\to\infty}\ln y &= \lim_{x\to\infty}\frac{\dfrac{-3}{\left(x+2\right)\left(x-1\right)}}{-\dfrac{1}{3x^{2}}}\\ &= \lim_{x\to\infty}\frac{-3}{\left(x+2\right)\left(x-1\right)}\cdot\left(-3x^{2}\right)\\ &= \lim_{x\to\infty}\frac{9x^{2}}{\left(x+2\right)\left(x-1\right)}\\ &= \lim_{x\to\infty}\frac{9}{\left(1+\dfrac{2}{x}\right)\left(1-\dfrac{1}{x}\right)}\\ &= \frac{9}{1\cdot 1}=9 \end{aligned}$$ **Step 5 — undo the logarithm.** Since $\exp$ is continuous and $\ln y\to 9$, $$\lim_{x\to\infty}\left(\frac{x+2}{x-1}\right)^{3x}=e^{9}$$ **Answer: $e^{9}\approx 8103.1$.** *Independent check.* Write $\frac{x+2}{x-1}=1+\frac{3}{x-1}$, so the expression is $\left(1+\frac{3}{x-1}\right)^{3x}$. Using $\left(1+\frac{a}{n}\right)^{n}\to e^{a}$, the exponent behaves like $3x\cdot\frac{3}{x-1}\to 9$, giving $e^{9}$. Numerically at $x=1000$: $\ln y=3000\ln\left(\frac{1002}{999}\right)=3000\times 0.0029985=8.9955$, and $e^{8.9955}\approx 8067$, closing in on $e^{9}\approx 8103$.Evaluate . Explain why the limit is only posed as a right-hand limit.
Show hint
Name the form first. Then take logarithms and choose which of the two factors to send into the denominator — the one whose reciprocal is easy to differentiate.
Show answer
Step 1 — why one-sided. The power with a variable exponent is defined by , which requires . Since exactly for (near the origin), the expression is undefined for close to . Only the right-hand limit can be asked.
Step 2 — identify the form. As : the base and the exponent . The form is : indeterminate.
Step 3 — take logarithms. Let for . Then
which has form : still indeterminate.
Step 4 — make a quotient. Keep the logarithm upstairs (it differentiates well) and send downstairs:
As the numerator and the denominator , so this is the case. On both functions are differentiable, and the denominator's derivative is never zero there.
Step 5 — apply the rule.
\lim_{x\to 0^{+}}\frac{\ln\left(\sin x\right)}{\dfrac{1}{x}} &= \lim_{x\to 0^{+}}\frac{\dfrac{\cos x}{\sin x}}{-\dfrac{1}{x^{2}}}\\ &= \lim_{x\to 0^{+}}\left(-\frac{x^{2}\cos x}{\sin x}\right)\\ &= \lim_{x\to 0^{+}}\left(-x\cos x\right)\cdot\frac{x}{\sin x} \end{aligned}$$ Now $-x\cos x\to 0$ and $\frac{x}{\sin x}\to 1$ (the reciprocal of the standard limit, allowed because the limit $1$ is nonzero), so by the Product Law $$\lim_{x\to 0^{+}}\ln y=0\cdot 1=0$$ **Step 6 — undo the logarithm.** By continuity of $\exp$, $$\lim_{x\to 0^{+}}\left(\sin x\right)^{x}=e^{0}=1$$ **Answer: $1$.** *Independent check.* $\ln y=x\ln\left(\sin x\right)=x\ln\left(\dfrac{\sin x}{x}\right)+x\ln x$. The first term tends to $0\cdot\ln 1=0$ and the second is the standard limit $x\ln x\to 0$, so $\ln y\to 0$ and $y\to 1$. Numerically at $x=0.01$: $\ln y=0.01\times\ln\left(0.0099998\right)\approx-0.04605$, so $y\approx 0.955$ — creeping up towards $1$ (the approach is slow because $x\ln x$ decays slowly).Show that repeated use of l'Hopital's Rule on produces an endless loop, and then evaluate the limit correctly.
Show hint
Apply the rule once, simplify the result into a single quotient, and then apply it to that quotient — compare what you get with where you started.
Show answer
Step 1 — the form is genuinely indeterminate. As , and : form . So the rule is at least allowed; the trouble is that it is useless here.
Step 2 — first application. With and ,
so the rule gives
provided the right-hand limit exists — and it does: it equals , by the same division-by- that Step 4 performs. So this first application is perfectly valid; the trap is what happens if you refuse to evaluate and reach for the rule again just because the new quotient still looks like .
Step 3 — second application. The new quotient is again . Differentiating numerator and denominator separately:
so the rule sends us to
which is the limit we started with (only the harmless has gone). A third application returns to , and so on forever. The rule loops: it is valid at each step but never terminates.
Step 4 — evaluate correctly by algebra. For we have , so
(the positive root, since ). Hence for
As , and ; the denominator tends to , so the Quotient Law applies:
Answer: .
