Definition

A function takes an input and produces an output. Its inverse runs the process backward: given the output, it recovers the input. If f(a)=bf(a) = b, then f−1(b)=af^{-1}(b) = a.

For an inverse to exist as a function, ff must be one-to-one: each output comes from exactly one input. If two different inputs gave the same output, the inverse would not know which one to return.

Trig functions fail this! Consider sin⁡θ=12\sin\theta = \tfrac{1}{2}. This is true for θ=π6\theta = \tfrac{\pi}{6}, but also 5π6\tfrac{5\pi}{6}, and π6+2π\tfrac{\pi}{6} + 2\pi, and infinitely many more. Asking "what angle has sine 12\tfrac{1}{2}?" has no single answer. So sine, cosine, and tangent have no inverse on their full domains.

We will use both inverse (e.g., sin⁡−1\sin^{-1}) and arc (e.g., arcsin⁡\arcsin) notation throughout this module; both mean the same thing.

Principal Ranges

We repair the above caveat using principal ranges, i.e, an interval where the function is one-to-one. On that interval the function passes the horizontal-line test, and an inverse exists. The chosen interval is called the principal range, and its outputs are the principal values.

For sinsin, we restrict to [−π2,π2][-\tfrac{\pi}{2}, \tfrac{\pi}{2}], where the function covers its entire range without repetition. This gives the inverse sine, written arcsin⁡\arcsin (or sin⁡−1\sin^{-1}):

 [−1,1]→[−π2,π2].\: [-1, 1] \to \left[-\tfrac{\pi}{2}, \tfrac{\pi}{2}\right].

For cosine, [−π2,π2][-\tfrac{\pi}{2}, \tfrac{\pi}{2}] would not work (cosine is symmetric there), so we use [0,π][0, \pi].

arccos⁡:[−1,1]→[0,π].\arccos: [-1, 1] \to [0, \pi].

For tangent, which covers every real value on (−π2,π2)(-\tfrac{\pi}{2}, \tfrac{\pi}{2}), we restrict to that open interval:

arctan⁡:R→(−π2,π2).\arctan: \mathbb{R} \to \left(-\tfrac{\pi}{2}, \tfrac{\pi}{2}\right).

Don't Get Trapped!

It is tempting to think arcsin⁡(sin⁡θ)=θ\arcsin(\sin\theta) = \theta always. It does not. The output must land in the principal range, so if θ\theta starts outside it, you must reduce. For example:

arcsin⁡(sin⁡3π4)=π4,\arcsin\left(\sin\tfrac{3\pi}{4}\right) = \tfrac{\pi}{4},

because sin⁡3π4=22\sin\tfrac{3\pi}{4} = \tfrac{\sqrt{2}}{2}, and the angle in [−π2,π2][-\tfrac{\pi}{2}, \tfrac{\pi}{2}] with that sine is π4\tfrac{\pi}{4}, not 3π4\tfrac{3\pi}{4}. Going the other way, sin⁡(arcsin⁡x)=x\sin(\arcsin x) = x always holds for x∈[−1,1]x \in [-1,1], because that composition does not leave the range.

Evaluating Compositions

Mixed compositions like cos⁡(arcsin⁡x)\cos(\arcsin x) look intimidating but are actually easy to deal with. Set θ=arcsin⁡x\theta = \arcsin x, so sin⁡θ=x\sin\theta = x with θ∈[−π2,π2]\theta \in [-\tfrac{\pi}{2}, \tfrac{\pi}{2}]. Recall the Pythagorean identity:

sin⁡2θ+cos⁡2θ=1\sin^{2}\theta + \cos^{2}\theta = 1

Substitute xx for sin⁡θ\sin\theta:

x2+cos⁡2θ=1x^2 + \cos^{2}\theta = 1
cos⁡θ=1−x2\cos\theta = \sqrt{1 - x^2}

We take the positive root because θ∈[−π2,π2]\theta \in [-\tfrac{\pi}{2}, \tfrac{\pi}{2}], cosine is never negative. Therefore:

