Differentiating an equation in x and y without solving for y, the chain-rule factor every y term produces, tangent lines to curves that fail the vertical line test, reading horizontal and vertical tangents off an implicit derivative, and computing a second derivative implicitly.
Subject: Calculus I · 68 slides · symbolic lesson
Open the interactive version of this deck
Title
Calculus I · Section 3.8
Slopes of curves you never solve for y
Objectives
Every derivative so far started from a formula y equals something in x. Many curves come as an equation linking x and y instead, and some cannot be untangled at all. This lesson differentiates them as they stand.
OpenStax Calculus Volume 1, §3.8 Implicit Differentiation §3.8, pp. 268-274 — learning objectives 3.8.1 and 3.8.2
So far every derivative you have taken started from a formula with y alone on one side. That is a luxury. Circles, ellipses, the folium of Descartes and plenty of curves from physics arrive as an equation that ties x and y together, and some of those equations cannot be solved for y by any formula at all.
Implicit differentiation is the way round that. You differentiate the equation exactly as it is written, remembering that y depends on x, and then solve for the derivative with ordinary algebra. There is no new rule to learn; it is the chain rule, used carefully.
Once you have a slope formula, the rest of the lesson is about using it: tangent and normal lines, the places where a curve goes flat or stands upright, and the second derivative, which tells you which way the curve bends.
Warm-up
Discussion prompt
Suppose f is some differentiable function of x, and you do not know its formula. What is the derivative of f(x) squared with respect to x? Write it before you reveal, and name the rule you used.
Write your answer before revealing it. If you wrote two f of x and stopped, you have found exactly the slip this lesson is built to prevent, so it is well worth catching now.
The derivative of a square of something is two times that something, times the derivative of the something. When the something is plain x, its derivative is one and you never notice the extra factor. When it is a function f of x, the factor f prime of x is essential.
Hold on to that picture. In a few slides the function f will be replaced by y, and the factor f prime of x will be written dy by dx. Everything else in implicit differentiation is algebra wrapped around that single factor.
Section
Part 1
Concept
An explicit formula hands you y directly. An implicit one only states a relationship that y must satisfy.
\[ \text{explicit: } y = x^2 + 1 \qquad \text{implicit: } y - x^2 = 1 \]
Here the two say exactly the same thing: solve the second for y and you get the first.
defined implicitly — An equation in x and y defines a function y of x implicitly when that function, substituted for y, makes the equation true for every x in its domain.
OpenStax Calculus Volume 1, §3.8 Implicit Differentiation §3.8, p. 268 — explicit and implicit functions
An explicit formula is a recipe: give it x and it hands you y. An implicit equation is more like a condition: it tells you which pairs of numbers belong together, and leaves you to work out y if you can.
In this first example the two descriptions are interchangeable, because you can add x squared to both sides and recover the explicit form in one move. That will not always be the case, and that is the whole point of the lesson.
Notice the careful wording of the definition. An equation defines a function implicitly when the function makes the equation true. It does not say the equation defines only one function; the next slide shows that a single equation can hide a great many.
Picture it
Figure (svg): Four small panels. The first shows the whole circle x squared plus y squared equals 25. The second shows only its upper half, the third only its lower half, and the fourth the upper half on the left of the y-axis joined with the lower half on the right.
\[ y = \begin{cases} \sqrt{25-x^2} & -5 \le x < 0 \\ -\sqrt{25-x^2} & 0 \le x \le 5 \end{cases} \]
Each of the last three panels passes the vertical line test, and each satisfies the circle's equation at every point it contains. The equation of the circle defines all of them implicitly.
OpenStax Calculus Volume 1, §3.8 Implicit Differentiation §3.8, pp. 268-269 — Figure 3.30
Look at the four panels from left to right. The full circle fails the vertical line test: most vertical lines meet it twice. The upper half and the lower half each pass it, and each is the graph of a genuine function whose points all satisfy the circle's equation.
The fourth panel is the surprising one. It takes the upper half on the left of the y-axis and the lower half on the right. It is an odd-looking function, but it is a function, and it too satisfies the equation at every point.
The book prints the domain of the second piece as zero to twenty-five, a misprint; the figure and the formula on this slide use the correct interval, zero to five. The lesson of the slide is that the equation fixes the curve, not a particular function, and you choose the piece you need by the point you are looking at.
Prediction
\[ x^2 + y^2 = 25 \]
Predict first
How many different functions y of x, each defined for every x from −5 to 5, satisfy this equation?
Correct: Infinitely many
Why: At every x strictly between −5 and 5 you may choose the plus root or the minus root independently. The halves are the two tidy choices, the mixed panel is a third, and switching sign at any point gives another. The equation pins down the curve, not which function you mean.
Commit to an answer before you reveal. Most people say two, because the upper and lower halves are the obvious candidates.
But nothing forces you to choose the same sign everywhere. At each x between minus five and five you could independently take the positive or the negative square root, and every such choice gives a function satisfying the equation. That is infinitely many functions, most of them jumping wildly between the two halves.
Why does this matter for derivatives? Because only the smooth pieces have derivatives, and the slope you want at a point is the slope of the smooth piece through that point. Implicit differentiation automatically gives you that slope, whichever piece it is, without asking you to choose one first.
Concept
Figure (svg): The circle of radius 5 centred at the origin, cut by the dashed vertical line x equals 3 at the points (3, 4) and (3, minus 4). A short tangent segment is drawn at each point: falling at the top point, rising at the bottom one.
The vertical line through x equal to 3 meets the circle twice, so the circle is not the graph of a function. Near each crossing, though, the curve looks like the graph of one.
\[ \text{near } (3, 4): \; y = \sqrt{25-x^2}, \qquad \text{near } (3,-4): \; y = -\sqrt{25-x^2} \]
So a slope at a point makes sense: it is the slope of whichever half runs through that point.
OpenStax Calculus Volume 1, §3.8 Implicit Differentiation §3.8, p. 269 — the tangent lines at (3, 4) and (3, −4)
Look at the dashed line at x equal to three. It meets the circle at two points, so no single function of x describes the whole circle. And yet the picture clearly shows a tangent at each point: falling at the top, rising at the bottom.
The resolution is that tangent lines are local. Near the top point, the circle coincides with the upper half, a function you can differentiate. Near the bottom point it coincides with the lower half. Each point has its own local function, and the tangent is that function's tangent.
The slopes marked, minus three quarters and three quarters, are what the book's explicit approach would give you by differentiating the two square roots separately. In a few slides you will get both from one line of implicit work.
Intuition
Figure (svg): The curve y to the fifth plus y equals x for x from minus 5 to 5, an S-shaped rising curve through the origin, with a tangent segment of slope one sixth at the point (2, 1).
\[ y^5 + y = x \]
The left side increases with y, so each x gives exactly one y: this is a genuine function. But there is no formula in radicals for the solution of a general fifth-degree equation, so you will never write y equals something.
Solving first is sometimes merely awkward and sometimes impossible. A method for slopes that never solves for y is not a convenience; for curves like this it is the only way in.
