The first graphs in the course with vertical asymptotes, and every feature of them inherited from cosine and sine by reciprocation. Covers where the asymptotes come from and how to determine which way each branch runs, the two values reciprocation leaves fixed, the range being everything of absolute value at least one, the absence of any amplitude, and the practical method of drawing the parent sinusoid first and reciprocating it point by point.
Subject: Trigonometry · 63 slides · symbolic lesson
Open the interactive version of this deck
Title
Trigonometry · Chapter 10 — Foundations of Trigonometry
The first curves in this course with vertical asymptotes
Objectives
You can already draw a cosine or a sine curve and read its four parameters off a formula. The secant and the cosecant are one over those two curves, and every feature of their graphs is forced by that one fact.
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.5 Graphs of the Trigonometric Functions §10.5, pp. 800-804 — §10.5.2 and Theorem 10.24
Every graph you have drawn so far in this chapter has been a single unbroken curve trapped between two horizontal lines. These two are the opposite: they run off to infinity, they come in separate pieces, and they are missing points. All of that sounds like a lot of new behaviour to learn.
It is not. The secant is one over the cosine and the cosecant is one over the sine, and every single feature of the new graphs is what that division does to a curve you already know. If you find yourself trying to memorise a shape, stop and ask what the parent curve is doing there instead.
By the end you should be able to take any formula built from a secant or a cosecant, sketch one cycle of it, and state its period, its asymptotes, its phase shift and its range, while knowing exactly why an amplitude is not on that list.
Warm-up
\[ \cos(x): \quad 1, \;\tfrac{\sqrt2}{2}, \;0, \;-\tfrac{\sqrt2}{2}, \;-1 \qquad \text{at } x = 0, \tfrac{\pi}{4}, \tfrac{\pi}{2}, \tfrac{3\pi}{4}, \pi \]
Discussion prompt
Take the reciprocal of each of those five cosine values. Four of them work. Say what the fifth one does, and what that must mean for the graph at that value of x.
Write your five answers down before revealing anything. Four of them are ordinary arithmetic, and the whole lesson hangs on the fifth.
Turning over root two over two is worth doing carefully once: it is two over root two, and rationalising gives root two, about 1.414. Notice that the size went up, from about 0.707 to about 1.414, and the sign stayed the same. That happens to every value smaller than one in size.
The zero is the interesting one. There is no number you can multiply zero by to get one, so there simply is no value of the secant there. What matters more is what happens just beside it, where the cosine is small but not yet zero, and the reciprocal is enormous. That is the whole reason these graphs look the way they do.
Section
Part 1
Concept
These are not new functions. They were defined back in Section 10.3 as reciprocals, and the graphs simply make that visible.
\[ \sec(x) = \frac{1}{\cos(x)}, \qquad \csc(x) = \frac{1}{\sin(x)} \]
So there is never any reason to plot one of these from scratch. Draw the parent curve, then replace every height by one over that height. The whole lesson is the consequences of doing that.
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.5 Graphs of the Trigonometric Functions §10.5, p. 800 — the opening of §10.5.2
These definitions came from Section 10.3, long before any graphing. Nothing new is being introduced here; you are simply going to look at them.
The three bulleted rules are the entire lesson in miniature, and they are worth reading as claims you could check rather than as facts to memorise. A zero on the bottom of a fraction kills the fraction, so the zeros of the parent are exactly the places the reciprocal is missing. One and negative one are the only numbers that equal their own reciprocal, so those are the only points the two curves can share. Everything else is smaller than one in size, so its reciprocal is bigger, and the new curve sits further from the axis than the old one.
Get into the habit of drawing the parent first and lightly. It costs you ten seconds and it removes almost all of the work that follows.
Concept
Figure (svg): Two horizontal number lines, one above the other, both running from negative 3 to 3. Arrows join each value on the top line to its reciprocal on the bottom line: 1 and negative 1 go straight down to themselves, while 0.5, 0.8 and one third fan outward to 2, 1.25 and 3. A shaded strip on the bottom line between negative 1 and 1 is marked as never reached, and the value 0 on the top line is marked as having no reciprocal at all.
\[ 0 < |u| < 1 \;\Longrightarrow\; \left|\tfrac1u\right| > 1, \qquad |u| > 1 \;\Longrightarrow\; \left|\tfrac1u\right| < 1, \qquad \tfrac{1}{\pm 1} = \pm 1 \]
Three rules, and they are enough to sketch either curve. Notice that the sign never changes: one over a positive number is positive and one over a negative number is negative.
Look at the arrows one at a time. The two that go straight down are the ones from 1 and from negative 1: those are the fixed points. Every other arrow fans outward, because every other starting value shown is smaller than one in size.
The shaded strip on the lower line is the part of the number line no arrow ever reaches. To land inside it you would have to start from a value bigger than one in size, and the cosine never gives you one. That strip will reappear on every graph in this lesson as the empty band around the horizontal axis.
The red dot at zero is the exception that makes everything interesting. It has no arrow at all, because there is no number to draw the arrow to. Zero is where the graph stops existing, and just either side of zero the arrows are getting longer and longer without limit.
Intuition
The cosine spends its whole life between negative one and one, so its reciprocal spends its whole life outside that range. There is no value the cosine can take that gives a secant value of one half.
The cosine is at its smallest, in size, exactly where it crosses the axis, so that is where the secant is at its largest. The cosine is at its largest, in size, at its peaks and troughs, so that is where the secant is at its smallest.
Everything about these two graphs is one of those two sentences applied somewhere. Keep them in mind and you will never need to memorise a picture.
Two sentences to carry with you. The first: the cosine lives between negative one and one, so the secant lives outside that range. There is no input whose secant is one half, and no input whose secant is zero, and knowing that will stop you drawing a branch that dips too close to the axis.
The second: where the parent is smallest in size, the reciprocal is largest, and where the parent is largest in size, the reciprocal is smallest. The cosine is smallest in size at its zeros, which is why the asymptotes go there, and largest in size at its peaks and troughs, which is why the turning points go there.
The sign is a separate matter and it never changes. One over a positive number is positive; one over a negative number is negative. So a branch that starts out above the axis stays above the axis until it disappears at an asymptote, and that fact alone decides which way every branch opens.
Picture it
Figure (svg): The graph of y equals one over u drawn on square axes from negative 3 to 3, in two branches separated by the vertical axis, with the line y equals u dashed through the origin. The two curves meet at the points 1 comma 1 and negative 1 comma negative 1. A shaded horizontal strip between y equals negative 1 and y equals 1 is marked as the values the reciprocal never takes.
This is what reciprocation looks like as a graph in its own right. The dashed diagonal marks the inputs, and the two curves cross it at one and at negative one, the only two numbers left where they are. The shaded strip is the set of outputs that never occur.
This picture is the previous one turned into a graph. The horizontal axis carries the input and the vertical axis carries its reciprocal, so the dashed diagonal is the line where nothing changes.
The two places the curve meets that diagonal are the only two numbers that reciprocate to themselves, one and negative one, and they are marked. On the secant graph those two heights are exactly where the curve touches its parent.
The shaded band is the set of outputs the curve never produces when the input is at most one in size. Read the picture in that direction: feed in a value from the cosine, which is always between negative one and one, and the output is always outside the band. That is the range of the secant, drawn.
Concept
The secant is undefined wherever the cosine is zero, and the cosine is zero at every odd multiple of a quarter turn.
\[ \cos(x) = 0 \iff x = \frac{\pi}{2} + \pi k, \quad k \text{ an integer} \]
\[ \sin(x) = 0 \iff x = \pi k, \quad k \text{ an integer} \]
Those two lists are the asymptote locations of the secant and of the cosecant. Nothing else is undefined, because a fraction with a nonzero bottom is always a number.
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.5 Graphs of the Trigonometric Functions §10.5, pp. 800-801 — the domains of secant and cosecant
Two lists to know cold, and they come straight from the unit circle rather than from anything new. The cosine is the horizontal coordinate, which is zero at the top and bottom of the circle, so at a quarter turn and every half turn after that. The sine is the vertical coordinate, which is zero at the two ends of the horizontal axis, so at every half turn starting from zero.
Those two lists are offset from each other by a quarter turn, which is the first hint that the two graphs will be the same picture in two places.
Notice the second sentence as well. Nothing except a zero on the bottom can make one of these functions undefined, because the top is always the constant one and division is fine for every other value. So these two lists are complete: there are no other bad points hiding anywhere.
Notation
Annotate
On: \( \text{as } x \to \tfrac{\pi}{2}^{-}, \; \cos(x) \to 0^{+}, \; \text{so } \sec(x) \to \infty \)
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.5 Graphs of the Trigonometric Functions §10.5, p. 800 — the four one-sided statements
This notation is used constantly from here on, and the two small signs carry all the information, so it is worth slowing down for. The sign attached to the value x is heading for tells you which side you are coming from. The sign attached to the limit of the cosine tells you which side of zero the cosine is on when it gets there.
Step through the annotations and keep the unit circle in view. Just below a quarter turn you are still in the first quadrant, where the horizontal coordinate is positive but shrinking toward nothing. One over a shrinking positive number grows without bound, which is the last line.
The final annotation is the one that saves marks. A statement like this covers one side of one asymptote and nothing more. There are four such statements in the fundamental cycle of the secant, and they do not all say the same thing.
