7.1 Limits
Real functions can be added, multiplied, composed and inverted, and a handful of them — polynomials, rational functions, $|x|$, $[x]$, exponentials, logarithms and the trigonometric functions — build almost everything in calculus. A limit describes what a function does near a point; we evaluate limits by factorising, rationalising and the standard limits $\lim\limits_{x\to0}\frac{\sin x}{x}=1$, $\lim\limits_{x\to0}\frac{e^x-1}{x}=1$ and the $1^\infty$ rule.
A GPS tracker in a car records its position every second. Between $t=10$ s and $t=11$ s the car covered $14.2$ m; between $t=10$ s and $t=10.1$ s it covered $1.38$ m; between $t=10$ s and $t=10.01$ s, $0.1371$ m. The average speeds over these shrinking intervals are $14.2$, $13.8$ and $13.71\ \text{m/s}$, and they are settling down on a number — the speed shown by the speedometer at the instant $t=10$ s. Yet the formula "distance ÷ time" cannot be used at that instant: over an interval of length $0$ it gives $\tfrac00$. The number that the averages approach is called a limit, and limits are the one new idea on which all of calculus rests.
Limits also explain the most famous constant after $\pi$. Invest $\text{₹}\,1$ at $100\%$ annual interest compounded $n$ times a year; after a year you have $\left(1+\tfrac1n\right)^n$ rupees — $2.59$ for $n=10$, $2.705$ for $n=100$, $2.717$ for $n=1000$. More frequent compounding never pushes the amount past about $2.71828$: that limit is the number $e$. Before limits, this section collects the functions that calculus works with — their algebra, their graphs and their inverses — because a limit is always the limit of a function. In JEE Main, limits are among the most dependable questions of the paper: a $\tfrac00$ form to be cleaned up, a standard trigonometric or exponential limit, a $1^\infty$ form, or a one-sided limit of $[x]$ or $|x|$.
Real Functions, their Algebra and the Standard Functions
Section 1.2 defined a function $f:A\to B$ as a rule giving every element of $A$ exactly one image. In calculus we work almost entirely with functions whose inputs and outputs are real numbers.
A real function (real-valued function of a real variable) is a function $f:D\to\mathbb{R}$ with domain $D\subseteq\mathbb{R}$. When only a formula is given, $D$ is the largest set of real $x$ for which the formula gives a real number.
If $f$ and $g$ have domains $D_f$ and $D_g$, then on $D_f\cap D_g$ we define
$$(f\pm g)(x)=f(x)\pm g(x),\qquad (fg)(x)=f(x)\,g(x),\qquad (cf)(x)=c\,f(x)\ \ (c\in\mathbb{R}),$$and $\left(\dfrac fg\right)(x)=\dfrac{f(x)}{g(x)}$ on the smaller set $\{x\in D_f\cap D_g:\ g(x)\ne0\}$.
A real function is even if $f(-x)=f(x)$ for every $x$ in its (symmetric) domain — its graph is symmetric about the $y$-axis, like $x^2$, $\cos x$ and $|x|$ — and odd if $f(-x)=-f(x)$ — its graph is symmetric about the origin, like $x^3$, $\sin x$ and $\tfrac1x$. Products follow the rule of signs: even $\times$ even and odd $\times$ odd are even, even $\times$ odd is odd. Most functions, such as $x+x^2$, are neither. Every function on a symmetric domain is the sum of an even and an odd part, $f(x)=\tfrac{f(x)+f(-x)}{2}+\tfrac{f(x)-f(-x)}{2}$.
A polynomial function is $p(x)=a_nx^n+a_{n-1}x^{n-1}+\dots+a_1x+a_0$ with real coefficients; its domain is $\mathbb{R}$, and its degree is $n$ if $a_n\ne0$. A rational function is a quotient $r(x)=\dfrac{p(x)}{q(x)}$ of polynomials with $q$ not the zero polynomial; its domain is $\mathbb{R}$ minus the real zeros of $q$.
Let $a\gt0$, $a\ne1$. The exponential function with base $a$ is $x\mapsto a^x$, with domain $\mathbb{R}$ and range $(0,\infty)$. For $x\gt0$, the logarithm $\log_a x$ is the unique real $y$ with $a^y=x$:
$$y=\log_a x\iff a^y=x\qquad(x\gt0).$$The most important base is $e=2.71828\ldots$ (Theorem 7.8); $e^x$ is often written $\exp x$, and $\log_e x$ is the natural logarithm $\ln x$. In JEE Main and in this course, $\log x$ without a base means $\ln x$ in calculus problems.