This curve passes the vertical line test: as y increases, y to the fifth plus y increases too, so each x comes from exactly one y. So y really is a function of x. The trouble is that you cannot write it down. There is a famous theorem that general fifth-degree equations have no solution in radicals.
Yet the picture shows a perfectly smooth curve with a tangent at (2, 1), and you will be able to compute that slope, one sixth, in a single line once you know the method: differentiate to get five y to the fourth times dy by dx plus dy by dx equals one, and substitute y equal to 1.
Keep this example in mind whenever implicit differentiation feels like a roundabout way of doing something easy. For the circle, solving for y is merely awkward. For curves like this one, it is impossible, and the implicit method is the only door.
Intuition
Implicit differentiation rests on one assumption: near the point you care about, the curve is the graph of some differentiable function y of x.
\[ y = y(x) \quad \text{near the point, even if no formula is known} \]
On the circle that is true everywhere except the two side points, where the curve turns vertical. You will see in Part 5 that the method itself warns you when the promise fails, by asking you to divide by zero.
OpenStax Calculus Volume 1, §3.8 Implicit Differentiation §3.8, p. 268 — portions of y are functions that satisfy the equation
When you differentiate an equation in x and y, you are quietly assuming that near your point the curve is the graph of some differentiable function. You do not need its formula; you only need it to exist, so that the chain rule has something to act on.
For most points on most curves this promise holds. On the circle it fails only at the left and right extremes, where the curve is vertical and no function of x can follow it in both directions.
The pleasant surprise is that the method is honest about its own failures. At points where the promise breaks, the slope formula you derive will ask you to divide by zero. You will learn to read that as a signal rather than an error in Part 5.
Section
Part 2
Concept
Write y as y of x while you differentiate, and the chain rule shows itself.
\[ \frac{d}{dx}\left(y^2\right) = \frac{d}{dx}\left[y(x)\right]^2 \]
\[ = 2y(x)\cdot y'(x) \]
\[ = 2y\,\frac{dy}{dx} \]
An x term differentiates as usual; a y term always leaves a factor of dy by dx behind. That factor is the unknown you are hunting for.
\[ \frac{d}{dx}\left(\sin x\right) = \cos x \qquad \frac{d}{dx}\left(\sin y\right) = \cos y\,\frac{dy}{dx} \]
OpenStax Calculus Volume 1, §3.8 Implicit Differentiation §3.8, pp. 269-270 — Problem-Solving Strategy, step 1
This slide is the whole method in miniature. Write y as y of x for a moment, and the square of y becomes a composition: square the output of the function y. The chain rule then gives the outer derivative, two y, times the inner derivative, y prime.
Once you see it that way you can go back to writing plain y, as long as you remember the factor. Every time you differentiate an expression containing y with respect to x, the result carries a dy by dx.
The last line puts the two sine derivatives side by side. They look almost identical and differ only by that trailing factor. That factor is not a nuisance to be tolerated; it is the unknown slope you are trying to find, and it only appears because you kept it.
Picture it
Figure (svg): Left panel: the upper semicircle y equals the square root of 25 minus x squared, with its tangent at x equals 3, slope minus three quarters. Right panel: the parabola y squared equals 25 minus x squared, with its tangent at x equals 3, slope minus 6.
On the upper half of the circle, y is the square root of 25 minus x squared and y squared is 25 minus x squared. At x equal to 3 the first has slope minus three quarters and the second has slope minus 6.
\[ -6 = 2(4)\left(-\frac34\right) = 2y\,\frac{dy}{dx} \]
Both panels are drawn from the upper half of the circle. On the left is y itself, the semicircle. On the right is y squared, which is just the downward parabola twenty-five minus x squared.
At x equal to three the semicircle has height four and slope minus three quarters. The parabola has height sixteen and slope minus six. Now check the implicit power rule: two times y times dy by dx is two times four times minus three quarters, which is minus six. The numbers match exactly.
So the factor is not an algebraic trick. Squaring a quantity that is changing multiplies its rate of change by twice its current size. When y is four, y squared changes eight times as fast as y does.
Notation
Annotate
On: \( \frac{d}{dx}\left(y^{n}\right) = n\,y^{n-1}\,\frac{dy}{dx} \)
Step through the annotations and connect each piece to the chain rule. The d by dx at the front sets the variable you are differentiating with respect to. It is always x in this section, whatever letter appears inside.
The middle piece is the ordinary power rule, applied as if y were a variable. The final factor is the inner derivative, and it is the difference between a correct answer and a wrong one.
The last annotation is the one to generalise. Any function of y, whether a power, a sine, an exponential or a logarithm, is differentiated by its usual rule and then multiplied by dy by dx. If you can differentiate a function of x, you can differentiate the same function of y.
Concept
A term like x times y is a product of two functions of x, so the product rule comes first and the chain rule rides inside it.
\[ \frac{d}{dx}(xy) = 1\cdot y + x\cdot\frac{dy}{dx} \]
\[ \frac{d}{dx}\left(x^2y^3\right) = 2x\,y^3 + x^2\cdot 3y^2\frac{dy}{dx} \]
\[ \frac{d}{dx}\left(e^{xy}\right) = e^{xy}\left(y + x\frac{dy}{dx}\right) \]
In each line exactly one piece carries the factor: the piece where the y part was differentiated.
OpenStax Calculus Volume 1, §3.8 Implicit Differentiation §3.8, p. 271 — Example 3.69, step 1.2
Terms that mix x and y are where most errors happen, because two rules are needed at once. The product rule comes first: derivative of the first times the second, plus the first times the derivative of the second.
Inside the product rule, each piece is handled on its own merits. The x piece differentiates normally. The y piece differentiates with its chain-rule factor. So exactly one of the two terms carries dy by dx, the one in which the y part was differentiated.
The third line is a composition wrapped around a product: the exponential is differentiated first, and its inside, x times y, needs the product rule. If you slow down and name each rule as you use it, these become routine.
Matching
Match the pairs
Why: Only x cubed comes out with no factor, because it has no y in it. The pure y terms each get one factor; the products get the product rule, with the factor on the piece where y was differentiated.
Before matching, scan each term on the left for a y. If there is none, the derivative will have no dy by dx; if there is one, it will.
The two product terms, x y and x squared y, each give two pieces, and only the second piece of each carries the factor. Notice also that the derivative of cos y has a minus sign from the cosine and a dy by dx from the chain rule; people often keep one and lose the other.
If you made a mismatch, find which rule you skipped. It is almost always either the chain-rule factor on a y, or one half of a product rule.
Sorting
Sort into buckets
Differentiate each term with respect to x in your head. Does the result contain dy/dx?
This is the habit that makes implicit differentiation reliable: before differentiating any term, look for a y in it. Sort each item by that test alone.
The three with a y all produce the factor. The tangent of y gives secant squared y times dy by dx; three x y gives three y plus three x dy by dx; and the square root of y gives dy by dx over twice the root. The constant, the power of x and the sine of x differentiate exactly as they always have.
The mixed term three x y is the one to watch, because only one of its two pieces carries the factor. If you sorted it as producing a dy by dx, you were right; just remember that its derivative also contains a piece without one.