Worked example
Describe what the secant does on each side of each asymptote in the fundamental cycle, using nothing but the sign of the cosine.
\[ \text{asymptotes at } x = \tfrac{\pi}{2} \text{ and } x = \tfrac{3\pi}{2} \]
Approach pi over two from the left
Why: Just below a quarter turn is still quadrant one, where the cosine is positive.
\[ x \to \tfrac{\pi}{2}^{-}: \quad \cos(x) \to 0^{+} \;\Longrightarrow\; \sec(x) \to \infty \]
Approach pi over two from the right
Why: Just past a quarter turn is quadrant two, where the cosine is negative.
\[ x \to \tfrac{\pi}{2}^{+}: \quad \cos(x) \to 0^{-} \;\Longrightarrow\; \sec(x) \to -\infty \]
Approach three pi over two from the left
Why: Just below three quarters of a turn is quadrant three, cosine still negative.
\[ x \to \tfrac{3\pi}{2}^{-}: \quad \cos(x) \to 0^{-} \;\Longrightarrow\; \sec(x) \to -\infty \]
Approach three pi over two from the right
Why: Just past it is quadrant four, where the cosine is positive again.
\[ x \to \tfrac{3\pi}{2}^{+}: \quad \cos(x) \to 0^{+} \;\Longrightarrow\; \sec(x) \to \infty \]
Figure (svg): Two panels over the same short stretch of x either side of pi over two. The left panel shows the cosine passing through zero, positive on the left and negative on the right, on a vertical scale of about a third. The right panel shows the secant on a scale of sixteen: it climbs off the top of the frame on the left of the asymptote and comes up from the bottom of the frame on the right.
Check with two nearby numbers
Why: At 1.5 radians the cosine is 0.0707 and the secant is 14.14; at 1.6 radians the cosine is negative 0.0292 and the secant is negative 34.25. The two sides genuinely run opposite ways.
\[ \sec(1.5) = 14.14, \qquad \sec(1.6) = -34.25 \]
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.5 Graphs of the Trigonometric Functions §10.5, p. 800 — the analysis before the secant table
Four short arguments with the same shape, and none of them needs a calculation. In each one, ask which quadrant you are in, recall the sign of the cosine there, and reciprocate.
Read the four results together and a pattern appears: at each asymptote the two sides run in opposite directions, and the two asymptotes are mirror images of each other. That pattern is not a coincidence; a zero of a smooth function is where it changes sign, so the two sides must disagree.
The check at the end is the sort you should run whenever you are unsure. Pick a number just below the asymptote and one just above it and evaluate. Getting 14 and then negative 34 out of two inputs a tenth of a radian apart is a vivid way of seeing what a vertical asymptote does.
Picture it
Figure (svg): A close-up of the stretch from just inside negative pi over two to just inside pi over two. The cosine arch is dashed, rising to a peak of 1 at the origin. A single U-shaped branch of the secant sits above it, touching the cosine exactly at that peak and running up to the asymptotes on both sides.
Between two consecutive asymptotes the secant has exactly one turning point, and it sits directly on top of the cosine's turning point. That single shared point, together with the two asymptotes, fixes the whole branch.
This close-up shows one branch and the piece of the parent it sits on. The cosine arch rises to a peak of one at the origin, and the secant branch comes down to meet it there and then climbs away on both sides.
That single touching point, together with the two asymptotes, is all you need to sketch the branch. You know where it cannot go, you know the one point it must pass through, and you know it bends away from the axis in between.
It is worth noticing that the branch has no maximum. It has a minimum, at the touching point, and it is unbounded above. That asymmetry is exactly why an amplitude cannot be defined for these curves, which is a point the lesson returns to later.
Prediction
A function is one over some parent function, and all you are told is that the parent is zero at one particular input and nowhere near zero elsewhere.
Predict first
What must the reciprocal do at that input?
Correct: It has a vertical asymptote there
Why: A fraction with zero on the bottom has no value, so the reciprocal is undefined at that input. Just either side of it the bottom is tiny, so the fraction is enormous, and the graph runs off the frame. That is exactly what a vertical asymptote is. The zero of a function and the asymptote of its reciprocal are always at the same place.
Choose an answer before revealing. The question is deliberately stripped of trigonometry, because the fact being tested has nothing to do with trigonometry.
A zero of a function and a vertical asymptote of its reciprocal are always at the same input. That single sentence covers the secant, the cosecant, the tangent and the cotangent you will meet next lesson, and every rational function from Chapter 4.
If you were tempted by the first option, check what it would take to be true: the reciprocal would have to equal zero, which needs the top of the fraction to be zero, and the top is the constant one. It never happens.
Trap
A sketch that looks plausible and is wrong:
\[ x \to \tfrac{\pi}{2}^{-}: \sec x \to \infty \]
\[ x \to \tfrac{\pi}{2}^{+}: \sec x \to \infty \; ? \]
Wrong. That would need the cosine to be positive on both sides of its zero.
A zero is a place where a function changes sign, so the two sides of the asymptote carry opposite signs and the branches run opposite ways. Look at the parent, not at the reciprocal.
\[ \cos(1.5) > 0 > \cos(1.6) \]
\[ \sec(1.5) > 0 > \sec(1.6) \]
This mistake usually comes from sketching the reciprocal directly instead of reading the parent. The picture in your head is of a curve rushing off the page, and it feels natural to draw it rushing off the same way on both sides.
The defence is to look down at the parent every time. A zero is where a function crosses the axis, and crossing means changing sign, so the two sides of the asymptote carry opposite signs and the branches must run opposite ways. The only way both sides could run the same way is if the parent touched zero without crossing it, and the cosine and sine never do that.
Sorting
Sort into buckets
For each one-sided approach, decide whether the function runs up to positive infinity or down to negative infinity. Read the sign of the parent first.
Sort these by asking one question each time: is the parent positive or negative on that side of its zero? Do not try to picture the reciprocal at all until you have answered that.
The sine cases may take a moment. Just above zero the sine is positive, because you are entering the first quadrant; just below a half turn it is still positive, because you are finishing the second; just above a half turn it has gone negative. So the two sides of the half-turn asymptote of the cosecant disagree, exactly as they must.
If you got one wrong, do not just note the answer. Go back and say out loud which quadrant that approach lands in and what the sign of the parent is there. That is the step that was skipped.
Section
Part 2
Worked example
Build the table the book builds, on the eighth marks of one cycle. No cosine value is recomputed; each one is simply turned over.
\[ x = 0, \tfrac{\pi}{4}, \tfrac{\pi}{2}, \tfrac{3\pi}{4}, \pi, \tfrac{5\pi}{4}, \tfrac{3\pi}{2}, \tfrac{7\pi}{4}, 2\pi \]
Write the cosine row first
Why: These are the nine values you already know from the unit circle.
\[ \cos: \;1, \;\tfrac{\sqrt2}{2}, \;0, \;-\tfrac{\sqrt2}{2}, \;-1, \;-\tfrac{\sqrt2}{2}, \;0, \;\tfrac{\sqrt2}{2}, \;1 \]
Turn over the value at the eighth marks
Why: Rationalise: one over root two over two is two over root two, which is root two.
\[ \frac{1}{\sqrt2/2} = \frac{2}{\sqrt2} = \sqrt2 \approx 1.414 \]
Turn over the two extreme values
Why: One and negative one are their own reciprocals, so those two points do not move at all.
\[ \frac11 = 1, \qquad \frac{1}{-1} = -1 \]
Mark the two failures
Why: The cosine is zero twice in this cycle, and a zero has no reciprocal.
\[ x = \tfrac{\pi}{2}, \tfrac{3\pi}{2}: \quad \sec(x) \text{ undefined, asymptote} \]
Read off the finished row
Why: Nine slots, seven numbers and two asymptotes.
\[ \sec: \;1, \;\sqrt2, \;\text{--}, \;-\sqrt2, \;-1, \;-\sqrt2, \;\text{--}, \;\sqrt2, \;1 \]
Figure (svg): One cycle of the secant over the interval from 0 to two pi. The cosine is dashed behind it. Red dashed vertical asymptotes stand at pi over two and three pi over two. Three branches appear: an upward half-branch from the point 0 comma 1, a downward cap through the point pi comma negative 1, and an upward half-branch returning to the point two pi comma 1. Four more marked points at the eighth marks sit at heights plus and minus root two.
Check the largest and smallest heights
Why: Every listed secant value has size at least one, and the two values of size exactly one occur precisely where the cosine reached a peak or a trough. Nothing lands strictly inside.
\[ |\sec| = 1 \text{ at } x = 0, \pi, 2\pi; \qquad |{-\sqrt2}| = 1.414 > 1 \]
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.5 Graphs of the Trigonometric Functions §10.5, p. 800 — the table for the fundamental cycle of secant
This is the table the book builds, and the point of it is how little work it takes. There are no new evaluations anywhere in it; every entry is one of the nine cosine values turned upside down.
Two lines deserve care. Rationalising one over root two over two is a step people rush: it is two over root two, and multiplying top and bottom by root two gives root two. And the two extreme values, one and negative one, are their own reciprocals, so the secant passes through exactly the same two points the cosine did.
Look at the finished row and notice its shape: it starts at one, grows to root two, blows up, comes back from negative infinity through negative root two to negative one, and then retraces its steps. That symmetry is the cosine's symmetry, inherited.
Concept
Figure (svg): The secant graph drawn across three full periods, from negative two pi to four pi, with asymptotes at every odd multiple of pi over two and an upward branch alternating with a downward one. The interval from 0 to two pi is shaded to mark the fundamental cycle that is being copied.