Let $a,b\gt0$ with $a,b\ne1$, and $x,y\gt0$, $r\in\mathbb{R}$. Then $a^{u}a^{v}=a^{u+v}$, $(a^u)^v=a^{uv}$, and
- $\log_a(xy)=\log_a x+\log_a y$, $\ \log_a\dfrac xy=\log_a x-\log_a y$, $\ \log_a(x^r)=r\log_a x$;
- change of base: $\log_a x=\dfrac{\log_b x}{\log_b a}=\dfrac{\ln x}{\ln a}$; and $a^{\log_a x}=x$, $\log_a(a^x)=x$;
- if $a\gt1$, $a^x$ and $\log_a x$ are strictly increasing ($\log_a x\lt\log_a y\iff x\lt y$); if $0\lt a\lt1$ both are strictly decreasing, and inequalities reverse when logarithms are taken.
Proof. Put $u=\log_a x$ and $v=\log_a y$, so $x=a^u$ and $y=a^v$. Then $xy=a^{u+v}$, $\tfrac xy=a^{u-v}$ and $x^r=a^{ru}$, which are the three laws of (1). For (2), $x=a^u$ gives $\log_b x=u\log_b a$ by (1), so $u=\tfrac{\log_b x}{\log_b a}$. For (3) with $a\gt1$: $a^{v}-a^{u}=a^u\left(a^{v-u}-1\right)\gt0$ whenever $v\gt u$, because $a^{t}\gt1$ for $t\gt0$; and the inverse of an increasing function is increasing (Theorem 7.2). For $0\lt a\lt1$ apply this to $\tfrac1a\gt1$. $\blacksquare$
The trigonometric functions are developed in Chapter 14; for calculus we need their domains, ranges and periods, always with the angle $x$ measured in radians ($180^\circ=\pi$).
| $f$ | domain | range | period | parity |
|---|---|---|---|---|
| $\sin x$ | $\mathbb{R}$ | $[-1,1]$ | $2\pi$ | odd |
| $\cos x$ | $\mathbb{R}$ | $[-1,1]$ | $2\pi$ | even |
| $\tan x$ | $x\ne(2k+1)\tfrac\pi2$ | $\mathbb{R}$ | $\pi$ | odd |
| $\cot x$ | $x\ne k\pi$ | $\mathbb{R}$ | $\pi$ | odd |
| $\sec x$ | $x\ne(2k+1)\tfrac\pi2$ | $(-\infty,-1]\cup[1,\infty)$ | $2\pi$ | even |
| $\operatorname{cosec} x$ | $x\ne k\pi$ | $(-\infty,-1]\cup[1,\infty)$ | $2\pi$ | odd |
Let $f(x)=\sqrt x$ and $g(x)=x-1$. The domain of $\dfrac fg$ is
Graphs of Simple Functions and Inverse Functions
Four functions with corners and jumps appear in JEE Main far more often than their simple formulas suggest, because they are exactly where limits, continuity and derivatives can fail.
For real $x$:
- modulus $|x|=x$ if $x\ge0$ and $-x$ if $x\lt0$ — the distance from $x$ to $0$;
- signum $\operatorname{sgn}x=1$ if $x\gt0$, $0$ if $x=0$, $-1$ if $x\lt0$ (so $x=|x|\operatorname{sgn}x$);
- greatest integer $[x]$ (also written $\lfloor x\rfloor$, the floor of $x$) $=$ the greatest integer that is $\le x$: $[2.7]=2$, $[3]=3$, $[-2.3]=-3$;
- fractional part $\{x\}=x-[x]$, which always lies in $[0,1)$: $\{2.7\}=0.7$, $\{-2.3\}=0.7$.
The definition of $[x]$ gives the inequalities used in almost every problem about it: $[x]\le x\lt[x]+1$, equivalently $x-1\lt[x]\le x$, and $[x+n]=[x]+n$ for every integer $n$. So $\{x\}$ is periodic with period $1$, and $[x]+[-x]$ is $0$ when $x$ is an integer and $-1$ otherwise.
- $y=f(x-a)+b$: shift the graph of $f$ by $a$ to the right and $b$ upwards. $y=kf(x)$ stretches vertically; $y=f(kx)$ ($k\gt1$) squeezes horizontally towards the $y$-axis.
- $y=-f(x)$ reflects in the $x$-axis; $y=f(-x)$ reflects in the $y$-axis.
- $y=|f(x)|$: keep the part of the graph on or above the $x$-axis and reflect the part below it upwards.
- $y=f(|x|)$: keep the part for $x\ge0$ and reflect it in the $y$-axis (an even function).
- The number of real solutions of $f(x)=g(x)$ is the number of intersections of the two graphs — often the fastest route in an MCQ.