cos⁡(arcsin⁡x)=1−x2\cos(\arcsin x) = \sqrt{1 - x^2}

Similarly, for sin⁡(arccos⁡x)\sin(\arccos x): Set θ=arccos⁡x\theta = \arccos x, so cos⁡θ=x\cos\theta = x with θ∈[0,π]\theta \in [0, \pi]. Substitute into the identity:

sin⁡2θ+x2=1\sin^{2}\theta + x^2 = 1
sin⁡θ=1−x2\sin\theta = \sqrt{1 - x^2}

Here we take the positive root because θ∈[0,π]\theta \in [0, \pi], where sine is never negative. Therefore:

sin⁡(arccos⁡x)=1−x2\sin(\arccos x) = \sqrt{1 - x^2}

For tangent compositions we use tan⁡θ=sin⁡θcos⁡θ\tan\theta = \dfrac{\sin\theta}{\cos\theta}. To find cos⁡(arctan⁡x)\cos(\arctan x), set θ=arctan⁡x\theta = \arctan x, so tan⁡θ=x\tan\theta = x with θ∈(−π2,π2)\theta \in (-\tfrac{\pi}{2}, \tfrac{\pi}{2}). Divide the Pythagorean identity by cos⁡2θ\cos^{2}\theta:

tan⁡2θ+1=1cos⁡2θ\tan^{2}\theta + 1 = \frac{1}{\cos^{2}\theta}

Substitute xx for tan⁡θ\tan\theta and solve:

x2+1=1cos⁡2θx^2 + 1 = \frac{1}{\cos^{2}\theta}
cos⁡θ=11+x2\cos\theta = \frac{1}{\sqrt{1 + x^2}}

The positive root holds because cosine is positive on (−π2,π2)(-\tfrac{\pi}{2}, \tfrac{\pi}{2}). Then sin⁡θ=xcos⁡θ\sin\theta = x\cos\theta gives the following results (you might want to memorize these!):

cos⁡(arctan⁡x)=11+x2,sin⁡(arctan⁡x)=x1+x2\cos(\arctan x) = \frac{1}{\sqrt{1 + x^2}}, \qquad \sin(\arctan x) = \frac{x}{\sqrt{1 + x^2}}

sin⁡(arctan⁡x)=x1+x2,cos⁡(arctan⁡x)=11+x2.\sin(\arctan x) = \frac{x}{\sqrt{1 + x^2}}, \qquad \cos(\arctan x) = \frac{1}{\sqrt{1 + x^2}}.

Key Identities

Here are a couple nice identites that pop up from time to time in contests. This topic is pretty niche, but everything's worth knowing:

arcsin⁡x+arccos⁡x=π2,x∈[−1,1].\arcsin x + \arccos x = \frac{\pi}{2}, \qquad x \in [-1, 1].

Proof: if θ=arcsin⁡x\theta = \arcsin x then cos⁡(π2−θ)=sin⁡θ=x\cos\left(\tfrac{\pi}{2} - \theta\right) = \sin\theta = x, and π2−θ∈[0,π]\tfrac{\pi}{2} - \theta \in [0, \pi], so π2−θ=arccos⁡x\tfrac{\pi}{2} - \theta = \arccos x.

It is also worth mentioning

arctan⁡a+arctan⁡b=arctan⁡(a+b1−ab)+kπ,\arctan a + \arctan b = \arctan\left(\frac{a + b}{1 - ab}\right) + k\pi,

where the correction term is

k={0ab<11ab>1, a>0−1ab>1, a<0.k = \begin{cases} 0 & ab < 1 \\ 1 & ab > 1,\ a > 0 \\ -1 & ab > 1,\ a < 0. \end{cases}

We bring in kπk\pi as when ab>1ab > 1 the sum leaves the arctan range, so the output is off by π\pi. When ab<1ab < 1 with both terms positive, you can drop kk.

Telescoping with Arctangent

We use the above formula backwards to attain:

arctan⁡(1n2+n+1)=arctan⁡(n+1)−arctan⁡(n).\arctan\left(\frac{1}{n^2 + n + 1}\right) = \arctan(n + 1) - \arctan(n).

