The last lesson of Chapter 10. An inequality asks where one side exceeds the other, and the answer is a union of intervals rather than a list of points. Covers the sign diagram technique with its two kinds of landmark, choosing test values in awkward gaps, using the same machinery to find the domain of any function built from circular functions, the strategy of solving on one period and translating, and equations and inequalities in the inverse trigonometric functions, which behave in the opposite way to everything else in the chapter.
Subject: Trigonometry · 65 slides · symbolic lesson
Open the interactive version of this deck
Title
Trigonometry · Section 10.7 (second part)
Solve the equation first, then ask what happens between the answers
Objectives
The first half of this section asked where two expressions are equal, and the answers were points. Replace the equals sign by an inequality and the answer becomes a union of intervals. The same change turns a domain question, which is really an inequality in disguise, into something you can now finish.
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.7 Trigonometric Equations and Inequalities §10.7, pp. 866-874 — Examples 10.7.3, 10.7.4 and 10.7.5
You have spent the first half of this section learning to solve trigonometric equations, and you can now find every angle at which two expressions agree. This lesson changes one character in the question. Replace the equals sign with an inequality and you are no longer asking where the two expressions meet; you are asking which side of each meeting point you are on.
That change makes the answer a different kind of object. An equation returns a list of numbers, usually infinite families of them. An inequality returns stretches of the line: a union of intervals. All the hard work has already been done, because the endpoints of those intervals are exactly the solutions of the corresponding equation.
The second half of the lesson spends that same machinery on a question you have been answering informally since algebra: what is the domain of this formula? A denominator asks you an equation, a square root asks you an inequality, and a trigonometric formula asks you the same question infinitely many times over, once per period. The last part turns everything upside down with the inverse functions, which have no families at all.
Warm-up
Discussion prompt
You already know that twice the sine of x equals one at exactly two places inside one revolution: pi over six and five pi over six. Before reading on, write down what those two numbers do to the interval from zero to two pi, and what single further fact would tell you everywhere that twice the sine of x is at most one.
Write your answer before reading on. You are not being asked to solve anything new, only to notice what the solutions of an equation do to the line they sit on.
Two points inside an interval cut it into three pieces. That is the whole geometric content of the sign diagram, and everything else in this part of the lesson is bookkeeping around it. The question worth sitting with is the second one: what extra fact turns three pieces into an answer?
The answer is the sign of the expression on each piece, and here is the point of the whole technique: you only have to substitute one number per piece. Not three numbers, not a sample of ten, one. Why one is enough is the subject of the next two slides.
Section
Part 1
Concept
Figure (svg): The curve y equals twice the sine of x across one revolution, with a horizontal line at 1. The curve meets the line at pi over six and at five pi over six, and the two stretches where it lies on or below the line, from 0 to pi over six and from five pi over six to two pi, are shaded.
Solving an equation locates crossings. Solving an inequality asks which side of the crossing you are on, and the answer is everything between one crossing and the next.
\[ 2\sin(x) \le 1 \quad\text{on } [0, 2\pi) \]
The circular functions are continuous on their domains, so the same sign diagram you used on polynomials and rational expressions works here without a single change.
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.7 Trigonometric Equations and Inequalities §10.7, p. 866 — the paragraph introducing Example 10.7.3
Look at the picture rather than the algebra for a moment. The wavy curve is twice the sine of x and the flat line is at height one. They meet twice. The question asks where the wave is at or below the line, and the eye answers it instantly: at the two ends of the revolution, not in the middle.
Now notice what the two meeting points are for. They are not the answer; they are the boundaries of the answer. That distinction causes more lost marks than any calculation in this section, because it is very easy to solve the equation carefully, write down the two angles, and hand those in.
The reason the technique from your earlier algebra course transfers without modification is continuity. Sines, cosines and tangents are continuous everywhere they are defined, so they cannot jump from one side of a line to the other without touching it. That is the only property the whole method needs.
Concept
A continuous function cannot get from a positive value to a negative one without passing through zero on the way. So between two consecutive landmarks the sign never changes.
\[ f \text{ continuous on } (c, d), \; f(p) > 0, \; f(q) < 0 \;\Longrightarrow\; f(z) = 0 \text{ for some } z \text{ between} \]
Turn that around. If there is no zero and no break between two landmarks, there can be no sign change either, so whatever one substitution reports is true across the entire region.
\[ \text{no zero and no break in } (c,d) \;\Longrightarrow\; f \text{ keeps one sign on } (c,d) \]
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.7 Trigonometric Equations and Inequalities §10.7, p. 866 — why the sign diagram technique applies
This is the theorem that makes the sign diagram legal, and it is worth being able to state. If a continuous function is positive somewhere and negative somewhere else, then between those two places it must equal zero. You cannot get from above the axis to below it without crossing.
Read that backwards and you have exactly what the method needs. Take a stretch of the line with no zero of f inside it and no point where f breaks. Then f cannot change sign anywhere on that stretch, so whatever sign you find at one point is the sign at every point.
That is why the landmarks are the zeros and the undefined points, and nothing else. Those are the only two ways a sign change can happen. Anywhere else, the function is quietly keeping the sign it already had, and a single substitution catches it in the act.
Concept
Figure (svg): An abstract number line with two landmarks: a solid vertical mark labelled as a zero, and a dashed vertical mark labelled as an undefined point. The three regions between and beyond them carry the signs plus, minus and plus, and each has a dot in the middle labelled as a test value.
Zero is the one number whose sign is never in doubt, which is why the first move is always to compare against it rather than against the original right-hand side.
The four moves are worth memorising as a sequence, because each one is cheap and skipping any of them costs you the whole problem. The picture shows the finished object: a line, the landmarks that cut it, a sign per region and a test point in each.
Move one looks pedantic and is not. If you compare twice the sine of x with one directly, you have two quantities changing at once and no fixed reference. Compare with zero and there is exactly one quantity to track, with a sign that is never ambiguous.
Move four is deliberately separate from move three. Finding the regions is sign work; deciding the brackets at their ends is a different question, answered by looking at the inequality symbol and at what kind of landmark each endpoint is. Doing them together is how the endpoints get decided by habit instead of by thought.
Notation
Annotate
On: \( f(x) = 2\sin(x) - 1 \le 0 \;\text{ on }\; [0, 2\pi) \)
Step through the four annotations in order; each one corresponds to a decision you will make on every problem in this lesson. The first two are about form, the last two about the answer.
The annotation on the inequality symbol is the one to underline. That single character does no work at all until the very last line, and then it decides every bracket around every zero. Students who lose marks here almost never lose them on the sign work.
The stated window matters more than it looks. Trigonometric expressions repeat for ever, so without a window the answer would always be an infinite union. Later in this lesson, when you find domains, nobody gives you a window, and you will see what that costs.
Worked example
Solve on one revolution, and give the answer in interval notation.
\[ 2\sin(x) \le 1, \qquad 0 \le x < 2\pi \]
Collect everything on one side
Why: Zero is the only number whose sign is unambiguous.
\[ 2\sin(x) - 1 \le 0 \]
Name the left side f
Why: Now the question reads: where is f negative or zero?
\[ f(x) = 2\sin(x) - 1 \]
Look for points where f is undefined
Why: The sine is defined for every real number, so there are none, and the only landmarks will be zeros.
\[ \text{domain of } f = (-\infty, \infty) \]
Solve f(x) = 0
Why: A basic sine equation, from the first half of this section.
\[ 2\sin(x) - 1 = 0 \iff \sin(x) = \tfrac12 \]
Keep the members of each family that lie in the window
Why: Two families, one member each inside one revolution.
\[ x = \tfrac{\pi}{6} + 2\pi k, \; \tfrac{5\pi}{6} + 2\pi k \;\Longrightarrow\; x = \tfrac{\pi}{6}, \; \tfrac{5\pi}{6} \]
Evaluate f at one point inside each of the three regions
Why: Pick the easiest angles available: 0, a quarter turn, a half turn.
\[ f(0) = -1, \qquad f\left(\tfrac{\pi}{2}\right) = 1, \qquad f(\pi) = -1 \]
Figure (svg): A sign diagram on the interval from 0 to two pi for f of x equals twice the sine of x minus one. Zeros are marked at pi over six and five pi over six, the three regions carry the signs minus, plus and minus, and the outer two regions are shaded as the answer.
Collect the regions where f is negative, and add the zeros
Why: The inequality is not strict, so the two zeros satisfy it and belong to the answer.
\[ \left[0, \tfrac{\pi}{6}\right] \cup \left[\tfrac{5\pi}{6}, 2\pi\right) \]
Check one point inside each interval and one outside
Why: At zero and at three quarters of a turn the inequality holds; at a quarter turn, which the answer excludes, it fails.
\[ 2\sin(0) = 0 \le 1, \quad 2\sin\left(\tfrac{3\pi}{2}\right) = -2 \le 1, \quad 2\sin\left(\tfrac{\pi}{2}\right) = 2 \not\le 1 \]
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.7 Trigonometric Equations and Inequalities §10.7, p. 866 — Example 10.7.3, part 1
Follow the order of the steps, not just their content. Everything to one side, name the left side, hunt for undefined points, hunt for zeros, test, collect, decide the brackets. On this problem the third step is over in a line, but do not learn to skip it: the very next example has undefined points and they change the answer.
The three test values were chosen for laziness, and that is correct practice. Zero, a quarter turn and a half turn all have sines you know without thinking, and each lies inside one of the three regions. A test value has exactly two jobs: be in the right region, and be easy.
The zeros are included here because the inequality allows equality. Twice the sine of a sixth of pi is exactly one, and at most one includes one. Notice the right end of the answer, at two pi: that is a round bracket, not because of the sign work, but because the window itself was written with a round bracket there.
Prediction
An expression is defined everywhere on the stated interval, and inside that interval it has exactly three zeros.
Predict first
How many test values does its sign diagram need?
Correct: Four
Why: Three landmarks cut the interval into four regions, and each region needs exactly one test value. In general n landmarks strictly inside the interval make n plus one regions. Counting the regions before you start is a cheap way to know when you have finished testing.
Commit to a number before revealing. This is a counting question, not a trigonometry question, and getting into the habit of counting the regions before you start is a genuine time saver.
Three landmarks strictly inside an interval produce four regions, because each landmark adds one cut. If a landmark happened to sit at an end of the interval it would not cut anything, so it would not add a region; that is the only way the count changes.
Why bother? Because it tells you when you have finished. If you know there are four regions and you have tested three, you have not made a decision about a quarter of the interval yet. Most incomplete answers in this topic are a region that was never tested.