The cosine repeats every full turn, so the secant does too, and the whole graph is the shaded cycle stamped out again and again. Two angles have the same secant exactly when they have the same cosine, which is why the period cannot be anything smaller.
\[ \cos(x + 2\pi) = \cos(x) \;\Longrightarrow\; \sec(x + 2\pi) = \sec(x) \]
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.5 Graphs of the Trigonometric Functions §10.5, p. 800 — footnote 11, the period of secant
The shaded stretch is the cycle you have just built; everything else in the picture is that cycle stamped out again. Nothing about the graph beyond the shaded region contains any new information.
The short argument on the slide is worth reading rather than skimming. Shifting the input by a full turn leaves the cosine unchanged, so it leaves one over the cosine unchanged too. That establishes that a full turn is a period.
The book's footnote makes the sharper point, which is that no smaller number works either. Two angles have the same secant precisely when they have the same cosine, because reciprocating is reversible, so any period of the secant would be a period of the cosine, and the cosine has nothing smaller than a full turn.
Concept
range of the secant — Every real number whose size is one or more, and nothing else. In interval language, everything from minus infinity up to and including negative one, together with everything from one upward.
\[ \text{range}(\sec) = \{y : |y| \ge 1\} = (-\infty, -1] \cup [1, \infty) \]
Both ends are reached and both are closed, because the secant really does equal one and really does equal negative one. What is missing is the open interval strictly between them, and it is missing because the cosine is never larger than one in size.
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.5 Graphs of the Trigonometric Functions §10.5, p. 800 — footnote 12, and Theorem 10.24 on p. 802
The range is written here in two ways because you will meet both. The set description says it in words: every number whose size is at least one. The interval description says the same thing as two pieces glued together.
Both square brackets matter. The secant really does equal one, at the origin, and really does equal negative one, at a half turn, so both endpoints belong to the range. What is excluded is the open interval strictly between them, and nothing else.
Sanity-check the claim from the parent. For the secant to take a value of size less than one, the cosine would have to take a value of size more than one, and it never does. That is the whole proof.
Intuition
The construction runs both ways, and being able to reverse it is a good test of whether you really understand it. Given only a secant graph, recover the cosine.
Doing it in this direction also explains something that can look odd at first: the reciprocal curve and its parent always cross the horizontal axis a different number of times, and in fact the parent crosses exactly where the reciprocal is missing.
Running the construction in reverse is a genuinely useful skill, not just a curiosity. Exam questions often hand you a graph with asymptotes and ask for a formula, and the fastest route is to recover the parent sinusoid first, where you already know how to read off every parameter.
The three rules reverse cleanly. Asymptotes become zeros, turning points stay put, and every other height gets turned over, which drags the curve in toward the axis and back inside the band between negative one and one.
The last remark is worth sitting with. The parent crosses the horizontal axis exactly where the reciprocal is missing, so the two curves never cross the axis at the same place, and the reciprocal never crosses it at all. If your sketch shows a secant curve touching the axis, something has gone wrong several steps earlier.
Socratic
The cosine crosses the horizontal axis twice per cycle. The secant never crosses it at all.
Discussion prompt
Explain that from the definition rather than from the picture: set the secant equal to zero and see what the equation asks for.
Write an answer before revealing. The temptation is to say because the graph does not go there, which is a description rather than a reason.
The argument is one line of algebra and it is completely general. A fraction equals zero only when its numerator equals zero, so setting one over the cosine equal to zero forces the constant one to be zero, which is absurd. There are no solutions, so there are no crossings.
The pleasing part is where the crossings went. The cosine crossed twice per cycle, and those two inputs are exactly where the secant is undefined. Reciprocation traded each crossing for an asymptote, which is the same trade the previous slide described in reverse.
Fill the middle
One row of the table, with the last entry removed. Work left to right.
Fill in the blanks
\cos\left(\tfrac-\sqrt{2}___\right) = -\tfrac___}___ \quad\Longrightarrow\quad \sec\left(\tfrac______\right) = \frac______}___} = ___
Why: Turning over negative root two over two gives negative two over root two, and rationalising the bottom gives negative root two, about negative 1.414. The sign is unchanged, and the size has grown from 0.707 to 1.414, as it must for any parent value smaller than one in size.
Fill the blank before checking. Work along the line rather than jumping to the answer: you are turning over negative root two over two.
Two things should come out. The sign stays negative, because one over a negative number is negative. And the size grows, from about 0.707 to about 1.414, because the value you started from was smaller than one in size.
If you wrote the answer as negative two over root two, that is correct but unfinished. Rationalising the denominator gives negative root two, and that is the form the book uses in its table.
Section
Part 3
Worked example
Exactly the same construction with the other parent. The only thing that changes is where the zeros are.
\[ \sin: \;0, \;\tfrac{\sqrt2}{2}, \;1, \;\tfrac{\sqrt2}{2}, \;0, \;-\tfrac{\sqrt2}{2}, \;-1, \;-\tfrac{\sqrt2}{2}, \;0 \]
Find the zeros of the parent first
Why: The sine vanishes at both ends of the interval and in the middle.
\[ \sin(x) = 0 \text{ at } x = 0, \;\pi, \;2\pi \]
Mark those three as asymptotes
Why: Three asymptotes on this interval, not two, and two of them are at the very edges.
\[ x = 0, \pi, 2\pi: \quad \csc(x) \text{ undefined} \]
Turn over the remaining six values
Why: The extremes stay put, the eighth-mark values grow to root two in size.
\[ \csc: \;\text{--}, \;\sqrt2, \;1, \;\sqrt2, \;\text{--}, \;-\sqrt2, \;-1, \;-\sqrt2, \;\text{--} \]
Note where the turning points landed
Why: The sine peaks at a quarter turn and troughs at three quarters of a turn, so those are the two turning points of the cosecant.
\[ \left(\tfrac{\pi}{2}, 1\right) \text{ and } \left(\tfrac{3\pi}{2}, -1\right) \]
Figure (svg): One cycle of the cosecant over the interval from 0 to two pi, with the sine dashed behind it. Red dashed asymptotes stand at 0, pi and two pi. An upward U sits over the first half with its lowest point at pi over two comma 1, and a downward cap sits over the second half with its highest point at three pi over two comma negative 1.
Check the count of branches
Why: Three asymptotes cut the closed interval into two open pieces, so one cycle of the cosecant shows two full branches, against the secant's one full branch and two halves.
\[ (0, \pi) \text{ and } (\pi, 2\pi): \quad 2 \text{ branches} \]
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.5 Graphs of the Trigonometric Functions §10.5, p. 801 — the table for the fundamental cycle of cosecant
Exactly the same procedure as before, and the only real work is noticing where the new parent is zero. The sine vanishes at both ends of the interval and once in the middle, so there are three asymptotes here where the secant had two.
That changes the look of the picture quite a lot. One cycle of the cosecant shows two complete branches, a U and then a cap, whereas one cycle of the secant showed a half-branch, a full branch and another half-branch. The functions are the same shape; you are just looking at the cycle through a window placed differently.
The two turning points sit at the sine's peak and trough, at heights one and negative one, exactly as the fixed-point rule requires. Check that on the picture before moving on.
Concept
Figure (svg): The secant in blue and the cosecant in orange drawn on the same axes from just left of zero to just past two pi. The two curves have the same shape throughout; the cosecant's lowest point at pi over two comma 1 sits a quarter turn to the right of the secant's lowest point at 0 comma 1, and a horizontal arrow joins the two.
The sine is the cosine shifted by a quarter turn, so their reciprocals are shifted by that same quarter turn. The two graphs are one graph drawn in two places, which is why you only ever have to remember one shape.
\[ \sin(x) = \cos\left(x - \tfrac{\pi}{2}\right) \;\Longrightarrow\; \csc(x) = \sec\left(x - \tfrac{\pi}{2}\right) \]
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.5 Graphs of the Trigonometric Functions §10.5, p. 801 — secant and cosecant as phase shifts of each other
This is the payoff for all the work on the secant. You do not need a second shape in your head, only a second starting place.
The identity on the slide is the one you met when the sinusoids were introduced: the sine is the cosine shifted right by a quarter turn. Take reciprocals of both sides and the shift comes along untouched, because shifting the input of a function shifts the input of one over that function in exactly the same way.
Look at the figure and find the two marked lowest points. Both sit at height one; the only difference is that one is at the origin and the other a quarter turn to the right. Every other feature of the two curves is displaced by the same amount.
Worked example
Exercise 16 in the section's exercise set asks for one cycle of the function below. It turns out to be a function you have already drawn.
\[ y = \sec\left(x - \tfrac{\pi}{2}\right) \]
Write the secant as one over its parent
Why: Nothing is assumed about the shift yet.
\[ \sec\left(x - \tfrac{\pi}{2}\right) = \frac{1}{\cos\left(x - \tfrac{\pi}{2}\right)} \]
Expand the cosine of the difference
Why: Use the difference formula with cosine of a quarter turn equal to zero and sine of a quarter turn equal to one.
\[ \cos\left(x - \tfrac{\pi}{2}\right) = \cos x \cos\tfrac{\pi}{2} + \sin x \sin\tfrac{\pi}{2} = \sin x \]
Substitute
Why: The bottom was the sine all along.
\[ \sec\left(x - \tfrac{\pi}{2}\right) = \frac{1}{\sin x} = \csc(x) \]
Read off the graph
Why: So one cycle is the cosecant's cycle: asymptotes at the multiples of pi, turning points at a quarter and three quarters of a turn.
\[ \text{asymptotes } x = \pi k; \quad \text{period } 2\pi; \quad \text{no amplitude} \]
Check at two values
Why: At x equal to 1 the secant of 1 minus a quarter turn is 1.1884 and the cosecant of 1 is also 1.1884; at x equal to 4 both come out as negative 1.3213.
\[ \sec\left(1 - \tfrac{\pi}{2}\right) = 1.1884 = \csc(1), \qquad \sec\left(4 - \tfrac{\pi}{2}\right) = -1.3213 = \csc(4) \]
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.5 Graphs of the Trigonometric Functions §10.5.4, p. 809 — Exercise 16
This exercise looks as though it will need the full graphing procedure, and it does not, which is the lesson. Always check whether a shift has turned the function into one you already know.