This holds because, with a=n+1a = n+1 and b=−nb = -n, we get a+b1−ab=11+n(n+1)=1n2+n+1\frac{a + b}{1 - ab} = \frac{1}{1 + n(n+1)} = \frac{1}{n^2 + n + 1}, and ab=−n(n+1)<1ab = -n(n+1) < 1, as desired.

Telescoping Example

Evaluate

S=∑n=1∞arctan⁡(1n2+n+1).S = \sum_{n=1}^{\infty} \arctan\left(\frac{1}{n^2 + n + 1}\right).

We use the identity we just showed:

SN=∑n=1N[arctan⁡(n+1)−arctan⁡(n)]=arctan⁡(N+1)−arctan⁡(1).S_N = \sum_{n=1}^{N} \left[\arctan(n+1) - \arctan(n)\right] = \arctan(N+1) - \arctan(1).

As N→∞N \to \infty, arctan⁡(N+1)→π2\arctan(N+1) \to \tfrac{\pi}{2} and arctan⁡(1)=π4\arctan(1) = \tfrac{\pi}{4}, so

S=π2−π4=π4.S = \frac{\pi}{2} - \frac{\pi}{4} = \frac{\pi}{4}.

Worked Examples

2008 AIME I · Problem 8: Find the positive integer nn such that arctan⁡13+arctan⁡14+arctan⁡15+arctan⁡1n=π4.\arctan\frac{1}{3} + \arctan\frac{1}{4} + \arctan\frac{1}{5} + \arctan\frac{1}{n} = \frac{\pi}{4}.

Every term here is a positive acute angle, so each arctan⁡1k\arctan\frac{1}{k} is small. That means we can simply apply the addition formula!

arctan⁡a+arctan⁡b=arctan⁡(a+b1−ab)\arctan a + \arctan b = \arctan\left(\frac{a + b}{1 - ab}\right)

Combine the first two terms with a=13a = \frac{1}{3}, b=14b = \frac{1}{4}:

arctan⁡13+arctan⁡14=arctan⁡(13+141−112)=arctan⁡(7121112)=arctan⁡711.\arctan\frac{1}{3} + \arctan\frac{1}{4} = \arctan\left(\frac{\frac{1}{3} + \frac{1}{4}}{1 - \frac{1}{12}}\right) = \arctan\left(\frac{\frac{7}{12}}{\frac{11}{12}}\right) = \arctan\frac{7}{11}.

Combine third term, arctan⁡15\arctan\frac{1}{5}:

arctan⁡711+arctan⁡15=arctan⁡(711+151−755)=arctan⁡(46554855)=arctan⁡2324.\arctan\frac{7}{11} + \arctan\frac{1}{5} = \arctan\left(\frac{\frac{7}{11} + \frac{1}{5}}{1 - \frac{7}{55}}\right) = \arctan\left(\frac{\frac{46}{55}}{\frac{48}{55}}\right) = \arctan\frac{23}{24}.
arctan⁡2324+arctan⁡1n=π4.\arctan\frac{23}{24} + \arctan\frac{1}{n} = \frac{\pi}{4}.

Since arctan⁡1n=π4−arctan⁡2324\arctan\frac{1}{n} = \frac{\pi}{4} - \arctan\frac{23}{24}, take the tangent of both sides. Using tan⁡π4=1\tan\frac{\pi}{4} = 1 and the subtraction formula:

1n=tan⁡(π4−arctan⁡2324)=1−23241+2324=1244724=147.\frac{1}{n} = \tan\left(\frac{\pi}{4} - \arctan\frac{23}{24}\right) = \frac{1 - \frac{23}{24}}{1 + \frac{23}{24}} = \frac{\frac{1}{24}}{\frac{47}{24}} = \frac{1}{47}.

Therefore n=47n = \boxed{47}.

2013 AIME I · Problem 8: The domain of f(x)=arcsin⁡ ⁣(log⁡m(nx))f(x) = \arcsin\!\left(\log_m(nx)\right) is a closed interval of length 12013\frac{1}{2013}, where mm and nn are positive integers and m>1m > 1. Find the remainder when the smallest possible m+nm + n is divided by 10001000.

The whole problem hinges on the domain of arcsin⁡\arcsin. Its input must satisfy −1≤t≤1-1 \le t \le 1, so we need

−1≤log⁡m(nx)≤1.-1 \le \log_m(nx) \le 1.