Worked example
The exercise set states this window with both ends included, which changes the brackets at the ends of the answer.
\[ 2\cos(x) \ge 1, \qquad 0 \le x \le 2\pi \]
Move everything over and name f
Why: Same first move, every time.
\[ f(x) = 2\cos(x) - 1 \ge 0 \]
Solve f(x) = 0 inside the window
Why: The cosine takes the value one half once on the way down and once on the way back.
\[ \cos(x) = \tfrac12 \;\Longrightarrow\; x = \tfrac{\pi}{3}, \; \tfrac{5\pi}{3} \]
Test the three regions
Why: The ends of the window are themselves convenient test points here.
\[ f(0) = 1, \qquad f(\pi) = -3, \qquad f(2\pi) = 1 \]
Figure (svg): The curve y equals twice the cosine of x from 0 to two pi with a horizontal line at 1. The curve meets the line at pi over three and five pi over three, and the two stretches where it lies on or above the line, at the two ends of the window, are shaded.
Collect the positive regions
Why: Both zeros are included, because the inequality is not strict, and both ends of the window are included, because the window is closed.
\[ \left[0, \tfrac{\pi}{3}\right] \cup \left[\tfrac{5\pi}{3}, 2\pi\right] \]
Check both ends and the middle
Why: At each end the cosine is 1, so twice it is 2, comfortably at least 1; at a half turn the cosine is minus 1, so twice it is minus 2, which fails.
\[ 2\cos(0) = 2 \ge 1, \quad 2\cos(2\pi) = 2 \ge 1, \quad 2\cos(\pi) = -2 \not\ge 1 \]
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.7 Trigonometric Equations and Inequalities §10.7, p. 875 — Exercise 76
This is the cosine twin of the worked example, and it is here to make one point: read the window. The exercise set for this section states its inequalities on the closed interval from zero to two pi, with both ends in, so both ends can appear in the answer with square brackets.
The shape of the answer is different from the sine version, and it is worth seeing why. The cosine starts at its maximum, dips through the middle of the revolution and climbs back, so the stretch where it is large is split across the two ends of the window rather than sitting in the middle.
The test values are the two ends and the midpoint, all of them angles whose cosine you know. Notice that the ends of the window are themselves perfectly legitimate test points when no landmark sits there.
Trap
A tempting shortcut on the next example:
\[ \sin(2x) > \cos(x) \]
\[ 2\sin(x)\cos(x) > \cos(x) \]
\[ 2\sin(x) > 1 \]
Wrong. Two things were destroyed at once.
Dividing by the cosine deletes the places where it is zero, and reverses the inequality wherever it is negative. Factor instead, and let the sign diagram handle both.
\[ \cos(x)\bigl(2\sin(x) - 1\bigr) > 0 \]
The shortcut answers one interval; the truth is two. At three quarters of a turn the shortcut says yes and the original says no.
\[ 2\sin\left(\tfrac{3\pi}{4}\right) \approx 1.414 > 1 \]
Cancelling the common factor feels like ordinary algebra, and that is exactly why it is dangerous here. Dividing both sides of an inequality by a quantity is only safe if you know the sign of that quantity, and the cosine changes sign twice per revolution.
There is a second, quieter loss. Dividing by the cosine deletes the two angles where the cosine is zero, and both of them turn out to be boundaries of the real answer. So the shortcut does not merely risk a flipped inequality; it also destroys two of the four landmarks.
The counterexample at the bottom is the cheapest possible test of any shortcut. Pick an angle, run it through both the original inequality and your simplified version, and see whether they agree. At three quarters of a turn they do not.
Worked example
\[ \sin(2x) > \cos(x), \qquad 0 \le x < 2\pi \]
Move everything to one side and name f
Why: Never divide across an inequality by something whose sign you do not know.
\[ f(x) = \sin(2x) - \cos(x) \]
Apply the double angle identity for sine
Why: This makes both terms carry a cosine.
\[ \sin(2x) - \cos(x) = 2\sin(x)\cos(x) - \cos(x) \]
Factor out the common cosine
Why: A product is zero exactly when one of its factors is.
\[ f(x) = \cos(x)\bigl(2\sin(x) - 1\bigr) \]
Set the first factor to zero
Why: The cosine vanishes at the odd multiples of a quarter turn.
\[ \cos(x) = 0 \;\Longrightarrow\; x = \tfrac{\pi}{2}, \; \tfrac{3\pi}{2} \]
Set the second factor to zero
Why: The same basic sine equation as before.
\[ \sin(x) = \tfrac12 \;\Longrightarrow\; x = \tfrac{\pi}{6}, \; \tfrac{5\pi}{6} \]
Test the first three of the five regions
Why: Zero, an eighth of a turn and three eighths of a turn.
\[ f(0) = -1, \quad f\left(\tfrac{\pi}{4}\right) = \tfrac{2-\sqrt{2}}{2} \approx 0.293, \quad f\left(\tfrac{3\pi}{4}\right) = \tfrac{\sqrt{2}-2}{2} \approx -0.293 \]
Test the last two regions
Why: A half turn and seven eighths of a turn.
\[ f(\pi) = 1, \qquad f\left(\tfrac{7\pi}{4}\right) = -\tfrac{2+\sqrt{2}}{2} \approx -1.707 \]
Figure (svg): A sign diagram on the interval from 0 to two pi for the sine of two x minus the cosine of x. Zeros are marked at pi over six, pi over two, five pi over six and three pi over two; the five regions carry the signs minus, plus, minus, plus, minus, and the two positive regions are shaded.
Figure (svg): Two curves across one revolution: the sine of two x and the cosine of x. They cross at pi over six, pi over two, five pi over six and three pi over two. The two stretches where the sine of two x is higher, from pi over six to pi over two and from five pi over six to three pi over two, are shaded.
Collect the two positive regions
Why: The inequality is strict, so every landmark is excluded.
\[ \left(\tfrac{\pi}{6}, \tfrac{\pi}{2}\right) \cup \left(\tfrac{5\pi}{6}, \tfrac{3\pi}{2}\right) \]
Check a point inside each answer interval
Why: At an eighth of a turn and at a half turn the original inequality holds, and the graphs agree.
\[ \sin\left(\tfrac{\pi}{2}\right) = 1 > \tfrac{\sqrt{2}}{2} = \cos\left(\tfrac{\pi}{4}\right), \qquad \sin(2\pi) = 0 > -1 = \cos(\pi) \]
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.7 Trigonometric Equations and Inequalities §10.7, pp. 866-867 — Example 10.7.3, part 2
Two obstacles appear before the sign diagram can even start: two different functions, and two different arguments. Both are cleared by the double angle identity, which rewrites the sine of two x as a product containing a cosine, after which the common factor is visible.
Factoring is the move that dividing was pretending to be. A product is zero exactly when one of its factors is zero, so the two factors between them hand you all four landmarks at once, and none is lost.
Five regions, five test values, and the signs alternate. That alternation is not an accident: at every landmark exactly one factor changes sign, so the product changes sign too. If your test values ever give you two neighbouring regions with the same sign, either you have miscounted a landmark or you have made an arithmetic slip, and it is worth going back to find out which.
Fill the middle
The factored form of the last example is on the left. Its four landmarks inside one revolution are on the right, with two of them missing.
Fill in the blanks
\cos(x)\bigl(2\sin(x)-1\bigr) = 0 \;\Longrightarrow\; x = \tfrac{\pi}{6}, \; \tfrac\tfrac{5\pi}{6}___, \; ___, \; \tfrac______
Why: The two given landmarks come from the cosine factor, which is zero at a quarter turn and at three quarters of a turn. The two missing ones come from the other factor: the sine equals one half at a sixth of pi and at five sixths of pi. Listing all four in increasing order is what makes the regions easy to read off.
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.7 Trigonometric Equations and Inequalities §10.7, p. 866 — Example 10.7.3, part 2
Fill both blanks before checking. The two given landmarks come from one factor and the two missing ones from the other, so the exercise is really asking whether you can keep two families apart in your head.
Listing the four in increasing order is not cosmetic. The sign diagram is a picture of the line, so the landmarks have to go on it in the order they actually occur, and a list written in the order you happened to find them will put them in the wrong places.
Sorting
Take the expression given by the tangent of x minus one, on one revolution.
Sort into buckets
Sort each angle by the part it plays on the sign diagram.
Work out the tangent of each angle before sorting. Two of the six are zeros of the expression, two are points where the tangent does not exist, and two are ordinary interior points.
The third bucket is the one that teaches. An ordinary point is not useless; it is precisely what you want as a test value. So the sorting exercise is really a sorting of jobs: two kinds of landmark, which cut the line, and everything else, which reports on it.
If you put a quarter turn in the zero bucket, check what the tangent actually does there. The cosine underneath is zero, so the fraction has no value at all, which is a different failure from being equal to one.
Section
Part 2
Concept
Figure (svg): Two panels. On the left, the curve twice the sine of x minus one crosses the axis cleanly at pi over six, with a solid dot at the crossing. On the right, the curve of the tangent of x minus three shoots up to the left of pi over two and comes back from below on the right, with a dashed vertical line where it is undefined.
landmark — A point where the expression is zero, or a point where it is undefined. Both cut the interval into regions; only a zero can ever belong to the answer.
Every function with a tangent, cotangent, secant or cosecant in it has undefined points, and they must go on the diagram. Leaving one off merges two regions of opposite sign into one, and the test value then reports the wrong answer for half of it.
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.7 Trigonometric Equations and Inequalities §10.7, p. 867 — the disclaimer on the sign diagram in Example 10.7.3, part 3
Look at the two panels side by side. On the left the curve strolls through the axis; on the right it vanishes upward and reappears from below. In both cases the sign on one side differs from the sign on the other, so in both cases the line must be cut.
But the two cuts behave completely differently at the end of the problem. A zero is a genuine value of the function and can satisfy an inequality that allows equality. An undefined point is not a value of anything, so it cannot satisfy any inequality, ever.
The cost of forgetting an undefined point is worth being precise about. Two regions of opposite sign get merged into one, your single test value reports the sign of only half of it, and you confidently hand in an answer that is exactly half right, with no arithmetic error anywhere in your work.
Worked example
\[ \tan(x) \ge 3, \qquad 0 \le x < 2\pi \]
Move everything over and name f
Why: As always.
\[ f(x) = \tan(x) - 3 \]
Find where f is undefined
Why: The tangent fails at the odd multiples of a quarter turn.
\[ x = \tfrac{\pi}{2} + \pi k \;\Longrightarrow\; x = \tfrac{\pi}{2}, \; \tfrac{3\pi}{2} \]
Solve f(x) = 0
Why: The value 3 belongs to no common angle, so the arctangent is the only name for the answer.
\[ \tan(x) = 3 \;\Longrightarrow\; x = \arctan(3) + \pi k \]
Locate the two zeros in the window
Why: Since 3 is positive, the arctangent of 3 sits strictly between zero and a quarter turn, which fixes its place on the diagram.
\[ 0 < \arctan(3) \approx 1.249 < \tfrac{\pi}{2}, \qquad \arctan(3) + \pi \approx 4.391 \]
Test the first three regions
Why: The middle one is the hard case: the only numbers between the arctangent of 3 and a quarter turn are arctangents of numbers bigger than 3.
\[ f(0) = -3, \qquad f\bigl(\arctan(117)\bigr) = 117 - 3 = 114, \qquad f(\pi) = -3 \]
Test the last two regions
Why: Adding a half turn to the awkward test value keeps it awkward but legal, and the tangent has period pi so the value repeats.
\[ f\bigl(\arctan(117) + \pi\bigr) = 114, \qquad f\left(\tfrac{7\pi}{4}\right) = -1 - 3 = -4 \]
Figure (svg): A sign diagram from 0 to two pi for the tangent of x minus three. Two zeros are marked, at the arctangent of three and at the arctangent of three plus pi, and two undefined points at pi over two and three pi over two. The five regions carry the signs minus, plus, minus, plus, minus, and the two positive regions are shaded.
