A function and its inverse undo each other, with conditions
Quick review (2 minutes)
Picture a Ferris wheel of radius 1 turning counterclockwise.
sin(arcsin x) = x for x in [−1, 1] (undefined for other x); arcsin(sin x) = x for x in [−, ], and for any other x it is the angle in [−, ] with the same sine.
cos(arccos x) = x for x in [−1, 1] (undefined for other x); arccos(cos x) = x for x in [0, π], and for any other x it is the angle in [0, π] with the same cosine.
Say what the problem asks, reading from the inside out: arcsin(sin θ) means 'take the sine of θ, then find the angle in [−, ] with that sine'.
Name the outside function; it decides the test. A trig function outside needs the number inside to be one its inverse accepts: from −1 to 1 for sin and cos, any number for tan. An inverse outside needs the angle inside to lie in its range: [−, ] for arcsin, [0, π] for arccos, (−, ) for arctan.
If the test passes, the two undo each other, and the answer is the number or angle you started with.
If a trig function is outside and the test fails, the inverse does not exist there: write undefined.
If an inverse is outside and the test fails, do not cancel. First make sure the inside value exists: tan() does not, so arctan(tan()) is undefined. Then find the inside value exactly from the unit circle, then find the angle in the range with that value by the four-step method of the last lesson.
Check: with an inverse outside, the answer lies in that inverse's range, and its sine, cosine or tangent equals the inside value; with sin, cos or tan outside, the answer is the number inside the inverse, which that inverse had to accept.
Say what it asks, reading from the inside out: the inner function acts first.
Name the outside function and run its test.
If the test passes, the answer is the value you started with.
If the test fails, a trig function outside means undefined; an inverse outside means you first check that the inside value exists (tan has none at ), then find it exactly, then the angle in the range with that value.
Check: with an inverse outside, the answer lies in its range and has the inside value as its sine, cosine or tangent; with sin, cos or tan outside, the answer equals the number you started with.
Find sin(arcsin(−)).
Canceling on sight. arcsin and sine look as if they cross out, so it is tempting to write arcsin(sin()) = . Before you cancel, test the angle against the range of the outside inverse. is past = , so it fails; work from the inside instead: sin() = , then arcsin() = .
one-to-one function
A function in which no two inputs share an output, so every output comes from exactly one input. Only a one-to-one function can be run backward without confusion.
horizontal line test
A picture check for one-to-one: if some horizontal line crosses a graph more than once, two inputs share an output and the function has no inverse as it stands.
restricted domain
A smaller set of inputs chosen on purpose, keeping only a piece of the graph that passes the horizontal line test, so the function gets an inverse.
inverse function
The function that runs another one backward: if f sends a to b, the inverse sends b back to a. Its graph is the mirror image across the line y = x.
principal value
The one answer an inverse trig function gives, picked by an agreed rule from the infinitely many angles that share a sine, cosine or tangent. Principal means main.
inverse sine (arcsin)
The function that takes a number from −1 to 1 and gives back the angle from − to whose sine is that number. Written arcsin x or x; said ark-SINE.
inverse cosine (arccos)
The function that takes a number from −1 to 1 and gives back the angle from 0 to π whose cosine is that number. Written arccos x or x; said ark-KOH-sine.
inverse tangent (arctan)
The function that takes any real number and gives back the angle strictly between − and whose tangent is that number. Written arctan x or x; said ark-TAN-jent.
reference angle
The angle between an angle's ending side and the nearest part of the x-axis, always from 0 to . Angles with the same reference angle share the same trig values, up to sign.
1 In plain words
Picture a Ferris wheel of radius 1 turning counterclockwise. Your seat's height above the center is the sine of your angle. A helper hears only your height and must name your seat. Most heights belong to two seats, one on the rising right half and one on the falling left half, so the helper has a fixed rule: always name the seat on the right half. Turn by an angle from − to , which puts you on the right half, and the helper names your seat by that very angle, so you get your angle back. Ride an extra lap first, as with , and the helper still names your seat, but by its first-lap angle, . Sit on the left half and the helper names a different seat at the same height. Starting from a height always works, as long as the wheel reaches it: the seat the helper names sits at that very height.