The middle step is the difference formula for cosine, with the cosine of a quarter turn equal to zero and the sine of a quarter turn equal to one. Those two values annihilate the first term and leave the second, so the whole bottom collapses to the sine.
Once the bottom is the sine, the function is the cosecant, and you can write down everything about it without drawing a single point: asymptotes at the multiples of a half turn, turning points at heights one and negative one, period a full turn, no amplitude. The numerical check at the end is there because collapsing an identity is exactly the kind of step where a sign slips.
Matching
Match the pairs
Why: Every asymptote is a zero of the parent, so the question is always where the parent vanishes. Cosine vanishes at the odd multiples of a quarter turn, sine at the multiples of a half turn, and the doubled sine at half of those, which is every multiple of a quarter turn. The cosine itself is defined everywhere and so has no asymptotes at all.
Match these by asking, every time, which function is on the bottom and where that function is zero. The answer never depends on anything else.
The fourth one is the only one with any work in it. The parent is the sine of twice the input, and that is zero when twice the input is a multiple of a half turn, so the input itself is a multiple of a quarter turn. Doubling the input halves the spacing of everything, asymptotes included.
The cosine is in the list as a reminder that having no asymptotes is also an answer. A function is only in danger of an asymptote where something it is divided by vanishes, and the cosine is not a quotient of anything.
Section
Part 4
Concept
Theorem 10.24 — Secant and cosecant are each continuous and smooth on their domain, each has range all values of size at least one, and each has period two pi. The secant is even and its domain omits the odd multiples of a quarter turn; the cosecant is odd and its domain omits the multiples of a half turn.
\[ \text{dom}(\sec) = \left\{x : x \ne \tfrac{\pi}{2} + \pi k\right\} = \bigcup_{k=-\infty}^{\infty} \left(\tfrac{(2k+1)\pi}{2}, \tfrac{(2k+3)\pi}{2}\right) \]
\[ \text{dom}(\csc) = \{x : x \ne \pi k\} = \bigcup_{k=-\infty}^{\infty} \left(k\pi, (k+1)\pi\right) \]
The two ways of writing each domain say the same thing. The set description removes the bad points one at a time; the union collects up the open intervals that survive, which are exactly the stretches between consecutive asymptotes.
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.5 Graphs of the Trigonometric Functions §10.5, p. 802 — Theorem 10.24
Theorem 10.24 is a summary rather than a new result. Every line of it has already been derived, or is about to be on the next few slides, and the point of collecting them is that they are all consequences of one division.
The two ways of writing each domain are equivalent, and it is worth being able to move between them. The set-builder version removes the bad inputs by naming them; the union version lists the good intervals that are left. The intervals are exactly the stretches between consecutive asymptotes, which is to say exactly the branches of the graph.
The cosecant's domain is easier to read at a glance, because the bad points are just the multiples of a half turn and the good intervals run from one multiple to the next. The secant's looks worse only because the odd multiples of a quarter turn need a formula to describe.
Notation
Annotate
On: \( \bigcup_{k=-\infty}^{\infty} \left(\frac{(2k+1)\pi}{2}, \; \frac{(2k+3)\pi}{2}\right) \)
\[ k = 0: \left(\tfrac{\pi}{2}, \tfrac{3\pi}{2}\right), \quad k = 1: \left(\tfrac{3\pi}{2}, \tfrac{5\pi}{2}\right), \quad k = -1: \left(-\tfrac{\pi}{2}, \tfrac{\pi}{2}\right) \]
Take the annotations in order, and read the whole expression as a recipe for listing intervals. Put in a whole number for k, work out the two endpoints, and you get one branch.
The trick inside is the expression two k plus one, which is odd for every whole number k, so it generates exactly the odd multiples of a quarter turn. Two k plus three is the next one up, one half turn further along, which is why each interval has the right width.
Set k to zero, then one, then negative one and see the three intervals listed at the bottom of the slide appear. Doing that once by hand is the fastest way to make this notation stop looking intimidating.
Concept
It is tempting to call these graphs broken. They are not broken; they are simply not defined at the asymptotes, and a function cannot fail to be continuous at a point where it has no value in the first place.
On each open interval between consecutive asymptotes the curve is a single unbroken piece with no corners, because the cosine and the sine are unbroken and cornerless everywhere and the reciprocal of a nonzero smooth function is smooth.
This is the same language used for rational functions back in Chapter 4: continuous and smooth on their domains. Saying it that way keeps the asymptote where it belongs, outside the domain.
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.5 Graphs of the Trigonometric Functions §10.5, p. 801 — footnote 13, and the remark above Theorem 10.24
Students often want to call these graphs discontinuous, and the word is understandable but wrong. A function can only fail to be continuous at a point of its domain, and the asymptotes are not in the domain at all.
On each branch the curve really is as well behaved as anything you have drawn: no jumps, no corners, no sudden changes of direction. That is inherited too, because the cosine and the sine are smooth everywhere, and dividing one by a smooth function that is not zero keeps things smooth.
The phrase to use, and the phrase the book uses, is continuous and smooth on their domains. It is the same language used for rational functions in Chapter 4, and it is precise: it tells you where the good behaviour holds and quietly excludes the places it could not possibly hold.
Worked example
Parity passes straight through a reciprocal, and the two-line proof is worth writing out once.
\[ \cos(-x) = \cos(x) \quad\text{(even)}, \qquad \sin(-x) = -\sin(x) \quad\text{(odd)} \]
Feed a negative input to the secant
Why: Replace x by negative x in the definition.
\[ \sec(-x) = \frac{1}{\cos(-x)} \]
Use the evenness of the cosine
Why: The bottom is unchanged, so the whole fraction is unchanged.
\[ \sec(-x) = \frac{1}{\cos(x)} = \sec(x) \]
Do the same for the cosecant
Why: Now the bottom picks up a minus sign, and a minus sign on the bottom comes out to the front.
\[ \csc(-x) = \frac{1}{\sin(-x)} = \frac{1}{-\sin(x)} = -\csc(x) \]
Figure (svg): Two panels. On the left the secant over one period either side of the origin, with two marked points at x equal to 1.1 and negative 1.1 at the same height, joined by a dashed horizontal line: mirror symmetry in the vertical axis. On the right the cosecant with two marked points at x equal to 1.1 and negative 1.1 at opposite heights, joined by a dashed line through the origin: half-turn symmetry about the origin.
Check at a specific angle
Why: At 1.1 radians the secant is 2.2046 and at negative 1.1 radians it is 2.2046 again, while the cosecant is 1.1221 and negative 1.1221.
\[ \sec(\pm 1.1) = 2.2046; \qquad \csc(1.1) = 1.1221, \; \csc(-1.1) = -1.1221 \]
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.5 Graphs of the Trigonometric Functions §10.5, p. 802 — Theorem 10.24, parity
Parity is one of the properties that passes straight through a reciprocal, and the proof is short enough to reconstruct in an exam rather than memorise.
Both halves start the same way, by feeding a negative input into the definition. For the secant the bottom does not change, because the cosine is even, so the whole fraction does not change. For the cosecant the bottom picks up a minus sign, and a minus sign on the bottom of a fraction is the same as a minus sign in front of it, so the function flips.
Look at the two panels. Even means the left half is a mirror image of the right half in the vertical axis, so the two marked points sit at the same height. Odd means the left half is the right half turned through half a turn about the origin, so the two marked points sit at opposite heights. Check each against the algebra you just did.
Worked example
The period is inherited from the parent, and the reason is an if-and-only-if rather than a picture.
\[ \text{Show: } \sec \text{ has period } 2\pi. \]
Show two pi is a period
Why: Shift the input by a full turn inside the definition and use the cosine's periodicity.
\[ \sec(x + 2\pi) = \frac{1}{\cos(x + 2\pi)} = \frac{1}{\cos(x)} = \sec(x) \]
Show nothing smaller works
Why: Where both are defined, equal secants force equal cosines, because taking reciprocals is reversible.
\[ \sec(\alpha) = \sec(\beta) \iff \cos(\alpha) = \cos(\beta) \]
Conclude from the parent
Why: A period smaller than a full turn for the secant would be a period smaller than a full turn for the cosine, and the cosine has none.
\[ \text{period}(\sec) = \text{period}(\cos) = 2\pi \]
Check against the graph
Why: One cycle of the secant runs from an upward branch at 0, through the downward branch centred at pi, back to an upward branch at two pi, and only then does the picture repeat; the half-turn shift sends an upward branch to a downward one, so pi is not a period.
\[ \sec(0) = 1 \ne -1 = \sec(0 + \pi) \]
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.5 Graphs of the Trigonometric Functions §10.5, p. 800 — footnote 11
Showing that a full turn is a period is the easy half and takes one line. The harder and more interesting half is showing that nothing smaller works, which needs the if-and-only-if in the middle step.