Section 1.2 showed that a bijection $f:A\to B$ pairs the elements of $A$ and $B$ perfectly. Reading the pairing backwards gives a new function.
Let $f:A\to B$ be a bijection (one-one and onto). Its inverse $f^{-1}:B\to A$ sends each $y\in B$ to the unique $x\in A$ with $f(x)=y$:
$$f^{-1}(y)=x\iff f(x)=y,\qquad\text{so}\quad f^{-1}\big(f(x)\big)=x\ (x\in A),\quad f\big(f^{-1}(y)\big)=y\ (y\in B).$$- $f:A\to B$ has an inverse — a function $g:B\to A$ with $g\circ f=I_A$ and $f\circ g=I_B$ — if and only if $f$ is a bijection, and then the inverse is unique.
- The graph of $f^{-1}$ is the mirror image of the graph of $f$ in the line $y=x$.
- $\left(f^{-1}\right)^{-1}=f$, and for bijections $f:A\to B$, $g:B\to C$: $(g\circ f)^{-1}=f^{-1}\circ g^{-1}$.
- If $f$ is strictly increasing on an interval, so is $f^{-1}$.
Proof. (1) If such a $g$ exists, $f(x_1)=f(x_2)$ gives $x_1=g(f(x_1))=g(f(x_2))=x_2$, so $f$ is one-one; and every $y\in B$ equals $f(g(y))$, so $f$ is onto. Conversely, for a bijection Definition 7.5 gives such a $g$. If $g$ and $h$ both work, $g=g\circ(f\circ h)=(g\circ f)\circ h=h$. (2) $(p,q)$ lies on the graph of $f$ $\iff q=f(p)\iff p=f^{-1}(q)\iff(q,p)$ lies on the graph of $f^{-1}$, and $(q,p)$ is the reflection of $(p,q)$ in $y=x$. (3) $f$ is an inverse of $f^{-1}$ by the two identities in Definition 7.5; and $(f^{-1}\circ g^{-1})\circ(g\circ f)$ $=f^{-1}\circ(g^{-1}\circ g)\circ f$ $=f^{-1}\circ f=I_A$, similarly in the other order. (4) If $y_1\lt y_2$ but $x_1=f^{-1}(y_1)\ge x_2=f^{-1}(y_2)$, then $y_1=f(x_1)\ge f(x_2)=y_2$, a contradiction. $\blacksquare$
Two inverse pairs matter most in calculus. $e^x:\mathbb{R}\to(0,\infty)$ and $\ln x:(0,\infty)\to\mathbb{R}$ are inverse to each other: $\ln(e^x)=x$ for all real $x$ and $e^{\ln x}=x$ for $x\gt0$ — their graphs are mirror images in $y=x$. And $\sin x$ is not one-one on $\mathbb{R}$, but it is a bijection from $\left[-\tfrac\pi2,\tfrac\pi2\right]$ onto $[-1,1]$; its inverse there is $\sin^{-1}x$, one of the inverse trigonometric functions of Section 14.2. To invert a function that is not one-one, first restrict its domain; to make it onto, take the codomain to be its range. Note that $f^{-1}(x)$ is not $\dfrac{1}{f(x)}$.
In the lab below choose $f(x)=x^2$ with no shift ($a=b=0$, plain graph) and press mirror in y = x. What does the mirror image look like?
Which of these functions $\mathbb{R}\to\mathbb{R}$ has an inverse function $\mathbb{R}\to\mathbb{R}$?
Limits: Meaning, One-sided Limits and the Algebra of Limits
The function $f(x)=\dfrac{x^2-4}{x-2}$ is not defined at $x=2$ — the formula gives $\tfrac00$. But near $2$ it is perfectly well behaved: $f(1.9)=3.9$, $f(1.99)=3.99$, $f(1.999)=3.999$ and $f(2.01)=4.01$, $f(2.001)=4.001$. For $x\ne2$, $f(x)=x+2$, so $f(x)$ is as close to $4$ as we please once $x$ is close enough to $2$. We say the limit of $f(x)$ as $x$ tends to $2$ is $4$. What happens at $2$ plays no part.
Let $f$ be defined on an open interval around $a$, except possibly at $a$ itself. We write $\displaystyle\lim_{x\to a}f(x)=L$ if $f(x)$ can be made as close to $L$ as we like by taking $x$ close enough to $a$, with $x\ne a$. Precisely: for every $\varepsilon\gt0$ there is a $\delta\gt0$ such that
$$0\lt|x-a|\lt\delta\ \Rightarrow\ |f(x)-L|\lt\varepsilon .$$If only $x\lt a$ is allowed we get the left-hand limit (LHL) $\displaystyle\lim_{x\to a^-}f(x)=\lim_{h\to0^+}f(a-h)$; if only $x\gt a$, the right-hand limit (RHL) $\displaystyle\lim_{x\to a^+}f(x)=\lim_{h\to0^+}f(a+h)$.