Trap
A tempting first line for the circle:
\[ \frac{d}{dx}\left(x^2 + y^2\right) = 2x + 2y \]
\[ 2x + 2y = 0 \Rightarrow y = -x \]
Wrong. That is the derivative with respect to each letter separately, and the result is a line, not a slope.
The y term needs its chain-rule factor. With it, the equation contains the unknown slope and can be solved for it.
\[ 2x + 2y\,\frac{dy}{dx} = 0 \]
\[ \frac{dy}{dx} = -\frac{x}{y} \]
This is the most common error in the whole topic, and it is tempting because the power rule is such a strong habit. Two x plus two y looks tidy and complete.
Follow where it leads. Setting it equal to zero gives y equal to minus x, which is the equation of a line, not a slope. There is no derivative left anywhere in the answer, and that is your warning sign: an implicit differentiation that does not produce dy by dx has lost it.
The corrected version keeps the factor on the y term, and the equation becomes something you can solve for the slope. Minus x over y is the answer, and you will verify it properly in Example 3.68.
Concept
\[ \underbrace{A\,\frac{dy}{dx} + B\,\frac{dy}{dx}}_{\text{left}} = \underbrace{R}_{\text{right}} \;\Longrightarrow\; \frac{dy}{dx} = \frac{R}{A + B} \]
After step 1 the problem is just algebra: a linear equation in the unknown dy by dx.
OpenStax Calculus Volume 1, §3.8 Implicit Differentiation §3.8, pp. 269-270 — Problem-Solving Strategy: Implicit Differentiation
These four steps are the book's strategy, and they are worth memorising in order. Step 1 is the only step that involves calculus. Everything after it is solving a linear equation for one unknown.
The equation after step 1 is linear in dy by dx because the chain rule only ever produces a single factor of dy by dx per term. So you can always gather those terms, factor the unknown out, and divide.
Occasionally step 3 is not needed, when only one term contains the derivative, as in the circle. And occasionally the dividing bracket turns out to be zero at a point you care about; that is not a failure of the method but important information about the curve, as Part 5 will show.
Section
Part 3
Worked example
Assume the equation defines y implicitly as a function of x, and find dy by dx.
\[ x^2 + y^2 = 25 \]
Differentiate both sides
Why: Step 1 of the strategy: apply the derivative to each side.
\[ \frac{d}{dx}\left(x^2 + y^2\right) = \frac{d}{dx}(25) \]
Use the sum rule
Why: Split the left side; the derivative of the constant is zero.
\[ \frac{d}{dx}\left(x^2\right) + \frac{d}{dx}\left(y^2\right) = 0 \]
Differentiate each term
Why: Power rule on the x term; power rule plus chain-rule factor on the y term.
\[ 2x + 2y\frac{dy}{dx} = 0 \]
Keep the dy/dx term on the left
Why: Step 2: move 2x to the right.
\[ 2y\frac{dy}{dx} = -2x \]
Divide by 2y
Why: Step 3 is not needed, since only one term holds the derivative.
\[ \frac{dy}{dx} = -\frac{x}{y} \]
Check at the point (3, 4)
Why: The point is on the circle, since 9 plus 16 is 25. The slope there is minus three quarters, and the radius to the point has slope four thirds; the product is minus one, so the tangent is perpendicular to the radius, as it must be for a circle.
\[ \left.\frac{dy}{dx}\right|_{(3,4)} = -\frac34, \qquad -\frac34\cdot\frac43 = -1 \]
OpenStax Calculus Volume 1, §3.8 Implicit Differentiation §3.8, p. 270 — Example 3.68
Follow the strategy line by line. Differentiate both sides, split the left with the sum rule, and differentiate each term. The only moment requiring thought is the y squared term, where the chain-rule factor appears.
After that it is algebra. Move two x across, divide by two y, and the slope formula is minus x over y. Step 3 of the strategy is skipped because there is only one term containing the derivative, so there is nothing to factor.
The check uses a fact you know from geometry: a tangent to a circle is perpendicular to the radius at the point of contact. The radius to (3, 4) has slope four thirds, the formula gives minus three quarters, and the product is minus one. When a derivative agrees with an independent fact like that, you can trust it.
Concept
On the upper half the circle is the graph of a formula you can differentiate directly. Do it, and compare.
\[ y = \sqrt{25 - x^2} \]
\[ \frac{dy}{dx} = \frac{-2x}{2\sqrt{25-x^2}} = -\frac{x}{\sqrt{25-x^2}} = -\frac{x}{y} \]
On the lower half y is the negative root, and the same work gives the same final form.
\[ y = -\sqrt{25-x^2} \;\Longrightarrow\; \frac{dy}{dx} = \frac{x}{\sqrt{25-x^2}} = -\frac{x}{y} \]
One formula covers both halves. The point you substitute tells it which half you mean.
OpenStax Calculus Volume 1, §3.8 Implicit Differentiation §3.8, p. 269 — the derivatives of the two halves
For the circle you can solve for y, so it is worth doing once to see that the implicit answer is right. On the upper half y is the positive root, and the chain rule gives minus x over that root, which is exactly minus x over y.
On the lower half y is the negative root. The derivative has the opposite sign, but y also has the opposite sign, so the ratio minus x over y comes out the same. That is not a coincidence: the implicit formula was derived without ever choosing a half.
This is one of the real advantages of the method. The explicit route needs a separate calculation for each branch, and you must first decide which branch your point is on. The implicit route produces one formula, and the point itself carries the information about which branch you mean.
Concept
Figure (svg): The circle of radius 5 with twelve short tangent segments drawn at evenly spaced points; each segment is perpendicular to the radius at its point. Three are labelled with their slopes minus x over y.
\[ \frac{dy}{dx} = -\frac{x}{y} \]
The slope formula needs both coordinates of a point. That is not unfinished work: above and below x equal to 3 the circle has two different slopes, so no formula in x alone could be right.
OpenStax Calculus Volume 1, §3.8 Implicit Differentiation §3.8, p. 270 — Example 3.68, Analysis
Look at the twelve tangent segments round the circle. Each one is perpendicular to its radius, and the formula minus x over y gives its slope. On the right half, where x and y have the same sign in the upper part, the slopes are negative; on the left they are positive.
Students often feel an answer containing y is unfinished and try to remove it. Resist that. At x equal to three the circle passes through two points with slopes of opposite sign, so any formula in x alone would have to give two answers at once. The y in the formula is exactly what distinguishes them.
In practice you will almost always evaluate the formula at a specific point, and then both coordinates are known numbers. The two-variable answer is the right form, not a halfway stage.
Counterexample
\[ x^2 + y^2 = -1 \]
Discussion prompt
Differentiate this equation implicitly exactly as in Example 3.68. What formula do you get, and what is wrong with using it?
Write your answer before revealing. The calculation is identical to the circle's, so the formula is identical too: minus x over y.
But this equation has no solutions. The sum of two squares cannot be negative, so there are no points on this curve and no function y of x that satisfies it. The implicit differentiation ran perfectly and produced a slope formula for an empty set.
The lesson is that the method assumes, rather than proves, that a differentiable function exists. That is why the strategy on the next few slides always begins a tangent-line problem by checking that the given point actually satisfies the equation. If it does not, the slope formula is describing nothing.