Read that step carefully. Two secant values are equal exactly when the two cosine values are equal, because taking reciprocals is reversible wherever both sides are defined. So any repetition in the secant is a repetition in the cosine, and the cosine has none shorter than a full turn.
The final check gives a concrete reason to reject the obvious wrong guess. A half turn looks plausible from the picture, because the asymptotes come half a turn apart. But a half-turn shift sends the upward branch at the origin onto the downward branch at a half turn, and one is not negative one, so a half turn is not a period.
Comparison
Comparison matrix
| property | cosine | secant |
|---|---|---|
| domain | all real numbers | all except the odd multiples of π/2 |
| range | from −1 to 1 | everything of size at least 1 |
| period | 2π | 2π |
| parity | even | even |
| zeros | odd multiples of π/2 | none |
| amplitude | 1 | none: the graph is unbounded |
Fill every blank from what you know rather than from the earlier slides. Each entry on the right is a consequence of the entry on its left, and being able to say which consequence is the real test.
Two rows are the ones people get wrong. The domain shrinks, because the parent's zeros drop out. The range turns inside out, because reciprocation swaps small for large. Meanwhile the period and the parity simply pass through untouched.
The last two rows are the ones worth remembering as a pair. The parent's zeros become no zeros at all, and the parent's amplitude becomes no amplitude at all. Both losses come from the same place: the graph is now unbounded and never reaches the axis.
Error analysis
Annotate
On: \( y = 3\sec(2x) + 1: \quad \text{amplitude } 3, \; \text{period } \pi, \; \text{midline } y = 1 \)
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.5 Graphs of the Trigonometric Functions §10.5, p. 804 — the remark that secant and cosecant have no amplitude
Read the line first and decide which of the three claims you would have written yourself. Two of them are wrong and one is right, and the right one may not be the one you expected.
Amplitude is defined as half the distance from the lowest point to the highest point. This graph has no highest point, so the definition has nothing to work with. Saying the amplitude is three is not slightly off; it is a category error.
The number three is still doing something, and naming it correctly is the fix. It is a vertical stretch by a factor of three, and you can see it in the turning points, which move out to heights three and negative three. The period is the one claim that survives, because the period depends only on the argument and the argument is a sinusoid's argument like any other.
Trap
A very common first line:
\[ y = \csc(x): \quad x = \tfrac{\pi}{2} + \pi k \;? \]
Wrong. That is where the cosecant is at a turning point, not where it is undefined.
Ask which parent is on the bottom. The cosecant is one over the sine, so its asymptotes are the zeros of the sine, at the multiples of a half turn. A quarter turn is where the sine peaks, and a peak of the parent is a perfectly good point of the reciprocal.
\[ \csc\left(\tfrac{\pi}{2}\right) = \frac{1}{1} = 1 \]
Under time pressure the two functions blur together, and the odd multiples of a quarter turn are the more memorable list, so they get used for both. The result is a graph with the branches in exactly the wrong places.
The one-question defence is: what is on the bottom? Cosecant is one over the sine, so the sine's zeros are the asymptotes, and those are the multiples of a half turn. A quarter turn is where the sine peaks, and a peak of the parent is a perfectly ordinary point of the reciprocal, at height one. Evaluating the cosecant there and getting a number is proof enough that it is not an asymptote.
Section
Part 5
Intuition
For a transformed secant or cosecant there is a procedure, and it is the quarter-mark procedure from the last lesson with one extra rule bolted on.
Set the argument equal to each of the five quarter marks and solve for x, exactly as before. Then substitute each of those five values back into the function. Where the value exists you have a point; where it does not you have found an asymptote.
Sketching the associated sinusoid, dotted, costs almost nothing and makes the rest automatic: its zeros are your asymptotes, its peaks and troughs are your turning points, and its shape tells you which way every branch opens.
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.5 Graphs of the Trigonometric Functions §10.5, p. 802 — the method of Example 10.5.4
The procedure is the quarter-mark method from the last lesson, with one extra branch in the middle: after you solve for the five values of x, substituting can now fail, and a failure is information rather than a dead end.
Resist the urge to work with transformations one at a time. Setting the argument equal to each quarter mark and solving handles the period, the phase shift and any reflection in one step, and it does not care whether the coefficient of x is negative.
The dotted sinusoid is the part students skip and then regret. Once it is on the page, its zeros tell you where the asymptotes go, its peaks and troughs tell you where the turning points go, and its shape tells you which way every branch opens. You are no longer sketching from memory; you are tracing.
Worked example
Graph one cycle and state the period.
\[ f(x) = 1 - 2\sec(2x) \]
Set the argument equal to the quarter marks
Why: Five equations in x, one per quarter mark of the parent cycle.
\[ 2x = 0, \;\tfrac{\pi}{2}, \;\pi, \;\tfrac{3\pi}{2}, \;2\pi \]
Solve each for x
Why: Divide every quarter mark by two.
\[ x = 0, \;\tfrac{\pi}{4}, \;\tfrac{\pi}{2}, \;\tfrac{3\pi}{4}, \;\pi \]
Substitute the first, third and fifth
Why: At those three the secant of the argument is plus or minus one.
\[ f(0) = 1 - 2(1) = -1, \quad f\left(\tfrac{\pi}{2}\right) = 1 - 2(-1) = 3, \quad f(\pi) = -1 \]
Substitute the second and fourth
Why: There the argument is an odd multiple of a quarter turn, so the secant does not exist.
\[ x = \tfrac{\pi}{4}, \tfrac{3\pi}{4}: \quad f(x) \text{ undefined, asymptote} \]
Read off the period
Why: One cycle was graphed from the first quarter mark to the last.
\[ \text{period} = \pi - 0 = \pi \]
Figure (svg): One cycle of y equals 1 minus 2 secant of 2x over the interval from 0 to pi, with the cosine curve y equals 1 minus 2 cosine of 2x dashed behind it. Asymptotes stand at pi over four and three pi over four. The outer two half-branches fall away from the points 0 comma negative 1 and pi comma negative 1, and the middle branch is an upward U with its lowest point at pi over two comma 3.
Check the direction of the outer branches
Why: At 0.1 radians the value is negative 1.09, below the point at the origin, so the branch really does fall away rather than rise: the minus two has turned the usual picture over.
\[ f(0.1) = 1 - 2\sec(0.2) = -1.0405 < -1 \]
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.5 Graphs of the Trigonometric Functions §10.5, pp. 802-803 — Example 10.5.4, part 1
This is the book's own worked example and it is the model answer to copy. Notice the order: quarter marks first, then solve, then substitute, and only then draw.
The five values of x come out evenly spaced a quarter of the period apart, which is always true and is a useful check that you solved correctly. Substituting the first, third and fifth gives numbers because the secant of a quarter mark is plus or minus one there; the second and fourth land on an odd multiple of a quarter turn, so they give asymptotes.
Pay attention to the minus two in front. It flips the picture over, so the branch that would ordinarily open upward at the origin now opens downward, and the branch that would open downward now opens upward with its lowest point at height three. The final check does that flip numerically, so you can see it rather than take it on trust.
Tweak it
Parameter explorer
The curve is the secant of a multiple of x. The picture has no scale, so judge by shape only: as you turn the frequency up, what happens to the number of branches in the window, and does any branch ever dip into the empty strip through the middle?
\[ y = \sec({w}x) \]
Start with the frequency at one, which is the plain secant, and then push it upward slowly. The plot has no axis labels, so do not try to read heights off it; watch the shape instead.
Two things happen and one does not. The branches multiply and crowd together, because raising the frequency shortens the period and so brings the asymptotes closer. The width of every branch shrinks in proportion.
What never changes is the empty strip running through the middle. No branch ever dips into it at any frequency, because the frequency only rescales the input, and the strip is a statement about outputs. That is the cleanest way to see that period and range are independent of each other.
Step zero
\[ g(x) = \frac{\csc(\pi - \pi x) - 5}{3} \]
Discussion prompt
Do not graph anything yet. Set the argument of the cosecant equal to each of the five quarter marks and solve for x. What is strange about the list you get, and what does that tell you before you plot a single point?
Write the five values down before revealing, and resist the urge to graph anything. This slide exists to slow you down for about thirty seconds at the point where most errors are made.
When you have them, look at the order. The largest quarter mark of the argument gave the smallest value of x, and the smallest gave the largest. That reversal is the negative coefficient of x showing itself, and it means the cycle you are about to draw runs from negative one to one rather than the other way.
The period falls out of the same list, as the distance from the smallest value to the largest, which is two. Knowing the period before you have plotted a point is a good position to be in, because it tells you immediately whether your finished picture is the right width.