$\displaystyle\lim_{x\to a}f(x)$ exists and equals $L$ if and only if both one-sided limits exist and $\displaystyle\lim_{x\to a^-}f(x)=\lim_{x\to a^+}f(x)=L$.
This is how limits fail. For $\dfrac{|x|}{x}$ at $0$: the LHL is $-1$ and the RHL is $1$, so there is no limit. For $[x]$ at an integer $n$: LHL $=n-1$, RHL $=n$. For $\dfrac1{x^2}$ at $0$ the values grow beyond every bound; we write $\displaystyle\lim_{x\to0}\frac1{x^2}=\infty$, but $\infty$ is not a real number and the limit does not exist in the sense of Definition 7.6. For $\sin\dfrac1x$ near $0$ the values swing between $-1$ and $1$ infinitely often, however small the interval — no limit, not even a one-sided one.
$\displaystyle\lim_{x\to\infty}f(x)=L$ means that $f(x)$ is as close to $L$ as we like for all sufficiently large $x$ (similarly $x\to-\infty$). $\displaystyle\lim_{x\to a}f(x)=\infty$ means that $f(x)$ exceeds any given number for all $x\ne a$ close enough to $a$ (similarly $-\infty$). Basic facts: $\dfrac1{x^n}\to0$ as $x\to\pm\infty$ ($n\gt0$); $e^{x}\to0$ as $x\to-\infty$; $\ln x\to-\infty$ as $x\to0^+$.
If $\displaystyle\lim_{x\to a}f(x)=L$ and $\displaystyle\lim_{x\to a}g(x)=M$ with $L,M$ real, then
$$\lim(f\pm g)=L\pm M,\qquad \lim(fg)=LM,\qquad \lim(cf)=cL,\qquad \lim\frac fg=\frac LM\ \ (M\ne0).$$Consequently $\displaystyle\lim_{x\to a}p(x)=p(a)$ for every polynomial $p$, and $\displaystyle\lim_{x\to a}\frac{p(x)}{q(x)}=\frac{p(a)}{q(a)}$ when $q(a)\ne0$. Moreover, the standard functions ($x^{r}$, $a^x$, $\log_a x$, the trigonometric functions, $|x|$) may be evaluated by substitution at any point inside their domains, and $\displaystyle\lim_{x\to a}h\big(g(x)\big)=h(M)$ for such a function $h$ defined around $M$. The same rules hold for one-sided limits and for $x\to\pm\infty$.
The rules follow from the $\varepsilon$–$\delta$ definition (for a sum: if $|f-L|\lt\tfrac\varepsilon2$ and $|g-M|\lt\tfrac\varepsilon2$ then $|(f+g)-(L+M)|\lt\varepsilon$); we use them freely. They do not apply directly to $\tfrac00$, $\tfrac\infty\infty$, $\infty-\infty$, $0\cdot\infty$ or $1^\infty$ — the indeterminate forms, whose value depends on the functions involved. Most limit questions are about these forms, and the first tool is the following factorisation result.
for every positive integer $n$ and real $a$; the same formula holds for every rational $n$ when $a\gt0$.
Proof (positive integer $n$).
$$x^n-a^n=(x-a)\left(x^{n-1}+x^{n-2}a+\dots+a^{n-1}\right),$$as multiplying out shows (the terms telescope). For $x\ne a$ cancel $x-a$; the bracket has $n$ terms, each tending to $a^{n-1}$ as $x\to a$ (Theorem 7.4), so the limit is $n\,a^{n-1}$. $\blacksquare$ (For $n=\tfrac pq$ put $x=t^q$, $a=b^q$ and apply the integer case twice: $\tfrac{t^p-b^p}{t^q-b^q}\to\tfrac{p\,b^{p-1}}{q\,b^{q-1}}=\tfrac pq\,a^{\frac pq-1}$.)
- Substitute first. If you get a real number (with a non-zero denominator), that is the limit (Theorem 7.4).
- $\tfrac00$: factorise and cancel $x-a$; or rationalise (multiply top and bottom by the conjugate $\sqrt{\ }+\sqrt{\ }$); or reduce to a standard limit (Theorems 7.5, 7.7, 7.8).
- $x\to\pm\infty$ with polynomials: divide top and bottom by the highest power of $x$ in the denominator. Equal degrees give the ratio of leading coefficients; a higher top degree gives $\pm\infty$; a lower one gives $0$. For $x\to-\infty$ remember $\sqrt{x^2}=|x|=-x$.
