11.6a Rotation of Axes

The conics again, this time tilted. Derives the rotation equations from polar coordinates and the sum formulas, uses them to convert both points and equations into a rotated frame, and finds the angle that eliminates the cross term — the one whose cotangent of twice it equals A minus C over B. Closes with the discriminant, which classifies a general second-degree equation as hyperbola, parabola or ellipse without performing any rotation at all.

Subject: Trigonometry · 61 slides · symbolic lesson

Open the interactive version of this deck

What this lesson covers

The lesson, slide by slide

1. Rotation of Axes

Title

Trigonometry · Chapter 11 — Applications of Trigonometry

Turning the frame until the cross term disappears

2. What this lesson gives you

Objectives

Chapter 7 handled conics whose axes lined up with the coordinate axes. A second-degree equation with a product of the two variables in it describes the same four shapes, tilted, and none of the Chapter 7 machinery fits. This lesson turns the frame instead of the curve.

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.6 Hooked on Conics Again §11.6, pp. 973-981 — learning targets for section 11.6.1

Everything you did with conics in Chapter 7 assumed the curve was lined up with the axes: an ellipse whose long axis ran along one of them, a parabola opening straight up or straight sideways. Tilt the curve and the equation grows a term with both variables multiplied together, and every standard form you know stops fitting.

The idea of this lesson is that you do not have to straighten the curve. You straighten the frame. Turn both axes through exactly the right angle and the tilted curve becomes a standard one, measured against the new axes, with the awkward product term gone.

Two things make that practical. First, a short derivation in polar coordinates gives you formulas for moving between the two frames. Second, one small equation tells you which angle to turn through. The last part of the lesson then gives you a shortcut that skips the rotation entirely when all you want is the name of the shape.

3. A curve you already know, in disguise

Warm-up

Discussion prompt

Sketch the graph of x times y equals 2 from a table of values. What shape is it, and can you match it against any standard conic equation from Chapter 7?

Figure (svg): The two branches of the curve x times y equals 2, with a dashed pair of axes turned by forty-five degrees. The branches sit symmetrically about the turned x-axis, with vertices marked at about 1.41, 1.41 and its opposite.

Turn the axes through forty-five degrees and the same curve satisfies a standard hyperbola equation. Nothing moved; the description changed.

Plot a few points before reading on: one and two, two and one, four and one half, and their negatives. The picture is unmistakably a hyperbola, with the two coordinate axes as asymptotes.

Now try to write it in a Chapter 7 standard form and you will fail, because every standard hyperbola equation has a squared term in each variable and this one has neither. Nothing is wrong with the curve; the frame is simply at the wrong angle to it.

In the picture, the dashed axes are the original ones turned through forty-five degrees. Against those axes the curve is a perfectly ordinary hyperbola with its vertices on the turned first axis. By the end of this lesson you will be able to produce that turned equation yourself, for this curve and for any other with a product term.

4. Two frames for one point

Section

Part 1

5. Rotating the axes relabels points; it does not move them

Concept

Keep the origin where it is and turn both axes counter-clockwise through an angle. Every point in the plane now has two names: the pair measured along the original axes and the pair measured along the turned ones.

Figure (svg): Original x and y axes with a second, dashed pair of axes turned counter-clockwise by thirty degrees. A ray from the origin reaches a point P; the angle from the original x-axis to the turned x-axis is theta, and the angle from the turned x-axis to the ray is phi.

The point never moves. Only the pair of axes measuring it does, and the two angles differ by exactly the rotation.

Nothing about the point has changed. Its distance from the origin is the same, and only the angle is measured from a different starting line, differing by exactly the rotation.

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.6 Hooked on Conics Again §11.6, p. 973 — the set-up before Theorem 11.9

This is the point to be completely clear about before any algebra. Nothing in the plane moves. You are drawing a second pair of axes through the same origin and asking what numbers the same, stationary point gets when it is measured against those.

In the picture, the point sits at the end of the ray. The angle marked at the origin between the original first axis and the dashed one is the rotation. The second angle, between the dashed axis and the ray, is what the point's direction looks like from the new frame.

Notice the one quantity that both frames agree about: the distance from the origin to the point. The origin did not move, so that distance cannot change. That single shared quantity is what makes the derivation on the next slides short.

6. Polar coordinates separate what changes from what does not

Intuition

Doing this in rectangular coordinates means a page of similar triangles. In polar coordinates it takes four lines, because polar coordinates split the description of a point into a distance and an angle, and a rotation touches only the angle.

\[ \text{same origin} \;\Longrightarrow\; \text{same } r; \qquad \text{turned axes} \;\Longrightarrow\; \text{the angle drops by } \theta \]

So the two descriptions share the distance and differ by a known amount in the angle. The sum formulas for sine and cosine are built to unpick exactly that.

If you try to do this with rectangular coordinates directly you end up building similar triangles and chasing signs through four cases depending on which quadrant the point is in. It can be done, and it is unpleasant.

Polar coordinates avoid all of that because they describe a point with exactly the two pieces you need: a distance, which the rotation leaves alone, and an angle, which the rotation changes by a known fixed amount. So the two descriptions of the point differ in one controlled way.

That is the moment where the sum formulas for sine and cosine become the right tool. They exist precisely to expand a trigonometric function of an angle that is written as one angle plus another, which is exactly the shape of what you have.

7. Deriving the first rotation equation

Worked example

Call the angle from the turned x-axis to the point phi, so the angle from the original x-axis is the rotation plus phi.

\[ x = r\cos(\theta + \varphi), \qquad y = r\sin(\theta + \varphi) \]

Describe the same point in the turned frame

Why: Match the polar axis with the turned x-axis and keep the same distance, since the origin is shared.

\[ x' = r\cos(\varphi), \qquad y' = r\sin(\varphi) \]

Expand the first with the sum formula for cosine

Why: The angle inside is a sum, which is what the formula is for.

\[ x = r\cos(\theta)\cos(\varphi) - r\sin(\theta)\sin(\varphi) \]

Regroup so each distance sits with a function of phi

Why: Only the brackets change; the product is the same.

\[ x = \bigl(r\cos(\varphi)\bigr)\cos(\theta) - \bigl(r\sin(\varphi)\bigr)\sin(\theta) \]

Recognise the two brackets

Why: They are exactly the turned coordinates named two lines ago.

\[ x = x'\cos(\theta) - y'\sin(\theta) \]

Run the same three moves on the sine

Why: Sum formula, regroup, recognise.

\[ y = x'\sin(\theta) + y'\cos(\theta) \]

Check with no rotation at all

Why: Put the angle to zero: the cosine is one and the sine is zero, so each formula collapses to the statement that the two frames agree, which is what a zero rotation should do.

\[ \theta = 0: \quad x = x'(1) - y'(0) = x', \qquad y = x'(0) + y'(1) = y' \]

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.6 Hooked on Conics Again §11.6, pp. 973-974 — the derivation of Theorem 11.9

Follow the four middle lines closely; they are the whole theorem. You write the old first coordinate in polar form, expand with the sum formula, regroup the four factors into two pairs, and then notice that each pair is one of the new coordinates.

The regrouping step is the one that looks like nothing is happening. It is just the associative law: the product of the distance, a cosine of the rotation and a cosine of the second angle can be bracketed either way. The reason for bracketing it the second way is that it makes the new coordinates appear.

The check at the end costs nothing and catches sign errors. Put the rotation angle to zero, which means not turning the axes at all, and each formula must say that the old coordinate equals the new one. If a sign has slipped anywhere, this test fails immediately.

8. Theorem 11.9: the two systems

Concept

Rotation of axes (Theorem 11.9) — If the positive x and y axes are turned counter-clockwise through an angle to produce the turned axes, then the two pairs of coordinates of any point are related by the two systems below.

\[ \begin{cases} x = x'\cos(\theta) - y'\sin(\theta) \\ y = x'\sin(\theta) + y'\cos(\theta) \end{cases} \qquad \begin{cases} x' = x\cos(\theta) + y\sin(\theta) \\ y' = -x\sin(\theta) + y\cos(\theta) \end{cases} \]

you want touse
convert an equation from the old variables to the newthe left system: substitute for x and y
find the new coordinates of a known pointthe right system
find the old coordinates of a point given in the turned framethe left system

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.6 Hooked on Conics Again §11.6, p. 974 — Theorem 11.9

Two systems, and the whole difficulty is remembering which is which. Look at where the primes are. In the left system the primed letters are on the right-hand side, so that system tells you old coordinates in terms of new ones. In the right system it is the other way round.

The table gives you the working rule. To convert an equation you are substituting for the unprimed letters, so you need them expressed in the primed ones: the left system. To convert a point whose ordinary coordinates you already know, you want the primed letters given in terms of the unprimed ones: the right system.

Notice also that the two systems differ only in the placement of the minus sign, and that the right system is the left one with the angle replaced by its negative. That is not a coincidence, and the next slide but one proves it.

9. Reading the old-to-new system piece by piece

Notation

Annotate

On: \( x' = x\cos(\theta) + y\sin(\theta) \)

  • The coordinate measured along the TURNED x-axis. The prime always means: read this off the new axes.
  • How much of the old x-coordinate points along the new x-axis. At a zero rotation this is all of it.
  • The old y-coordinate also leans along the new x-axis once the axes have turned, and this term is that share. It is a PLUS because the turn was counter-clockwise.
  • The rotation angle, not the angle of the point. It is the same number in all four formulas.

Work through the annotations in order. The important one is the third. It is tempting to think that the new first coordinate should depend only on the old first coordinate, but once the axes have turned, a point that lies entirely along the old second axis has a real, nonzero shadow along the new first axis. That shadow is the second term.

The sign in that term is the one thing to commit to memory, and the safest way is not to memorise it but to test it. Take the point one, zero and a turn of a right angle: the new first axis now points where the old second axis used to, so the point should have new coordinates zero and minus one. Check that the formulas give that.

The last annotation is worth a second of attention every time. The angle in these formulas is the rotation of the axes, a single fixed number for the whole problem. It has nothing to do with the direction of any particular point.