Worked example
Graph one cycle and state the period.
\[ g(x) = \frac{\csc(\pi - \pi x) - 5}{3} \]
Use the quarter marks found on the last slide
Why: Five values of x, listed in increasing order this time.
\[ x = -1, \;-\tfrac12, \;0, \;\tfrac12, \;1 \]
Substitute the values where the argument is a multiple of pi
Why: The sine is zero there, so the cosecant does not exist.
\[ x = -1, 0, 1: \quad g(x) \text{ undefined, asymptote} \]
Substitute x equal to one half
Why: The argument is a quarter turn, where the cosecant equals one.
\[ g\left(\tfrac12\right) = \frac{\csc\left(\tfrac{\pi}{2}\right) - 5}{3} = \frac{1 - 5}{3} = -\frac43 \]
Substitute x equal to negative one half
Why: The argument is three quarters of a turn, where the cosecant equals negative one.
\[ g\left(-\tfrac12\right) = \frac{-1 - 5}{3} = -2 \]
Read off the period
Why: From the smallest quarter mark to the largest.
\[ \text{period} = 1 - (-1) = 2 \]
Figure (svg): One cycle of y equals the quantity cosecant of pi minus pi x, minus 5, all over 3, over the interval from negative 1 to 1, with its sine curve dashed behind it. Asymptotes stand at negative 1, 0 and 1. The branch between 0 and 1 opens upward from the point one half comma negative four thirds, and the branch between negative 1 and 0 opens downward from the point negative one half comma negative 2.
Check against the associated sine curve
Why: The dotted curve is the same formula with the sine in place of the cosecant, so it runs between negative two and negative four thirds; the two turning points of the cosecant sit exactly on its trough and its peak, as they must.
\[ \frac{\sin(\pi - \pi x) - 5}{3} \in \left[-2, -\tfrac43\right] \]
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.5 Graphs of the Trigonometric Functions §10.5, p. 803 — Example 10.5.4, part 2
With the quarter marks already in hand this is mostly substitution, but each substitution is worth doing slowly, because there is a fraction outside the cosecant as well as a shift.
Three of the five values of x send the argument to a multiple of a half turn, where the sine is zero, so they give asymptotes. The other two send it to a quarter turn and to three quarters of a turn, where the cosecant is one and negative one, and the arithmetic outside converts those into negative four thirds and negative two.
The whole picture sits below the horizontal axis, which can be disorienting the first time. The reason is the vertical shift: dividing by three after subtracting five leaves a shift of negative five thirds, which drags everything down past the axis. The final check confirms that the two turning points land exactly on the dotted sine curve's trough and peak, which is where they must be.
Worked example
From the section's exercise set. Graph one cycle and state the period.
\[ y = \csc(2x - \pi) \]
Set the argument equal to the quarter marks
Why: The argument is the whole of two x minus pi, not just the two x.
\[ 2x - \pi = 0, \;\tfrac{\pi}{2}, \;\pi, \;\tfrac{3\pi}{2}, \;2\pi \]
Solve each for x
Why: Add pi, then halve.
\[ x = \tfrac{\pi}{2}, \;\tfrac{3\pi}{4}, \;\pi, \;\tfrac{5\pi}{4}, \;\tfrac{3\pi}{2} \]
Locate the asymptotes
Why: The first, third and fifth quarter marks are where the parent sine vanishes.
\[ x = \tfrac{\pi}{2}, \;\pi, \;\tfrac{3\pi}{2}: \quad \text{asymptotes} \]
Find the two turning points
Why: At the second and fourth the parent is at a peak and a trough.
\[ \left(\tfrac{3\pi}{4}, 1\right) \text{ and } \left(\tfrac{5\pi}{4}, -1\right) \]
Read off the period
Why: From the first quarter mark to the last.
\[ \text{period} = \tfrac{3\pi}{2} - \tfrac{\pi}{2} = \pi \]
Figure (svg): One cycle of y equals cosecant of the quantity 2x minus pi, from pi over two to three pi over two, with its sine curve dashed behind. Asymptotes stand at pi over two, pi and three pi over two, so one period of the curve is only pi wide. The first branch opens upward from three pi over four comma 1 and the second opens downward from five pi over four comma negative 1.
Check the period against the rule
Why: The period of a sinusoid is two pi divided by the coefficient of x, and that rule carries over unchanged to a reciprocal because it depends only on the argument.
\[ \frac{2\pi}{2} = \pi \;\checkmark \]
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.5 Graphs of the Trigonometric Functions §10.5.4, p. 809 — Exercise 19
Take the argument as a single object. The quarter marks are set equal to the whole of two x minus a half turn, not just to the two x, and getting that wrong is the most common slip in these problems.
Solving gives five values a quarter of a period apart, and the period turns out to be a half turn rather than a full one. That is the doubling inside showing up: two full turns of the argument are squeezed into one full turn of x.
The check at the end connects this to the rule you already have. The period of a sinusoid is a full turn divided by the coefficient of x, and that rule carries straight over to the reciprocal, because the period depends only on the argument and the argument is identical.
Worked example
Graph one cycle, and watch what the vertical shift does and does not move.
\[ y = \sec(3x - 2\pi) + 4 \]
Set the argument equal to the quarter marks
Why: The plus four sits outside and plays no part here.
\[ 3x - 2\pi = 0, \;\tfrac{\pi}{2}, \;\pi, \;\tfrac{3\pi}{2}, \;2\pi \]
Solve each for x
Why: Add two pi, then divide by three.
\[ x = \tfrac{2\pi}{3}, \;\tfrac{5\pi}{6}, \;\pi, \;\tfrac{7\pi}{6}, \;\tfrac{4\pi}{3} \]
Evaluate at the first, third and fifth
Why: The secant of the quarter mark is 1, negative 1, 1, and each value is then lifted by four.
\[ y = 1 + 4 = 5, \quad -1 + 4 = 3, \quad 1 + 4 = 5 \]
Mark the second and fourth as asymptotes
Why: The lift does not rescue an undefined value.
\[ x = \tfrac{5\pi}{6}, \tfrac{7\pi}{6}: \quad \text{asymptotes} \]
State the period and the range
Why: The period is two pi over three, and the excluded strip of heights moves up with the curve.
\[ \text{period} = \tfrac{2\pi}{3}, \qquad \text{range} = (-\infty, 3] \cup [5, \infty) \]
Figure (svg): One cycle of y equals secant of the quantity 3x minus two pi, plus 4, from two pi over three to four pi over three, with its cosine curve dashed. The excluded strip of heights, shaded, now runs from 3 to 5 rather than from negative 1 to 1, and the whole picture sits well above the horizontal axis. Asymptotes stand at five pi over six and seven pi over six.
Check that the asymptotes did not move
Why: Adding four changes every height by four, and an undefined height stays undefined; the asymptotes depend only on the argument, so they sit at exactly the same places as for the unshifted secant of three x minus two pi.
\[ \sec(3x - 2\pi) \text{ undefined} \iff \sec(3x - 2\pi) + 4 \text{ undefined} \]
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.5 Graphs of the Trigonometric Functions §10.5.4, p. 809 — Exercise 20
The plus four is outside the secant, so it plays no part at all in the first three steps. Ignore it until you are computing heights, and it will not confuse the search for the asymptotes.
When the heights do arrive, everything moves up by four together. The turning points go from one and negative one to five and three, and the strip of heights the curve cannot reach goes from between negative one and one to between three and five. The shading in the figure makes that move visible.
The check is the point of the whole example. Adding a constant changes every output and no input, and an asymptote is an input, so the asymptotes cannot have moved. If a vertical shift ever seems to move your asymptotes, you have put the shift inside the function by accident.
Worked example
One cycle of a secant with a fractional coefficient in front and a fractional coefficient inside.
\[ y = -\tfrac13\sec\left(\tfrac12 x + \tfrac{\pi}{3}\right) \]
Set the argument equal to the quarter marks
Why: Half of x plus a third of pi is the whole argument.
\[ \tfrac12 x + \tfrac{\pi}{3} = 0, \;\tfrac{\pi}{2}, \;\pi, \;\tfrac{3\pi}{2}, \;2\pi \]
Solve each for x
Why: Subtract pi over three, then double.
\[ x = -\tfrac{2\pi}{3}, \;\tfrac{\pi}{3}, \;\tfrac{4\pi}{3}, \;\tfrac{7\pi}{3}, \;\tfrac{10\pi}{3} \]
Evaluate at the first, third and fifth
Why: The secant there is 1, negative 1, 1, and each is multiplied by negative one third.
\[ y = -\tfrac13, \quad -\tfrac13(-1) = \tfrac13, \quad -\tfrac13 \]
Mark the other two as asymptotes
Why: The second and fourth quarter marks put the argument at an odd multiple of a quarter turn.
\[ x = \tfrac{\pi}{3}, \tfrac{7\pi}{3}: \quad \text{asymptotes} \]
State the period and the excluded heights
Why: The period is two pi divided by one half, and the stretch by one third shrinks the gap in the range.
\[ \text{period} = \frac{2\pi}{1/2} = 4\pi, \qquad \text{range} = \left(-\infty, -\tfrac13\right] \cup \left[\tfrac13, \infty\right) \]
Figure (svg): One cycle of y equals negative one third times secant of the quantity x over 2 plus pi over 3, running from negative two pi over three to ten pi over three, a period of four pi. Its cosine curve is dashed behind it, oscillating only between negative one third and one third. Asymptotes stand at pi over three and seven pi over three. The outer branches hang below the axis and the middle branch sits above it.
Check the flip against a nearby value
Why: Just to the right of the first quarter mark the value is negative 0.334, further below negative one third rather than above it, so the negative sign really has turned the outer branches downward.
\[ y(-2) = -\tfrac13\sec\left(-1 + \tfrac{\pi}{3}\right) = -0.3345 < -\tfrac13 \]
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.5 Graphs of the Trigonometric Functions §10.5.4, p. 809 — Exercise 18
Three separate things are happening here, and the quarter-mark method deals with all of them at once, which is exactly why it is preferred to applying transformations one at a time.