- $\infty-\infty$: combine fractions, or rationalise $\sqrt{A}-B$ into $\dfrac{A-B^2}{\sqrt A+B}$.
- A formula that changes at $a$ — piecewise definitions, $|x-a|$, $[x]$, $\{x\}$, $\operatorname{sgn}$ — work out the LHL and the RHL separately (Theorem 7.3).
In the lab below choose $(1+x)^{1/x}$ at $x=0$ and drag $h$ down to $0.001$. The base $1+x$ tends to $1$, and $1$ to any power is $1$. What do the two readouts approach?
Suppose $\displaystyle\lim_{x\to2}f(x)=5$. Which statement must be true?
Standard Limits: Trigonometric, Exponential, Logarithmic and $1^\infty$
Factorising cannot remove the $\tfrac00$ in $\dfrac{\sin x}{x}$ or $\dfrac{e^x-1}{x}$. These limits are found once, with a squeeze argument or a substitution, and then used as building blocks.
If $g(x)\le f(x)\le h(x)$ for all $x\ne a$ in an open interval around $a$, and $\displaystyle\lim_{x\to a}g(x)=\lim_{x\to a}h(x)=L$, then $\displaystyle\lim_{x\to a}f(x)=L$. (Also for one-sided limits and $x\to\pm\infty$.)
Why. Given $\varepsilon\gt0$, close enough to $a$ both $g(x)$ and $h(x)$ lie in $(L-\varepsilon,L+\varepsilon)$, and $f(x)$ is trapped between them. $\blacksquare$ Example: $-|x|\le x\sin\dfrac1x\le|x|$ and $\pm|x|\to0$, so $\displaystyle\lim_{x\to0}x\sin\frac1x=0$ even though $\sin\tfrac1x$ itself has no limit. Likewise $\dfrac{\sin x}{x}\to0$ as $x\to\infty$, because $\left|\dfrac{\sin x}x\right|\le\dfrac1{|x|}$.
Hence $\displaystyle\lim_{x\to0}\frac{\sin px}{qx}=\frac pq$, and also $\displaystyle\lim_{x\to0}\frac{\sin^{-1}x}{x}=\lim_{x\to0}\frac{\tan^{-1}x}{x}=1$.
Proof. Take $0\lt x\lt\tfrac\pi2$ and the unit circle of Figure 7.2, with $A=(1,0)$, $P=(\cos x,\sin x)$ and $T=(1,\tan x)$ on the tangent at $A$. Comparing areas, $\tfrac12\sin x\lt\tfrac12x\lt\tfrac12\tan x$ (the area of a sector of angle $x$ in a unit circle is $\tfrac x2$). Dividing by $\tfrac12\sin x\gt0$ and taking reciprocals: $\cos x\lt\dfrac{\sin x}{x}\lt1$. As $x\to0^+$, $\cos x\to1$, so the sandwich theorem gives the RHL $1$; and $\dfrac{\sin x}x$ is even, so the LHL is also $1$. Then $\dfrac{\tan x}x=\dfrac{\sin x}{x}\cdot\dfrac1{\cos x}\to1$, and $\dfrac{1-\cos x}{x^2}=\dfrac{2\sin^2(x/2)}{x^2}=\dfrac12\left(\dfrac{\sin(x/2)}{x/2}\right)^2\to\dfrac12$. For $\sin^{-1}$ put $x=\sin t$: $\dfrac{\sin^{-1}x}{x}=\dfrac{t}{\sin t}\to1$. $\blacksquare$ In degrees the answer changes: $\sin x^\circ=\sin\tfrac{\pi x}{180}$, so $\dfrac{\sin x^\circ}{x}\to\dfrac\pi{180}$.
- $\displaystyle\lim_{x\to0}(1+x)^{1/x}=e$ and $\displaystyle\lim_{x\to\infty}\left(1+\frac1x\right)^{x}=e$, where $e=2.718281828\ldots$;
- $\displaystyle\lim_{x\to0}\frac{\ln(1+x)}{x}=1$;
- $\displaystyle\lim_{x\to0}\frac{e^x-1}{x}=1$, and for $a\gt0$: $\displaystyle\lim_{x\to0}\frac{a^x-1}{x}=\ln a$.