Independent check. At : , and at the value is — converging to . Moral: looping is a signal to abandon the rule, not to keep differentiating. Warning for : there and the limit is .
Consider . Show that l'Hopital's Rule fails to settle it — naming precisely which hypotheses break — and then find the limit by another method.
Show hint
Differentiate top and bottom and look hard at the resulting expression: ask both whether it has a limit and whether its denominator is ever zero far out.
Show answer
Step 1 — the form. Since , we have , so ; likewise . The form is : indeterminate, so the rule is worth considering.
Step 2 — what the rule produces. With and ,
so l'Hopital would point at .
Step 3 — hypothesis 2 fails. whenever for an integer . These points occur arbitrarily far out, so there is no interval on which is nonvanishing. The theorem's second hypothesis is violated on every neighbourhood of .
Step 4 — hypothesis 4 fails too. Even ignoring the zeros, the quotient has no limit. Along it equals ; along it equals . Two subsequences with different limits, so does not exist (and is not ).
Step 5 — the correct conclusion about the rule. Because hypotheses 2 and 4 both fail, l'Hopital's Rule says nothing at all. In particular it is a logical error to conclude that the original limit does not exist.
Step 6 — evaluate by algebra and the Squeeze Theorem. For , divide numerator and denominator by :
Since we get for , and both bounds tend to , so by the Squeeze Theorem ; identically . The denominator therefore tends to , so the Quotient Law applies:
Answer: the limit is ; l'Hopital's Rule is simply inapplicable here.
Independent check. At : , and the deviation from is at most about , which tends to .
Find all constants and for which
Show hint
Apply the rule one stage at a time, and at each stage ask what the numerator must do for the quotient to stay indeterminate — a nonzero numerator over a vanishing denominator forces an infinite limit.
Show answer
Write and . Both are differentiable everywhere, and for all , so hypotheses 1 and 2 hold on any interval around for every choice of .
Stage 0. and , and both are continuous, so the form is for every and .
Stage 1. and , so the rule points at
The denominator and the numerator .
Case . Then the form is , which is determinate and infinite: the quotient tends to if and to if . Since hypothesis 4 permits , l'Hopital's Rule then tells us the original limit is also — certainly not . So we must have
Stage 2. With the quotient is , whose numerator : the form is again, and the new denominator derivative is for . Differentiating:
Stage 3 — re-check. Numerator and denominator : still , and the next denominator derivative is . Differentiating once more:
This last expression is continuous, so no indeterminacy remains and the chain of applications is valid from the bottom up. The original limit therefore equals , and setting this to gives
Answer: and .
Independent check with the Maclaurin series. Since
the numerator is , so
For the limit to exist and be we need and : the same answer. Numerically with , and , the numerator is , and dividing by gives — matching the predicted , which indeed tends to .
Evaluate .
Show hint
Work out what the base and the exponent each tend to before touching anything. After taking logarithms you will meet a product — turning into downstairs is the cleanest move.
Show answer
Step 1 — identify the form. As : the base and the exponent . The form is : indeterminate. (Do not write ; compare , which also has form but equals .)
Step 2 — take logarithms. For the base is positive and is defined. Let ; then
with form : indeterminate, so make it a quotient.
Step 3 — make a quotient. Keep the logarithm upstairs and send downstairs as :
As : and , so this is the case. On both functions are differentiable, and is never there, so hypotheses 1 and 2 hold.
Step 4 — apply the rule.
\lim_{x\to 0^{+}}\frac{-\ln x}{\cot x} &= \lim_{x\to 0^{+}}\frac{-\dfrac{1}{x}}{-\csc^{2}x}\\ &= \lim_{x\to 0^{+}}\frac{1}{x}\cdot\sin^{2}x\\ &= \lim_{x\to 0^{+}}\frac{\sin x}{x}\cdot\sin x\\ &= 1\cdot 0=0 \end{aligned}$$ (The splitting in the last line is allowed because both factors have limits: $\frac{\sin x}{x}\to 1$ and $\sin x\to 0$.) So $\lim_{x\to 0^{+}}\ln y=0$. **Step 5 — undo the logarithm.** Since $\exp$ is continuous, $$\lim_{x\to 0^{+}}\left(\frac{1}{x}\right)^{\tan x}=e^{0}=1$$ **Answer: $1$.** *Independent check.* Write $\ln y=-\tan x\ln x=-\dfrac{\tan x}{x}\cdot x\ln x$. Here $\frac{\tan x}{x}\to 1$ and $x\ln x\to 0$ (a standard limit), so $\ln y\to-1\cdot 0=0$ and $y\to 1$, as found. Numerically at $x=0.001$: $\tan\left(0.001\right)\ln\left(1000\right)\approx 0.0069078$, so $y\approx e^{0.0069}\approx 1.0069$ — approaching $1$ from above.