10. Inverting the system with a rotation matrix

Worked example

The left system gives the old coordinates in terms of the new. To go the other way, solve it.

\[ \begin{cases} x'\cos(\theta) - y'\sin(\theta) = x \\ x'\sin(\theta) + y'\cos(\theta) = y \end{cases} \]

Write the system as a matrix equation

Why: The unknowns are the two turned coordinates.

\[ AX' = X, \qquad A = \begin{bmatrix} \cos\theta & -\sin\theta \\ \sin\theta & \cos\theta \end{bmatrix} \]

Take the determinant

Why: Down the diagonal minus up the diagonal, and the result is the Pythagorean Identity.

\[ \det(A) = \cos^2\theta - (-\sin\theta)(\sin\theta) = \cos^2\theta + \sin^2\theta = 1 \]

Invert

Why: A two by two inverse swaps the diagonal entries, negates the other two, and divides by the determinant, which is one here.

\[ A^{-1} = \begin{bmatrix} \cos\theta & \sin\theta \\ -\sin\theta & \cos\theta \end{bmatrix} \]

Multiply out

Why: Read the two components off the product.

\[ \begin{bmatrix} x' \\ y' \end{bmatrix} = A^{-1}X = \begin{bmatrix} x\cos\theta + y\sin\theta \\ -x\sin\theta + y\cos\theta \end{bmatrix} \]

Check against the original matrix

Why: Replacing the angle by its negative flips the sign of both sines, because cosine is even and sine is odd, and that is precisely the inverse just computed: undoing a turn is turning back.

\[ A(-\theta) = \begin{bmatrix} \cos\theta & \sin\theta \\ -\sin\theta & \cos\theta \end{bmatrix} = A^{-1} \]

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.6 Hooked on Conics Again §11.6, p. 974 — the matrix derivation of the second system

You could solve this system by elimination, multiplying the first line by a cosine and the second by a sine and adding. The matrix route is cleaner and it produces a fact worth having: the matrix of a rotation has determinant one.

That determinant is the Pythagorean Identity in disguise, which is a pleasing place for it to turn up. Because it equals one rather than some awkward expression, the inverse is free: for a two by two matrix you swap the two diagonal entries, negate the other two, and divide by the determinant, and here the division does nothing.

The final line is the check and also the insight. Negating the angle flips the sign of both sines and leaves both cosines alone, because sine is odd and cosine is even. Doing that to the original matrix produces exactly the inverse, which says in symbols what you already believed: undoing a counter-clockwise turn is a clockwise turn of the same size.

11. What does turning and turning back do?

Prediction

\[ A(\theta)\,A(-\theta) = \;? \]

Predict first

Apply the old-to-new system with a given angle, then apply it again with the opposite angle. What is the overall effect on a point?

  • A turn through twice the angle
  • Nothing at all: every point keeps its coordinates
  • A reflection in the x-axis
  • It depends on where the point is

Correct: Nothing at all: every point keeps its coordinates

Why: The second system is the first with the angle negated, so the two are inverse to each other. Composing them returns every point to its original coordinates, which is the matrix statement that the product of the two matrices is the identity.

Commit to an option before revealing. Most people get this right by instinct, and the value of the slide is in connecting the instinct to the algebra.

Turning the axes one way and then the same amount back leaves you with the axes you started from, so every point must end with the coordinates it began with. In matrix language, the product of the two matrices is the identity matrix, which is another way of saying each is the inverse of the other.

If you were tempted by a turn through twice the angle, you were composing two turns in the same direction. That is also a true and useful fact, and it is worth checking for yourself: multiplying two rotation matrices gives the rotation matrix for the sum of the two angles, and the sum formulas fall out of the multiplication.

12. Which system does the job?

Sorting

Sort into buckets

Each task needs one of the two systems. Which one?

old to new: x′ = x cos θ + y sin θ
find the turned coordinates of the point (2, −4); decide whether the point (1, 1) lies on the turned x-axis
new to old: x = x′ cos θ − y′ sin θ
convert 21x² + 10√3xy + 31y² = 144 into the turned variables; find the original coordinates of the turned point (0, 3); write the turned equation of the line y = 3x; locate, in original coordinates, a vertex found in the turned frame
new
Each of these starts from a point whose ORIGINAL coordinates are known and asks for the turned ones, so the old-to-new system answers directly. For the last, compute the turned y and see whether it is zero.
old
Converting an equation means substituting for x and y, so you need x and y written in the turned variables, which is the new-to-old system. Turning a known primed point back into original coordinates uses the same system.

Sort all six before checking. The single question to ask of each task is: am I starting from unprimed information or from primed information? You always use the system whose left-hand side is the thing you want.

The equation items are the ones that catch people. Converting an equation feels like going from the old frame to the new one, so the instinct is to reach for the old-to-new system. But converting an equation means replacing every unprimed letter that appears in it, and for that you need unprimed letters written in terms of primed ones, which is the other system.

The last item is a small twist. Asking whether a point lies on the turned first axis is asking whether its turned second coordinate is zero, so you compute that one number with the old-to-new system and look at it.

13. Example 11.6.1(1): moving a point into the turned frame

Worked example

The axes are turned counter-clockwise through pi over three. Find the coordinates of the point two, minus four in the turned frame.

\[ P(x, y) = (2, -4), \qquad \theta = \tfrac{\pi}{3} \]

Pick the system that goes old to new

Why: The turned coordinates are wanted, so use the second pair.

\[ x' = x\cos\theta + y\sin\theta, \qquad y' = -x\sin\theta + y\cos\theta \]

Put the numbers into the first

Why: Cosine of pi over three is one half; sine is root three over two.

\[ x' = 2\cdot\tfrac12 + (-4)\cdot\tfrac{\sqrt3}{2} = 1 - 2\sqrt3 \]

Put them into the second

Why: The same two values, in the other places.

\[ y' = -2\cdot\tfrac{\sqrt3}{2} + (-4)\cdot\tfrac12 = -\sqrt3 - 2 \]

Write the decimals for the sketch

Why: Root three is about 1.7321.

\[ P(x', y') = \left(1 - 2\sqrt3,\; -2 - \sqrt3\right) \approx (-2.4641,\; -3.7321) \]

Figure (svg): The point two, minus four plotted on the original axes, with a dashed pair of axes turned by sixty degrees. Dashed perpendiculars run from the point to each turned axis, meeting them at the new coordinates about minus 2.46 and about minus 3.73.

Each new coordinate is read along a turned axis: drop a perpendicular onto it and measure. Both come out negative, which the picture makes obvious.

Check by converting back

Why: Feed the turned pair into the new-to-old system: the root threes cancel in pairs and the halves add to two, and the same computation on the second line returns minus four. The round trip lands on the point we started from.

\[ x = \left(1-2\sqrt3\right)\tfrac12 - \left(-2-\sqrt3\right)\tfrac{\sqrt3}{2} = \tfrac12 - \sqrt3 + \sqrt3 + \tfrac32 = 2 \]

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.6 Hooked on Conics Again §11.6, p. 975 — Example 11.6.1 number 1

This is the model solution for a point, and it is four lines long. Identify the system, substitute, simplify, and then, because the numbers are irrational, write the decimals so you can sketch.

Both new coordinates come out negative, and the picture explains why. The point sits low in the fourth quadrant of the original frame, but the axes have turned sixty degrees counter-clockwise, which swings the positive turned first axis up and to the right, well away from the point. Measured against the turned axes the point is behind the origin in both directions.

Copy the check. Converting back is the only verification that catches every possible error, including using the wrong system entirely, and it takes one line. Watch the root threes cancel: that cancellation is the signature of a correct round trip.

14. Turn the point (3, 5) through half a right angle

Fill the middle

The axes are turned counter-clockwise through pi over four, where both the cosine and the sine are root two over two. Fill in the two whole numbers.

Fill in the blanks

x' = 3\cos\tfrac82 + 5\sin\tfrac______ = ___\cdot\tfrac______, \qquad y' = -3\sin\tfrac______ + 5\cos\tfrac______ = ___\cdot\tfrac______

Why: Both trigonometric values are the same, so each line is just a combination of the two coordinates with the signs the system dictates: three plus five is eight, and minus three plus five is two.

\[ (x', y') = \left(4\sqrt2,\; \sqrt2\right) \approx (5.657,\; 1.414) \]

Fill in both blanks before checking. At forty-five degrees the arithmetic is as easy as it ever gets, because the cosine and the sine are the same number, so each coordinate is just a combination of three and five with the signs the system dictates.

That is worth noticing as a general fact about this angle: whenever you rotate through forty-five degrees, the new first coordinate is proportional to the sum of the old two and the new second coordinate to their difference. Half the textbook problems in this section use that angle.

If you wrote minus two for the second blank, you used the wrong order in the subtraction. The system puts the minus sign on the first coordinate, so it is minus three plus five, not three minus five.

15. Trap: pouring a point into the equation system

Trap

The trap

Asked for the turned coordinates of the point two, minus four, a common wrong first line is:

\[ x' = 2\cos\tfrac{\pi}{3} - (-4)\sin\tfrac{\pi}{3} \]

Wrong. That is the new-to-old system, with the point's coordinates poured into the primed slots.

The fix

The system carrying the minus sign expresses the old coordinates in terms of the new ones. A point whose old coordinates you already know needs the other system.

\[ x' = x\cos\theta + y\sin\theta \]

A quick guard: after any conversion, convert back. A wrong system will not return the point you started from.

This is far and away the most common error in the first half of this section, and it is not a slip of the pen. It happens because both systems have an unprimed letter on one side and primed letters on the other, so at a glance either looks like it could do the job.

The discipline that removes the error is to say out loud what you are given and what you want. Given the ordinary coordinates, wanting the turned ones, you need the system whose left-hand sides are the turned ones. Write that system down before you substitute anything.

And keep converting back. A round trip that does not return the original point is a failed calculation, and it costs one line to find that out rather than discovering it three pages later.

16. Rotating an equation

Section

Part 2

17. The cross term is the record of a mismatch

Concept

A conic in standard position has no product of the two variables in its equation. Turn the axes away from the curve and that product appears. So the presence of a cross term says one thing: the curve and the frame are out of line.

\[ Ax^2 + Bxy + Cy^2 + Dx + Ey + F = 0 \]

The plan is therefore to turn the frame back until the product vanishes, and then read the standard form off the result. This part does the turning with a given angle; the next part works out what the angle must be.

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.6 Hooked on Conics Again §11.6, p. 975 — the set-up for Example 11.6.1 number 2

Here is the diagnostic that organises the whole rest of the lesson. A conic in standard position has an equation with no product of the two variables in it. So if the product appears, the curve and the axes are out of alignment, and the size of its coefficient measures how badly.

The general form written here is the object of study for the rest of the lesson. Get used to reading the six coefficients off in order, including their signs, because two of the three tools you are about to meet need nothing but the first three of them.

The strategy from here is a two-stage one. This part shows what happens when somebody hands you the angle. The next part works out what the angle has to be, which turns out to be a two-line calculation once you know where to look.