Step zero
\[ x^3\sin y + y = 4x + 3 \]
Discussion prompt
Look at each of the four terms before you touch them. Which rule, or rules, does each one need, and which ones will produce a factor of dy by dx?
Take thirty seconds to plan before you write. Look at each term and name the rule it needs. Planning like this is what separates a clean solution from a tangle of half-corrected lines.
The first term is the hard one: it is a product, x cubed times sine y, so the product rule applies, and within the product rule the sine of y needs the chain rule. Only one of the two pieces will carry dy by dx.
The lone y differentiates to dy by dx, which is easy to overlook because there is nothing around it. The right side is ordinary. Once you have this plan, the differentiation on the next slide is mechanical.
Worked example
\[ x^3\sin y + y = 4x + 3 \]
Differentiate both sides
Why: The right side is ordinary; its derivative is 4.
\[ \frac{d}{dx}\left(x^3\sin y + y\right) = \frac{d}{dx}(4x + 3) \]
Use the sum rule on the left
Why: One term at a time.
\[ \frac{d}{dx}\left(x^3\sin y\right) + \frac{d}{dx}(y) = 4 \]
Apply the product rule
Why: x cubed and sin y are both functions of x.
\[ \frac{d}{dx}\left(x^3\right)\sin y + x^3\,\frac{d}{dx}(\sin y) + \frac{dy}{dx} = 4 \]
Differentiate the pieces
Why: Power rule on x cubed; chain rule on sin y.
\[ 3x^2\sin y + x^3\cos y\,\frac{dy}{dx} + \frac{dy}{dx} = 4 \]
Move the term without dy/dx
Why: Step 2 of the strategy.
\[ x^3\cos y\,\frac{dy}{dx} + \frac{dy}{dx} = 4 - 3x^2\sin y \]
Factor out dy/dx
Why: Step 3: both left-hand terms share it.
\[ \frac{dy}{dx}\left(x^3\cos y + 1\right) = 4 - 3x^2\sin y \]
Divide
Why: Step 4.
\[ \frac{dy}{dx} = \frac{4 - 3x^2\sin y}{x^3\cos y + 1} \]
Figure (svg): The curve x cubed sine y plus y equals 4x plus 3 for x from minus 0.7 to 0.72, rising from near the origin to about y equals 6, found by solving the equation numerically. The tangent at (0, 3) has slope 4, and a second point (0.5, 5.115) is marked.
Check against the numerically solved curve
Why: At x equal to one half, solving the equation numerically gives y about 5.1150. The formula gives slope 4.4710 there, and a finite difference of the solved curve gives 4.4710 too. At (0, 3) the formula gives 4 over 1, the slope drawn.
\[ y(0.5) \approx 5.1150: \quad \frac{dy}{dx} \approx 4.4710 \text{ both ways} \]
\[ \left.\frac{dy}{dx}\right|_{(0,3)} = \frac{4 - 0}{0 + 1} = 4 \]
OpenStax Calculus Volume 1, §3.8 Implicit Differentiation §3.8, pp. 270-271 — Example 3.69
Follow the plan from the previous slide. After the sum rule, the product rule splits the first term into two pieces: three x squared times sine y, with no factor, and x cubed times cosine y times dy by dx. The lone y contributes one more dy by dx.
Now the strategy takes over. Move the term without the derivative to the right, factor dy by dx out of the two terms that have it, and divide by the bracket. The answer, like the circle's, depends on both x and y.
The figure is the strongest check available. Nobody can solve this equation for y, but a computer can solve it numerically at each x and trace the curve. At x equal to one half the traced curve has height about 5.115, and both the formula and a finite-difference slope of the traced curve give 4.4710. At (0, 3) the formula gives 4, and the drawn tangent matches.
Error analysis
A first attempt at the first term of Example 3.69:
Annotate
On: \( \frac{d}{dx}\left(x^3\sin y\right) = 3x^2\cos y\,\frac{dy}{dx} \)
Read the line before revealing the notes and try to find what is wrong. The chain-rule factor is present, which is often the part people forget, so the error lies elsewhere.
The mistake is treating the derivative of a product as the product of the derivatives. Three x squared is the derivative of the first factor and cosine y dy by dx is the derivative of the second, and they have simply been multiplied. The product rule never does that; it produces two separate terms.
The corrected derivative has one piece without the factor and one piece with it. If you ever find a single term where you expected two, go back and check whether a product rule was collapsed.
Worked example
Find dy by dx for y defined implicitly by the equation below.
\[ 4x^5 + \tan y = y^2 + 5x \]
Differentiate both sides
Why: The derivative of tan y is sec squared y, times the chain-rule factor.
\[ 20x^4 + \sec^2 y\,\frac{dy}{dx} = 2y\,\frac{dy}{dx} + 5 \]
Gather the dy/dx terms on the left
Why: Subtract the y term; subtract 20x to the fourth.
\[ \sec^2 y\,\frac{dy}{dx} - 2y\,\frac{dy}{dx} = 5 - 20x^4 \]
Factor out dy/dx
Why: Both left-hand terms share it.
\[ \frac{dy}{dx}\left(\sec^2 y - 2y\right) = 5 - 20x^4 \]
Divide
Why: The answer, in both variables.
\[ \frac{dy}{dx} = \frac{5 - 20x^4}{\sec^2 y - 2y} \]
Check at the origin
Why: The origin is on the curve, since both sides are 0 there, and the formula gives 5 over 1. Solving the equation numerically at x equal to plus and minus 0.001 gives y about 0.005025 and minus 0.004975, a difference quotient of 5.0002.
\[ \left.\frac{dy}{dx}\right|_{(0,0)} = \frac{5 - 0}{1 - 0} = 5, \qquad \frac{0.005025 - (-0.004975)}{0.002} \approx 5.000 \]
OpenStax Calculus Volume 1, §3.8 Implicit Differentiation §3.8, p. 271 — Checkpoint 3.48
This checkpoint has y terms on both sides of the equation, which is a good test of step 2. The derivative of tan y is secant squared y times dy by dx, and the derivative of y squared is two y times dy by dx.
Gather both derivative terms on the left, move everything else to the right, factor, and divide. The answer is five minus twenty x to the fourth, over secant squared y minus two y.
The check picks the origin, which lies on the curve because both sides are zero there. At the origin secant squared is one and y is zero, so the formula gives five. Solving the equation numerically just to the left and right of zero and taking the difference quotient gives five to three decimal places. Agreement like that is convincing evidence the algebra is right.