Proof. (1) The number $e$ is defined as this limit: one shows (using the binomial theorem of Chapter 5) that $\left(1+\tfrac1n\right)^n$ increases with $n$ and stays below $3$, so it settles on a number; numerically $2.5937$, $2.7048$, $2.7169$, $2.71815$ for $n=10,100,10^3,10^4$. The form with $x\to0$ is the same limit with $x$ replaced by $\tfrac1x$. (2) $\dfrac{\ln(1+x)}{x}=\ln\left[(1+x)^{1/x}\right]\to\ln e=1$, by (1) and Theorem 7.4 for the function $\ln$. (3) Put $t=e^x-1$, so $x=\ln(1+t)$ and $t\to0$ as $x\to0$: $\dfrac{e^x-1}{x}=\dfrac{t}{\ln(1+t)}\to1$ by (2). Finally $a^x=e^{x\ln a}$, so $\dfrac{a^x-1}{x}=\ln a\cdot\dfrac{e^{x\ln a}-1}{x\ln a}\to\ln a$ (for $a\ne1$; for $a=1$ both sides are $0$). $\blacksquare$
Suppose that, as $x\to a$ (or $x\to\pm\infty$), $f(x)\to1$ with $f(x)\ne1$, $g(x)\to\pm\infty$, and $g(x)\big(f(x)-1\big)\to\lambda$ (real). Then
$$\lim f(x)^{g(x)}=e^{\lambda}=e^{\,\lim g(x)\,(f(x)-1)} .$$Proof. Write $t=f(x)-1\to0$. Then $f^{g}=\left[(1+t)^{1/t}\right]^{\,g\,t}$. The bracket tends to $e$ by Theorem 7.8(1) and the exponent $g\,t$ tends to $\lambda$; so $\ln\left(f^g\right)=g\,t\cdot\ln\left[(1+t)^{1/t}\right]\to\lambda\cdot1$, and $f^g\to e^\lambda$. $\blacksquare$ For example $\displaystyle\lim_{x\to0}(1+3x)^{2/x}=e^{\lim\frac2x\cdot3x}=e^6$, and $\displaystyle\lim_{x\to0}(\cos x)^{1/x^2}=e^{\lim\frac{\cos x-1}{x^2}}=e^{-1/2}$.
L'Hôpital's rule is listed in the JEE Advanced 2026 syllabus but not in the JEE Main 2026 list. Every JEE Main limit can be done with the standard limits of this section. This box is for interest only; nothing in it is used in the exercises, the Exam corner or the quiz. It needs derivatives, which are defined in Section 7.2 and developed in Section 7.3.
If $f$ and $g$ are differentiable near $a$ (except possibly at $a$), $g'(x)\ne0$ there, $f(x)\to0$ and $g(x)\to0$ (or both $\to\pm\infty$) as $x\to a$, and $\displaystyle\lim_{x\to a}\frac{f'(x)}{g'(x)}$ exists, then $\displaystyle\lim_{x\to a}\frac{f(x)}{g(x)}=\lim_{x\to a}\frac{f'(x)}{g'(x)}$. Example: $\displaystyle\lim_{x\to0}\frac{x-\sin x}{x^3}=\lim_{x\to0}\frac{1-\cos x}{3x^2}=\frac16$, using $\dfrac{1-\cos x}{x^2}\to\dfrac12$. Always check that the form really is $\tfrac00$ or $\tfrac\infty\infty$ before differentiating.
1. Radians, not degrees. $\dfrac{\sin x}{x}\to1$ only for $x$ in radians; $\dfrac{\sin x^\circ}{x}\to\dfrac{\pi}{180}$. And $\dfrac{\sin x}x\to1$ only as $x\to0$: as $x\to\infty$ it tends to $0$.
2. $1^\infty$ is not $1$. $(1+\tfrac1n)^n\to e$. Whenever the base $\to1$ and the exponent $\to\infty$, use $e^{\lim g(f-1)}$ (Theorem 7.9).
3. The limit ignores $f(a)$. $\displaystyle\lim_{x\to2}([x]+[-x])=-1$ although the value at $2$ is $0$; and $\dfrac{x^2-4}{x-2}$ has a limit at $2$ although it is undefined there.
4. Check both sides near jumps and corners. $\displaystyle\lim_{x\to0}[x]$ does not exist (LHL $-1$, RHL $0$), but $\displaystyle\lim_{x\to0}[x^2]=0$ and $\displaystyle\lim_{x\to0}\left[\frac{\sin x}{x}\right]=0$ because $\tfrac{\sin x}x$ approaches $1$ from below. For $x\to-\infty$, $\sqrt{x^2}=-x$.
5. Replacing by equivalents inside a difference. $\tan x-\sin x$ is not "$x-x=0$": it is about $\tfrac{x^3}2$. Substitute $\sin u\approx u$ only in products and quotients.
6. Inverses need bijections. State the domain and codomain; $f^{-1}(x)\ne\dfrac1{f(x)}$; and the domain of $f^{-1}$ is the range of $f$.