18. The three expansions you will need

Worked example

Converting an equation means substituting for both variables and squaring. For a turn of pi over three the substitutions are these.

\[ x = \tfrac{x'}{2} - \tfrac{y'\sqrt3}{2}, \qquad y = \tfrac{x'\sqrt3}{2} + \tfrac{y'}{2} \]

Square the first

Why: A binomial square: first squared, twice the product, last squared.

\[ x^2 = \tfrac{(x')^2}{4} - \tfrac{x'y'\sqrt3}{2} + \tfrac{3(y')^2}{4} \]

Multiply the two together

Why: Outer and inner do not cancel here, because the two brackets differ.

\[ xy = \tfrac{(x')^2\sqrt3}{4} - \tfrac{x'y'}{2} - \tfrac{(y')^2\sqrt3}{4} \]

Square the second

Why: The same pattern, with a plus in the middle.

\[ y^2 = \tfrac{3(x')^2}{4} + \tfrac{x'y'\sqrt3}{2} + \tfrac{(y')^2}{4} \]

Notice the pattern in the cross terms

Why: The two squares contribute cross terms of opposite sign and equal size, and the product contributes one of its own. That is what will be made to cancel.

\[ -\tfrac{\sqrt3}{2}, \qquad -\tfrac12, \qquad +\tfrac{\sqrt3}{2} \]

Check the three at one turned point

Why: Take the turned point one, zero. The new-to-old system puts it at one half, root three over two, so directly x squared is a quarter, the product is root three over four and y squared is three quarters; the three formulas, at that point, give exactly those.

\[ (x',y') = (1,0) \Rightarrow (x,y) = \left(\tfrac12, \tfrac{\sqrt3}{2}\right): \; x^2 = \tfrac14, \; xy = \tfrac{\sqrt3}{4}, \; y^2 = \tfrac34 \]

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.6 Hooked on Conics Again §11.6, p. 976 — the expansions the book asks the reader to verify

These three lines are what the book asks you to verify yourself, and doing it once by hand is worth more than reading it ten times. They are ordinary binomial expansions; the only thing that makes them feel heavy is the number of fractions.

Look at the middle column, the three cross-term coefficients. The two squares contribute cross terms of equal size and opposite sign, and the product of the two substitutions contributes one of its own. Whether the total cancels depends entirely on how those three are weighted by the original coefficients, and that is the whole mechanism of the next part.

The check at the end is a technique to reuse whenever an expansion is long. Pick a convenient point, evaluate both sides of each line numerically, and see that they agree. It will not catch an error that happens to vanish at your chosen point, so pick a point where nothing is zero if you can.

19. Example 11.6.1(2): a tilted ellipse straightened out

Worked example

Convert the equation into the turned variables and identify the curve.

\[ 21x^2 + 10xy\sqrt3 + 31y^2 = 144, \qquad \theta = \tfrac{\pi}{3} \]

Collect the coefficient of the new x squared

Why: Twenty-one quarters, plus ten root three times root three over four, plus thirty-one times three quarters.

\[ \tfrac{21}{4} + \tfrac{30}{4} + \tfrac{93}{4} = \tfrac{144}{4} = 36 \]

Collect the coefficient of the cross term

Why: All three contributions are multiples of root three, and they add to nothing.

\[ \sqrt3\left(-\tfrac{21}{2} - 5 + \tfrac{31}{2}\right) = \sqrt3\,(0) = 0 \]

Collect the coefficient of the new y squared

Why: Sixty-three quarters, minus thirty quarters, plus thirty-one quarters.

\[ \tfrac{63}{4} - \tfrac{30}{4} + \tfrac{31}{4} = \tfrac{64}{4} = 16 \]

Write the reduced equation

Why: No cross term survives, which is the whole point.

\[ 36(x')^2 + 16(y')^2 = 144 \]

Divide by one hundred and forty-four

Why: Standard form for an ellipse centred at the origin.

\[ \frac{(x')^2}{4} + \frac{(y')^2}{9} = 1 \]

Figure (svg): A tilted ellipse drawn over the original x and y axes, with a dashed pair of axes turned by sixty degrees. The ellipse's long axis lies exactly along the turned y-axis and its short axis along the turned x-axis; four dots mark the ends.

The cross term was only the record of the mismatch between the curve and the frame. Line the frame up with the curve and the term is gone.

Check the shape against the coefficients

Why: Ten root three squared is 300, and four times 21 times 31 is 2604, so the quantity B squared minus 4AC is minus 2304, which is negative; the classification theorem later in this lesson calls that an ellipse, and an ellipse is what came out.

\[ B^2 - 4AC = 300 - 2604 = -2304 < 0 \]

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.6 Hooked on Conics Again §11.6, pp. 975-976 — Example 11.6.1 number 2

Rather than write out the whole substitution, this solution collects one coefficient at a time from the three expansions on the previous slide. That is how you should organise the work in practice: three short columns of arithmetic instead of one enormous expression.

The second step is the one to enjoy. Every contribution to the cross term carries a factor of root three, and the three numbers inside the bracket add to zero. That is not luck; the angle was chosen to make it happen, and the next part shows how.

In the picture, the long axis of the ellipse lies exactly along the dashed turned second axis, and the short axis along the turned first one. The reduced equation says the same thing: the larger denominator sits under the turned second variable, so the curve reaches three units that way and only two the other. Reading which axis is the long one off the denominators is the standard Chapter 7 skill, and it still works, as long as you remember the answer is in the turned frame.

20. Where the cross term went

Picture it

Figure (svg): The coefficient of the new cross term plotted against the rotation angle from zero to pi. It is a sine wave of amplitude twenty that crosses zero at pi over three and again at five pi over six.

Two angles in any half turn kill the cross term, and they differ by a right angle: the same pair of axes with the names swapped.

For a fixed equation, the coefficient of the new cross term is a function of the rotation angle, and the picture shows it is a sine wave. Removing the cross term means landing on one of its zeros, and there are two of them in any half turn.

The two zeros are a right angle apart, which is the same pair of axes with the names of the two directions exchanged. Either is a correct answer; the acute one is the conventional choice.

This picture is worth a minute. For one fixed equation, ask what the cross coefficient of the converted equation would be for every possible rotation angle. The answer is a clean sine wave, and the whole task of this section is to land on one of its zeros.

Two features matter. First, the wave crosses zero, so a solution exists; this picture is really a proof of Theorem 11.10 for this particular equation. Second, it crosses twice in a half turn, and the two crossings are a right angle apart.

Those two crossings give the same pair of lines, with the roles of the two axes exchanged. So both are correct answers and they produce reduced equations that differ only by swapping the two variables. Textbooks pick the acute one by convention, and so should you, so that your answer matches everyone else's.

21. Find the error: squaring the substitution term by term

Error analysis

A student converting an equation writes this line and then wonders why the cross term never cancels.

Annotate

On: \( x^2 = (x')^2\cos^2\theta - (y')^2\sin^2\theta \)

  • Squaring a difference of two things gives three terms, not two. The middle term, twice the product, is exactly the cross term that was dropped.
  • The whole method lives or dies on the cross terms. Throw the middle term away and there is nothing left to cancel, so no angle will ever clear the product.
  • x squared equals the first squared, minus twice the product of cosine and sine and of the two turned variables, plus the last squared.
  • Set the angle to zero. The correct expansion must give x squared equal to the turned x squared; the wrong line above happens to as well, so test at pi over four instead, where the correct line keeps a cross term and the wrong one does not.

\[ x^2 = (x')^2\cos^2\theta - 2x'y'\cos\theta\sin\theta + (y')^2\sin^2\theta \]

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.6 Hooked on Conics Again §11.6, p. 976 — the expansions above Theorem 11.10

Look at the line before opening the annotations. The error is a Beginning Algebra one, not a trigonometry one, which is exactly why it survives into this section: attention is on the angles, so the squaring goes on autopilot.

Squaring a difference of two things gives three terms. Dropping the middle one is bad here in a very specific way. The middle term is the cross term, and the cross terms are the entire object of the exercise. Lose them and no angle will ever clear the product, so you will conclude the method does not work.

The last annotation gives you a test that actually discriminates. Checking at a zero angle does not help, because the wrong line passes that test too. Check at forty-five degrees instead, where the correct expansion still carries a cross term and the wrong one does not.

22. What a rotation changes, and what it cannot

Comparison

Fill in the right-hand column for the tilted ellipse, turned through pi over three.

Comparison matrix

quantitybefore the turnafter the turn
coefficient of the squared first variable2136
coefficient of the cross term10√3 ≈ 17.320
coefficient of the squared second variable3116
sum of the two squared coefficients5252
B² − 4AC−2304−2304

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.6 Hooked on Conics Again §11.6, pp. 979-980 — the invariants used to prove Theorem 11.11

Fill in the right-hand column before checking. The first three rows come straight from the worked example; the last two are the interesting ones.

The sum of the two squared coefficients is fifty-two before the turn and fifty-two after. The combination made of the cross coefficient squared minus four times the product of the squared ones is minus 2304 before and minus 2304 after. Both survive the rotation completely unchanged.

Hold on to the second of those. A quantity that every frame agrees about is exactly what you want if you are trying to say something about the curve itself rather than about your choice of axes, and the last part of the lesson turns it into a classification rule.

23. Finding the angle

Section

Part 3

24. Can the cross term always be removed?

Concept

In the last example the angle was handed to us. In a real problem it is not. So ask the general question: given any second-degree equation whose cross coefficient is not zero, is there an angle that clears the product?

\[ Ax^2 + Bxy + Cy^2 + Dx + Ey + F = 0, \qquad B \ne 0 \]

The answer is yes, and the route to it is short: substitute, gather only the terms that carry the product of the two turned variables, and set that coefficient to zero.

Notice that the linear terms cannot contribute. A linear term substituted stays linear, so only the three quadratic terms are in play.

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.6 Hooked on Conics Again §11.6, p. 976 — the question that opens the derivation of Theorem 11.10

Up to now the angle has been a gift. Real problems do not come with one, so the question has to be settled in general: for any equation with a product term, is there an angle that clears it?

The plan is short and you should be able to state it before seeing it worked. Substitute the two rotation formulas, collect only the terms carrying the product of the two turned variables, set that coefficient to zero, and solve for the angle.

One observation saves half the work. A linear term, when you substitute, stays linear: it can never produce a product of two variables. So only the three quadratic terms contribute, and the constant and the two linear coefficients play no part in choosing the angle at all.

25. Deriving the condition on the angle

Worked example

Collect the cross-term contribution from each of the three quadratic terms and set the total to zero.

\[ x^2 \to -2\cos\theta\sin\theta, \quad xy \to \cos^2\theta - \sin^2\theta, \quad y^2 \to 2\cos\theta\sin\theta \]

Weight each by its coefficient and add

Why: This is the coefficient of the product in the converted equation.

\[ -2A\cos\theta\sin\theta + B\bigl(\cos^2\theta - \sin^2\theta\bigr) + 2C\cos\theta\sin\theta = 0 \]

Rewrite with the double angle identities

Why: Twice sine times cosine is the sine of twice the angle; the difference of the squares is the cosine of twice the angle.

\[ -A\sin(2\theta) + B\cos(2\theta) + C\sin(2\theta) = 0 \]

Gather the sines on one side

Why: Both sine terms carry the same factor.

\[ B\cos(2\theta) = (A - C)\sin(2\theta) \]

Divide by the sine and by B

Why: Dividing by B is safe because B is not zero; dividing by the sine is safe because a zero sine would force the cosine to zero too, and no angle has both.

\[ \cot(2\theta) = \frac{A - C}{B} \]

Check on the equation already solved

Why: For the tilted ellipse A is 21, B is ten root three and C is 31, so the right-hand side is minus one over root three; twice pi over three is two pi over three, and the cotangent of that is indeed minus one over root three.

\[ \frac{21 - 31}{10\sqrt3} = -\frac{1}{\sqrt3} = \cot\left(\tfrac{2\pi}{3}\right) \]

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.6 Hooked on Conics Again §11.6, pp. 976-977 — the derivation of Theorem 11.10

The first line collects what each quadratic term contributes to the product, straight from the three expansions you verified earlier, written now for a general angle. Weight each by its coefficient, add, and set the total to zero. That is the whole demand.