Worked example
\[ xy - \cos(xy) = 1 \]
Differentiate both sides
Why: Product rule on xy; chain rule on the cosine, whose inside is also xy.
\[ y + x\frac{dy}{dx} + \sin(xy)\left(y + x\frac{dy}{dx}\right) = 0 \]
Spot the common factor
Why: The bracket y plus x dy/dx appears twice.
\[ \left(y + x\frac{dy}{dx}\right)\left(1 + \sin(xy)\right) = 0 \]
Divide by the second factor
Why: It is not zero on this curve; see the check.
\[ y + x\frac{dy}{dx} = 0 \]
Solve
Why: Subtract y and divide by x.
\[ \frac{dy}{dx} = -\frac{y}{x} \]
Check by solving for the product
Why: The function u minus cos u is increasing, so the equation forces xy to be the single constant c solving c minus cos c equal to 1, about 1.2834. Then y is c over x, and differentiating that gives minus y over x again. Also sine of 1.2834 is 0.959, so the factor divided out was 1.959, not zero.
\[ xy = c \approx 1.2834 \;\Rightarrow\; y = \frac{c}{x}, \quad y' = -\frac{c}{x^2} = -\frac{y}{x} \]
OpenStax Calculus Volume 1, §3.8 Implicit Differentiation §3.8, p. 274 — Exercise 304
Differentiate carefully. The x y term gives y plus x dy by dx. The cosine of x y differentiates to minus sine of x y times the derivative of its inside, and since the equation subtracts it, the result is plus sine of x y times that same bracket.
Now look for structure instead of expanding everything. The bracket y plus x dy by dx appears in both terms, so it factors out, and the other factor is one plus sine of x y. Dividing by that leaves a very simple equation, and the slope is minus y over x.
The check explains why the answer is so simple. The equation involves x and y only through their product, and the function u minus cosine u is increasing, so the product is forced to be one constant, about 1.2834. The curve is just the hyperbola x y equal to that constant, whose slope is minus y over x. It also confirms that the factor divided out was never zero.
Fill the middle
Differentiate the term below with respect to x. The first piece is given; supply the coefficient of dy by dx.
Fill in the blanks
\frac2xy___\left(x y^___\right) = y^___ + ___\,\frac______
Why: Product rule: the derivative of x is 1, giving y squared; then x times the derivative of y squared, which is 2y times dy/dx. Together the coefficient of dy/dx is 2xy.
Apply the product rule to x times y squared. The first piece, the derivative of x times y squared, is given: it is just y squared.
The second piece is x times the derivative of y squared. That derivative is two y times dy by dx, by the implicit power rule. So the coefficient of dy by dx is two x y.
If you typed two y, you remembered the chain rule but lost the x that multiplies the whole second piece. If you typed two x, you lost the y from the power rule. Both are common; the fix is to write out the product rule in full before simplifying.
Ranking
Put in order
Order the moves of a complete implicit-differentiation solution.
Why: Differentiate first and substitute last. Substituting a point before differentiating freezes x and destroys the relationship you are differentiating.
Drag the moves into order before checking. The first four are the book's strategy. The fifth is the step that turns a slope formula into a number.
The only order worth debating is where substitution goes. It must come last. If you substitute a point before differentiating, you fix x and y at numbers and there is nothing left to differentiate; the trap slide later in Part 4 shows exactly what goes wrong. Substituting after dividing is always safe.
Notice also that checking the point lies on the curve is part of the substitution step. It costs one line and catches typing errors and empty equations alike.
Section
Part 4
Concept
The implicit derivative gives a slope formula. Evaluate it at the point, then use the point-slope form.
\[ m = \left.\frac{dy}{dx}\right|_{(x_0,\,y_0)} \]
\[ y - y_0 = m\,(x - x_0) \]
Before anything else, confirm the point satisfies the equation. A point off the curve gives a meaningless slope, as the empty equation showed.
OpenStax Calculus Volume 1, §3.8 Implicit Differentiation §3.8, p. 271 — Finding Tangent Lines Implicitly
A tangent line needs a point and a slope. The point is given to you; implicit differentiation supplies the slope formula, and substituting both coordinates of the point turns it into a number.
From there it is the point-slope form you have used since algebra. Most examples in this part end by solving for y to get slope-intercept form, which makes the line easy to graph and compare.
Make a habit of checking that the point is really on the curve before anything else. It takes a line of arithmetic, and it protects you both from copying mistakes and from the empty-curve situation you saw earlier, where the formula happily produces slopes for points that do not exist.
Worked example
\[ x^2 + y^2 = 25 \quad \text{at} \quad (3, -4) \]
Confirm the point
Why: Nine plus sixteen is twenty-five.
\[ 3^2 + (-4)^2 = 25 \]
Reuse the derivative from Example 3.68
Why: No need to differentiate again.
\[ \frac{dy}{dx} = -\frac{x}{y} \]
Substitute the point
Why: x is 3 and y is minus 4.
\[ m = -\frac{3}{-4} = \frac34 \]
Write the point-slope form
Why: Through (3, minus 4) with slope three quarters.
\[ y - (-4) = \frac34(x - 3) \]
Solve for y
Why: Distribute, then subtract 4.
\[ y = \frac34 x - \frac94 - 4 = \frac34 x - \frac{25}{4} \]
Figure (svg): The circle x squared plus y squared equals 25 and the line y equals three quarters x minus 25 over 4 touching it at (3, minus 4). A dashed radius runs from the origin to the point, at right angles to the line.
Check the distance from the centre
Why: A line tangent to a circle sits exactly one radius from the centre. Rewrite the line and measure.
\[ 3x - 4y - 25 = 0: \quad \frac{|0 - 0 - 25|}{\sqrt{3^2 + 4^2}} = \frac{25}{5} = 5 \]
OpenStax Calculus Volume 1, §3.8 Implicit Differentiation §3.8, p. 272 — Example 3.71 and Figure 3.31
The slope formula was already found in Example 3.68, so this example is pure substitution. With x equal to three and y equal to minus four, minus x over y is three quarters, positive, as you would expect at the bottom right of the circle where the curve is rising.
Point-slope form and a little arithmetic give the book's answer, y equals three quarters x minus twenty-five quarters. The figure shows the line touching the circle at one point and continuing below it.
The check uses the distance from a point to a line. A line tangent to a circle is exactly one radius from the centre. Rewriting the tangent as three x minus four y minus twenty-five equals zero and measuring from the origin gives twenty-five over five, which is five, the radius. A line any closer would cut the circle twice, and any further would miss it.
Concept
The implicit derivative proves the geometry fact you may have taken on trust. At a point off the axes, compare the two slopes.
\[ \text{radius: } \frac{y}{x} \qquad \text{tangent: } -\frac{x}{y} \]
\[ \frac{y}{x}\cdot\left(-\frac{x}{y}\right) = -1 \]
Slopes that multiply to minus one belong to perpendicular lines. The same derivation works for a circle of any radius, because the radius only affects the constant, which differentiates to zero.
You have probably been told since school geometry that a tangent to a circle is perpendicular to the radius. Here is a proof in one line of calculus.
The radius from the origin to a point has slope y over x. The tangent's slope, from implicit differentiation, is minus x over y. Multiply them and everything cancels to minus one, which is the condition for perpendicular lines.
The argument never used the value twenty-five, only the fact that the right-hand side is a constant, whose derivative is zero. So it holds for a circle centred at the origin of any radius. The points where x or y is zero, where one of those slopes is undefined, are the horizontal and vertical tangents, and the result holds there too if you read it geometrically.