The half in front of the x stretches the graph horizontally, so the period is a full turn divided by one half, which is two full turns. The third in front of the secant squashes it vertically, so the turning points move in to heights one third and negative one third, and the excluded strip shrinks with them. The minus sign turns the whole thing over.
The final check tests the flip specifically, by evaluating a point just to the right of the first turning point and confirming that the curve has gone further below negative one third rather than climbing back toward it. That is the one feature of the picture that a sign error would silently destroy.
Worked example
The last of the exercise-set graphs, and the one where parity earns its keep.
\[ y = \csc\left(-x - \tfrac{\pi}{4}\right) - 2 \]
Pull the minus sign out of the argument
Why: Factor negative one out of the argument first; the cosecant is odd, so the minus sign comes out to the front.
\[ \csc\left(-x - \tfrac{\pi}{4}\right) = \csc\left(-\left(x + \tfrac{\pi}{4}\right)\right) = -\csc\left(x + \tfrac{\pi}{4}\right) \]
Rewrite the whole function
Why: Now the coefficient of x inside is positive one, which is the standard form.
\[ y = -\csc\left(x + \tfrac{\pi}{4}\right) - 2 \]
Set the argument equal to the quarter marks and solve
Why: Subtract a quarter of pi from each.
\[ x = -\tfrac{\pi}{4}, \;\tfrac{\pi}{4}, \;\tfrac{3\pi}{4}, \;\tfrac{5\pi}{4}, \;\tfrac{7\pi}{4} \]
Evaluate at the second and fourth
Why: Those are the turning points; the outer three are the asymptotes.
\[ y\left(\tfrac{\pi}{4}\right) = -1 - 2 = -3, \qquad y\left(\tfrac{5\pi}{4}\right) = -(-1) - 2 = -1 \]
Figure (svg): One cycle of y equals cosecant of the quantity negative x minus pi over 4, minus 2, from negative pi over four to seven pi over four, with its sine curve dashed. Asymptotes stand at negative pi over four, three pi over four and seven pi over four. The first branch opens downward to the point pi over four comma negative 3, the second opens upward to the point five pi over four comma negative 1.
Check the rewrite numerically
Why: At x equal to 1 the original formula gives negative 0.9134 and the rewritten one gives the same, so factoring the minus sign out changed nothing but the appearance.
\[ \csc\left(-1 - \tfrac{\pi}{4}\right) - 2 = -0.9134 = -\csc\left(1 + \tfrac{\pi}{4}\right) - 2 \]
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.5 Graphs of the Trigonometric Functions §10.5.4, p. 809 — Exercise 21
The first step is the one that matters, and it is worth learning as a reflex: when the coefficient of x inside is negative, factor the minus sign out of the whole argument before doing anything else.
Once it is factored out, the oddness of the cosecant lets it escape to the front, and you are left with a function in the standard form with a positive coefficient of x. From there it is the ordinary procedure, with a minus sign in front that turns the branches over and a shift of negative two that lowers everything.
The numerical check exists because that first step is where sign errors live. Evaluating both the original and the rewritten formula at the same input and getting the same number is cheap insurance, and it is the habit that separates a reliable answer from a hopeful one.
Ranking
Put in order
Put the steps of graphing one cycle of a transformed secant or cosecant into the order you would actually carry them out.
Why: The sinusoid comes first because it is the reference the rest of the work leans on. The quarter marks are set on the argument, not on x, so the equations come before their solutions. Only after substituting do you know which of the five is an asymptote, and only after that can the branches be drawn. The period is read off the finished cycle at the end.
Drag the steps into order before checking. Most of the sequence is forced, and the two places worth thinking about are where the sinusoid goes and where the period is read off.
The sinusoid goes first because everything afterwards refers back to it. The quarter marks are set on the argument, so the equations must be written before they can be solved, and you cannot know which of the five gives an asymptote until you have substituted.
The period comes last for a practical reason: by then you have the five values in front of you in order, and the period is just the distance from the first to the last. Computing it earlier from the coefficient is fine too, but doing it at the end doubles as a check on the whole list.
Error analysis
Annotate
On: \( y = \csc(x) + 3: \quad \text{asymptotes at } y = 3 \text{, range } [3, \infty) \)
Two different errors are stacked in this one line, and separating them is the exercise. One is about notation and one is about what a vertical shift actually does.
The notation error is writing an asymptote of these graphs with a y. An asymptote here is a vertical line, so it is described by an equation in x. Writing y equals something describes a horizontal line, and these curves do not have one.
The substantive error is thinking the shift moved anything horizontal. Adding three changes every height and no input, so the undefined points are still undefined and still in the same places. What the shift does move is the range, and the corrected range at the end of the annotations has two pieces, not one, because the original range had two.
Trap
Drawing the dotted sinusoid and then misreading it:
\[ y = 2\sec(x) \]
\[ \text{asymptotes at } x = 0, \pi, 2\pi \;? \]
Wrong. Those are the cosine's peaks and troughs, not its zeros.
The dotted curve is a map with two kinds of landmark. Its zeros are the asymptotes; its peaks and troughs are the turning points that lie on the graph. Mark them in two different ways as you draw.
\[ 2\sec(0) = 2, \quad 2\sec(\pi) = -2 \]
This one happens to people who have taken the advice about drawing the dotted sinusoid, which is frustrating, because the advice is good. The problem is reading the dotted curve with only one kind of landmark in mind.
The dotted curve carries two kinds of information and they must be marked differently. Its zeros are where the reciprocal disappears; its peaks and troughs are where the reciprocal touches it. Draw the asymptotes as vertical dashed lines and the turning points as filled dots, and the two can never be confused again. And if you are unsure which is which, evaluate: a peak of the parent gives a perfectly good number.
Section
Part 6
Concept
A sinusoid came with four numbers: amplitude, period, phase shift and vertical shift. Three of them still make sense for a secant or a cosecant, and one of them does not.
| parameter | still applies? | why |
|---|---|---|
| period | yes | it is decided by the argument alone, so it is two pi over the coefficient of x, exactly as before |
| phase shift | yes | same computation on the same argument: the cycle starts wherever the argument reaches zero |
| vertical shift | yes | every height moves by the same amount, and so does the strip of excluded heights |
| amplitude | no | there is no highest point and no lowest point, so there is no half-distance between them to measure |
The coefficient in front is still worth naming: it is a vertical stretch, it moves the two turning points of each cycle, and its sign decides which way the branches open. It is just not an amplitude.
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.5 Graphs of the Trigonometric Functions §10.5, p. 804 — the closing remark of §10.5.2
This table is the answer to the question students ask as soon as they meet these graphs, which is whether all the parameter work from the last lesson still applies.
Three of the four survive untouched, and for the same reason in each case: period and phase shift are determined entirely by the argument, and the argument is a sinusoid's argument, while a vertical shift moves every output by the same amount whatever the function is.
Amplitude is the casualty, and the reason is worth stating precisely rather than waving at. Amplitude is half the distance between the lowest and highest points, and these graphs have neither. What they do have is a coefficient in front, which stretches the curve and moves the turning points. Call it a stretch, report the turning points, and you have said everything that number does.
Worked example
A single formula, and a full description of its graph without drawing it.
\[ y = 4\csc\left(3x - \tfrac{\pi}{2}\right) + 1 \]
Read the period from the coefficient of x
Why: Two pi divided by three.
\[ \text{period} = \frac{2\pi}{3} \]
Find the phase shift
Why: Set the argument to zero and solve; the cycle begins there.
\[ 3x - \tfrac{\pi}{2} = 0 \;\Longrightarrow\; x = \tfrac{\pi}{6} \text{ (a shift right)} \]
Locate the asymptotes
Why: The parent sine is zero when the argument is a multiple of pi.
\[ 3x - \tfrac{\pi}{2} = \pi k \;\Longrightarrow\; x = \tfrac{\pi}{6} + \tfrac{\pi}{3}k \]
Find the range
Why: The bare cosecant has size at least one, the four stretches that to at least four, and the plus one lifts it.
\[ |y - 1| \ge 4 \;\Longrightarrow\; y \le -3 \text{ or } y \ge 5 \]
Say what does not apply
Why: No amplitude and no midline, because the graph is unbounded above and below.
\[ \text{amplitude: none}, \qquad \text{vertical shift} = 1 \]
Check the range against one point
Why: At the first turning point after the start, a quarter of a period along, the argument is a quarter turn and the cosecant is one, so the value is five, which is the bottom of the upper piece of the range.
\[ x = \tfrac{\pi}{6} + \tfrac{\pi}{6} = \tfrac{\pi}{3}: \quad y = 4(1) + 1 = 5 \]
Stitz & Zeager, College Trigonometry, Ch. 10 Foundations of Trigonometry — §10.5 Graphs of the Trigonometric Functions §10.5, p. 804 — period, phase shift and vertical shift for these curves
This is the kind of question that appears on tests with no graph attached, and it can be answered entirely from the formula if you work through the parts in a fixed order.
Take the period from the coefficient of x, the phase shift from setting the argument to zero, and the asymptotes from setting the argument equal to the parent's zeros. Notice that the asymptote formula has the phase shift in it as the starting point and the period over the number of asymptotes per cycle as the spacing, which is a useful pattern to recognise.
The range takes a moment's thought because two things happen to it. The stretch by four multiplies the size threshold from one to four, and then the shift by one moves both pieces up. The final check evaluates the function at the first turning point after the start and confirms that the boundary value five is actually attained.