The second step is where trigonometry takes over from algebra. Twice a sine times a cosine is the sine of twice the angle, and a difference of the two squares is the cosine of twice the angle. Both double angle identities appear, which is why the condition that comes out is about twice the angle rather than the angle itself.

The division at the end needs a moment's care, and the book takes it. Dividing by the cross coefficient is safe because the whole discussion assumes it is not zero. Dividing by the sine of twice the angle is safe because if that sine were zero, the equation would force the cosine of twice the angle to be zero as well, and no angle has both of those at once.

26. Theorem 11.10, and why it never fails

Concept

Theorem 11.10 — A second-degree equation whose cross coefficient is not zero can be turned into one with no cross term, by rotating the axes counter-clockwise through an angle whose cotangent of twice it equals A minus C, all over B.

\[ \cot(2\theta) = \frac{A - C}{B} \]

Figure (svg): The graph of the cotangent of twice the angle for angles from zero to a right angle. It falls from very large positive values through zero at forty-five degrees to very large negative values, hitting every real number exactly once.

Between zero and a right angle the cotangent of twice the angle runs once through every real number, so there is always exactly one acute answer.

As the angle runs from zero to a right angle, twice it runs from zero to a half turn, and the cotangent falls once through every real number. So whatever the right-hand side is, there is always exactly one acute angle that works.

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.6 Hooked on Conics Again §11.6, p. 977 — Theorem 11.10 and its footnote

The theorem is one equation, and the picture is the reason it always has a solution. As the rotation angle runs from zero to a right angle, twice it runs from zero to a half turn, and the cotangent falls from arbitrarily large positive values, through zero at the halfway point, to arbitrarily large negative values.

That means it takes every real number exactly once. Whatever the combination of coefficients on the right-hand side turns out to be, there is one and only one acute angle matching it. You never have to worry about a case where no rotation works.

The marked point in the middle is the case where the two squared coefficients are equal, which makes the right-hand side zero and the angle exactly half a right angle. That case comes up constantly, so it is worth recognising on sight.

27. Reading the angle condition piece by piece

Notation

Annotate

On: \( \cot(2\theta) = \frac{A - C}{B} \)

  • Cotangent, not tangent. If it is easier to work with, flip both sides at once: the tangent of twice the angle is B over A minus C.
  • Twice the angle, because the derivation used the double angle identities. Solve for twice the angle first, then halve. Forgetting to halve is the single most common slip here.
  • The two squared coefficients, in the order they appear. If they are equal this is zero, the cotangent is zero, and the angle is pi over four, whatever B may be.
  • The cross coefficient, and the reason there is anything to do. The theorem assumes it is not zero; if it is, the equation is already free of the product.

The annotation to take most seriously is the second one. The equation gives you twice the angle, not the angle. Solve for twice the angle first, write that value down, and only then halve it. Skipping the halving is by a wide margin the most common error in this section.

The first annotation offers a practical convenience. Cotangent is awkward on most calculators, so invert both sides and work with the tangent of twice the angle instead. Just be careful that inverting swaps the roles of numerator and denominator.

The fourth annotation states the assumption. If the cross coefficient were zero, the right-hand side would be undefined, and rightly so: there would be no product term to remove and no rotation to perform.

28. Equal squared coefficients always mean forty-five degrees

Intuition

When the two squared coefficients agree, the numerator is zero, so the cotangent of twice the angle is zero, so twice the angle is a right angle and the angle itself is half of one.

\[ A = C \;\Longrightarrow\; \cot(2\theta) = 0 \;\Longrightarrow\; 2\theta = \tfrac{\pi}{2} \;\Longrightarrow\; \theta = \tfrac{\pi}{4} \]

That covers a surprising share of textbook problems, and it explains the opening example: the graph of a product equal to a constant has both squared coefficients equal to zero, so the frame must be turned through forty-five degrees.

This case deserves its own slide because it comes up so often and because it is completely mechanical. If the two squared coefficients agree, the numerator is zero, so the cotangent of twice the angle is zero, so twice the angle is a right angle, so the angle is half of one.

Notice how little that argument used. It did not look at the cross coefficient at all, and it did not look at the linear terms or the constant. Any equation whose squared coefficients match needs a forty-five degree turn, full stop.

That explains the curve you met in the warm-up. A product equal to a constant has both squared coefficients equal to zero, which certainly makes them equal, so the frame must be turned through forty-five degrees. You could have predicted that before knowing any of this.

29. Before you touch the algebra

Step zero

\[ 5x^2 + 26xy + 5y^2 - 16x\sqrt2 + 16y\sqrt2 - 104 = 0 \]

Discussion prompt

Do not convert anything yet. Write down the three numbers you need from this equation, the angle they give, and the one arithmetic fact that makes the angle easy here. Then predict what shape the reduced equation will describe.

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.6 Hooked on Conics Again §11.6, p. 977 — Example 11.6.2 number 1

Write your three numbers and your prediction before revealing. This slide exists to build a habit that will save you a great deal of wasted work: look at the coefficients and think for thirty seconds before starting to substitute.

The two squared coefficients here are both five, so the angle falls out instantly with no calculation at all. Spotting that is worth a minute of the time you would have spent reaching for a calculator.

The prediction is the part that catches people. Both squared coefficients are positive and equal, which looks exactly like a circle or an ellipse, and it is not. The quantity formed from the three quadratic coefficients comes out positive, which means hyperbola, and the worked example on the next slide confirms it. That mismatch between instinct and fact is what the last part of this lesson is about.

30. Example 11.6.2(1): a hyperbola behind a cross term

Worked example

Graph the first equation of Example 11.6.2 by reducing it to standard form.

\[ 5x^2 + 26xy + 5y^2 - 16x\sqrt2 + 16y\sqrt2 - 104 = 0 \]

Find the angle

Why: The two squared coefficients are equal, so the cotangent is zero.

\[ \cot(2\theta) = \frac{5-5}{26} = 0 \;\Longrightarrow\; \theta = \tfrac{\pi}{4} \]

Write the substitutions

Why: Both the cosine and the sine of pi over four are root two over two.

\[ x = \tfrac{\sqrt2}{2}(x' - y'), \qquad y = \tfrac{\sqrt2}{2}(x' + y') \]

Reduce the quadratic part

Why: The cross terms of the two squares cancel against each other, and the product supplies the difference of the two squares.

\[ 5x^2 + 26xy + 5y^2 = 18(x')^2 - 8(y')^2 \]

Reduce the linear part

Why: Each root two meets a root two over two, leaving whole numbers; the turned first variable then cancels.

\[ -16\sqrt2\,x + 16\sqrt2\,y = -16(x' - y') + 16(x' + y') = 32y' \]

Assemble and complete the square

Why: Take minus eight out of the two terms in the turned second variable first.

\[ 18(x')^2 - 8\bigl[(y')^2 - 4y'\bigr] = 104 \;\Longrightarrow\; 18(x')^2 - 8(y'-2)^2 = 72 \]

Divide by seventy-two

Why: Standard form for a hyperbola opening along the turned first axis.

\[ \frac{(x')^2}{4} - \frac{(y'-2)^2}{9} = 1 \]

Figure (svg): A hyperbola drawn over the original axes with a dashed pair of axes turned by forty-five degrees. Its centre sits on the turned y-axis two units from the origin, and the two branches open along the turned x-direction.

Read every feature in the turned frame first: centre, vertices, axis direction. Only then translate back if the question asks for original coordinates.

Check the centre and the vertices

Why: The centre sits at zero, two in the turned frame and the vertices two units either side of it, at plus and minus two, two; the asymptotes have turned-frame slopes plus and minus three halves. In the picture the curve passes through both marked vertices, and both squared coefficients of the original equation were positive even though this is a hyperbola.

\[ B^2 - 4AC = 676 - 100 = 576 > 0 \]

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.6 Hooked on Conics Again §11.6, p. 977 — Example 11.6.2 number 1

This is the first full reduction in the lesson where you were not handed the angle, and it runs through every stage: find the angle, write the substitutions, reduce the quadratic part, reduce the linear part, assemble, complete the square, divide.

Do not skip past the linear part. It is the reason the centre is not at the origin. The two root twos in the original equation were put there precisely so that they would meet the root two over two from the substitution and leave whole numbers, which is a common piece of textbook kindness.

Completing the square is the same move you made all through Chapter 7, now applied to a turned variable. And look at the picture: the centre sits on the dashed turned second axis, two units up, exactly where the reduced equation puts it. The two branches open along the turned first direction because the positive term is the one in that variable.

31. Every answer you report belongs to a frame

Concept

The reduced equation describes the curve in the turned frame, so every feature read off it is stated in turned coordinates. A vertex at two, two means two, two measured along the turned axes, and that is a completely different place from two, two on the original ones.

\[ \text{turned } (2, 2) \;\Longrightarrow\; \text{original } \left(2\cos\tfrac{\pi}{4} - 2\sin\tfrac{\pi}{4},\; 2\sin\tfrac{\pi}{4} + 2\cos\tfrac{\pi}{4}\right) = \left(0, 2\sqrt2\right) \]

If a question asks where something is, translate with the new-to-old system and say which frame the numbers belong to. If it only asks for the shape, or a length, no translation is needed: lengths and angles between features survive a rotation untouched.

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.6 Hooked on Conics Again §11.6, p. 977 — the features listed for Example 11.6.2 number 1

This is a small point that costs a lot of marks. Once you have reduced an equation, every feature you read off it, every vertex, centre, focus and asymptote, is stated in turned coordinates. Those numbers name a completely different place on the page than the same numbers would in the original frame.

So finish a problem by asking what was actually requested. If the question wants the curve identified, or a length, you are done: lengths and the angles between features are unchanged by a rotation. If it wants a location, run it back through the new-to-old system and say which frame you are quoting.

The safest habit is to label everything as you write it. Two words, in the turned frame, beside a coordinate pair is enough, and it makes the difference between an answer a reader can use and one they have to guess at.

32. Match each equation to its rotation angle

Matching

Match the pairs

  • a. 5x² + 26xy + 5y² − 104 = 0
  • b. 21x² + 10√3xy + 31y² = 144
  • c. 13x² − 34√3xy + 47y² − 64 = 0
  • d. 16x² + 24xy + 9y² + 15x − 20y = 0
  • w. π/4, because the two squared coefficients agree
  • x. π/3, from a cotangent of minus one over root three
  • y. π/6, from a cotangent of one over root three
  • z. not a common angle: the cotangent is 7/24

Why: Each answer comes from one division. Equal squared coefficients give zero and therefore pi over four. Minus ten over ten root three is minus one over root three, so twice the angle is two pi over three. Minus thirty-four over minus thirty-four root three is one over root three, so twice the angle is pi over three. Seven over twenty-four is not the cotangent of any common angle, so that one needs the double angle identity for tangent.