Collect the two positive regions
Why: The zeros are included, because the inequality is not strict; the asymptotes are excluded, because nothing exists there.
\[ \left[\arctan(3), \tfrac{\pi}{2}\right) \cup \left[\arctan(3)+\pi, \tfrac{3\pi}{2}\right) \]
Check the two kinds of endpoint
Why: At a zero the tangent equals 3 exactly, which satisfies an at-least-three condition; at an asymptote the tangent has no value at all to compare.
\[ \tan\bigl(\arctan(3)\bigr) = 3 \ge 3, \qquad \tan\left(\tfrac{\pi}{2}\right) \; \text{undefined} \]
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.7 Trigonometric Equations and Inequalities §10.7, pp. 867-868 — Example 10.7.3, part 3
This is the hardest problem in the first half of the lesson, and it earns that place by being awkward in two separate ways. It has undefined points, and its zeros are not common angles.
The second difficulty is handled the way you handled it when solving equations: name the angle with the arctangent and carry the name. What is new is that you must also know where that name sits on the line, and the argument given is the one to copy. Three is positive, the arctangent of a positive number lies between zero and a quarter turn, and that is enough to place the landmark correctly.
The two answer intervals are narrow, and the picture on the next slide shows why: the tangent only manages to exceed three in the last stretch before each asymptote. Notice that the two intervals are exactly a half turn apart, which is the period of the tangent, and that is a free check on your work.
Picture it
Figure (svg): The tangent curve across one revolution with its two vertical asymptotes and a horizontal line at 3. The two stretches where the curve is at or above the line, each just to the left of an asymptote, are shaded.
Each answer interval starts where the curve meets the level and runs up to the asymptote on its right. The picture makes the two different brackets obvious: the curve reaches the level, so that end is closed, and it never reaches the asymptote, so that end is open.
The graph is the fastest way to confirm a sign diagram, and it is what the book means by verifying your answer graphically. You are looking for the stretches where the tangent curve is on or above the flat line.
The two brackets are visible rather than remembered here. On the left of each shaded strip the curve actually touches the line, so that value is a solution and the bracket closes. On the right the curve races off without ever arriving anywhere, so there is nothing at that end to include.
One more thing to notice: the stretch to the right of the second asymptote, from three halves of pi onward, contributes nothing at all. There the tangent climbs only from far below zero up to zero, and never gets near three.
Concept
Figure (svg): Two panels. On the left, the arctangent curve rising from 0 and flattening under a dashed line at pi over two, with marked points at 3 and at 117. On the right, a zoomed number line from 1.20 to 1.60 showing the arctangent of three at 1.249, the arctangent of 117 at 1.5622 and pi over two at 1.5708, with the gap between the first and the last shaded.
The middle region of the last example runs from the arctangent of three to a quarter turn. No common angle lies inside it, and a decimal would leave you evaluating a tangent by calculator.
\[ \arctan \text{ is increasing} \;\Longrightarrow\; t > 3 \;\Longrightarrow\; \arctan(t) > \arctan(3) \]
\[ \arctan(t) < \tfrac{\pi}{2} \text{ for every } t \;\Longrightarrow\; \arctan(3) < \arctan(117) < \tfrac{\pi}{2} \]
And the evaluation is free, because the tangent undoes the arctangent on the spot. Any number bigger than three would have done; the book picks 117 to make the point that it does not matter which.
\[ f\bigl(\arctan(117)\bigr) = \tan\bigl(\arctan(117)\bigr) - 3 = 114 \]
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.7 Trigonometric Equations and Inequalities §10.7, p. 867 — the choice of test value and its footnote
This is the cleverest small idea in the section, and it generalises well beyond this problem. You need a number in a gap that contains no angle you recognise. Rather than hunting for a decimal, you build one out of the function that created the gap.
Two facts do the work, and both are in the picture. The arctangent is increasing, so feeding it a bigger number gives a bigger output, which puts the arctangent of any number above three to the right of the arctangent of three. And the arctangent never reaches a quarter turn, so the output stays to the left of the asymptote. The new number is trapped exactly where you need it.
Then the evaluation is free, which is the real payoff. The tangent of the arctangent of one hundred and seventeen is one hundred and seventeen, no calculator required. The book's choice of one hundred and seventeen over, say, four is deliberate mischief: any number bigger than three would do, and the absurd one makes that obvious.
Error analysis
A student hands in the answer below to the last worked example. The regions are right. The brackets are not.
Annotate
On: \( \left[\arctan(3), \tfrac{\pi}{2}\right] \cup \left[\arctan(3)+\pi, \tfrac{3\pi}{2}\right] \)
Try to spot both wrong brackets before revealing the annotations. The regions here are correct, which is what makes the error so easy to miss when you are checking your own work.
The key realisation is that the inequality symbol and the bracket at an undefined point have nothing to do with each other. At a zero, the symbol decides. At an asymptote, the symbol has no vote, because there is no value there to compare with anything.
The habit suggested in the last annotation is worth adopting from today. Draw zeros and undefined points differently on your diagram, and the closing bracket becomes a matter of looking rather than remembering.
Faded example
Exercise 70 of the section, and the same shape as the worked example, except that the zeros are common angles this time. The picture gives them away.
Figure (svg): The tangent curve from 0 to two pi with its two asymptotes and a horizontal line at the square root of three. The curve meets the line at pi over three and at four pi over three, and the two stretches from each crossing up to the next asymptote are shaded.
Fill in the blanks
\tan(x) \ge \sqrt\tfrac{\pi}{3} \;\text\tfrac{4\pi}{3}\; [0, 2\pi] \;\Longrightarrow\; \left[___, \tfrac______\right) \cup \left[___, \tfrac______\right)
Why: The tangent equals root three at a third of pi, and the tangent has period pi, so the other zero in the window is a third of pi plus pi, which is four thirds of pi. Each answer interval then runs from its zero up to the next asymptote, closed at the zero and open at the asymptote. Notice that nothing past three halves of pi qualifies: on that last stretch the tangent climbs only from far below zero up to zero.
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.7 Trigonometric Equations and Inequalities §10.7, p. 875 — Exercise 70
Fill both blanks from the picture before checking. This is the same problem as the worked example with the awkwardness removed, so you should be able to read the zeros straight off the graph.
The tangent equals root three at a third of pi, an angle you know. The second zero is then a half turn further along, because the tangent repeats every half turn, and that gives four thirds of pi. Both shaded strips run from a zero up to the asymptote on the right, closed then open.
Check the last stretch of the window as well, from three halves of pi to two pi. Nothing there qualifies, and confirming that for yourself is what stops you from assuming the answer must have the same number of pieces as there are asymptotes.
Comparison
Fill the blanks. The whole table is two rules, and the second one has no exceptions.
Comparison matrix
| the landmark | the inequality | is that endpoint in the answer? |
|---|---|---|
| a zero of f | strict: f(x) > 0 | no |
| a zero of f | not strict: f(x) ≥ 0 | yes |
| a point where f is undefined | strict: f(x) > 0 | no |
| a point where f is undefined | not strict: f(x) ≥ 0 | no |
| the end of a window written [a, b] | the sign there is the one you want | yes |
Fill in each blank before checking, and notice as you go that you are only using two rules. The first rule has a case for each inequality symbol; the second rule has no cases at all.
Rows three and four are the whole point of the table. They are identical in their answer even though the inequality symbol differs, which is the fact people find hardest to accept under exam pressure.
The last row is a reminder about the window itself. Whether the window was stated with a square bracket or a round one is a separate decision made by whoever wrote the problem, and it survives into your answer. The exercise set for this section uses closed windows; the worked examples in the text use half-open ones.
Worked example
Both kinds of landmark appear here, and one whole branch of the curve joins the answer at once.
\[ \sec(x) \le \sqrt{2}, \qquad 0 \le x \le 2\pi \]
Move everything over and name f
Why: As always.
\[ f(x) = \sec(x) - \sqrt{2} \]
Find where f is undefined
Why: The secant fails wherever the cosine is zero.
\[ x = \tfrac{\pi}{2}, \; \tfrac{3\pi}{2} \]
Solve f(x) = 0
Why: Turn the secant back into a cosine before solving.
\[ \sec(x) = \sqrt{2} \iff \cos(x) = \tfrac{\sqrt{2}}{2} \;\Longrightarrow\; x = \tfrac{\pi}{4}, \; \tfrac{7\pi}{4} \]
Test the first three regions
Why: Zero, a sixth of pi past it, and a half turn.
\[ f(0) = 1-\sqrt{2} \approx -0.414, \; f\left(\tfrac{\pi}{3}\right) = 2-\sqrt{2} \approx 0.586, \; f(\pi) = -1-\sqrt{2} \]
Test the last two regions
Why: Five thirds of pi, and the right end of the window.
\[ f\left(\tfrac{5\pi}{3}\right) = 2-\sqrt{2} \approx 0.586, \qquad f(2\pi) = 1-\sqrt{2} \approx -0.414 \]
Figure (svg): The secant curve from 0 to two pi, in three branches separated by asymptotes at pi over two and three pi over two, with a horizontal line at the square root of two. The stretches at or below the line, from 0 to pi over four, the whole middle branch, and from seven pi over four to two pi, are shaded.
Collect the regions where f is negative or zero
Why: Both zeros join the answer; both asymptotes stay out, which is why the middle interval is open at both ends.
\[ \left[0, \tfrac{\pi}{4}\right] \cup \left(\tfrac{\pi}{2}, \tfrac{3\pi}{2}\right) \cup \left[\tfrac{7\pi}{4}, 2\pi\right] \]
Check the middle interval in one line
Why: There the cosine is negative, so the secant is negative, and any negative number is below root two.
\[ \cos(\pi) = -1 \;\Longrightarrow\; \sec(\pi) = -1 \le \sqrt{2} \]
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.7 Trigonometric Equations and Inequalities §10.7, p. 875 — Exercise 79
This one repays a careful look before you start calculating, because the answer has three pieces rather than two, and the middle piece is the whole branch between the asymptotes.