Undo the logarithm by raising each side to the power of m:

m−1≤nx≤m1.m^{-1} \le nx \le m^{1}.

Divide by nn:

1mn≤x≤mn.\frac{1}{mn} \le x \le \frac{m}{n}.

This is the domain. Its length is the right end minus the left:

mn−1mn=m2−1mn.\frac{m}{n} - \frac{1}{mn} = \frac{m^2 - 1}{mn}.

Set that equal to 12013\frac{1}{2013}:

m2−1mn=12013⟹n=2013(m2−1)m.\frac{m^2 - 1}{mn} = \frac{1}{2013} \quad\Longrightarrow\quad n = \frac{2013(m^2 - 1)}{m}.

For nn to be an integer, mm must divide 2013(m2−1)2013(m^2 - 1). But mm shares no factor with m2−1m^2 - 1, so mm must divide 2013=3⋅11⋅612013 = 3 \cdot 11 \cdot 61.

Now minimize the sum. Substitute the expression for nn:

m+n=m+2013(m2−1)m=2014m−2013m.m + n = m + \frac{2013(m^2 - 1)}{m} = 2014m - \frac{2013}{m}.

This grows as mm grows, so we want the smallest divisor of 20132013 with m>1m > 1, namely m=3m = 3. Then

n=2013(9−1)3=2013⋅83=5368,n = \frac{2013(9 - 1)}{3} = \frac{2013 \cdot 8}{3} = 5368,

giving m+n=5371m + n = 5371. The remainder when divided by 10001000 is 371\boxed{371}.

Practice Problems

StatusSourceProblem NameDifficultyTags
CustomNormal
Show TagsArctan Identity, Inverse Functions, Trigonometry
CustomHard
Show TagsInverse Functions, Substitution, Trigonometry
CustomHard
Show TagsAlgebraic Manipulation, Inverse Functions
Berkeley Math Circle: Monthly Contest 8Hard
Show TagsTrigonometric functions
MathNetHard
Show TagsFunctions, Trigonometric functions
Harvard-MIT Mathematics TournamentHard
Show TagsTrigonometric functions
Harvard-MIT Mathematics TournamentHard
Show TagsTrigonometric functions
Harvard-MIT Mathematics TournamentHard
Show TagsTrigonometric functions
Harvard-MIT Math TournamentHard
Show TagsSingle-variable, Trigonometric functions
Harvard-MIT Math TournamentHard
Show TagsTrigonometric functions
Harvard-MIT Math TournamentHard
Show TagsApplications, Single-variable, Trigonometric functions
Harvard-MIT Math TournamentHard
Show TagsApplications, Derivatives, Trigonometric functions
Harvard-MIT Math TournamentHard
Show TagsTrigonometric functions
Harvard-MIT Mathematics TournamentHard
Show TagsTrigonometric functions
Harvard-MIT Mathematics TournamentHard
Show TagsLimits, Trigonometric functions
Harvard-MIT Mathematics TournamentHard
Show TagsODEs, Trigonometric functions
Harvard-MIT Mathematics TournamentHard
Show TagsApplications, Single-variable, Trigonometric functions
Harvard-MIT Mathematics TournamentHard
Show TagsLimits, Trigonometric functions
Harvard-MIT Mathematics TournamentHard
Show TagsTrigonometric functions
Harvard-MIT Mathematics TournamentHard
Show TagsSingle-variable, Trigonometric functions
Harvard-MIT Mathematics TournamentHard
Show TagsLimits, Trigonometric functions
Harvard-MIT Mathematics TournamentHard
Show TagsLimits, ODEs, Trigonometric functions
Harvard-MIT Mathematics TournamentHard
Show TagsDerivatives, Trigonometric functions
Harvard-MIT November TournamentHard
Show TagsFunctions, Trigonometric functions
Harvard-MIT Mathematics TournamentHard
Show TagsDerivatives, Limits, Trigonometric functions

Module Progress:

Join the Discord Community!

Stuck on a problem, or don't understand a module? Join the Discord and get help with your doubts while making more math friends.