Match all four before checking. Each one is a single division followed by recognising a cotangent value, so the whole exercise should take a couple of minutes.

The third is the one where signs matter. Both the numerator and the denominator are negative, and the negatives cancel to give a positive cotangent, which puts twice the angle in the first quadrant. Getting a sign wrong here moves the answer by a right angle, which is a visible error on a sketch but an invisible one in the algebra.

The fourth has no common value, and that is not a failure. It is simply the case where you go after the tangent of the angle directly, which is the technique on the next slide.

33. Example 11.6.2(2): an angle that is not a common one

Worked example

Now the cotangent is not a value you recognise. Rather than take an inverse cotangent and then halve the angle with a half angle identity, go after the tangent of the angle itself.

\[ 16x^2 + 24xy + 9y^2 + 15x - 20y = 0 \]

Find the cotangent of twice the angle

Why: A is 16, B is 24 and C is 9.

\[ \cot(2\theta) = \frac{16-9}{24} = \frac{7}{24} \]

Flip it and expand with the double angle identity for tangent

Why: Write t for the tangent of the angle itself.

\[ \tan(2\theta) = \frac{24}{7} = \frac{2t}{1 - t^2} \]

Clear the fractions

Why: Cross-multiply, then bring everything to one side.

\[ 14t = 24 - 24t^2 \;\Longrightarrow\; 24t^2 + 14t - 24 = 0 \]

Factor and keep the acute root

Why: A positive tangent means an acute angle, which is the conventional choice.

\[ 2(3t + 4)(4t - 3) = 0 \;\Longrightarrow\; t = \tfrac34 \;\text{ or }\; t = -\tfrac43 \]

Figure (svg): A right triangle with the horizontal leg labelled 4, the vertical leg labelled 3 and the hypotenuse labelled 5, with the angle theta at the left-hand vertex and the tangent of theta equal to three quarters.

The tangent fixes the two legs up to scale; Pythagoras supplies the hypotenuse, and the cosine and sine follow with no inverse function anywhere.

Read the cosine and the sine off a right triangle

Why: Legs three and four give a hypotenuse of five, so no inverse function is needed anywhere.

\[ \cos\theta = \tfrac45, \; \sin\theta = \tfrac35 \;\Longrightarrow\; x = \tfrac{4x' - 3y'}{5}, \; y = \tfrac{3x' + 4y'}{5} \]

Substitute and collect

Why: Every contribution to the squared second variable cancels, along with the cross term, and the linear part collapses to a single term.

\[ 25(x')^2 - 25y' = 0 \;\Longrightarrow\; y' = (x')^2 \]

Check the classification

Why: Twenty-four squared is 576, and four times 16 times 9 is 576 as well, so B squared minus 4AC is exactly zero, which is the parabola case, and a parabola is what the reduced equation describes.

\[ B^2 - 4AC = 576 - 576 = 0 \]

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.6 Hooked on Conics Again §11.6, p. 978 — Example 11.6.2 number 2

There are two routes from here, and the book explains why it takes the second. You could take an inverse cotangent, halve it, and then dig the cosine and sine out with half angle identities. That works and it is horrible.

The better route never names the angle at all. Flip the cotangent to a tangent, expand with the double angle identity for tangent, and you have an ordinary quadratic in the tangent of the angle itself. Solve it and you have what you actually need, since the cosine and sine follow from the tangent with a triangle.

Keep the positive root, and notice what the other one is: the negative reciprocal. Two numbers whose product is minus one describe perpendicular directions, so the discarded root is the turned second axis, offering you the same pair of axes with the names swapped. Choosing the positive root is choosing an acute angle, which is the convention.

34. The parabola, drawn in both frames

Picture it

Figure (svg): A parabola drawn over the original axes with a dashed pair of axes turned by about thirty-seven degrees. Its vertex is at the shared origin and it opens along the positive turned y-direction.

A parabola has one squared coefficient and one linear one after the turn, which is why both squared coefficients of the original equation being positive told you nothing.

Both squared coefficients of the original equation were positive, and the graph is a parabola. After the turn one of them is gone entirely, which is exactly what a parabola looks like in standard position: one squared variable and one linear one.

The other root of the tangent equation, minus four thirds, is the negative reciprocal of three quarters, so it points along the turned second axis. Choosing it would simply exchange the names of the two axes.

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.6 Hooked on Conics Again §11.6, p. 978 — the footnote about the discarded root

Look at the reduced equation against the original one. The original had two squared terms, both with positive coefficients, and this is the graph: a parabola. After the turn, one of the squared coefficients has gone entirely, which is what a parabola in standard position looks like.

That should feel slightly shocking, and it is meant to. In Chapter 7 the number of squared terms was a reliable label for the shape. Once a cross term is in the equation, a rotation can create or destroy a squared term, so the count tells you nothing until after the rotation.

The dashed axes in the picture rise three units for every four across, which is the tangent of three quarters you solved for. The discarded root, minus four thirds, points along the other dashed line, at right angles to the first.

35. Estimate the tilt before you compute it

Estimation

\[ 16x^2 + 24xy + 9y^2 + 15x - 20y = 0 \]

Predict first

Look back at the drawing of this parabola. Without using the answer, roughly how far have the axes been turned?

  • Under twenty degrees
  • Between thirty and forty-five degrees
  • Between fifty and seventy degrees
  • More than eighty degrees

Correct: Between thirty and forty-five degrees

Why: The exact angle has tangent three quarters, which is about thirty-seven degrees, so the turned first axis rises three units for every four it runs. A tangent below one always means under forty-five degrees, and a quick sketch of the rise over the run is usually enough to catch an answer that is out by a whole quadrant.

Estimating first is cheap insurance. The commonest failure in these problems is an angle that is wrong by a right angle, and that error is glaring against a rough sketch even when it is invisible in the algebra.

Guess before revealing, and use the previous picture rather than the algebra. This is a habit rather than a technique: a rough sense of the answer, formed before the work, catches the errors that the work itself will not.

A tangent is a rise over a run, so a tangent below one means the turned axis rises less than it runs, which means the angle is under forty-five degrees. Three quarters gives about thirty-seven. If your algebra ever hands you an angle over sixty for this equation, you know something went wrong without having to find out what.

The specific error this guards against is being out by a right angle, which happens when you take the wrong root of the tangent equation, or when you forget to halve. Both produce answers that look perfectly reasonable on paper and absurd against a sketch.

36. Trap: keeping the negative root of the tangent

Trap

The trap

From the quadratic in the tangent, both roots are genuine solutions of the angle condition:

\[ t = \tfrac34 \;\text{ or }\; t = -\tfrac43 \]

Taking the negative one and pressing on gives cosine minus three fifths and sine four fifths, and then a reduced equation of

\[ x' = (y')^2 \]

The fix

Not wrong, exactly, but not the conventional answer. The two roots multiply to minus one, so the two candidate angles differ by a right angle: the same pair of axes with the two directions swapped.

\[ \tfrac34 \cdot \left(-\tfrac43\right) = -1 \]

Pick the positive root, which gives an acute angle. Then a reader comparing your answer with anyone else's sees the same one.

This one is only half a mistake, which is what makes it worth discussing. The negative root genuinely satisfies the condition, and the reduction you get from it genuinely has no cross term. If you carried it all the way through you would produce a correct description of the same parabola.

What you would not produce is the answer in the back of the book, because the two roots differ by a right angle and so the two descriptions exchange the roles of the two turned axes. Their product being minus one is the algebraic fingerprint of that perpendicularity.

So take the positive root and get an acute angle. It costs nothing, it is the universal convention, and it means that when you compare your work with a classmate's you are comparing the same thing.

37. Finish the double angle step

Fill the middle

Fill in the missing coefficient and the missing root for the equation from the last worked example.

Fill in the blanks

\frac14-4/3 = \frac______ \;\Longrightarrow\; 24t^2 + ___\,t - 24 = 0 \;\Longrightarrow\; t = \tfrac34 \;\text___\; t = ___

Why: Cross-multiplying gives fourteen t on the left and twenty-four minus twenty-four t squared on the right, so the middle coefficient is fourteen. The product of the two roots is minus twenty-four over twenty-four, which is minus one, so the partner of three quarters is minus four thirds.

Fill in both blanks before checking. The first is pure cross-multiplication, and the only way to get it wrong is to drop a sign when bringing terms across.

The second blank you can get without solving anything. For a quadratic, the product of the two roots is the constant term over the leading coefficient, which here is minus twenty-four over twenty-four, or minus one. One root is three quarters, so the other must be minus four thirds.

That shortcut is worth keeping. Whenever this quadratic in the tangent appears, its two roots multiply to minus one, because the two candidate angles are always perpendicular. Notice that the leading coefficient and the constant term are always negatives of each other in this construction, which is exactly why.

38. Exercise 8: an ellipse whose angle has tangent two

Worked example

Exercise 8 from the section, worked the same way. The cotangent is again not a common value, but the arithmetic is kinder.

\[ 8x^2 + 12xy + 17y^2 - 20 = 0 \]

Find the cotangent of twice the angle, then the tangent

Why: Eight minus seventeen, over twelve.

\[ \cot(2\theta) = \frac{8-17}{12} = -\frac34 \;\Longrightarrow\; \tan(2\theta) = -\frac43 \]

Expand with the double angle identity

Why: Write t for the tangent of the angle itself and cross-multiply.

\[ \frac{2t}{1-t^2} = -\frac43 \;\Longrightarrow\; 6t = -4 + 4t^2 \;\Longrightarrow\; 4t^2 - 6t - 4 = 0 \]

Halve and factor

Why: Keep the positive root for an acute angle.

\[ 2t^2 - 3t - 2 = (2t+1)(t-2) = 0 \;\Longrightarrow\; t = 2 \]

Read the cosine and the sine

Why: Legs one and two give a hypotenuse of root five.

\[ \cos\theta = \tfrac{1}{\sqrt5}, \quad \sin\theta = \tfrac{2}{\sqrt5} \;\Longrightarrow\; x = \tfrac{x' - 2y'}{\sqrt5}, \; y = \tfrac{2x' + y'}{\sqrt5} \]

Collect the three new coefficients

Why: Fifths throughout: the cross contributions are minus 32, minus 36 and plus 68, which add to zero.

\[ A' = \tfrac{8 + 24 + 68}{5} = 20, \quad B' = \tfrac{-32 - 36 + 68}{5} = 0, \quad C' = \tfrac{32 - 24 + 17}{5} = 5 \]

Write the standard form

Why: Divide by twenty.

\[ 20(x')^2 + 5(y')^2 = 20 \;\Longrightarrow\; (x')^2 + \frac{(y')^2}{4} = 1 \]

Figure (svg): A tilted ellipse drawn over the original axes with a dashed pair of axes turned by about sixty-three degrees. The long axis lies along the turned y-axis, twice as long as the short axis.