Sorting
Sort into buckets
Each row changes y = sec(x) in one way. Sort by whether the asymptotes move.
Sort these with one question: does the change touch the input or the output? Changes to the output can never move an asymptote, because an asymptote is a statement about which inputs are forbidden.
Replacing x by three x squeezes the whole input axis, so the asymptotes crowd together three times as closely. Replacing x by x minus a third of a turn slides them all along by that amount. Those are the two genuine movers.
Replacing x by negative x is the one worth thinking hardest about. It does touch the input, so it does move each individual asymptote, to its mirror image. But the secant's asymptote set is symmetric about the origin already, so the set as a whole lands back on itself and nothing appears to have changed. That is a nice reminder that a symmetric set can be moved without looking moved.
Counterexample
Discussion prompt
Someone generalises from this lesson: taking one over a function always produces vertical asymptotes. Give a counterexample built from functions in this chapter, and then say precisely what the correct statement is.
Write your counterexample before revealing. Overgeneralising from a lesson is a natural mistake and catching yourself doing it is a skill worth practising.
The neatest counterexample is sitting in this lesson. The secant is a function, and one over it is the cosine, which has no asymptotes anywhere. The claim collapses immediately.
The second example shows what the real condition is. Lifting the cosine by two gives a function that stays between one and three and so never reaches zero, and its reciprocal is defined everywhere and bounded. The correct statement is conditional: a reciprocal has an asymptote at each zero of the original, and a function with no zeros gives a reciprocal with none.
Real world
Figure (svg): Left: a lamp below a long straight horizontal wall, with five beams drawn at increasing angles from the perpendicular; each lit spot lies further along the wall than the last, and the beams themselves grow longer and longer. Right: the beam length plotted against the angle, a curve that starts at the perpendicular distance and runs off the top of the frame as the angle approaches a right angle.
Discussion prompt
A lamp sits a fixed distance d from a long straight wall and swings through an angle measured from the perpendicular. Write the length of the beam from the lamp to the lit spot in terms of d and the angle, and say what the asymptote means physically.
OpenStax Algebra and Trigonometry 2e, §8.2 Graphs of the Other Trigonometric Functions §8.2 — the same reciprocal graphs in OpenStax
Work out the formula before revealing. The right triangle is drawn for you: the distance to the wall is the side next to the angle, and the beam is the slanted side.
The cosine of the angle is the near side over the slanted side, so rearranging gives the beam length as the distance divided by the cosine, which is the distance times the secant. So a lamp turning at a steady rate traces out a secant curve as the length of its own beam, which is about as physical as this function gets.
The asymptote has a plain meaning here. As the angle approaches a right angle the beam becomes parallel to the wall and never meets it, so there is no lit spot and no length to report. Undefined is not a mathematical inconvenience in this picture; it is a real thing failing to happen.
Section
Part 7
Pattern
Figure (svg): A flow diagram of the graphing procedure: name the parent sinusoid, set the argument equal to the five quarter marks and solve, then split into two branches. Quarter marks where the parent is zero become vertical asymptotes; quarter marks where the parent is at a peak or a trough become points on the graph. Both feed into a final box: draw one branch between consecutive asymptotes.
This is the order to work in every time, and the diagram is the same order drawn. Notice that the whole procedure forks at exactly one point, when you substitute the quarter marks back in, and the fork is what makes these graphs different from sinusoids.
The step people are most tempted to skip is the first one, naming and sketching the parent. It is also the step that makes the last one possible: you cannot reliably decide which way a branch opens without something to look at.
The final line is the one to carry into an exam. Report period, phase shift and vertical shift, and do not report an amplitude. If a question asks for one, the correct answer is that the function does not have one, together with a one-sentence reason.
Check
Check your understanding
Where are the vertical asymptotes of y = csc(2x)?
Answer: A
Why: The cosecant is undefined where its parent sine is zero, so solve sin(2x) = 0. That gives 2x = πk, hence x = πk/2: every multiple of a quarter turn. Doubling the input halves the spacing of the asymptotes.
The question is about the parent, as always. Write down what the parent is, set it equal to zero, and solve; do not try to reason from the shape of the cosecant graph.
Two of the wrong answers are the right answers to nearby questions, which is why they are tempting. One of them gives the asymptotes of the plain cosecant, before the doubling; another gives the asymptotes of the secant, whose parent is the cosine instead.
The remaining wrong answer puts the asymptotes where the turning points are. If you chose it, evaluate the function at one of those inputs: you get a number, so it cannot be an asymptote.
Check
Check your understanding
What is the range of y = −4 csc(x) + 1?
Answer: B
Why: The bare cosecant has size at least 1. Multiplying by −4 makes the size at least 4, so the values are at most −4 or at least 4; the minus sign only swaps which branch is which. Adding 1 slides both pieces up by one, giving at most −3 or at least 5.
Build the range in the same order the function is built, from the inside out. Start with the bare cosecant, whose values have size at least one, then apply the stretch, then the shift.
The minus sign in front needs a moment's care but does less than it looks. Multiplying by negative four makes the size at least four, and the minus only swaps which branch of the graph is which; the set of values is unchanged by that swap. Then adding one slides both pieces up by one.
The most tempting wrong answer is the interval the graph avoids rather than the values it takes. Whenever an answer to a range question about these functions is a bounded interval, something has gone wrong: these curves are unbounded in both directions.
Check
Check your understanding
What does sec(x) do as x approaches 3π/2 from below?
Answer: B
Why: Just below three quarters of a turn the angle is in quadrant three, where the cosine is negative, so the cosine approaches zero from below. One over a small negative number is a large negative number, so the secant runs to negative infinity.
Answer this the way the worked example did. Name the quadrant, recall the sign of the parent there, and then reciprocate. Do not try to remember the picture.
Just below three quarters of a turn you are in the third quadrant, where both coordinates are negative, so the cosine is negative and approaching zero from below. One over a small negative number is a large negative number.
If you picked the other infinity, you have almost certainly described the far side of the same asymptote. Just past three quarters of a turn you are in the fourth quadrant, where the cosine is positive again, and there the branch does run upward.
Explain it to yourself
Discussion prompt
In three or four sentences, explain to someone who has drawn a cosine but never a secant why the secant graph looks the way it does. Use the words zero, asymptote, peak and turning point, and do not describe the picture without giving a reason for it.
Write your explanation out in full before revealing, and time yourself. Being able to give this account in under a minute is a good sign that the lesson has landed.
The strongest versions name a cause for every feature. The asymptotes are caused by the parent's zeros; the turning points are caused by one and negative one being their own reciprocals; the bending away from the axis is caused by every other parent value being smaller than one in size.
The sign point is the one most often left out and it does real work. Because reciprocation never changes a sign, a branch that starts above the axis stays above it for its whole length, which is why branches never wander across the middle of the picture.
Exit ticket
\[ y = 2\sec\left(\pi x - \tfrac{\pi}{2}\right) \]
Discussion prompt
Give the period, the locations of all the vertical asymptotes, the coordinates of the turning point nearest the origin with a positive first coordinate, and the range. Then say which of the four sinusoid parameters does not apply, and why.
This one problem uses every part of the lesson, so work through it in the order the pattern slide gave rather than jumping to whichever piece looks easiest.
The coefficient of x is a half turn's worth per unit, so the period is two, which is the kind of number that feels wrong the first time and is not. Setting the argument equal to the cosine's zeros gives asymptotes at the whole numbers, and setting it to zero gives the first turning point at one half, at height two because of the stretch.
The last part is the conceptual one. Amplitude does not apply, and the reason must be given: there is no highest point and no lowest point, so there is no half-distance between them to measure. The two in front is a vertical stretch, and its effect is visible in the turning points rather than in any amplitude.
Recap
| feature of the reciprocal | where it comes from | secant | cosecant |
|---|---|---|---|
| vertical asymptotes | the zeros of the parent | odd multiples of π/2 | multiples of π |
| turning points | the peaks and troughs of the parent | at heights 1 and −1 | at heights 1 and −1 |
| range | reciprocation throws small values out | size at least 1 | size at least 1 |
| period | inherited unchanged | 2π | 2π |
| parity | inherited from cosine or sine | even | odd |
| amplitude | needs a highest and a lowest point | none | none |
\[ \sec(x) = \frac{1}{\cos(x)}, \qquad \csc(x) = \sec\left(x - \tfrac{\pi}{2}\right) \]
Next, the tangent and the cotangent. They also have asymptotes, but they break two of the habits built here: their period is a half turn rather than a whole one, and each branch runs straight past every height instead of turning around.
OpenStax Algebra and Trigonometry 2e, §8.2 Graphs of the Other Trigonometric Functions §8.2 — Graphs of the Other Trigonometric Functions
Every row of the table has the same first move behind it. Ask what the parent is doing, then turn the answer over. That single habit replaces memorising two graphs.
Two things are inherited unchanged, the period and the parity, and two things are transformed, the domain and the range. One thing is lost entirely, the amplitude, because an unbounded graph has no top and no bottom. The identity at the bottom of the slide says the rest: the cosecant is the secant moved a quarter turn, so there was only ever one picture here.
Next lesson the tangent and the cotangent arrive, and they will break two of the habits you have just built. Their period is a half turn rather than a whole one, and their branches do not turn around at all: each one climbs or falls past every height between one asymptote and the next. What does carry over is the first move, which is still to find where the bottom of the fraction is zero.
Want this taught 1-on-1? Alexander tutors Trigonometry — $55/session, free consultation.