The reason the middle branch joins the answer in one go is that the cosine is negative throughout it, so the secant is negative throughout it, and every negative number is less than root two. No test value in that region could have said anything else. Recognising a whole branch like this is often faster than testing it.
Look closely at the brackets. The two outer pieces close at their zeros because equality is allowed, and open at nothing, because their other ends are the ends of the window. The middle piece is open at both ends because both are asymptotes. Three pieces, five brackets, and every one decided by a different consideration.
Tweak it
Parameter explorer
The curve is twice the sine of x with a constant c subtracted, drawn across one revolution. The solution set of the inequality asking for f to be at least zero is exactly where the curve sits on or above the axis. Slide c: at which value does that set shrink to a single point, and below which value is it the whole revolution?
\[ f(x) = 2\sin x - c, \quad c = {c} \]
Start with the level at one, which is the first worked example turned upside down, and move it slowly upward. Watch the two crossings slide toward each other as the level rises.
At the moment the level reaches two, the two crossings collide at a quarter turn and the solution set is a single point: the curve touches the axis and immediately turns back. Push the level a fraction higher and there is nothing left. Every inequality has a level at which its answer collapses, and it is always where the function reaches its extreme value.
Now go the other way. As the level drops toward minus two the excluded middle shrinks to nothing, and below minus two every angle in the revolution qualifies. Between the two extremes the answer is always exactly one interval, which is worth knowing: a simple sine inequality on one revolution can have one interval or two, and which of those you get depends on where the level cuts.
Section
Part 3
Concept
Finding a domain means listing the inputs a formula refuses. Each feature of the formula refuses for a different reason, and each reason is one of the two problems you can now solve.
| what the formula contains | what you must solve | what you get back |
|---|---|---|
| a denominator | an equation: set it to zero | points to throw away |
| a tangent, cotangent, secant or cosecant | an equation: set its own denominator to zero | points to throw away |
| an even root | an inequality: the inside is at least zero | intervals to keep |
| a logarithm | an inequality: the inside is strictly positive | intervals to keep |
A trigonometric formula almost always refuses infinitely many inputs, spaced by a period, so the answer needs a notation that can say and so on in both directions.
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.7 Trigonometric Equations and Inequalities §10.7, p. 868 — the sentence introducing Example 10.7.4
Domains have been in the background of every course you have taken, usually as a rule to recite. Here they finally become a calculation, and the table says which calculation.
Sort the four rows into two groups as you read. Denominators and the reciprocal functions hand you an equation, and its solutions are points you throw away. Roots and logarithms hand you an inequality, and its solution is a set of intervals you keep. Knowing which kind of question you face before you start is most of the work.
What makes the trigonometric versions different is the repetition. A rational function might lose two points; a formula built from a cosine loses two infinite families of points, spaced by a period, and you need a notation that can write down infinitely many intervals at once. That notation is the subject of the next few slides.
Worked example
\[ f(x) = \csc\left(2x + \tfrac{\pi}{3}\right) \]
Rewrite the cosecant as a reciprocal
Why: The cosecant is not a new obstacle; it is a denominator in disguise.
\[ f(x) = \frac{1}{\sin\left(2x + \tfrac{\pi}{3}\right)} \]
Set the denominator to zero
Why: The sine is defined everywhere, so its zeros are the only trouble.
\[ \sin\left(2x + \tfrac{\pi}{3}\right) = 0 \]
Solve for the argument
Why: The sine vanishes at every integer multiple of pi.
\[ 2x + \tfrac{\pi}{3} = \pi k \]
Unwind to x
Why: Subtract the shift, then halve both the angle and the step.
\[ x = -\tfrac{\pi}{6} + \tfrac{\pi}{2}k \]
Write the excluded points over one denominator
Why: Keeping every denominator at 6 is what makes the pattern legible.
\[ x_k = -\tfrac{\pi}{6} + \tfrac{\pi}{2}k = \frac{(3k-1)\pi}{6} \]
Figure (svg): A number line carrying six hollow circles at minus seven pi over six, minus four pi over six, minus pi over six, two pi over six, five pi over six and eight pi over six, with the open intervals between them drawn as solid bars.
Name the interval between consecutive exclusions
Why: The domain is everything else, which is the open gap between one excluded point and the next.
\[ (x_k, x_{k+1}) = \left(\frac{(3k-1)\pi}{6}, \frac{(3k+2)\pi}{6}\right) \]
Assemble the answer in extended interval notation
Why: One formula, one index, every gap.
\[ \bigcup_{k=-\infty}^{\infty} \left(\frac{(3k-1)\pi}{6}, \frac{(3k+2)\pi}{6}\right) \]
Check two values of k against the number line
Why: At k equal to zero the formula gives the gap from minus a sixth of pi to two sixths of pi, and at k equal to one the next gap along; both match the drawing.
\[ k=0: \left(-\tfrac{\pi}{6}, \tfrac{2\pi}{6}\right), \qquad k=1: \left(\tfrac{2\pi}{6}, \tfrac{5\pi}{6}\right) \]
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.7 Trigonometric Equations and Inequalities §10.7, p. 868 — Example 10.7.4, part 1
The first move dissolves the apparent difficulty. A cosecant is not a new kind of obstacle; it is a fraction with a sine on the bottom, and you already know exactly what to do with a bottom.
The unwinding is the same skill you built in the first half of the section. Solve for the whole argument first, then subtract the shift and divide by the coefficient, remembering that the step gets divided too. Halving the step from pi to half of pi is what doubles the number of points you lose.
Keeping every fraction over the denominator six is a small trick with a large payoff. Written that way the excluded points march along in steps of three sixths of pi, the pattern is visible at a glance, and the formula for the general excluded point almost writes itself.
Notation
Annotate
On: \( \bigcup_{k=-\infty}^{\infty} \left(\frac{(3k-1)\pi}{6}, \frac{(3k+2)\pi}{6}\right) \)
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.7 Trigonometric Equations and Inequalities §10.7, p. 868 — extended interval notation, introduced on p. 756
This notation looks more forbidding than it is. Read the big union sign as the words and so on, in both directions, and everything else is an ordinary interval with a k in it.
The practical skill is checking such a formula, and the method is always the same: substitute k equal to zero, then one, then minus one, and compare what comes out with the points you drew on your number line. If they match, the formula is right; if the first one is off, you have usually made a sign slip in the general term.
Notice how the right-hand endpoint was obtained. It is not a new calculation: it is the left-hand endpoint formula with k replaced by k plus one, then simplified. Any domain built from one evenly spaced family will have this shape.
Worked example
\[ f(x) = \frac{\sin(x)}{2\cos(x) - 1} \]
Set the denominator to zero
Why: The numerator is defined everywhere and a zero numerator is harmless, so only the bottom matters.
\[ 2\cos(x) - 1 = 0 \iff \cos(x) = \tfrac12 \]
Solve: this time there are two families
Why: The cosine takes the value one half twice per revolution.
\[ x = \tfrac{\pi}{3} + 2\pi k \quad\text{or}\quad x = \tfrac{5\pi}{3} + 2\pi k \]
Give each family a name and a single fraction
Why: Two letters keep the bookkeeping straight.
\[ a_k = \tfrac{\pi}{3} + 2\pi k = \frac{(6k+1)\pi}{3}, \qquad b_k = \tfrac{5\pi}{3} + 2\pi k = \frac{(6k+5)\pi}{3} \]
Figure (svg): A number line carrying eight hollow circles at minus eleven pi over three, minus seven pi over three, minus five pi over three, minus pi over three, pi over three, five pi over three, seven pi over three and eleven pi over three. The gaps alternate between wide and narrow, and the two gaps making up one period are drawn in different colours.
List a few in increasing order
Why: The two families interleave, and the gaps alternate between long and short.
\[ \dots < a_{-1} = -\tfrac{5\pi}{3} < b_{-1} = -\tfrac{\pi}{3} < a_0 = \tfrac{\pi}{3} < b_0 = \tfrac{5\pi}{3} < a_1 = \tfrac{7\pi}{3} < \dots \]
Group the gaps in pairs, one pair per period
Why: Each k now contributes a full period of length two pi, cut into two intervals by the b in the middle.
\[ (a_k, b_k) \cup (b_k, a_{k+1}) \]
Write it out
Why: Substituting the two formulas gives the domain in extended interval notation.
\[ \bigcup_{k=-\infty}^{\infty} \left(\frac{(6k+1)\pi}{3}, \frac{(6k+5)\pi}{3}\right) \cup \left(\frac{(6k+5)\pi}{3}, \frac{(6k+7)\pi}{3}\right) \]
Check with the other route
Why: Work on the single period from a third of pi to seven thirds of pi, where the only excluded interior point is five thirds of pi, then add every multiple of two pi; the two answers agree once the fractions are put over 3.
\[ \left(\tfrac{\pi}{3}+2\pi k, \tfrac{5\pi}{3}+2\pi k\right) \cup \left(\tfrac{5\pi}{3}+2\pi k, \tfrac{7\pi}{3}+2\pi k\right) \]
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.7 Trigonometric Equations and Inequalities §10.7, p. 869 — Example 10.7.4, part 2
Everything changes when the excluded points come from two families rather than one, and the first thing to change is your ability to write them as a single arithmetic progression. Look at the number line: the gaps alternate between long and short, so no single step size describes them.
The book's first route is bookkeeping done carefully. Name the two families, list enough members of each to see the interleaving, and then notice that grouping them in pairs makes each value of k cover exactly one full period. That is why the final formula has two intervals joined by a union inside the big union.
The second route, in the check, is the one most people find easier, and it generalises better. The function repeats every two pi, so find the domain on any single interval of length two pi and then slide that answer left and right. Choosing an interval whose ends are themselves excluded points is what makes the description short.
Intuition
The second solution to that example is worth keeping as a general method, because it works whenever the function repeats.
\[ f(x + p) = f(x) \text{ for all } x \;\Longrightarrow\; \text{the domain repeats with period } p \]
So describe the domain on one interval of length p, then add every integer multiple of p. The hard part is choosing the interval: pick one whose ends are themselves excluded points, and the description inside it becomes short.
\[ \text{domain} = \bigcup_{k=-\infty}^{\infty} \bigl( \text{the answer on one period} + pk \bigr) \]
The first route is better when a single evenly spaced family is excluded. The second is better when several families interleave, as they did a moment ago.
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.7 Trigonometric Equations and Inequalities §10.7, p. 869 — the second approach and its footnote
This idea is worth separating from the example that introduced it, because you will use it for the rest of the lesson and throughout the next chapter. If a function repeats, then so does everything about it, including its domain and the solution set of any inequality in it.
The one place care is needed is choosing which period. The period of the function inside the formula is what matters, not two pi out of habit. A cotangent repeats every half turn, so a cotangent problem should be solved on an interval of length pi, and translating by two pi would leave you with half the answer.
There is also a subtlety hidden in the footnote of the textbook here. A formula built from functions of period two pi certainly repeats every two pi, but its own period might be shorter. That never causes a wrong answer, only a longer description than necessary.