A tangent of two is not a common angle, yet the cosine and sine are exact fractions, so the substitution stays exact all the way to the standard form.

Check the classification and the long axis

Why: The discriminant is 144 minus 544, which is minus 400 and negative, so an ellipse; the larger denominator sits under the turned second variable, so the long axis runs along the turned second axis, twice the length of the short one, and the picture agrees.

\[ B^2 - 4AC = 144 - 4(8)(17) = -400 < 0 \]

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.6 Hooked on Conics Again §11.6, p. 986 — Exercise 8

This is the same procedure as the last worked example, run at full speed, and it is the one to imitate under exam conditions. Nothing here is new; the only thing being tested is whether you can keep the arithmetic straight.

The cotangent is negative this time, which means twice the angle is obtuse, which means the angle itself is more than half a right angle. The picture bears that out: the dashed turned first axis is steeper than the diagonal. If you had ended up with a shallow angle, that would be a signal to go back.

The step that collects all three new coefficients at once is worth copying. Writing the cross coefficient's three contributions out and seeing them add to zero is your proof that the angle was right, and it is cheaper to check there than to discover the problem after completing the square.

39. The discriminant

Section

Part 4

40. The squared coefficients tell you nothing

Concept

In Chapter 7 you could name the conic at a glance: two squared terms with the same sign gave an ellipse, opposite signs a hyperbola, one squared term a parabola. The last two examples destroy that habit.

\[ \underbrace{5x^2 + 26xy + 5y^2 - 16x\sqrt2 + 16y\sqrt2 - 104 = 0}_{\text{a hyperbola}} \]

\[ \underbrace{16x^2 + 24xy + 9y^2 + 15x - 20y = 0}_{\text{a parabola}} \]

Both have two squared terms, both with positive coefficients, and one is a hyperbola while the other is a parabola. Once a cross term is present, the signs of the squared coefficients are no longer evidence of anything.

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.6 Hooked on Conics Again §11.6, p. 979 — the remark after Example 11.6.2

Put the two equations side by side and read only the squared coefficients. Both equations have two of them, both positive. One equation is a hyperbola and the other is a parabola.

That is a genuine loss. In Chapter 7 the sign pattern of the squared coefficients was a complete and reliable classifier, and it was the first thing you looked at. With a cross term present it is worth nothing at all, because a rotation redistributes those coefficients freely, as you watched it do twice in the last part.

So a replacement is needed: some combination of the coefficients that a rotation cannot touch. You have already met one in the matrix exercise a few slides back, and the next slide names it.

41. Theorem 11.11: one number decides

Concept

Discriminant of a conic — For a second-degree equation in two variables, the quantity B squared minus four A C, formed from the three quadratic coefficients.

Figure (svg): Three small panels side by side: a hyperbola with positive discriminant, a parabola with zero discriminant and an ellipse with negative discriminant, each drawn from an equation that has a cross term.

One number, computed from three coefficients, separates the three shapes. No rotation was performed to draw any of these conclusions.
sign of the discriminantthe graph is
positivea hyperbola
zeroa parabola
negativean ellipse, possibly a circle

The theorem assumes the equation actually describes a conic and not something degenerate. That assumption earns its own slide shortly.

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.6 Hooked on Conics Again §11.6, p. 979 — Theorem 11.11

One number, built from three coefficients, and it separates the three shapes completely. Compute it and you know what you are dealing with before you have done a single line of the rotation.

The three panels are three different equations, each with a cross term, drawn so you can see that the rule really does correspond to the pictures. The one in the middle sits on the boundary, with the discriminant exactly zero, and it is the parabola.

Two words in the theorem are doing real work: non-degenerate conic. The theorem tells you which conic you have, assuming you have one. It does not promise that you do, and a trap slide shortly shows what happens when you forget that.

42. Reading the general second-degree equation

Notation

Annotate

On: \( Ax^2 + Bxy + Cy^2 + Dx + Ey + F = 0 \)

  • The three quadratic coefficients. These alone fix the SHAPE: the discriminant is built from them and nothing else.
  • The cross term. It is not part of any Chapter 7 standard form, and its coefficient is the one the rotation is designed to kill.
  • The linear coefficients. They shift the curve away from the origin, and after the rotation they are what you complete the square on.
  • The constant. It scales the curve but cannot change its shape, which is why it never appears in the discriminant.

Read the letters off in this exact order. Writing the equation with the cross term out of place, or forgetting that a coefficient is negative, is the commonest source of a wrong classification.

Everything in this lesson starts with reading six coefficients off an equation, so it is worth being deliberate about it. Put the equation in exactly this order, with everything on one side and zero on the other, before you read anything off.

The first annotation is the one to remember. Only the three quadratic coefficients determine the shape. The linear terms shift the curve around the plane and the constant scales it, but neither can turn an ellipse into a hyperbola, which is why neither appears in the discriminant.

Most wrong classifications come from a misread coefficient rather than a misunderstood theorem. A minus sign in front of a squared term, or a cross coefficient written without its root, will flip the answer. Write the three numbers down separately, signs included, before you multiply anything.

43. Example 11.6.3: classify three conics without rotating

Worked example

All three of the section's worked equations, decided by arithmetic alone.

\[ \text{(a) } 21x^2 + 10xy\sqrt3 + 31y^2 = 144 \]

Compute the discriminant for the first

Why: Ten root three squared is three hundred.

\[ (10\sqrt3)^2 - 4(21)(31) = 300 - 2604 = -2304 < 0 \;\Rightarrow\; \text{ellipse} \]

Compute it for the second

Why: Both squared coefficients are five.

\[ 26^2 - 4(5)(5) = 676 - 100 = 576 > 0 \;\Rightarrow\; \text{hyperbola} \]

Compute it for the third

Why: Twenty-four squared is 576, and so is four times sixteen times nine.

\[ 24^2 - 4(16)(9) = 576 - 576 = 0 \;\Rightarrow\; \text{parabola} \]

Check against the rotations already performed

Why: The first reduced to an ellipse in standard form, the second to a hyperbola centred at zero, two, and the third to a parabola with its vertex at the origin. Three sign tests reproduced the verdicts of three pages of algebra.

\[ -2304 < 0, \quad 576 > 0, \quad 0 = 0 \]

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.6 Hooked on Conics Again §11.6, pp. 980-981 — Example 11.6.3

Three arithmetic lines settle three problems that cost the book several pages each. That is the point of the theorem, and it is worth pausing to appreciate how much the discriminant buys for how little.

The middle one is the case that would have fooled you an hour ago: both squared coefficients positive and equal, and the answer is a hyperbola. The third is the one sitting exactly on the boundary, which is the parabola case, and a boundary case is always worth double-checking arithmetically because being out by one changes the answer entirely.

Use this as a first move on every problem from now on. It takes thirty seconds and it tells you what the two pages of rotation should produce, so if the reduced equation disagrees with the prediction you know to go looking for the error immediately.

44. Why no rotation can change the discriminant

Worked example

The theorem works because the discriminant is blind to the frame. Here is that claim, proved in five lines.

Write the two new squared coefficients

Why: Apply the power reduction formulas: each is half the sum of the old two, plus or minus the same wobble.

\[ 2A' = (A+C) + u, \quad 2C' = (A+C) - u, \quad u = (A-C)\cos 2\theta + B\sin 2\theta \]

Write the new cross coefficient the same way

Why: Sine and cosine have exchanged places, and a sign has flipped.

\[ B' = -(A-C)\sin 2\theta + B\cos 2\theta \]

Add the squares of those two quantities

Why: The cross products cancel and the Pythagorean Identity closes it.

\[ u^2 + (B')^2 = (A-C)^2 + B^2 \]

Multiply the two squared coefficients

Why: A difference of squares, then substitute for the wobble.

\[ 4A'C' = (A+C)^2 - u^2 = (A+C)^2 - (A-C)^2 - B^2 + (B')^2 \]

Simplify the first two brackets

Why: Their difference is four times the product of the old squared coefficients.

\[ 4A'C' = 4AC - B^2 + (B')^2 \]

Move everything to one side

Why: The new cross coefficient cancels itself, and only the old letters remain.

\[ (B')^2 - 4A'C' = (B')^2 - \bigl[4AC - B^2 + (B')^2\bigr] = B^2 - 4AC \]

Figure (svg): Three curves against the rotation angle from zero to pi: four times the product of the new squared coefficients wobbling between about 2300 and 2700, the square of the new cross coefficient wobbling between 0 and 400, and their difference, a perfectly flat line at 2304.

Two quantities move as the frame turns; their difference does not. That flat green line, shifted up to fit on the same axes, is the discriminant.

Check at the angle that clears the cross term

Why: There the new cross coefficient is zero, so the discriminant is just minus four times the product of the two new squared coefficients. For the tilted ellipse those were 36 and 16, and minus four times 576 is minus 2304, the number computed straight from the original coefficients.

\[ -4(36)(16) = -2304 = (10\sqrt3)^2 - 4(21)(31) \]

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.6 Hooked on Conics Again §11.6, pp. 979-980 — the proof of Theorem 11.11

This is the proof, and it is shorter than the book's because it keeps the new cross coefficient in the argument instead of setting it to zero straight away. Follow the shape of it: write both new squared coefficients as a fixed part plus or minus a wobble, notice that the new cross coefficient is built from the same two pieces with sine and cosine swapped, and then add the squares.

That third line is where the work happens. Squaring both quantities and adding, the cross products cancel exactly and the Pythagorean Identity collapses the rest, leaving a combination of the original coefficients only. From there it is a difference of squares and a subtraction.

The picture shows the same claim numerically. Two of the three curves move as the angle turns, sometimes a great deal, and their difference is a perfectly flat line. That flat line, shifted upward so it fits on the same axes, is the discriminant, and it sits at the same value at every angle.

45. Slide the cross coefficient across the boundary

Tweak it

Parameter explorer

The curve is the upper branch of the graph of the equation shown. Slide the cross coefficient upward from zero and watch the arch flatten. At which value does it straighten into a perfect line, and what is the discriminant there?

\[ x^2 + {B}\,xy + y^2 = 4 \]

  • B — from 0 to 3: cross coefficient

Start at zero, where the graph is the top half of a circle of radius two, and push the slider up slowly. The arch tips and flattens as the circle stretches into a tilted ellipse, but it is still curved.

Somewhere it stops being curved. Watch for the value where the arch straightens into a perfect line: that is a cross coefficient of two, where the discriminant is exactly zero and the whole graph has degenerated into a pair of parallel lines. Past it the curve bends the other way, and the graph is a hyperbola.

The lesson is that nothing dramatic happens to the equation at the boundary. One coefficient passes through one value and the shape changes category. That is exactly the kind of switch a single sign test is good at detecting, which is why the discriminant works.