2 Picture it
Reminder
Reminder: Sine and cosine on the unit circle, Tangent on the unit circle, Comparing fractions of π, Subtracting from π, Full turns, Reference angles, Signs by quadrant, The three ranges, The four-step method, Square and round brackets
- Sine and cosine on the unit circle. The point for an angle has x = cosine and y = sine. The point for is (−, ), so cos() = − and sin() = .
- Tangent on the unit circle. Tangent is y ÷ x for the point: for , ÷ = 1.
- Comparing fractions of π. Give them the same bottom, then compare the tops: = , and 5 > 3, so > .
- Subtracting from π. Write π with the same bottom first: π − = − = .
- Full turns. Adding or subtracting 2π, one full turn, lands on the same point, so sine and cosine do not change: − 2π = − = .
- Reference angles. The angle back to the nearest side of the x-axis. First swap θ for β, the angle from 0 up to 2π that ends at the same point. The reference angle is then β in Quadrant I, π − β in Quadrant II, β − π in Quadrant III and 2π − β in Quadrant IV, and 0 or when β lies on an axis. For : 2π − = . For −: β = , so the reference angle is 2π − = .
- Signs by quadrant. All Silver Tea Cups: all three are positive in Quadrant I, only sine in II, only tangent in III, only cosine in IV. So sin(), in Quadrant III, is negative.
- The three ranges. arcsin answers in [−, ], arccos in [0, π], arctan in (−, ). For example arccos(−1) = π is allowed, because the bracket at π is square.
- The four-step method. Reference angle, sign of the input, the quadrant the range allows, write the angle. arcsin(−): reference angle , negative input, the part of the range below 0, so −.
- Square and round brackets. A square bracket includes the end and a round one leaves it out: 1 is in [−1, 1], but is not in (−, ).
3 Why
arcsin x is, by definition, an angle whose sine is x. Take the sine of that angle and you must get x back: that is what the angle was chosen for. The only condition is that arcsin x exists, which needs x in [−1, 1], since every height on the unit circle is between −1 and 1. The other order loses information. Sine gives the same number for many angles, such as and , and arcsin answers with only one of them, the principal value in [−, ]. So arcsin(sin x) equals x only when x already lies in that interval.
4 The rule
sin(arcsin x) for x in (undefined for other x); arcsin for x in , and for any other x it is the angle in with the same sine.
cos(arccos x) for x in (undefined for other x); arccos for x in , and for any other x it is the angle in with the same cosine.
The same idea, five ways
The sine of the arcsine of x is x, for x from negative 1 to 1. The arcsine of the sine of x is x only when x is from negative pi over 2 to pi over 2.
Sine undoes its inverse function, arcsin, for every number arcsin accepts, but arcsin undoes sine only for angles in its own range; for any other angle it returns the angle in that range with the same sine.
- sin(arcsin x) = x when −1 ≤ x ≤ 1
- arcsin(sin x) = x when − ≤ x ≤
- Interval: arcsin(sin x) = x for x in [−, ]
- Set-builder: arcsin(sin x) = x on {x | − ≤ x ≤ }
- Other notation: sin(sin⁻¹ x) = x on [−1, 1] and sin⁻¹(sin x) = x on [−, ]
- Cosine: cos(arccos x) = x on [−1, 1]; arccos(cos x) = x on [0, π]
A round trip through a translator: Spanish to English and back always returns your word, but English to Spanish and back can turn 'large' into 'big'.
The same idea, other ways
Draw the unit circle. arcsin answers only on its right half: from the bottom point, −, up through the rightmost point, 0, to the top point, . Every height from −1 to 1 appears exactly once on that half. An angle from − to survives the round trip arcsin(sin x). An angle on the left half, such as , comes back as the angle of its mirror point across the y-axis, : the point on the right half at the same height. For arccos the allowed half is the top half, from 0 to π, and the mirror is across the x-axis.
Translate the English word 'large' into Spanish and you get 'grande'. Translate 'grande' back and the dictionary gives its first entry, 'big'. Several English words share that one Spanish word, so the trip back cannot know which one you meant and hands back the standard one. Sine works the same way: , and all have sine , and arcsin hands back the standard angle, the principal value . Start on the Spanish side and the trip always works: 'grande' becomes 'big' and comes back as 'grande', the way sin(arcsin()) = .