Matching
Match the pairs
Why: The tangent of half x fails when half x is a quarter turn plus a multiple of pi, so x is pi plus a multiple of two pi: the odd multiples of pi. The cosecant of two x fails when the sine of two x is zero, so two x is a multiple of pi and x is a multiple of half pi. The third loses the single family where the sine equals one. Only the fourth is an inequality, and only it gives back intervals rather than a list of points.
Work out each one before matching, and notice as you go which of the four is a different kind of question from the other three.
The first two are unwinding exercises: find where the inner function misbehaves, then solve for x, dividing or multiplying the step as the coefficient demands. Halving the argument doubles the step, so the tangent of half x loses points only every two pi; doubling the argument halves the step, so the cosecant of two x loses them every half of pi.
The fourth is the odd one out and the reason it is here. A square root asks an inequality, so it does not throw away isolated points at all; it keeps whole intervals and throws away whole intervals. If you matched it by elimination, go back and check that you can say why its answer has a different shape.
Trap
From the cosine equal to one half:
\[ x \ne \tfrac{\pi}{3} + \pi k \]
Wrong. That family is evenly spaced.
\[ \tfrac{\pi}{3}, \; \tfrac{4\pi}{3}, \; \tfrac{7\pi}{3}, \dots \]
The true exclusions alternate between a long gap and a short one, so no single arithmetic progression can list them. Either keep two formulas, or pair them one period at a time.
\[ \tfrac{\pi}{3}, \; \tfrac{5\pi}{3}, \; \tfrac{7\pi}{3}, \; \tfrac{11\pi}{3}, \dots \]
A quick test: write the first four terms of your formula and compare them with the first four you found by hand.
This mistake comes from a correct memory applied in the wrong place. When the sine or cosine of x equals plus or minus one, or when a tangent is involved, one family with a step of pi really does describe everything. For a general value it does not.
The reliable defence takes about ten seconds. Write the first three or four members of your proposed formula and compare them with the angles you actually found. Here the proposed formula produces four thirds of pi, and a moment's thought about the unit circle shows that the cosine there is minus one half, not plus one half.
When two families genuinely interleave with unequal gaps, you have two honest options and neither is a compromise: keep two separate formulas, or pair them one period at a time as the worked example did.
Estimation
\[ f(x) = \csc(3x) \]
Predict first
How many points does this formula refuse inside one revolution, from zero up to but not including two pi?
Correct: Six
Why: The cosecant of three x is one over the sine of three x, which is zero when three x is any multiple of pi, so x is any multiple of a third of pi. Inside one revolution those are zero, a third of pi, two thirds, pi, four thirds and five thirds: six of them. The coefficient three compresses the usual two exclusions per revolution into six. Infinitely many is true of the whole line, but not of one revolution.
Answer before revealing, and answer by counting rather than by listing if you can.
The cosecant fails wherever its sine is zero. The sine of three x is zero when three x is a whole multiple of pi, so x is a whole multiple of a third of pi, and a third of pi fits into one revolution six times. The coefficient three is doing exactly what it did to solution families in the first half of this section: compressing the pattern, so three times as many of them fit in a revolution.
If you answered infinitely many, you were right about the whole number line and wrong about the question. Getting into the habit of noticing whether a question is about one revolution or about everything is worth more here than any single calculation.
Section
Part 4
Concept
An even root needs its inside to be at least zero. That is an inequality, so the sign diagram returns; the new difficulty is that no interval was handed to you.
\[ \sqrt{g(x)} \text{ is defined} \iff g(x) \ge 0 \]
The fix is the periodicity of everything in sight. Solve the inequality on one full period, then add every integer multiple of that period. Choose the period that belongs to the function inside the root, not automatically two pi.
\[ \text{tangent, cotangent} \;\longrightarrow\; \text{period } \pi \]
\[ \text{sine, cosine, secant, cosecant} \;\longrightarrow\; \text{period } 2\pi \]
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.7 Trigonometric Equations and Inequalities §10.7, p. 870 — the strategy stated in Example 10.7.4, part 3
Here is where the two halves of the lesson meet. A square root needs its inside to be at least zero, which is an inequality, and you now know how to solve those with a sign diagram.
The new difficulty is a real one. Every inequality you have solved so far came with a window attached, which is what kept the answer finite. A domain question has no window, so a literal sign diagram would need infinitely many landmarks and infinitely many test values.
Periodicity rescues it, and the rescue is exact rather than approximate. Solve the inequality on one period, and the answer on every other period is a translated copy, because the function itself is a translated copy. Choose the period belonging to the function inside the root, not two pi by reflex.
Worked example
\[ f(x) = \sqrt{1 - \cot(x)} \]
Record the inner function's own domain first
Why: The cotangent is undefined at every integer multiple of pi, and no root can rescue that.
\[ x \ne \pi k \]
Write the condition the radical imposes
Why: The inside must be at least zero.
\[ 1 - \cot(x) \ge 0 \]
Name the inside g and choose one period
Why: The cotangent has period pi, and the open interval from zero to pi carries one full cycle of it.
\[ g(x) = 1 - \cot(x), \qquad 0 < x < \pi \]
Solve g(x) = 0
Why: One member of the family lands inside this period.
\[ \cot(x) = 1 \;\Longrightarrow\; x = \tfrac{\pi}{4} + \pi k \;\Longrightarrow\; x = \tfrac{\pi}{4} \]
Test the two regions
Why: A sixth of pi and a half of pi are both easy cotangents.
\[ g\left(\tfrac{\pi}{6}\right) = 1 - \sqrt{3} \approx -0.732, \qquad g\left(\tfrac{\pi}{2}\right) = 1 - 0 = 1 \]
Figure (svg): A sign diagram on the open interval from 0 to pi for one minus the cotangent of x. Both ends are marked undefined, a zero sits at pi over four, the left region carries a minus and the right region a plus, and the right region is shaded.
Read the answer on this one period
Why: The inequality is not strict, so the zero is included; the right end is excluded because the cotangent dies there.
\[ \left[\tfrac{\pi}{4}, \pi\right) \]
Translate by the period
Why: Add an integer multiple of pi to both ends, then put the left end over one denominator.
\[ \left[\tfrac{\pi}{4} + \pi k, \; \pi + \pi k\right) = \left[\frac{(4k+1)\pi}{4}, (k+1)\pi\right) \]
Check the answer inside a translated copy
Why: In the next period along, five quarters of pi is the left endpoint and the root is zero there, while nine eighths of pi sits just before it and is correctly excluded.
\[ \cot\left(\tfrac{5\pi}{4}\right) = 1 \Rightarrow f = 0; \qquad \cot\left(\tfrac{9\pi}{8}\right) \approx 2.414 \Rightarrow 1 - \cot < 0 \]
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.7 Trigonometric Equations and Inequalities §10.7, p. 870 — Example 10.7.4, part 3
Notice what comes first, before any inequality is written down. The cotangent has its own domain, and no amount of work on the square root can rescue an angle where the cotangent does not exist. Two lists, always: what the outer operation forbids and what the inner function forbids.
The interval from zero to pi is chosen because it carries exactly one cycle of the cotangent, running from far above the axis down to far below it. Both ends are undefined points, which is ideal: the landmarks at the ends of your working interval come for free.
The final answer is closed at the left and open at the right, and the two brackets have different reasons. The left end is a genuine zero of the inside, so the root of zero is zero and the function is defined. The right end is where the cotangent dies, so the formula has no value at all. A single sloppy bracket here turns a correct answer into a wrong one.
Picture it
Figure (svg): The curve one minus the cotangent of x drawn over three periods, from minus pi to two pi, breaking at every multiple of pi. In each period the stretch from a quarter of pi past the left asymptote to the next asymptote is shaded, so three identical shaded strips appear.
Three periods of the same picture. The shaded strip in each one begins a quarter of pi past the left asymptote and runs to the right asymptote, and the pattern repeats in both directions without end. Solving the inequality once and translating is not a shortcut; it is the only way to write down infinitely many intervals.
Three periods of the same picture, drawn so you can see that the translation really is a translation. The curve between any two consecutive asymptotes is an exact copy of the curve between the previous two, shifted by half a turn.
Trace one shaded strip with your finger and then find the corresponding strip in the period on either side. Each begins a quarter of pi past its own left asymptote and runs to its own right asymptote. That is the geometric content of the phrase add any multiple of the period.
This is also the answer to a reasonable objection: is translating a shortcut, or is it the real thing? It is the real thing. The alternative would be to write down infinitely many intervals one at a time, which is not a method at all.
Error analysis
The sign work was right and the translation was right. One bracket carries a mistake that the sign diagram alone cannot catch.
Annotate
On: \( \bigcup_{k=-\infty}^{\infty} \left[\frac{(4k+1)\pi}{4}, \; (k+1)\pi\right] \)
Read the answer and try to find the flaw before the annotations appear. Every region is correct, the translation is correct, and the formula for the left endpoint is correct. Only one character is wrong.
The closing square bracket claims that the right end of each interval belongs to the domain. But that point is a multiple of pi, where the cotangent does not exist, so the whole formula does not exist there either, whatever the inequality on the inside happens to say.
The way this error is produced is by treating the problem as one question instead of two. Solving the inequality for the inside of the root is necessary but not sufficient; the inner function's own domain is a separate condition, and the domain is the overlap of the two.
Worked example
A function of the same shape, where both restrictions bite in the same period.
\[ f(x) = \sqrt{\tan(x)} \]
Record where the tangent itself fails
Why: The tangent is undefined at the odd multiples of a quarter turn.
\[ x \ne \tfrac{\pi}{2} + \pi k \]
Write the radical's condition
Why: The inside must be at least zero.
\[ \tan(x) \ge 0 \]
Choose one period
Why: The tangent has period pi; take the half-open interval from zero to pi, with the asymptote removed.
\[ 0 \le x < \pi, \quad x \ne \tfrac{\pi}{2} \]
Find the landmarks on that period
Why: One zero at the left end, one undefined point in the middle.
\[ \tan(0) = 0, \qquad \tan\left(\tfrac{\pi}{2}\right) \; \text{undefined} \]
Test the two regions
Why: A quarter of pi and three quarters of pi.
\[ \tan\left(\tfrac{\pi}{4}\right) = 1 > 0, \qquad \tan\left(\tfrac{3\pi}{4}\right) = -1 < 0 \]
Figure (svg): The tangent curve over three periods from minus pi to two pi. In each period the half-open stretch from the zero up to the next asymptote is shaded, with a solid dot at each zero and a dashed line at each asymptote.