46. Trap: trusting a zero discriminant blindly

Trap

The trap

A zero discriminant is supposed to mean a parabola. Here is an equation with one:

\[ x^2 + 2xy + y^2 = 0 \]

\[ B^2 - 4AC = 4 - 4 = 0 \]

But the graph is a single straight line, not a parabola.

The fix

The left side factors as a perfect square, so the equation says that the sum of the two variables is zero. That is one line, and Theorem 11.11 does not cover it: the theorem opens with the words non-degenerate conic.

\[ (x+y)^2 = 0 \;\Longleftrightarrow\; y = -x \]

So the discriminant classifies given that the graph is a conic. Always look for a factorisation, an empty graph or a single point before reporting the verdict.

The discriminant here really is zero, and the graph really is not a parabola. The quadratic part is a perfect square, so the equation says that a square equals zero, which pins the two variables to a single line.

The theorem was stated for a non-degenerate conic, and this is what that phrase was guarding against. Degenerate cases arise exactly when the quadratic part factors, and then the graph can be a single line, a pair of lines, a single point or nothing at all.

In practice: when the discriminant comes out zero, spend ten seconds checking whether the quadratic part is a perfect square, and if it is, look at the linear part. If the linear part survives the rotation you have a genuine parabola; if it does not, you have lines.

47. Two graphs with a zero discriminant, neither a parabola

Picture it

Figure (svg): Two panels. On the left, the single line y equals minus x, the whole graph of x plus y all squared equals zero. On the right, two parallel lines, the graph of x plus y all squared equals two.

Both have a zero discriminant and neither is a parabola. Theorem 11.11 assumes a non-degenerate conic, and that assumption is doing real work.

Change only the constant on the right and the same quadratic part gives a single line, a pair of parallel lines, or nothing at all when the constant is negative. The discriminant is zero in every case, because the constant never enters it.

The degenerate cases are exactly the ones where the quadratic part is a perfect square and the linear part fails to produce a genuine bend. Theorem 11.11 is a classification of conics, not a test for being one.

The two panels differ only in the constant term, and the constant never appears in the discriminant, so the discriminant is zero for both. One is a single line and the other is a pair of parallel lines. Make the constant negative and the graph is empty.

Compare this with the parabola you reduced earlier, which also had a perfect square for its quadratic part. The difference was the linear part: there, a linear term survived the rotation and produced the bend. Here there is no linear term at all, so nothing bends.

So the honest statement of the rule is: a zero discriminant means the quadratic part is a perfect square. Whether that gives a parabola depends on what the rest of the equation does, and the fastest way to find out is to carry out the rotation.

48. Sort six equations by their discriminant

Sorting

Sort into buckets

Compute B² − 4AC for each and sort by the sign. Assume each is non-degenerate.

positive: hyperbola
3x² − 5xy + 2y² + x = 0; x² − 2√3xy − y² + 8 = 0
zero: parabola
x² + 2√3xy + 3y² + 2√3x − 2y − 16 = 0; x² − 4xy + 4y² − 2√5x − √5y = 0
negative: ellipse
8x² + 12xy + 17y² − 20 = 0; 5x² + 6xy + 5y² − 4√2x + 4√2y = 0
hyp
25 − 4(3)(2) = 1, and 12 − 4(1)(−1) = 16. Both positive. Notice the second one: a negative coefficient on the squared second variable makes the subtracted product positive, so it adds to the discriminant instead of taking away.
par
12 − 4(1)(3) = 0, and 16 − 4(1)(4) = 0. In both, the quadratic part is a perfect square, which is always what a zero discriminant means.
ell
144 − 4(8)(17) = −400, and 36 − 4(5)(5) = −64. Both negative, so both are ellipses; the second has equal squared coefficients, so its long axis lies at forty-five degrees.

Sort all six before checking. Each is one subtraction after two multiplications, so the arithmetic is quick; what is being practised is reading the coefficients correctly.

The fourth item is the one that needs care. Its squared second coefficient is negative, so the product of the two squared coefficients is negative, and subtracting four times a negative number adds. Miss that and you will report a negative discriminant for a hyperbola.

Notice that the two items in the middle bucket both have a perfect square for their quadratic part. That is always what a zero discriminant means, and having seen the degenerate examples you should now regard those two as parabolas only until the rotation confirms it.

49. Exercise 6: a rectangular hyperbola in disguise

Worked example

Exercise 6 from the section. This time the two squared coefficients have opposite signs.

\[ x^2 - 2\sqrt3\,xy - y^2 + 8 = 0 \]

Predict with the discriminant

Why: Minus two root three squared is twelve, and the product of the squared coefficients is negative, so subtracting four times it adds.

\[ B^2 - 4AC = 12 - 4(1)(-1) = 16 > 0 \;\Rightarrow\; \text{hyperbola} \]

Find the angle

Why: One minus minus one is two, over minus two root three.

\[ \cot(2\theta) = \frac{1-(-1)}{-2\sqrt3} = -\frac{1}{\sqrt3} \;\Longrightarrow\; 2\theta = \tfrac{2\pi}{3}, \; \theta = \tfrac{\pi}{3} \]

Write the substitutions

Why: Cosine one half, sine root three over two.

\[ x = \tfrac{x' - \sqrt3\,y'}{2}, \qquad y = \tfrac{\sqrt3\,x' + y'}{2} \]

Collect the two new squared coefficients

Why: Quarters throughout, and the constant is untouched by a rotation.

\[ A' = \tfrac{1 - 6 - 3}{4} = -2, \qquad C' = \tfrac{3 + 6 - 1}{4} = 2 \]

Write the reduced equation and tidy it

Why: Multiply through by minus one, then divide by eight.

\[ -2(x')^2 + 2(y')^2 + 8 = 0 \;\Longrightarrow\; \frac{(x')^2}{4} - \frac{(y')^2}{4} = 1 \]

Figure (svg): A rectangular hyperbola drawn over the original axes with a dashed pair of axes turned by sixty degrees. Its two branches open along the turned x-direction and its asymptotes bisect the turned axes.

Equal denominators make the asymptotes cut the turned axes at forty-five degrees, so the whole picture is the standard rectangular hyperbola, tipped.

Check the prediction and the asymptotes

Why: The discriminant said hyperbola and a hyperbola is what appeared, with vertices two units along the turned first axis, at one, root three and its opposite in the original frame. Equal denominators make the asymptotes cut the turned axes at forty-five degrees, which is what the dotted lines in the picture do.

\[ \text{vertices } (\pm 2, 0) \text{ turned} \;\Longrightarrow\; (\pm 1, \pm\sqrt3) \text{ original} \]

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.6 Hooked on Conics Again §11.6, p. 986 — Exercise 6

This one starts with the discriminant, which is the order you should be working in from now on. It also has a negative squared coefficient, which is the case that trips people up, so read the sign carefully before multiplying.

The angle calculation has two negatives in it and they do not cancel: the numerator is positive two and the denominator is negative, so the cotangent is negative and twice the angle is obtuse. Halving gives sixty degrees.

The two denominators in the reduced equation are equal, which makes this a rectangular hyperbola: its asymptotes meet at right angles, and they sit at forty-five degrees to the turned axes. Those are the faint dotted lines in the picture, and the branches hug them as they run out. Notice this is the same kind of curve as the warm-up, just turned through a different angle.

50. Exercise 1: the parabola case, end to end

Worked example

Exercise 1. Both squared coefficients are one, and the quadratic part is a perfect square, so the degenerate warning is worth keeping in mind.

\[ x^2 + 2xy + y^2 - x\sqrt2 + y\sqrt2 - 6 = 0 \]

Take the discriminant first

Why: Two squared, minus four times one times one.

\[ B^2 - 4AC = 4 - 4 = 0 \;\Rightarrow\; \text{parabola, if it is not degenerate} \]

Find the angle

Why: The squared coefficients agree, so the cotangent is zero.

\[ \cot(2\theta) = \frac{1-1}{2} = 0 \;\Longrightarrow\; \theta = \tfrac{\pi}{4} \]

Reduce the quadratic part

Why: It is a perfect square, and the sum of the two variables is root two times the turned first variable.

\[ x^2 + 2xy + y^2 = (x+y)^2 = \left(\sqrt2\,x'\right)^2 = 2(x')^2 \]

Reduce the linear part

Why: Each root two meets a root two over two; the turned first variable cancels and the second survives twice.

\[ -\sqrt2\,x + \sqrt2\,y = -(x' - y') + (x' + y') = 2y' \]

Assemble and solve for the turned second variable

Why: Divide by two and move the constant across.

\[ 2(x')^2 + 2y' - 6 = 0 \;\Longrightarrow\; (x')^2 = -(y' - 3) \]

Check the vertex in the original equation

Why: The vertex sits at zero, three in the turned frame, which is minus three over root two, three over root two in the original. Substituting those gives four and a half, minus nine, four and a half, three, three and minus six, and those add to zero, so the point really is on the curve; a genuine bend, so not degenerate after all.

\[ \tfrac92 - 9 + \tfrac92 + 3 + 3 - 6 = 0 \]

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.6 Hooked on Conics Again §11.6, p. 986 — Exercise 1

The discriminant is zero and the quadratic part is a perfect square, so by the rule you just learnt you should be suspicious. Carry out the rotation and see.

The reduction is unusually pleasant because you never need the three expansions: the quadratic part is a perfect square, and the sum of the two variables is just root two times the turned first variable, so the whole thing collapses in one line. Look for that shortcut whenever the quadratic part factors.

The linear part survives and produces a single turned second variable, which is the bend. So this is a genuine parabola, opening in the negative turned second direction with its vertex three units along the positive one. The check substitutes that vertex back into the original equation and gets zero, which confirms both the vertex and the whole chain of algebra.

51. Break the claim

Counterexample

Discussion prompt

Someone claims: if an equation has a cross term, then rotating the axes through the right angle always produces one of the three standard conic forms. Give an equation that breaks the claim, and say in one sentence what the counterexample shows about Theorem 11.11.

Write an equation before revealing. Any perfect square set equal to a positive constant will do; the one given is the smallest.

What makes it a good counterexample is that the rotation itself works flawlessly. The cross term does vanish, the reduced equation is clean, and the answer is still not a conic. So the failure is not in the method but in the assumption that a second-degree equation must describe a conic.

The general lesson is one worth carrying past this lesson: a classification theorem tells you which member of a family you have, given that you have one. It is not a membership test, and treating it as one is how you end up confidently calling a pair of straight lines a parabola.

52. Putting it together

Section

Part 5

53. Pattern: reducing an equation with a cross term

Pattern

Figure (svg): A flow diagram: read off A, B and C; if B is zero there is nothing to rotate; otherwise take the discriminant to predict the shape, solve the cotangent condition for the angle, find its cosine and sine, substitute, and complete the square.