The table sends four angles whose sine is through sine and then arcsin. Every column ends at , so only the column that started at came back to where it began: is the only one of the four inside [−, ]. Now start from numbers instead: sin(arcsin 0.3) = 0.3 and sin(arcsin(−1)) = sin(−) = −1. Every number from −1 to 1 comes back unchanged.
arcsin was built as the inverse function of sine on the restricted domain [−, ]. There sine is a one-to-one function: it passes the horizontal line test, so each height comes from exactly one angle. An inverse function undoes its function on exactly those inputs, so arcsin(sin x) = x for every x in [−, ]. For any other x, the output still has to lie in the range [−, ], so it cannot be x. In the other order, every number in [−1, 1] is a legal input to arcsin, and arcsin x is by definition an angle with sine x, so sin(arcsin x) = x for all of them.
Outside trig, trust; outside inverse, test. When sin, cos or tan is on the outside, trust the cancel, as long as the number inside is one the inverse accepts. When arcsin, arccos or arctan is on the outside, first test the angle against that inverse's range. Pass, and the answer is the angle. Fail, and the answer is the angle in the range with the same sine, cosine or tangent, provided the inside has a value at all: tan x has none when x is an odd multiple of .
Outside trig, trust: sin(arcsin x) = x for x in [−1, 1]
Outside inverse, test: arcsin(sin x) = x for x in [−π/2, π/2]
Test failed: if the inside value exists, the angle in the range with that value
| Pair | Trig on the outside | True for | Inverse on the outside | True for |
|---|---|---|---|---|
| sine and inverse sine | sin(arcsin x) = x | x in [−1, 1] | arcsin(sin x) = x | x in [−, ] |
| cosine and inverse cosine | cos(arccos x) = x | x in [−1, 1] | arccos(cos x) = x | x in [0, π] |
| tangent and inverse tangent | tan(arctan x) = x | every real x | arctan(tan x) = x | x in (−, ) |
Part 1: Sine and the inverse sine (arcsin)
Sine on the outside: arcsin x is an angle whose sine is x, so the sine of that angle is x again. That works for every x from −1 to 1; past them no such angle exists, and sin(arcsin x) is undefined. Inverse sine on the outside: arcsin answers only on the right half of the unit circle, with angles from − at the bottom to at the top. If your angle ends at a point on the right half, arcsin gives the angle in that range that ends at the same point, which is your angle itself when it was already in the range. If your angle ends on the left half, arcsin gives the angle of the mirror point across the y-axis, at the same height.
- sin(arcsin x) = x for every x in [−1, 1]
- sin(arcsin x) is undefined when x < −1 or x > 1
- arcsin(sin x) = x for every x in [−, ]
- Any other x: arcsin(sin x) is the angle in [−, ] with the same sine as x
- Mirror: the points for θ and π − θ are mirror images across the y-axis, so sin(π − θ) = sin θ
Worked example
Sine on the outside: sin(arcsin(−))
Find sin(arcsin(−)).
Show the steps and the answer
- Say what it asks: find the angle whose sine is , then take the sine of that angle.arcsin is inside, so it acts first, and sine acts on its answer.
- Test the number: , so is less than 1, and lies between and 0, inside .Sine on the outside gives back the number whenever arcsin of it exists, and arcsin accepts exactly the numbers from −1 to 1.
- Cancel: sin(arcsin()) .arcsin(−) is an angle whose sine is −, so its sine is −.
Counterexamples
What it is not, and the tempting wrong moves.
Tips and tricks
- The number inside sin(arcsin x) does not have to be a special value: sin(arcsin 0.37) = 0.37, and you never need the angle.
Part 2: Cosine and the inverse cosine (arccos)
Cosine is the left-right position of the point on the unit circle, its x. Cosine on the outside: arccos x is an angle whose cosine is x, so cos(arccos x) = x for every x from −1 to 1, and it is undefined past them. Inverse cosine on the outside: arccos answers only on the top half of the circle, from 0 at the right end, over the top, to π at the left end. Every x from −1 to 1 appears exactly once on that half. If your angle ends on the top half, arccos gives the angle in [0, π] that ends at that same point, your angle itself when it was already in [0, π]. If it ends on the bottom half, such as , arccos gives the angle of the mirror point across the x-axis, which has the same x.