Read the answer and translate
Why: Closed at the zero, open at the asymptote, then add every multiple of pi.
\[ \left[0, \tfrac{\pi}{2}\right) \;\longrightarrow\; \bigcup_{k=-\infty}^{\infty}\left[k\pi, \; \frac{(2k+1)\pi}{2}\right) \]
Check both endpoint types numerically
Why: At a multiple of pi the tangent is zero and its root is zero, so that end is genuinely in; just short of the asymptote the values are large but still perfectly defined, so nothing is lost by stopping there.
\[ \sqrt{\tan(0)} = 0, \qquad \tan(1.5) \approx 14.10, \quad \sqrt{14.10} \approx 3.755 \]
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.7 Trigonometric Equations and Inequalities §10.7, p. 870 — the method of Example 10.7.4, part 3
This is the same method on a function where both restrictions bite inside a single period, which makes it a good test of whether the procedure has really stuck.
On the period from zero to pi, the tangent is zero at the left end and undefined in the middle. Those are your two landmarks, one of each kind, and they immediately split the period into the half where the tangent is positive and the half where it is negative. Only the first half can sit under a square root.
So the answer on one period is closed at the left and open in the middle, and the translated family inherits both brackets. The numerical check at the end is there to make the open end feel right: just short of the asymptote the tangent is large but perfectly ordinary, and its root is a real number. Nothing goes wrong until you arrive at the asymptote itself.
Prediction
\[ f(x) = \sqrt{\cot(x)} \]
Predict first
Working on the period from zero to pi, which describes the part of the domain that lies there?
Correct: From zero to pi over two, open at the left and closed at the right
Why: The cotangent is positive on the first half of the period and negative on the second, so only the first half can sit under a square root. At the left end the cotangent is undefined, so that end is open; at a quarter turn the cotangent is exactly zero, its root is zero, and the function is defined, so that end is closed. This is the mirror image of the one minus cotangent example, where the open end was on the right.
Predict before revealing. This is the mirror image of the worked example, and the mirror is exactly what the question is testing.
The cotangent is positive on the first half of the period from zero to pi and negative on the second half, so the root can only exist on the first half. At the left end the cotangent is undefined, so that end is open; at a quarter turn the cotangent is exactly zero, so the root is zero and the function is happily defined, and that end is closed.
If you chose the second half, you have probably pictured the tangent instead of the cotangent. They are reflections of each other on this interval, which is precisely why it is worth sketching the inner function rather than recalling its sign.
Section
Part 5
Concept
Figure (svg): Two panels. On the left, the sine curve over two revolutions crossed by a horizontal line at root three over two in eight places. On the right, the curve of the arcsine of two x over its short domain, crossed by a horizontal line at pi over three in exactly one place.
Everything in this chapter so far produced families, because the circular functions repeat. The inverse functions do not repeat: each one hands back a single value from a restricted range, so an equation in one of them has one solution or none at all.
| function | its range | what that range rules out |
|---|---|---|
| arcsine | from minus a quarter turn to a quarter turn, inclusive | anything outside, such as two thirds of pi |
| arccosine | from zero to pi, inclusive | anything negative, and anything above pi |
| arctangent | strictly between minus a quarter turn and a quarter turn | the two endpoints themselves |
| arccotangent | strictly between zero and pi | zero and pi themselves |
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.7 Trigonometric Equations and Inequalities §10.7, p. 870 — the sentence introducing Example 10.7.5
Everything in this chapter has trained you to expect families of solutions, and this slide asks you to unlearn that for one specific setting. The picture shows why. On the left, a horizontal line cuts the sine curve again and again, for ever. On the right, the same kind of line cuts the arcsine curve exactly once.
The structural reason is that an inverse function is a function with a restricted range. It returns one value, chosen from a deliberately narrowed set, and it never returns the same value twice. So an equation asking it to produce a particular number has at most one answer.
The table is the fact you will use constantly. Learn the four ranges, and in particular learn which endpoints are included: the arcsine and arccosine reach their endpoints, the arctangent and arccotangent do not. The very next slide shows that this distinction decides whether a solution exists.
Notation
Annotate
On: \( \sin\bigl(\arcsin(u)\bigr) = u \qquad \text{for } -1 \le u \le 1 \)
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.7 Trigonometric Equations and Inequalities §10.7, pp. 870-871 — Theorems 10.26 to 10.29 as used in Example 10.7.5
The identity on this slide is the engine of every solve in the rest of the lesson, and its one-sidedness is what the annotations are about.
Applying the sine to the arcsine of something always gives that something back, on the whole domain. Applying the arcsine to the sine of an angle gives the angle back only when the angle was already in the arcsine's narrow range. Since the number on the right-hand side of your equation is going to be fed backwards through the inverse, that is the direction you must check.
Make the range check a reflex, written on the page before any algebra. It costs one line, it catches every impossible equation before you waste effort on it, and skipping it produces answers that look completely convincing and are completely wrong.
Worked example
\[ \arcsin(2x) = \tfrac{\pi}{3} \]
Check the right side against the range
Why: The arcsine only ever returns a value between minus a quarter turn and a quarter turn, and a third of pi is inside, so a solution can exist.
\[ -\tfrac{\pi}{2} \le \tfrac{\pi}{3} \le \tfrac{\pi}{2} \]
Apply the sine to both sides
Why: The two sides are equal numbers, so their sines are equal.
\[ \sin\bigl(\arcsin(2x)\bigr) = \sin\left(\tfrac{\pi}{3}\right) \]
Collapse the left side by the inverse property
Why: The sine undoes the arcsine on the spot.
\[ 2x = \tfrac{\sqrt{3}}{2} \]
Solve the linear equation
Why: Divide both sides by 2.
\[ x = \tfrac{\sqrt{3}}{4} \]
Check by substituting back
Why: Twice root three over four is root three over two, whose arcsine is a third of pi. Numerically the answer is about 0.4330; the printed text says 0.4430, which is a transposition of the digits.
\[ \arcsin\left(2\cdot\tfrac{\sqrt{3}}{4}\right) = \arcsin\left(\tfrac{\sqrt{3}}{2}\right) = \tfrac{\pi}{3}, \qquad \tfrac{\sqrt{3}}{4} \approx 0.4330 \]
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.7 Trigonometric Equations and Inequalities §10.7, pp. 870-871 — Example 10.7.5, part 1
The first line of the solution is the range check, and the book writes it out even though the answer is obviously yes. Copy that habit; on the next slide but one, the same check will be the entire content of the problem.
The rest is mechanical, and pleasantly so after the sign diagrams. Apply the sine to both sides, the composition collapses on the left, and what remains is a linear equation. Every part of Example 10.7.5 follows this same shape: isolate, check, strip, solve.
The book's decimal here is misprinted. Root three over four is about 0.4330, not 0.4430, and the two digits have been swapped. Substituting back is how you would catch it: root three over four doubled is root three over two, whose arcsine really is a third of pi, so the exact answer is right and only the decimal is wrong.
Worked example
\[ 4\arccos(x) - 3\pi = 0 \]
Isolate the inverse function
Why: Nothing can be undone while the arccosine is still wrapped in arithmetic.
\[ \arccos(x) = \tfrac{3\pi}{4} \]
Check the right side against the range
Why: The arccosine returns values from zero to pi, and three quarters of pi is inside.
\[ 0 \le \tfrac{3\pi}{4} \le \pi \]
Apply the cosine to both sides
Why: Then the inverse property collapses the left side.
\[ \cos\bigl(\arccos(x)\bigr) = \cos\left(\tfrac{3\pi}{4}\right) \]
Read off x
Why: The cosine of three quarters of pi is negative root two over two.
\[ x = -\tfrac{\sqrt{2}}{2} \approx -0.7071 \]
Check by substituting back
Why: Four times three quarters of pi is three pi, and subtracting three pi leaves zero.
\[ 4\arccos\left(-\tfrac{\sqrt{2}}{2}\right) - 3\pi = 4\cdot\tfrac{3\pi}{4} - 3\pi = 0 \]
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.7 Trigonometric Equations and Inequalities §10.7, p. 871 — Example 10.7.5, part 2
Isolating comes before everything, for a reason that is easy to overlook: the range check is a statement about the value of the inverse function itself, not about whatever arithmetic has been wrapped around it. Until the arccosine stands alone, there is nothing to compare with its range.
Three quarters of pi is inside the arccosine's range, which runs from zero to pi, so the equation is solvable. Notice that this value would have failed the arcsine's range test, which reaches only to a quarter turn. The check is specific to the function in front of you.
The answer is negative, and that should feel right. The arccosine returns angles in the second quadrant precisely when its input is negative, so an answer of three quarters of pi demands a negative x.
Trap
An arccosine set equal to a familiar angle:
\[ \arccos(x) = \tfrac{4\pi}{3} \]
\[ x = \cos\left(\tfrac{4\pi}{3}\right) = -\tfrac12 \]
Wrong. There is no solution at all.
The arccosine never returns a value bigger than pi, and four thirds of pi is bigger. The equation was impossible before any algebra started.
\[ \tfrac{4\pi}{3} > \pi \]
The proposed answer fails at once when you substitute it back.
\[ \arccos\left(-\tfrac12\right) = \tfrac{2\pi}{3} \ne \tfrac{4\pi}{3} \]
Everything in the wrong column is valid algebra, which is what makes this so effective at catching people. The cosine of four thirds of pi really is minus one half. The problem is that the equation being solved was never solvable.
The arccosine hands back angles from zero to pi and nothing else. Four thirds of pi is past the end of that range, so no input could produce it, and the correct answer is the sentence there is no solution rather than a number.
Substituting back exposes it immediately, which is the practical moral. The arccosine of minus one half is two thirds of pi, not four thirds. If you make substituting back the last line of every inverse problem, this trap cannot catch you even if you forget the range check.
Worked example
\[ 3\operatorname{arcsec}(2x-1) + \pi = 2\pi \]
Isolate the inverse function
Why: Subtract pi, then divide by 3.
\[ 3\operatorname{arcsec}(2x-1) = \pi \;\Longrightarrow\; \operatorname{arcsec}(2x-1) = \tfrac{\pi}{3} \]
Check the range, both versions of it
Why: Section 10.6 offered two competing ranges for the arcsecant, and a third of pi lies in both, so the answer will not depend on the convention.
\[ \tfrac{\pi}{3} \in \left[0, \tfrac{\pi}{2}\right) \]
Apply the secant to both sides
Why: The inverse property collapses the left side.
\[ \sec\bigl(\operatorname{arcsec}(2x-1)\bigr) = \sec\left(\tfrac{\pi}{3}\right) \]
Solve the linear equation that is left
Why: The secant of a third of pi is 2, because the cosine there is one half.
\[ 2x - 1 = 2 \;\Longrightarrow\; x = \tfrac32 \]
Check by substituting back
Why: The arcsecant of 2 is the arccosine of one half, which is a third of pi; three times that plus pi is two pi.
\[ \operatorname{arcsec}(2) = \arccos\left(\tfrac12\right) = \tfrac{\pi}{3}, \qquad 3\cdot\tfrac{\pi}{3} + \pi = 2\pi \]
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.7 Trigonometric Equations and Inequalities §10.7, p. 871 — Example 10.7.5, part 3
The arcsecant is the awkward member of the family, because there are two competing conventions for its range and different books choose differently. Section 10.6 set both out.