$\displaystyle\lim_{x\to\infty}x\sin\frac1x$ equals
Exercises
Logarithms without a base are natural logarithms. Give limits exactly where you can, and say clearly when a limit does not exist.
- 1 Find the domain of $f(x)=\sqrt{x+3}+\dfrac{1}{\ln(2-x)}$.
- 2 Find the domain of $f(x)=\log_2\big(\log_3(x-1)\big)$.
- 3 $f(x)=x^2-1$ and $g(x)=\sqrt{x+1}$. Find the domain and a simplified formula for $\tfrac fg$, and the value $\left(\tfrac fg\right)(3)$.
- 4 Decide whether each is even, odd or neither: $x^3\cos x$; $\ \ln\dfrac{1-x}{1+x}$; $\ 2^x+2^{-x}$; $\ x+x^2$.
- 5 Evaluate $\log_2 48-\log_2 3+9^{\log_3 5}$.
- 6 Solve $\log_2x+\log_2(x-2)=3$.
- 7 $f:\mathbb{R}-\{-4\}\to\mathbb{R}-\{3\}$, $f(x)=\dfrac{3x-2}{x+4}$. Find $f^{-1}(x)$ and $f^{-1}(1)$.
- 8 $f:[2,\infty)\to[-1,\infty)$, $f(x)=x^2-4x+3$. Find $f^{-1}(x)$ and $f^{-1}(8)$.
- 9 Show that $f:\mathbb{R}\to(-1,1)$, $f(x)=\dfrac{x}{1+|x|}$, is a bijection, find $f^{-1}$, and evaluate $f^{-1}\!\left(\tfrac12\right)$.
- 1 Evaluate $[-2.5]+[2.5]+\{-2.5\}+[\pi]$.
- 2 Solve $[x]^2-5[x]+6=0$.
- 3 Find the number of real solutions of $|x^2-6x+5|=3$.
- 4 Sketch $y=|x-1|+|x+2|$. Find its least value and the set of $x$ where it is attained.
- 5 Find the number of real solutions of $\{x\}=\dfrac x4$.
- 6 Sketch $y=\operatorname{sgn}(x^2-1)$ and state where it equals $-1$.
- 7 The graph of $y=x^2$ is shifted $2$ units to the right and $3$ units up, and the result is reflected in the $x$-axis. Write the new equation and find where it meets the $y$-axis.
- 8 Find the domain of $f(x)=\dfrac{1}{\sqrt{[x]^2-[x]-6}}$.
- 1 $\displaystyle\lim_{x\to2}\frac{x^3-8}{x^2-4}$
- 2 $\displaystyle\lim_{x\to0}\frac{\sqrt{4+x}-2}{x}$
- 3 $\displaystyle\lim_{x\to1}\frac{x^7-1}{x^3-1}$
- 4 $\displaystyle\lim_{x\to\infty}\frac{5x^3-2x+1}{2x^3+x^2}$
- 5 $\displaystyle\lim_{x\to\infty}\left(\sqrt{x^2+6x+1}-x\right)$
- 6 $\displaystyle\lim_{x\to-\infty}\frac{\sqrt{4x^2+1}}{x+3}$
- 7 $\displaystyle\lim_{x\to3}\frac{|x-3|}{x-3}$ (find both one-sided limits).
- 8 $\displaystyle\lim_{x\to2^-}[x]$, $\displaystyle\lim_{x\to2^+}[x]$ and $\displaystyle\lim_{x\to2.5}[x]$.
- 9 $f(x)=x^2+k$ for $x\lt2$ and $f(x)=3x-k$ for $x\ge2$. Find $k$ so that $\displaystyle\lim_{x\to2}f(x)$ exists.
- 10 $\displaystyle\lim_{x\to1}\left(\frac1{x-1}-\frac2{x^2-1}\right)$
- 1 $\displaystyle\lim_{x\to0}\frac{\sin7x}{2x}$
- 2 $\displaystyle\lim_{x\to0}\frac{\tan3x-\sin x}{x}$
- 3 $\displaystyle\lim_{x\to0}\frac{1-\cos6x}{x^2}$
- 4 $\displaystyle\lim_{x\to0}\frac{e^{5x}-1}{\sin2x}$
- 5 $\displaystyle\lim_{x\to0}\frac{4^x-1}{2^x-1}$
- 6 $\displaystyle\lim_{x\to0}\frac{\ln(1+3x)}{e^x-1}$
- 7 $\displaystyle\lim_{x\to0}(1+2x)^{3/x}$ — write the answer as $e^k$ and give $k$.