- cos(arccos x) = x for every x in [−1, 1]; undefined for other x
- arccos(cos x) = x for every x in [0, π]
- Any other x: arccos(cos x) is the angle in [0, π] with the same cosine as x
- Mirror: the points for θ and −θ are mirror images across the x-axis, so cos(−θ) = cos θ
- A negative angle never comes out of arccos, because arccos never answers below 0
Worked example
An angle on the bottom half: arccos(cos())
Find the exact value of arccos(cos()).
Show the steps and the answer
- Say what it asks: take the cosine of , then find the angle in whose cosine is that number.The outside function is arccos, and arccos answers only with angles from 0 to π.
- Test the angle: , and , so is outside .With the same bottom, compare the tops: 7 > 4. The test fails, so arccos and cosine may not be canceled.
- Find the inside value. is in Quadrant IV, its reference angle is , and cosine is positive there, so . lies between = and 2π = , which is Quadrant IV. There the point is right of the y-axis, so its x, the cosine, is positive.
- Replace the inside by its value: arccos arccos().What remains is an inverse cosine of a special value.
- Find arccos(): the reference angle is , because ; the input is positive, so the answer is in Quadrant I; the angle is .arccos answers a positive input with an angle between 0 and .
Counterexamples
What it is not, and the tempting wrong moves.
Tips and tricks
- Think 'arccos lives on the top half': an answer from arccos that is negative or larger than π is always a mistake.
Part 3: Tangent and the inverse tangent (arctan)
Tangent is the height divided by the left-right position, y ÷ x, for the point on the unit circle. Tangent on the outside: arctan accepts every real number, so tan(arctan x) = x for every x, with no condition at all. Inverse tangent on the outside: arctan answers only with angles strictly between − and , the right half without its top and bottom points. If your angle ends on the right half, arctan gives the angle in the range that ends at that same point. If it ends on the left half, arctan gives the angle of the opposite point, half a turn away: a half turn flips the sign of both x and y, so y ÷ x stays the same.
- tan(arctan x) = x for every real number x
- arctan(tan x) = x for every x in (−, )
- Any other x where tan x exists: arctan(tan x) is the angle in (−, ) with the same tangent as x
- Half turn: the points for θ and θ − π are opposite each other, so tan(θ − π) = tan θ
- tan() does not exist, because x = 0 at the top of the circle, so arctan(tan()) is undefined
Worked example
An angle on the left half: arctan(tan())
Find the exact value of arctan(tan()).
Show the steps and the answer
- Say what it asks: take the tangent of , then find the angle in whose tangent is that number.The outside function is arctan, and arctan answers only with angles strictly between − and .
- Test the angle: , and , so is outside the range.With the same bottom, compare the tops: 3 > 2. The test fails, so arctan and tangent may not be canceled.
- Find the inside value: the point for is , so .Tangent is y ÷ x, and a number divided by its own negative is −1.
- Replace the inside by its value: arctan arctan().What remains is an inverse tangent of a special value.
- Find arctan(): the reference angle is , because ; the input is negative, so the answer is in the part of the range below 0; the angle is .arctan answers a negative input with a negative angle, between − and 0.
Counterexamples
What it is not, and the tempting wrong moves.
Tips and tricks
- tan(arctan x) never fails, whatever the number: tan(arctan 50) = 50.
5 How
- Say what the problem asks, reading from the inside out: arcsin(sin θ) means 'take the sine of θ, then find the angle in [−, ] with that sine'.
- Name the outside function; it decides the test. A trig function outside needs the number inside to be one its inverse accepts: from −1 to 1 for sin and cos, any number for tan. An inverse outside needs the angle inside to lie in its range: [−, ] for arcsin, [0, π] for arccos, (−, ) for arctan.
- If the test passes, the two undo each other, and the answer is the number or angle you started with.
- If a trig function is outside and the test fails, the inverse does not exist there: write undefined.