Here that awkwardness costs nothing, and the book says why in one line: a third of pi lies in both candidate ranges, so whichever convention your course uses, the equation is solvable and the answer is the same. Checking that explicitly is exactly the right instinct.
After the strip, all that is left is a linear equation in the inside of the arcsecant. That pattern is worth noticing across all six parts of this example: the inverse function is a shell, and once you strip it you are back to algebra you have done for years.
Worked example
\[ 4\arctan^{2}(x) - 3\pi\arctan(x) - \pi^{2} = 0 \]
Substitute a single letter for the arctangent
Why: The same move that turned trigonometric equations into polynomials in the first half of this section.
\[ u = \arctan(x) \;\Longrightarrow\; 4u^{2} - 3\pi u - \pi^{2} = 0 \]
Factor the quadratic
Why: Treat pi as an ordinary constant and the numbers are small.
\[ (4u + \pi)(u - \pi) = 0 \]
Expand to confirm the factorisation
Why: The two middle terms combine to the required coefficient.
\[ 4u^{2} - 4\pi u + \pi u - \pi^{2} = 4u^{2} - 3\pi u - \pi^{2} \]
Read the two candidate values
Why: One from each factor.
\[ u = -\tfrac{\pi}{4} \quad\text{or}\quad u = \pi \]
Reject the candidate outside the range
Why: The arctangent returns values strictly between minus a quarter turn and a quarter turn, and pi is far outside, so that branch yields nothing.
\[ \pi \notin \left(-\tfrac{\pi}{2}, \tfrac{\pi}{2}\right) \]
Undo the surviving arctangent
Why: Apply the tangent to both sides.
\[ \arctan(x) = -\tfrac{\pi}{4} \;\Longrightarrow\; x = \tan\left(-\tfrac{\pi}{4}\right) = -1 \]
Check by substituting back
Why: The arctangent of minus one is minus a quarter of pi, and the two positive pieces cancel the third.
\[ 4\left(-\tfrac{\pi}{4}\right)^{2} - 3\pi\left(-\tfrac{\pi}{4}\right) - \pi^{2} = \tfrac{\pi^{2}}{4} + \tfrac{3\pi^{2}}{4} - \pi^{2} = 0 \]
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.7 Trigonometric Equations and Inequalities §10.7, p. 872 — Example 10.7.5, part 4
The substitution is the same trick that turned trigonometric equations into polynomials in the first half of this section. Give the repeated expression a single letter and the shape of the equation becomes visible, pi and all.
Do not be intimidated by the constant. Pi behaves exactly like any other number in the factoring, and if you would rather, replace it by the letter c, factor, and put pi back at the end. The expansion step is there so you can confirm the factorisation rather than trusting it.
Then comes the part that makes this an inverse problem rather than a quadratic problem. Two candidate values come out and only one of them is a possible output of an arctangent, because the arctangent's range stops strictly short of a quarter turn. Discarding the impossible one is not an optional tidy-up; it is the difference between one solution and two.
Worked example
The same shape as the last example, and the range check comes out the other way.
\[ 9\arcsin^{2}(x) - \pi^{2} = 0 \]
Isolate the square
Why: There is no linear term, so no factoring is needed.
\[ \arcsin^{2}(x) = \tfrac{\pi^{2}}{9} \]
Take both square roots
Why: A squared quantity has two roots, and both must be carried forward.
\[ \arcsin(x) = \tfrac{\pi}{3} \quad\text{or}\quad \arcsin(x) = -\tfrac{\pi}{3} \]
Test each against the range
Why: The arcsine's range runs from minus a quarter turn to a quarter turn, and a third of pi is smaller than a half of pi in both directions, so neither candidate is rejected.
\[ -\tfrac{\pi}{2} \le -\tfrac{\pi}{3} < \tfrac{\pi}{3} \le \tfrac{\pi}{2} \]
Undo each arcsine
Why: Apply the sine to both sides of each equation.
\[ x = \sin\left(\tfrac{\pi}{3}\right) = \tfrac{\sqrt{3}}{2}, \qquad x = \sin\left(-\tfrac{\pi}{3}\right) = -\tfrac{\sqrt{3}}{2} \]
Check both at once
Why: Squaring kills the sign, so the same arithmetic settles both.
\[ 9\left(\pm\tfrac{\pi}{3}\right)^{2} - \pi^{2} = 9\cdot\tfrac{\pi^{2}}{9} - \pi^{2} = 0 \]
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.7 Trigonometric Equations and Inequalities §10.7, p. 875 — Exercise 65
This exercise is here as a deliberate contrast with the example before it. The structure is the same, the range check is the same, and the outcome is the opposite.
Taking both square roots gives a third of pi and minus a third of pi. Both sit inside the arcsine's range, which runs from minus a quarter turn to a quarter turn, so neither can be discarded and the equation genuinely has two solutions.
That is the lesson worth carrying: a range check is a check, not a rule that always eliminates something. Run it every time, and let it decide. The final line is a small economy worth copying, since squaring removes the sign and one calculation verifies both answers at once.
Sorting
Decide each one by the range alone, before any algebra.
Sort into buckets
Sort each equation by whether a solution exists.
Sort all six before checking, and do it purely from the ranges, without solving anything. That is the skill being trained.
The two that fail for size are easy to feel: an arccosine cannot be negative, and an arcsine cannot exceed a quarter turn. The subtle one is the arctangent set equal to a quarter turn exactly. Its range is an open interval, so the arctangent climbs toward a quarter turn as its input grows but never arrives, and no real number is a solution.
That open-versus-closed distinction is the single most common source of mistakes in this part of the section. The arcsine and arccosine attain their endpoints; the arctangent and arccotangent do not.
Concept
The inverse functions are continuous on their domains, so a sign diagram works on them exactly as it did on the circular functions. Two things change, and both make the work shorter.
\[ \pi^{2} - 4\arccos^{2}(x) < 0 \quad \text{is a question about } -1 \le x \le 1 \]
Test values are easy too: the ends of the domain are usually the friendliest numbers available.
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.7 Trigonometric Equations and Inequalities §10.7, p. 872 — the opening of Example 10.7.5, part 5
An inequality in an inverse function sounds like a new topic and is not. The inverse functions are continuous on their domains, so a sign diagram works on them exactly as it does on anything else.
Two things actually get easier. The domain is usually tiny, because the arcsine and arccosine accept only inputs from minus one to one, so your whole diagram fits inside a short interval. And there is usually a single landmark, because an equation in an inverse function has at most one solution.
There is one thing to watch. Recording the domain is now part of the setup rather than a formality, because the domain restriction supplies the ends of your diagram and often appears in the final answer as an endpoint.
Worked example
\[ \pi^{2} - 4\arccos^{2}(x) < 0 \]
Name the left side f and record its domain
Why: Everything is already on one side. The arccosine accepts nothing outside minus one to one, so the whole question lives there.
\[ f(x) = \pi^{2} - 4\arccos^{2}(x), \qquad -1 \le x \le 1 \]
Solve f(x) = 0
Why: Isolate the square and take both roots.
\[ \arccos^{2}(x) = \tfrac{\pi^{2}}{4} \;\Longrightarrow\; \arccos(x) = \pm\tfrac{\pi}{2} \]
Discard the negative root
Why: The arccosine never returns a negative value, so only the positive branch can happen.
\[ \arccos(x) = \tfrac{\pi}{2} \;\Longrightarrow\; x = \cos\left(\tfrac{\pi}{2}\right) = 0 \]
Test the two regions
Why: One landmark, so two regions, and the ends of the domain are the natural test values.
\[ f(-1) = \pi^{2} - 4\pi^{2} = -3\pi^{2} \approx -29.61, \qquad f(1) = \pi^{2} - 0 = \pi^{2} \approx 9.87 \]
Figure (svg): The curve of pi squared minus four times the square of the arccosine of x, plotted over its whole domain from minus one to one. It starts near minus thirty at the left end, rises through zero at x equal to zero, and reaches about nine point nine at the right end. The part left of zero, where the curve is below the axis, is shaded.
Collect the region where f is negative
Why: The left end of the domain is included; the landmark is not, because the inequality is strict.
\[ [-1, 0) \]
Check the excluded landmark
Why: At zero the arccosine is a quarter turn, four times its square is pi squared, and the expression is exactly zero, which is not less than zero.
\[ f(0) = \pi^{2} - 4\left(\tfrac{\pi}{2}\right)^{2} = \pi^{2} - \pi^{2} = 0 \not< 0 \]
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.7 Trigonometric Equations and Inequalities §10.7, p. 872 — Example 10.7.5, part 5
The domain line is the most important line in this solution. Once you have written that the arccosine only accepts inputs from minus one to one, the whole question has shrunk to a short interval and you know your diagram will be small.
Solving for the landmark repeats the pattern from the quadratic example: two candidates come out of the square root and one is impossible, because an arccosine is never negative. One landmark survives, at zero, and one landmark means two regions.
The two test values are the ends of the domain, which is usually the best choice for this kind of problem. And the endpoints of the final answer have different reasons yet again: minus one is in because the domain includes it and the sign there is right, while zero is out because the inequality is strict and the expression is exactly zero there.
Worked example
\[ 4\operatorname{arccot}(3x) > \pi \]
Move everything to one side and name f
Why: Same first move as every other inequality in this lesson.
\[ f(x) = 4\operatorname{arccot}(3x) - \pi \]
Record the domain
Why: The arccotangent accepts every real number, so there is no restriction and no undefined landmark.
\[ \text{domain of } f = (-\infty, \infty) \]
Solve f(x) = 0
Why: Isolate the inverse function; a quarter of pi is inside the arccotangent's range.
\[ \operatorname{arccot}(3x) = \tfrac{\pi}{4} \]
Undo the arccotangent
Why: Apply the cotangent to both sides, then divide by 3.
\[ 3x = \cot\left(\tfrac{\pi}{4}\right) = 1 \;\Longrightarrow\; x = \tfrac13 \]
Choose test values that land on common angles
Why: Set three x equal to the cotangent of an angle you know: x equal to zero gives the arccotangent of zero, a quarter turn.
\[ f(0) = 4\cdot\tfrac{\pi}{2} - \pi = \pi > 0 \]
Test on the other side of the landmark
Why: Root three over three makes three x equal root three, whose arccotangent is a sixth of pi.
\[ f\left(\tfrac{\sqrt{3}}{3}\right) = 4\cdot\tfrac{\pi}{6} - \pi = -\tfrac{\pi}{3} < 0 \]
Figure (svg): The curve of four times the arccotangent of three x, falling from near four pi on the left toward zero on the right, crossed by a horizontal line at pi at x equal to one third. Everything left of one third, where the curve is above the line, is shaded.