Worked example
\[ y^3 + x^3 - 3xy = 0 \quad \text{at} \quad \left(\tfrac32, \tfrac32\right) \]
Differentiate term by term
Why: Chain rule on y cubed; product rule on 3xy.
\[ 3y^2\frac{dy}{dx} + 3x^2 - \left(3y + 3x\frac{dy}{dx}\right) = 0 \]
Collect the dy/dx terms on the left
Why: Move 3x squared and minus 3y across.
\[ 3y^2\frac{dy}{dx} - 3x\frac{dy}{dx} = 3y - 3x^2 \]
Factor out dy/dx
Why: Both terms share it.
\[ \frac{dy}{dx}\left(3y^2 - 3x\right) = 3y - 3x^2 \]
Divide
Why: The book's slope formula.
\[ \frac{dy}{dx} = \frac{3y - 3x^2}{3y^2 - 3x} \]
Substitute the point
Why: Three halves squared is nine quarters.
\[ m = \frac{\frac92 - \frac{27}{4}}{\frac{27}{4} - \frac92} = \frac{-\frac94}{\frac94} = -1 \]
Write the tangent line
Why: Point-slope, then simplify.
\[ y - \tfrac32 = -\left(x - \tfrac32\right) \;\Longrightarrow\; y = -x + 3 \]
Figure (svg): The folium of Descartes, y cubed plus x cubed minus 3xy equals 0: a loop in the first quadrant and two arms running off to the upper left and lower right. The line y equals minus x plus 3 touches the loop at (3/2, 3/2).
Check that the point is on the folium
Why: Both cubes are twenty-seven eighths, and 3xy is twenty-seven quarters. The folium is also symmetric in the line y equal to x, so at a point on that line the tangent must cross it at right angles: slope minus one.
\[ \tfrac{27}{8} + \tfrac{27}{8} - 3\cdot\tfrac32\cdot\tfrac32 = \tfrac{27}{4} - \tfrac{27}{4} = 0 \]
OpenStax Calculus Volume 1, §3.8 Implicit Differentiation §3.8, pp. 272-273 — Example 3.72 and Figure 3.32
The folium is a classic curve that cannot sensibly be solved for y: it is a cubic in y, and near its loop there are three y values for some x. Implicit differentiation handles it without complaint.
Differentiate term by term: y cubed gives three y squared dy by dx; x cubed is ordinary; and the product three x y gives three y plus three x dy by dx, with the whole thing subtracted. Gather the derivative terms, factor, and divide to get three y minus three x squared, over three y squared minus three x.
At the point, both coordinates are three halves. The numerator and denominator come out equal and opposite, so the slope is minus one, and the tangent is y equals minus x plus three. The check confirms the point is on the curve, and the symmetry of the folium, which is unchanged when you swap x and y, makes a slope of minus one exactly what you should expect on the line y equal to x.
Worked example
A rocket moves on the ellipse below and fires a missile along the tangent from the point (3, 8/5). Where does the missile cross the x-axis?
\[ 4x^2 + 25y^2 = 100 \]
Differentiate
Why: Chain-rule factor on the y term; the constant gives zero.
\[ 8x + 50y\frac{dy}{dx} = 0 \]
Solve for dy/dx
Why: Subtract 8x, divide by 50y, reduce.
\[ \frac{dy}{dx} = -\frac{8x}{50y} = -\frac{4x}{25y} \]
Substitute the firing point
Why: x is 3 and y is eight fifths.
\[ m = -\frac{4(3)}{25\left(\frac85\right)} = -\frac{12}{40} = -\frac{3}{10} \]
Write the tangent line
Why: Point-slope form, then solve for y.
\[ y - \frac85 = -\frac{3}{10}(x - 3) \;\Longrightarrow\; y = -\frac{3}{10}x + \frac52 \]
Set y equal to zero
Why: The x-axis is where y is zero.
\[ 0 = -\frac{3}{10}x + \frac52 \;\Longrightarrow\; x = \frac52\cdot\frac{10}{3} = \frac{25}{3} \]
Figure (svg): The ellipse 4x squared plus 25y squared equals 100, five units wide each side and two units tall. From the rocket at (3, 8/5) the tangent line runs down to the right and meets the x-axis at (25/3, 0), where an arrow shows the asteroid arriving along the positive x-axis.
Check both points
Why: The firing point is on the ellipse, since 36 plus 64 is 100, and the landing point is on the tangent line.
\[ 4(9) + 25\left(\tfrac{64}{25}\right) = 100, \qquad -\frac{3}{10}\cdot\frac{25}{3} + \frac52 = 0 \]
OpenStax Calculus Volume 1, §3.8 Implicit Differentiation §3.8, pp. 273-274 — Example 3.73
The story hides a familiar problem: find the tangent line at the firing point, then find where it crosses the x-axis. The missile travels along the tangent because at the moment of firing it is moving in the rocket's direction.
Differentiate the ellipse: eight x plus fifty y dy by dx equals zero, so the slope is minus four x over twenty-five y. At (3, eight fifths) that is minus three tenths, a gentle downward slope. The tangent line is y equals minus three tenths x plus five halves.
Setting y to zero gives the landing point, x equal to twenty-five thirds, a little past eight. The figure shows the line leaving the ellipse and meeting the asteroid's path there. The check confirms that the firing point really lies on the orbit and that the landing point really lies on the line.
Estimation
\[ 4x^2 + 25y^2 = 100, \quad \text{firing point } \left(4, \tfrac65\right) \]
Predict first
The rocket moves on to (4, 6/5), lower and further right, and fires along its tangent again. Where does the missile cross the x-axis now?
Correct: Closer in, at x = 6.25
Why: The slope there is −16/30 = −8/15, steeper than before, so the line reaches the axis sooner: y − 6/5 = −(8/15)(x − 4) gives x = 4 + (6/5)(15/8) = 6.25. In general the tangent at (x₀, y₀) lands at 25/x₀, so 25/4 = 6.25, and from (3, 8/5) it was 25/3.
\[ 4x_0x + 25y_0y = 100 \;\Longrightarrow\; x_{\text{land}} = \frac{25}{x_0} \]
Make a guess before you calculate. The new firing point is further right and lower down the orbit, where the ellipse is falling more steeply.
A steeper tangent reaches the x-axis sooner, so the missile lands closer in. The slope at (4, six fifths) is minus sixteen thirtieths, which is minus eight fifteenths, and the line lands at x equal to six and a quarter.
The formula under the question gives a shortcut worth noticing. The tangent to this ellipse at any point on it can be written with the point's coordinates in place of one x and one y, and its x-intercept is always twenty-five divided by the firing point's x-coordinate. From x equal to three that is twenty-five thirds, as in the example; from x equal to four it is six and a quarter.
Tweak it
Parameter explorer
The line drawn is the tangent to the ellipse 4x² + 25y² = 100 at the point on its upper half with x-coordinate a. Slide a from 1 to 4.9 and watch where the line meets the x-axis.
\[ x_{\text{land}} = \frac{25}{{a}} \]
Start with a at three, the book's firing point, and confirm the line crosses the axis a little past eight. Then slide a towards the top of the ellipse and watch the line flatten and its crossing point run far to the right.