- 8 $\displaystyle\lim_{x\to\infty}\left(\frac{x^2+2x}{x^2+1}\right)^{x}$
- 9 $\displaystyle\lim_{x\to0}\frac{\sin x^\circ}{x}$ (angle in degrees), to four decimal places.
- 10 (Modelling) $\text{₹}\,10\,000$ is invested at $8\%$ per year, compounded $n$ times a year, for $5$ years, so the amount is $10\,000\left(1+\frac{0.08}{n}\right)^{5n}$. Find the limit of the amount as $n\to\infty$ ("continuous compounding"), to the nearest rupee, and compare with yearly compounding.
E1. JEE Main · MCQ The domain of $f(x)=\dfrac{\sqrt{x+2}}{\log_{10}(3-x)}$ is (A) $[-2,3)$ (B) $[-2,2)\cup(2,3)$ (C) $(-2,2)\cup(2,3)$ (D) $[-2,2]$
E2. JEE Main · Numerical The function $f:[1,\infty)\to[2,\infty)$, $f(x)=x+\dfrac1x$, is a bijection. Find $f^{-1}\!\left(\dfrac{26}{5}\right)$. Give the answer to the nearest integer.
E3. JEE Main · MCQ $\displaystyle\lim_{x\to0}\frac{\sqrt{1+\sin x}-\sqrt{1-\sin x}}{\tan x}$ equals (A) $0$ (B) $\tfrac12$ (C) $1$ (D) $2$
E4. JEE Main · Numerical If $\displaystyle\lim_{x\to\infty}\left(\frac{x+5}{x+2}\right)^{2x+1}=e^{k}$, find $k$. Give the answer to the nearest integer.
E5. JEE Main · MCQ Let $f(x)=\dfrac{[x]\sin x}{x}$ for $x\ne0$. The left-hand and right-hand limits of $f$ at $x=0$ are, respectively, (A) $-1$ and $0$ (B) $0$ and $0$ (C) $-1$ and $1$ (D) $1$ and $0$
E6. JEE Main · Numerical Find $\displaystyle\lim_{x\to0}\frac{\ln(1+4x)\,(1-\cos2x)}{x^2\sin x}$. Give the answer to the nearest integer.
Research question. A computer stores numbers with about $16$ significant digits. Can it compute a limit by "plugging in a very small $h$" — and if not, where and why does it fail?
Why it matters. Engineering software, weather models, spacecraft navigation and financial pricing engines all evaluate expressions such as $(1+\tfrac1n)^n$ or $\tfrac{1-\cos x}{x^2}$ millions of times. A formula that is mathematically perfect can return nonsense in floating-point arithmetic; famous failures (a missile-defence clock drifting by a third of a second, a rocket lost to a number conversion) were numerical-precision errors. Numerical analysts study exactly this gap between a limit and its computer approximation.
What to do.
- In a spreadsheet or Python, tabulate $(1+\tfrac1n)^n$ for $n=10^1,10^2,\dots,10^{17}$ and the error compared with $e=2.718281828459045$. Where is the error smallest? What happens from $n=10^{12}$ onwards, and at $n=10^{16}$?
- Explain the failure: what is $1+10^{-16}$ in a machine that keeps about $16$ digits? Why does an error in the base matter so much when the exponent is huge?
- Tabulate $\dfrac{1-\cos x}{x^2}$ for $x=10^{-1},\dots,10^{-9}$. Then tabulate the equivalent form $\dfrac{2\sin^2(x/2)}{x^2}$ (Theorem 7.7's proof). Which one stays close to $\tfrac12$, and why?
- Find a third limit from this section whose direct evaluation fails for tiny $x$, and a rewriting that fixes it.
Tools: Spreadsheet or Python (ordinary floating-point numbers).
Hand in: The two tables, a sketch of error against $n$ (logarithmic scales), and a paragraph explaining "catastrophic cancellation" in your own words.
Going further. Many languages provide special functions such as expm1 and log1p for $e^x-1$ and $\ln(1+x)$. Which standard limits of Theorem 7.8 do they protect, and how could you use them to compute $(1+\tfrac1n)^n$ accurately for every $n$?
🧪 Self-check quiz
Q1. The domain of $f(x)=\ln(x-2)+\sqrt{5-x}$ is
Q2. $\displaystyle\lim_{x\to0}\frac{\sin4x}{\sin2x}$ equals
Q3. $\displaystyle\lim_{x\to\infty}\left(1+\frac2x\right)^{x}$ equals
Q4. $[-1.5]+\{-1.5\}$ equals
Q5. For the bijection $f:\mathbb{R}\to\mathbb{R}$, $f(x)=2x+3$, the inverse is $f^{-1}(x)=$