Collect the positive region
Why: Strict inequality, so the landmark itself is excluded, and the region runs away to the left without end.
\[ \left(-\infty, \tfrac13\right) \]
Check a decimal just inside the answer
Why: At three tenths, which is below a third, three x is nine tenths and the arccotangent of nine tenths is above a quarter of pi, so four times it is above pi.
\[ \operatorname{arccot}(0.9) \approx 0.8380 > 0.7854 \approx \tfrac{\pi}{4} \]
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.7 Trigonometric Equations and Inequalities §10.7, p. 873 — Example 10.7.5, part 6
Compare this setup with the previous one. The arccotangent accepts every real number, so there is no domain restriction, and the answer will run off to infinity in one direction instead of stopping at the end of an interval.
The test values deserve attention, because they were reverse-engineered rather than guessed. The book's footnote explains the method: decide which common angles you would like the arccotangent to produce, work out what three x must be for each, and take those as your test values. That is the same trick as the arctangent of one hundred and seventeen, used for convenience rather than necessity.
The arccotangent decreases, which is why the answer lies to the left of the landmark rather than the right. If you ever find yourself unsure, the graph settles it in one glance, and the decimal check at the end settles it in one line.
Counterexample
Discussion prompt
Everything in the first half of this section produced infinitely many solutions, spaced by a period. Write down an equation from this lesson that has exactly one solution, and say in one sentence what structural fact about its function makes the difference.
Write your answer before revealing. The second half of the question is the one that matters: not which equation, but why.
The honest answer is that the difference has nothing to do with trigonometry as such. It is periodicity. A periodic function takes each of its values over and over, so an equation in it has infinitely many solutions. An inverse function is not periodic; it is one-to-one by construction, so it takes each value at most once.
That is worth generalising. Whenever you meet a new function, asking whether it is periodic tells you in advance what shape its equations' answers will have, long before you try to solve one.
Ranking
Put in order
Put the steps of a complete solve of an equation in an inverse function into order.
Why: Isolating comes first because the range check is a statement about the isolated value, not about the whole left side. The check comes before any algebra, since a value outside the range means there is nothing to solve and any answer you produce will be false. Only then is it safe to apply the circular function, collapse the composition and finish with ordinary algebra, and the substitution at the end catches both arithmetic slips and forgotten range failures.
Drag the six steps into order before checking. Only two of the orderings are genuinely in question, and both are worth thinking about.
The first is whether isolating or checking comes first. Isolating must come first, because the range check compares the range with the isolated value, and before isolation there is no such value to compare.
The second is where the substitution goes. It is last, and it is not optional. It catches ordinary arithmetic slips, and it also catches the range failure you forgot to check for, because an answer produced from an impossible equation will not survive being put back in.
Section
Part 6
Pattern
Figure (svg): A flow diagram. A first box asks what kind of question it is. An inequality on a stated interval goes straight to the sign diagram. A denominator goes to an equation, then to an excluded family. A radical goes to an inequality on one period, then to a translation by the period. An inverse function goes to a range check and then to a single answer. The sign diagram box leads to a final box about endpoints.
This is the whole lesson compressed into one procedure, and the diagram shows how little actually differs between the four kinds of question. Everything funnels into a sign diagram or a single equation, and everything comes out as intervals or excluded families.
Work down the list in order when you are stuck. Most errors in this topic are not errors of calculation; they are a step skipped. The two most commonly skipped are hunting for undefined points, which is the second item, and choosing the right period when no window was given, which is the fourth.
The last item is deliberately last, because it is a separate decision made after all the sign work is done. Get into the habit of writing your regions first without any brackets at all, and then going back along the line assigning brackets one landmark at a time.
Real world
Figure (svg): The sun's elevation in degrees plotted against the hour of the day, rising from zero at six in the morning to sixty degrees at noon and back to zero at six in the evening, with a horizontal line at fifteen degrees. The stretch between the two crossings, from about seven to about five, is shaded.
Discussion prompt
A rooftop panel makes usable power only while the sun's elevation is above fifteen degrees. On a spring day the elevation in degrees is modelled by sixty times the sine of pi times the quantity t minus six, all over twelve, where t is the hour of the day and the model runs from six to eighteen. Find the hours of useful production, and say why the answer has to be an interval rather than a list of times.
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.7 Trigonometric Equations and Inequalities §10.7, p. 874 — the sinusoid models of the exercise set
Work the problem before revealing, at least as far as setting up the inequality. Everything you need is on the previous fifty slides; the only new thing is reading the model out of a sentence.
Notice what kind of answer the question demands. It does not ask when the elevation equals fifteen degrees; it asks when it exceeds fifteen degrees, and those two times are only the ends of the answer. Because the elevation changes continuously, everything between them qualifies, which is exactly why the answer is an interval and not a list.
The arcsine appears here for the same reason the arctangent appeared earlier: a quarter is not the sine of any common angle, so the boundary times have to be named by an inverse function and then converted to hours. About ten hours of useful production from twelve hours of daylight, and the missing two hours are the shallow ends of the day.
Check
Check your understanding
The solution of the tangent of x being at least 3, on one revolution, begins at the arctangent of 3 and stops at a quarter turn. Which pair of brackets is right?
Answer: A
Why: The left end is a zero of tan x − 3, and the inequality is not strict, so that value satisfies it and the bracket is square. The right end is an asymptote, where the tangent has no value at all, so it can never be included, whatever the inequality sign says.
This question is about the two rules for brackets and nothing else. The regions have already been handed to you.
The left end is a zero of the expression and the inequality allows equality, so that value really does satisfy the inequality and its bracket is square. The right end is an asymptote, where the tangent has no value, so it is excluded no matter what the inequality symbol says.
If you were tempted by the third option, notice what it claims: that the tangent of a quarter turn is a number at least as big as three. It is not a number at all.
Check
Check your understanding
Which numbers must be removed from the domain of the reciprocal of twice the sine of x plus one?
Answer: A
Why: Set the denominator to zero: twice the sine of x plus one is zero when the sine of x is minus one half, which happens at seven sixths of pi and at eleven sixths of pi, each repeating every two pi. Two families, because the sine takes a given value twice per revolution.
Set the denominator to zero and solve. The sine equals minus one half, which happens twice per revolution, in the third and fourth quadrants.
Two families come out, one stepping from seven sixths of pi and one from eleven sixths of pi, each by two pi. The most tempting wrong answer forgets the minus sign and reports the two angles where the sine is plus one half instead.
The last option is worth a second of thought even though it is clearly wrong, because it names the zeros of the sine itself. There the denominator equals one, which is a perfectly good denominator. Only a denominator of zero causes trouble.
Check
Check your understanding
Which of these equations has no solution?
Answer: C
Why: The arctangent's range is the open interval from minus a quarter turn to a quarter turn. It approaches a quarter turn as x grows without bound but never attains it, so no real x satisfies this equation. The other three targets all lie inside their function's range.
Answer from the ranges alone. Three of these targets sit inside the range of the inverse function involved, and one does not.
The arctangent's range is open at both ends. As x grows without bound the arctangent climbs toward a quarter turn and gets as close as you like, but it never arrives, so no real number is sent to a quarter turn exactly.
Compare that with the arccosine option, whose target of three quarters of pi is comfortably inside a range that includes its endpoints. Open versus closed is the entire difference, and it is the distinction to carry out of this lesson.
Explain it to yourself
Discussion prompt
In two or three sentences, explain why substituting a single number into f settles the sign of f across a whole region of the diagram, and say what property of the circular functions the argument depends on. Then say what would go wrong if you forgot to mark an undefined point.
Write your two or three sentences before revealing. Being able to say why the method works is what lets you rebuild it under pressure instead of half remembering it.
The argument has two halves and both are needed. Continuity says a sign change forces a crossing of zero; the landmarks say there is no crossing available inside a region. Together they force the sign to be constant, so one substitution reports on the whole stretch.
The second half of the question is the practical payoff. If an undefined point is left off the diagram, two regions of opposite sign are treated as one, and your single test value speaks for territory it has no right to. That is why hunting for undefined points is a separate step in the procedure rather than something you notice in passing.
Exit ticket
\[ f(x) = \frac{\sqrt{2\sin(x) - 1}}{\cos(x)} \]
Discussion prompt
Find the domain, in extended interval notation. Both of this lesson's tools are needed: one for the root, one for the denominator. State which endpoints are closed and why.
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.7 Trigonometric Equations and Inequalities §10.7, pp. 868-870 — the combined method of Example 10.7.4
This single problem uses both halves of the lesson, which is why it is the exit ticket. The numerator is a square root, so it asks an inequality; the denominator asks an equation. Do them separately and combine at the end.
The inequality gives a closed interval per revolution, because the inside of the root is allowed to be zero. The denominator then punches a single point out of the middle of that interval, at a quarter turn, where the cosine vanishes. A punched-out point always produces round brackets on both sides of the hole.
If your answer has only one interval per revolution, you have probably solved the root correctly and then forgotten that the removed point lies inside it rather than outside. Sketching the closed interval and marking the hole is the fastest way to see the shape of the answer before writing the notation.
Recap
| the question | what you actually solve | what the answer looks like |
|---|---|---|
| an inequality on a stated window | the equation, then one test value per region | a union of intervals inside that window |
| a denominator, or a cosecant, secant, tangent, cotangent | an equation: set the bottom to zero | everything except one or two families |
| an even root | an inequality on one period, then translate | a repeating union, spaced by the period |
| an equation in an inverse function | isolate it, check the range, strip it | one number, or none |
| an inequality in an inverse function | a sign diagram on the inverse's own domain | a subinterval of that domain |
\[ \bigcup_{k=-\infty}^{\infty} \bigl(\text{the answer on one period} + pk\bigr) \]
And the rule that survives every one of these: a zero belongs to the answer exactly when the inequality allows equality, while a point where the expression is undefined is excluded no matter what. Chapter 10 ends here; Chapter 11 turns the same functions on real triangles and real waves, starting with applications of sinusoids.
OpenStax Algebra and Trigonometry 2e, §9.5 Solving Trigonometric Equations §9.5 — the same material in OpenStax Algebra and Trigonometry
Five questions, one machine. Move everything to one side, find both kinds of landmark, test one point per region, and then, separately, decide the brackets. A domain question is the same machine with the window removed and periodicity supplying the missing structure.
Two facts deserve to be memorised rather than rederived. The first is that a point where the expression is undefined is excluded from the answer no matter which inequality symbol appears. The second is that an equation in an inverse function has at most one solution, and whether it has any is decided by a range check that costs one line.
That closes Chapter 10, which built the circular functions, their identities, their graphs, their inverses and now their equations and inequalities. Chapter 11 puts them to work: sinusoid models of real oscillations first, then the Law of Sines and the Law of Cosines on real triangles, then polar coordinates and vectors.
Want this taught 1-on-1? Alexander tutors Trigonometry — $55/session, free consultation.