Now slide a towards five, the right-hand tip. The tangent becomes steeper and steeper, and its crossing point moves in towards five. At the tip itself the tangent would be vertical and cross the axis exactly at five.
The readout shows the landing point, twenty-five over a, updating as you slide. Watching the line and the formula move together is a good way to see that the implicit derivative really does describe the tangent at every point, not just the one in the example.
Worked example
\[ x^2 - y^2 = 16 \quad \text{at} \quad (5, 3) \]
Confirm the point
Why: Twenty-five minus nine is sixteen.
\[ 5^2 - 3^2 = 16 \]
Differentiate
Why: The y term brings a factor and a minus sign.
\[ 2x - 2y\frac{dy}{dx} = 0 \]
Solve for dy/dx
Why: Add the y term to both sides and divide by 2y.
\[ \frac{dy}{dx} = \frac{x}{y} \]
Substitute the point
Why: Five over three.
\[ m = \frac53 \]
Write the tangent line
Why: Point-slope, then simplify.
\[ y - 3 = \frac53(x - 5) \;\Longrightarrow\; y = \frac53x - \frac{16}{3} \]
Figure (svg): The hyperbola x squared minus y squared equals 16, two branches opening left and right with dashed asymptotes y equals x and y equals minus x. The line y equals five thirds x minus sixteen thirds touches the right branch at (5, 3).
Check that the line touches rather than cuts
Why: Substitute the line into the hyperbola: a tangent must produce a repeated root at x equal to 5.
\[ x^2 - \left(\frac{5x-16}{3}\right)^2 = 16 \iff 16x^2 - 160x + 400 = 0 \]
\[ \iff 16(x - 5)^2 = 0 \]
OpenStax Calculus Volume 1, §3.8 Implicit Differentiation §3.8, p. 274 — Checkpoint 3.49
The hyperbola's equation has a minus sign in front of y squared, and that sign carries through: differentiating gives two x minus two y dy by dx, and solving gives the slope x over y. At (5, 3) that is five thirds.
The tangent line is y equals five thirds x minus sixteen thirds. In the figure it is a little steeper than the asymptote through the origin, because near the vertex the branch rises more steeply than it eventually does.
The check uses an algebraic characterisation of tangency. Substitute the line into the hyperbola and you get a quadratic in x. A line that cuts the curve would give two different roots; a tangent gives one repeated root. Here the quadratic is sixteen times x minus five, squared, so the line meets the curve only at x equal to five, doubly. That is exactly what a tangent should do.
Worked example
\[ x^2y^2 + 5xy = 14 \quad \text{at} \quad (2, 1) \]
Confirm the point
Why: Four plus ten is fourteen.
\[ (4)(1) + 5(2)(1) = 14 \]
Differentiate
Why: Product rule on both terms.
\[ 2xy^2 + 2x^2y\frac{dy}{dx} + 5y + 5x\frac{dy}{dx} = 0 \]
Collect and factor
Why: Terms with dy/dx on the left.
\[ \frac{dy}{dx}\left(2x^2y + 5x\right) = -\left(2xy^2 + 5y\right) \]
Divide
Why: The slope formula.
\[ \frac{dy}{dx} = -\frac{2xy^2 + 5y}{2x^2y + 5x} \]
Substitute the point
Why: Numerator 4 plus 5, denominator 8 plus 10.
\[ m = -\frac{4 + 5}{8 + 10} = -\frac12 \]
Write the tangent line
Why: Point-slope.
\[ y - 1 = -\tfrac12(x - 2) \;\Longrightarrow\; y = -\tfrac12x + 2 \]
Figure (svg): The equation x squared y squared plus 5xy equals 14 drawn as two hyperbolas: xy equals 2 in the first and third quadrants and xy equals minus 7 in the second and fourth. The line y equals minus half x plus 2 touches xy equals 2 at (2, 1).
Check by factoring
Why: Treat xy as one unknown u. The equation factors, so the curve through (2, 1) is xy equal to 2, whose slope is minus 2 over x squared.
\[ u^2 + 5u - 14 = (u + 7)(u - 2) = 0 \]
\[ y = \frac{2}{x} \;\Rightarrow\; y' = -\frac{2}{x^2} = -\frac{2}{4} = -\frac12 \]
OpenStax Calculus Volume 1, §3.8 Implicit Differentiation §3.8, p. 274 — Exercise 311
Both terms on the left are products, so the product rule is needed twice. Keep the pieces in order: two x y squared, plus x squared times two y dy by dx, then five y, plus five x dy by dx.
Gather the two derivative terms, factor, and divide. At (2, 1) the slope is minus nine over eighteen, which is minus one half, and the tangent line is y equals minus half x plus two.
The check reveals something the differentiation never needed. If you call x y a single unknown u, the equation is a quadratic in u that factors, so x y equals two or x y equals minus seven. The graph is two separate hyperbolas, drawn in two colours, and the point is on the first one. Differentiating y equal to two over x directly gives minus one half again.
Concept
normal line — The line through a point of a curve perpendicular to the tangent line there.
\[ m_{\text{normal}} = -\frac{1}{m_{\text{tangent}}} \]
It uses the same point and the same derivative, with the slope flipped and negated. When the tangent is horizontal the normal is vertical, and the other way round.
OpenStax Calculus Volume 1, §3.8 Implicit Differentiation §3.8, p. 274 — Exercises 316 and 317
The normal line at a point is the line through it at right angles to the tangent. It uses exactly the same ingredients: the point and the derivative. Only the final slope changes.
Perpendicular lines have slopes whose product is minus one, so the normal's slope is the negative reciprocal of the tangent's. A tangent slope of five quarters gives a normal slope of minus four fifths; a tangent slope of one gives a normal slope of minus one.
Keep the two special cases in mind. A horizontal tangent has a vertical normal, and a vertical tangent has a horizontal normal. In those cases the negative reciprocal formula breaks down, and you write the line directly as x or y equal to a constant.
Worked example
\[ 2x^3 + 2y^3 - 9xy = 0 \quad \text{at} \quad (2, 1) \]
Confirm the point
Why: Sixteen plus two minus eighteen.
\[ 2(8) + 2(1) - 9(2)(1) = 0 \]
Differentiate
Why: Chain rule on the y cube, product rule on 9xy.
\[ 6x^2 + 6y^2\frac{dy}{dx} - 9y - 9x\frac{dy}{dx} = 0 \]
Collect and factor
Why: Terms with dy/dx on the left.
\[ \frac{dy}{dx}\left(6y^2 - 9x\right) = 9y - 6x^2 \]
Divide and substitute
Why: Numerator 9 minus 24, denominator 6 minus 18.
\[ \frac{dy}{dx} = \frac{9y - 6x^2}{6y^2 - 9x}, \quad m = \frac{-15}{-12} = \frac54 \]
Tangent line (part a)
Why: Point-slope with slope five quarters.
\[ y - 1 = \tfrac54(x - 2) \;\Longrightarrow\; y = \tfrac54x - \tfrac32 \]
Normal line (part b)
Why: Slope minus four fifths through the same point.
\[ y - 1 = -\tfrac45(x - 2) \;\Longrightarrow\; y = -\tfrac45x + \tfrac{13}{5} \]
Figure (svg): The folium 2x cubed plus 2y cubed minus 9xy equals 0, a loop in the first quadrant, with the tangent line y equals five quarters x minus three halves and the normal line y equals minus four fifths x plus thirteen fifths crossing at right angles at (2, 1).