- If an inverse is outside and the test fails, do not cancel. First make sure the inside value exists: tan() does not, so arctan(tan()) is undefined. Then find the inside value exactly from the unit circle, then find the angle in the range with that value by the four-step method of the last lesson.
- Check: with an inverse outside, the answer lies in that inverse's range, and its sine, cosine or tangent equals the inside value; with sin, cos or tan outside, the answer is the number inside the inverse, which that inverse had to accept.
Evaluating a trig function and its inverse stacked together
- Say what it asks, reading from the inside out: the inner function acts first.
- Name the outside function and run its test.
- If the test passes, the answer is the value you started with.
- If the test fails, a trig function outside means undefined; an inverse outside means you first check that the inside value exists (tan has none at ), then find it exactly, then the angle in the range with that value.
- Check: with an inverse outside, the answer lies in its range and has the inside value as its sine, cosine or tangent; with sin, cos or tan outside, the answer equals the number you started with.
6 Worked example
Worked example
An angle outside the range: arcsin(sin())
Find the exact value of arcsin(sin()).
Show the steps and the answer
- Say what it asks: take the sine of , then find the angle in whose sine is that number.The inner function acts first. The outside function is arcsin, and arcsin answers only with angles from − to .
- Test the angle against the range. Give a bottom of 6: . Since , the angle is outside .With the same bottom, comparing fractions means comparing tops, and 5 > 3. Had the angle passed, the answer would be itself; it failed, so arcsin and sine may not be canceled.
- Find the inside value. is in Quadrant II, its reference angle is , and sine is positive there, so . lies between = and π = , which is Quadrant II. There the point on the unit circle is above the x-axis, so its height, the sine, is positive.
- Replace the inside by its value: arcsin arcsin().What remains is a plain inverse sine of a special value, the kind of problem the last lesson solved.
- Find arcsin() with the four-step method: the reference angle is , because ; the input is positive; so the answer is in Quadrant I, the part of the range from 0 to ; the angle is .arcsin answers a positive input with an angle between 0 and , and is that angle.
Ladder: from easy to exam-hard
Try each rung before you open its steps.
Rung 1 of 7
Trig outside, an allowed number inside
Find sin(arcsin()).
Show the steps and the answer
- Say what it asks: find the angle whose sine is , then take the sine of that angle.arcsin is inside, so it acts first, and sine acts on its answer.
- Test the number: , so is in .With sine on the outside, the test is whether arcsin accepts the number, and arcsin accepts exactly the numbers from −1 to 1.
- Cancel: sin(arcsin()) .arcsin() is an angle whose sine is , so its sine is . You never need to know which angle it is.
Rung 2 of 7
Trig outside, a number that is not allowed
Find cos(arccos 3).
Show the steps and the answer
- Say what it asks: find the angle whose cosine is 3, then take the cosine of that angle.arccos is inside, so it acts first.
- Test the number: 3 > 1, so 3 is not in .arccos accepts only numbers from −1 to 1.
- Stop and write undefined: arccos 3 does not exist, so cosine has nothing to act on.Cosine is the x of a point on the unit circle, and no point of a circle of radius 1 lies 3 units right of the center.
Rung 3 of 7
Inverse outside, an angle already in the range
Find arcsin(sin(−)).
Show the steps and the answer
- Say what it asks: take the sine of , then find the angle in with that sine.arcsin is on the outside, and arcsin answers only from − to .
- Test the angle: with a bottom of 6, the range runs from to , and lies between them.With equal bottoms, compare the tops: −3 ≤ −1 ≤ 3.
- Cancel: arcsin.On the restricted domain each sine value belongs to exactly one angle, so arcsin returns the angle you started with.
Rung 4 of 7
Inverse outside, an angle in Quadrant II
Find arcsin(sin()).
Show the steps and the answer
- Say what it asks: take the sine of , then find the angle in with that sine.arcsin is on the outside, and arcsin answers only from − to .
- Test the angle: with a bottom of 6, and . Since , the angle is outside the range.The test fails, so arcsin and sine may not be canceled.
- Find the inside value: is in Quadrant II, its reference angle is , and sine is positive there, so . = lies between = and π = . In Quadrant II the point is above the x-axis, so its height, the sine, is positive.