The discriminant comes early because it costs three multiplications and tells you what to expect from two pages of algebra.
  1. Read off A, B and C in the order the general form lists them, signs included. If B is zero there is nothing to rotate.
  2. Take the discriminant B squared minus four A C. It costs three multiplications and predicts the shape.
  3. Solve the angle condition for twice the angle first, then halve it. A common cotangent gives the angle directly; otherwise flip to the tangent and use the double angle identity, keeping the positive root.
  4. Get the cosine and the sine exactly, from a right triangle if the angle is not common, and write the two substitutions.
  5. Substitute and collect. If the cross term does not vanish, the angle or an expansion is wrong; go back rather than on.
  6. Complete the square and read the conic off, remembering that every feature you read is in the turned frame.

This is the order to work in, and it is arranged so the cheap steps come first. Reading the coefficients and taking the discriminant costs under a minute and gives you a prediction the expensive work has to match.

The fifth line is the checkpoint most people skip. After substituting, before completing the square, look at the cross coefficient and confirm it is zero. If it is not, either the angle is wrong or an expansion is, and finding that out now saves you from completing the square on a wrong equation.

The last line is where marks are lost rather than mathematics. Every feature you read off the reduced equation lives in the turned frame, so either translate it or label it. An unlabelled coordinate pair is an incomplete answer.

54. Put the procedure in order

Ranking

Put in order

Order the steps of a complete reduction.

  1. Read off A, B and C from the general form
  2. Compute the discriminant and predict the shape
  3. Solve the cotangent condition for the rotation angle
  4. Find the cosine and sine of that angle exactly
  5. Substitute for x and y, expand, and check the cross term is gone
  6. Complete the square and read the conic in the turned frame

Why: The discriminant goes second because it is cheap and it tells you what the two pages of algebra should produce, so a mistake announces itself. The angle must be found before the cosine and sine, and both before any substitution; the check that the cross term vanished belongs with the substitution, not after the answer is written down.

Drag the six steps into order before checking. Most of the sequence is forced, and the one genuinely free choice is where the discriminant goes.

It goes early, for the same reason you estimate before you compute. A prediction made before the algebra is an independent check on the algebra; a prediction made afterwards is just a restatement of whatever you already got.

The other thing to notice is that the angle, the cosine and the sine are three separate steps. Trying to find the cosine and sine while solving for the angle is where the half angle identities creep in and the work triples.

55. Where a turned frame earns its keep

Real world

Discussion prompt

A satellite dish is a parabola spun about its axis. The manufacturer's drawing puts the axis along a coordinate axis and the equation has no cross term. An installer measures the same dish against the wall of a building and gets a cross term. What single number should each of them compute to be sure they are describing the same shape, and why is that the right choice?

OpenStax Algebra and Trigonometry 2e, §12.4 Rotation of Axes §12.4 Rotation of Axes — the same identification of conics from a general equation

Write an answer before revealing. The question is really asking what it means for two descriptions of the same object to agree when the describers are using different coordinate systems.

The discriminant is the answer because it is frame-independent. The installer and the manufacturer will disagree about every single coefficient, and they will still get the same value from that one combination, because you proved a few slides ago that a rotation cannot change it.

This is the general shape of a great deal of applied mathematics. When two observers must agree on a fact about an object, you look for a quantity built from their measurements that is unchanged by whatever transformation relates their viewpoints. Here it is a rotation and the quantity is the discriminant; in other settings it is a different transformation and a different invariant, but the logic is the same.

56. Check: which system for which job

Check

Check your understanding

The axes are turned counter-clockwise through an angle. A point's ORIGINAL coordinates are known. Which expression gives its new first coordinate?

  • A. x cos θ − y sin θ
  • B. x cos θ + y sin θ (correct)
  • C. x′ cos θ − y′ sin θ
  • D. −x sin θ + y cos θ

Answer: B

Why: Going from the old coordinates to the new ones uses the inverse of the rotation matrix, whose first row is cosine then sine, both positive. The minus sign lives in the second line of that system, not the first.

Why A tempts people
This is the new-to-old system with the primes dropped: it gives the old first coordinate from the two new ones.
Why C tempts people
Correct as a formula, but it produces the OLD first coordinate from the two new ones, which is the opposite direction.
Why D tempts people
This is the right system but the wrong line: it gives the new SECOND coordinate.

The question is about direction of travel, not about the formulas themselves. Two of the four options are correct formulas from the theorem; the issue is which one answers the question asked.

Given ordinary coordinates and wanting turned ones, you want the system whose left-hand sides are the turned coordinates, and its first line has a plus sign. The option with the minus sign is the first line of the other system, which goes the wrong way.

If you chose the last option, you had the right system but the wrong line: that expression produces the turned second coordinate.

57. Check: finding the angle

Check

Check your understanding

Which rotation removes the cross term from 4x² + 5xy + 4y² = 9?

  • A. π/6
  • B. π/4 (correct)
  • C. π/3
  • D. arctan(5/8)

Answer: B

Why: The two squared coefficients are both four, so the numerator of the condition is zero, the cotangent of twice the angle is zero, twice the angle is a right angle, and the angle is pi over four. The value of the cross coefficient is irrelevant once the squared ones agree.

Why A tempts people
Twice pi over six is pi over three, whose cotangent is one over root three, not zero.
Why C tempts people
Twice pi over three is two pi over three, whose cotangent is minus one over root three, not zero.
Why D tempts people
This looks like a plausible use of the coefficients but no step of the derivation produces it; the condition involves the cotangent of TWICE the angle.

This should take about ten seconds once you look at the two squared coefficients rather than at the cross term. They are both four, so the numerator of the condition is zero and nothing else matters.

The cross coefficient being five, an odd and awkward number, is deliberately there to distract. It plays no part in choosing the angle whenever the squared coefficients agree.

The two wrong common angles both fail for the same reason: they make twice the angle something other than a right angle, so the cotangent is not zero. If you are unsure, compute twice the candidate angle and take its cotangent; it takes one line.

58. Check: classifying from the discriminant

Check

Check your understanding

Classify the non-degenerate conic x² + 4xy + y² − 9 = 0.

  • A. an ellipse
  • B. a circle
  • C. a hyperbola (correct)
  • D. a parabola

Answer: C

Why: The discriminant is sixteen minus four, which is twelve and positive, so the graph is a hyperbola. Rotating through pi over four confirms it: the equation becomes three times the new first variable squared minus the new second variable squared equal to nine.

Why A tempts people
Two positive squared coefficients suggest an ellipse, and that suggestion is exactly what the cross term destroys; compute the discriminant instead.
Why B tempts people
Equal squared coefficients would mean a circle only if there were no cross term at all.
Why D tempts people
A parabola needs a discriminant of exactly zero, which would require the cross coefficient to be two rather than four.

Read the three coefficients off first: one, four and one. Then the discriminant is sixteen minus four, which is twelve, positive, hyperbola. No rotation needed.

Two of the wrong options come from the Chapter 7 instinct that two positive squared coefficients mean an ellipse, and equal ones mean a circle. That instinct is exactly what the cross term destroys, and this question is here to make sure it has been retired.

If you want to see it, rotate through forty-five degrees. The equation becomes three times the turned first variable squared minus the turned second variable squared equal to nine, which is unmistakably a hyperbola.

59. Say why it has to be twice the angle

Explain it to yourself

Discussion prompt

Explain, in two or three sentences, why the condition on the rotation angle involves twice the angle rather than the angle itself. Refer both to the identities used in the derivation and to the fact that turning through a further right angle gives the same pair of axes.

Write two or three sentences before revealing. Being able to explain this means you can rebuild the condition rather than memorise it, which matters because misremembering it as a statement about the angle itself is the classic failure.

The algebraic half of the answer is that every cross-term contribution came out as twice a sine times a cosine, or as a difference of their squares, and those two expressions are precisely the double angle identities. Nothing else could have come out.

The geometric half is the more satisfying one. Adding a right angle to a correct rotation angle gives the same two lines with the two axis names exchanged, so it must also be correct. Doubling converts that right angle into a half turn, and the cotangent repeats exactly every half turn, so the condition automatically accepts both answers. The algebra and the geometry are telling you the same thing.

60. Exit ticket

Exit ticket

\[ 13x^2 - 34xy\sqrt3 + 47y^2 - 64 = 0 \]

Discussion prompt

Predict the shape from the discriminant, find the rotation angle, and say which way the curve opens in the turned frame. You do not have to carry out the full substitution.

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.6 Hooked on Conics Again §11.6, p. 986 — Exercise 5

Work all three parts before revealing. The point of the ticket is that you can now say a great deal about a conic without ever performing the substitution.

The discriminant comes out positive, so hyperbola, and that is settled in three multiplications. The angle condition gives a cotangent of one over root three, which is a common value: twice the angle is sixty degrees, so the angle is thirty. Watch the two minus signs cancel in that division.

In the reduced equation the positive squared term is the one in the turned second variable, so the curve opens along the turned second axis, with vertices one unit either side of the origin. If you carried out the full substitution and got something else, the discriminant you computed at the start is the check that tells you to go back.

61. Recap

Recap

toolwhat it needswhat it gives
the two rotation systemsthe angle, and which direction you are goingcoordinates in the other frame; the two are inverse
substitute and expandcosine and sine of the angle, exactlythe equation in the turned variables
the angle conditionthe three quadratic coefficientsthe one acute angle that removes the cross term
the discriminantthe same three coefficientshyperbola, parabola or ellipse, with no rotation

\[ \cot(2\theta) = \frac{A-C}{B}, \qquad B^2 - 4AC = (B')^2 - 4A'C' \]

Two habits are worth taking away. Compute the discriminant before you rotate, so the algebra has something to agree with. And name the frame of every feature you report, because a vertex at two, zero means nothing until you say which axes measured it.

Next, the same conics from a completely different starting point: one focus, one directrix and one number, which in polar coordinates produce all four shapes from a single equation.

OpenStax Algebra and Trigonometry 2e, §12.4 Rotation of Axes §12.4 Rotation of Axes — the same material in OpenStax

Four tools, and they divide into two jobs. The two rotation systems and the substitution are for converting things; the angle condition and the discriminant are for deciding things. Keep the jobs separate in your head and the section stays simple.

The discriminant is the single most useful line in this lesson. It costs three multiplications, it needs no angle, no substitution and no completing of the square, and it names the shape. Compute it first on every problem, even when you intend to do the full reduction, because it is the only independent check you have on two pages of algebra.

And label your frames. Half the marks lost in this material are lost not to wrong mathematics but to a coordinate pair reported without saying which axes measured it. Next comes a completely different route to the same four shapes: one focus, one directrix and one number, which in polar coordinates give all of them from a single equation.

Sources

  1. Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.6 Hooked on Conics Again — Stitz & Zeager, College Trigonometry, Version 3 Corrected Edition, July 4 2013, pp. 973-981
  2. OpenStax Algebra and Trigonometry 2e, §12.4 Rotation of Axes

Want this taught 1-on-1? Alexander tutors Trigonometry — $55/session, free consultation.

Book on Wyzant · Text (657) 465-8108