- Finish: arcsin() has reference angle and a positive input, so it is the Quadrant I angle .This is the four-step method: a positive input puts the answer between 0 and .
Rung 5 of 7
A negative angle inside arccos
Find arccos(cos(−)).
Show the steps and the answer
- Say what it asks: take the cosine of , then find the angle in with that cosine.arccos is on the outside, and arccos answers only from 0 to π.
- Test the angle: is negative, and the range starts at 0, so it fails.arccos never answers with a negative angle.
- Find the inside value: add a full turn, . That angle is in Quadrant III, its reference angle is , and cosine is negative there, so .Adding 2π, one full turn, lands on the same point, so the cosine is unchanged. = lies between π = and = , which is Quadrant III, where the point is left of the y-axis, so its x, the cosine, is negative.
- Finish: arccos() has reference angle and a negative input, so it is in Quadrant II: .arccos answers a negative input with an angle past , in Quadrant II, where cosine is negative.
Rung 6 of 7
Arcsin outside, an angle in Quadrant III
Find arcsin(sin()).
Show the steps and the answer
- Say what it asks: take the sine of , then find the angle in with that sine.arcsin is on the outside, and arcsin answers only from − to .
- Test the angle: with a bottom of 6, and . Since , the angle is outside the range.The test fails, so arcsin and sine may not be canceled.
- Find the inside value: is in Quadrant III, its reference angle is , and sine is negative there, so . lies between π = and = , which is Quadrant III. There the point is below the x-axis, so its height, the sine, is negative.
- Finish: arcsin() has reference angle and a negative input, so it is the angle below 0 in the range: .arcsin answers a negative input with a negative angle, between − and 0.
Rung 7 of 7
Arccos outside, an angle in Quadrant III
Find arccos(cos()).
Show the steps and the answer
- Say what it asks: take the cosine of , then find the angle in with that cosine.arccos is on the outside, and arccos answers only from 0 to π.
- Test the angle: , and , so the angle is outside .The test fails, so arccos and cosine may not be canceled.
- Find the inside value: is in Quadrant III, its reference angle is , and cosine is negative there, so . lies between π = and = , which is Quadrant III. There the point is left of the y-axis, so its x, the cosine, is negative.
- Finish: arccos() has reference angle and a negative input, so it is in Quadrant II: .arccos answers a negative input with an angle past , in Quadrant II, where cosine is negative.
Counterexamples
What it is not, and the tempting wrong moves.
Tips and tricks
- Memory device: outside trig, trust; outside inverse, test. sin(arcsin x) gives back x whenever x is in [−1, 1]; arcsin(sin x) gives back x only after x passes the range test.
- The outside function owns the test. In arccos(cos x), compare x with [0, π], the range of arccos, even if x would pass for arcsin.
- Write the middle step every time: arcsin(sin()) = arcsin() = . It shows a reader why you did not cancel, and it lets you catch a slip in the last step.
- With an inverse on the outside, check the answer twice: it must lie in that inverse's range, and its sine, cosine or tangent must equal the inside value. With sin, cos or tan on the outside, the answer is the number you started with, once the inverse has accepted it.
- Picture the halves: the right half of the unit circle for arcsin and arctan, the top half for arccos. An answer on the wrong half is wrong.
- Same angle, different answers: arcsin(sin()) = −, but arccos(cos()) = . Always look at which inverse is outside.
7 The trap
Canceling on sight. arcsin and sine look as if they cross out, so it is tempting to write arcsin(sin()) = . Before you cancel, test the angle against the range of the outside inverse. is past = , so it fails; work from the inside instead: sin() = , then arcsin() = .
Flash cards
Say the answer out loud, then flip.
Why can arcsin(sin x) give an angle different from x?
What is cos(arccos 2)?
- Undefined.
- arccos accepts only numbers from −1 to 1, so arccos 2 does not exist.
What is arcsin(sin(−))?
- The angle − lies in [−, ], the range of arcsin, so the two undo each other. You do not need to find sin(−).
What is arccos(cos())?
- The angle is past π, outside [0, π], so do not cancel. cos() = −, and the angle in [0, π] with cosine − is .