Check both lines
Why: Each passes through (2, 1), and their slopes multiply to minus one.
\[ \tfrac54(2) - \tfrac32 = 1, \quad -\tfrac45(2) + \tfrac{13}{5} = 1, \quad \tfrac54\cdot\left(-\tfrac45\right) = -1 \]
OpenStax Calculus Volume 1, §3.8 Implicit Differentiation §3.8, p. 274 — Exercise 316
This is the folium again, stretched by a factor of one and a half, so its loop reaches further out. The differentiation is the same pattern as Example 3.72: chain rule on the y cube, product rule on nine x y.
At (2, 1) the slope formula gives minus fifteen over minus twelve, which is five quarters. The tangent line follows from point-slope form, and the normal uses the negative reciprocal, minus four fifths.
In the figure the two lines cross at right angles at the point, with the tangent grazing the loop and the normal cutting straight through it. The check confirms both lines pass through (2, 1) and that their slopes multiply to minus one, which is the whole definition of the normal line.
Worked example
\[ x^2 + 2xy - 3y^2 = 0 \quad \text{at} \quad (1, 1) \]
Differentiate
Why: Product rule on 2xy, chain rule on the y square.
\[ 2x + 2y + 2x\frac{dy}{dx} - 6y\frac{dy}{dx} = 0 \]
Collect, factor, divide
Why: Dividing top and bottom by 2 at the end.
\[ \frac{dy}{dx}(2x - 6y) = -(2x + 2y) \;\Longrightarrow\; \frac{dy}{dx} = \frac{x + y}{3y - x} \]
Slope of the tangent, then of the normal
Why: At (1, 1) the tangent slope is 2 over 2.
\[ m = \frac{2}{2} = 1, \qquad m_{\text{normal}} = -1 \]
The normal line (part a)
Why: Through (1, 1) with slope minus one.
\[ y - 1 = -(x - 1) \;\Longrightarrow\; y = -x + 2 \]
Factor the curve (part b)
Why: The quadratic in x and y splits into two lines.
\[ x^2 + 2xy - 3y^2 = (x - y)(x + 3y) = 0 \]
Intersect the normal with the other line
Why: Substitute y equal to minus x plus 2 into x plus 3y equal to 0.
\[ x + 3(2 - x) = 0 \;\Longrightarrow\; x = 3, \; y = -1 \]
Figure (svg): The graph of x squared plus 2xy minus 3y squared equals 0, which is the pair of lines y equals x and y equals minus x over 3. The dashed normal line y equals minus x plus 2 passes through (1, 1) on the first line and meets the second line at (3, minus 1).
Check the new point
Why: It satisfies the original equation.
\[ 3^2 + 2(3)(-1) - 3(-1)^2 = 9 - 6 - 3 = 0 \]
OpenStax Calculus Volume 1, §3.8 Implicit Differentiation §3.8, p. 274 — Exercise 317
The differentiation is routine and gives a slope of one at (1, 1), so the normal has slope minus one and equation y equals minus x plus two.
Part b asks where else the normal meets the curve, and here the curve holds a surprise. The left side factors, so the graph is not a curved shape at all but the pair of lines y equal to x and y equal to minus x over three. The point (1, 1) is on the first line, which explains the slope of one, since the tangent to a line is the line itself.
The normal crosses the second line at (3, minus 1), found by substituting the normal into x plus three y equal to zero. The check confirms the new point satisfies the original equation. Factoring before differentiating would have made this exercise almost trivial; factoring afterwards at least makes it clear.
Trap
Asked for the slope of the circle at (3, minus 4), someone substitutes x first:
\[ 3^2 + y^2 = 25 \]
\[ 2y\,y' = 0 \Rightarrow y' = 0 \]
Wrong. Setting x equal to 3 froze it.
With x fixed at 3 the equation only describes two points, and a point has no slope. Differentiate the whole relationship first, and substitute last.
\[ 2x + 2y\,y' = 0 \]
\[ y' = -\frac{x}{y} = -\frac{3}{-4} = \frac34 \]
This mistake comes from a good instinct, wanting to simplify early, applied at the wrong moment. Once you replace x by three, the equation no longer describes a curve. It describes only the two values of y that go with x equal to three.
Differentiating that frozen equation gives nonsense, because x can no longer change. The derivative of three squared is zero, and the whole calculation collapses to a slope of zero, which is wrong: the picture in Example 3.71 clearly shows a rising tangent.
The rule is simple. Differentiate the whole relationship while x and y are still free to vary, solve for the slope formula, and only then substitute the numbers.
Section
Part 5
Concept
Figure (svg): The circle of radius 5 with horizontal tangent segments at its top and bottom points (0, 5) and (0, minus 5), and vertical tangent segments at its leftmost and rightmost points (minus 5, 0) and (5, 0).
Most implicit derivatives come out as one expression over another. A fraction is zero when its top is zero, and it blows up when its bottom is zero.
\[ \frac{dy}{dx} = -\frac{x}{y}: \quad x = 0 \Rightarrow \text{flat}, \quad y = 0 \Rightarrow \text{vertical} \]
On the circle the tangent is horizontal at (0, 5) and (0, minus 5), and vertical at (5, 0) and (minus 5, 0).
Implicit slope formulas usually come out as a fraction, and a fraction carries two pieces of information. Its top tells you where it is zero; its bottom tells you where it is undefined.
Look at the circle. The formula minus x over y is zero when x is zero, at the top and bottom points, where the tangent is flat. It is undefined when y is zero, at the far left and right, where the tangent is vertical. The figure shows all four.
Notice that the zeros have to be combined with the equation of the curve. Setting x equal to zero gives a whole vertical line; only the two places where that line meets the circle are actual points with horizontal tangents.
Notation
Annotate
On: \( \frac{dy}{dx} = \frac{N(x, y)}{D(x, y)} \)
Step through the four annotations. The first two are the useful cases: a zero on top with a nonzero bottom means a horizontal tangent, and a zero on the bottom with a nonzero top means a vertical tangent.
The third case is the one to treat with care. When both are zero, the fraction tells you nothing, and the curve might do almost anything there: cross itself, come to a sharp point, or even be perfectly smooth. You will meet the most famous example, the folium's origin, in two slides.
The fourth annotation is a reminder that every candidate must lie on the curve. Setting the numerator to zero gives a new equation; solve it together with the original to get actual points.
Worked example
\[ y^3 + x^3 - 3xy = 0, \qquad \frac{dy}{dx} = \frac{3y - 3x^2}{3y^2 - 3x} \]
Set the numerator to zero
Why: A horizontal tangent needs slope zero.
\[ 3y - 3x^2 = 0 \;\Longrightarrow\; y = x^2 \]
Substitute into the curve
Why: The point must also lie on the folium.
\[ \left(x^2\right)^3 + x^3 - 3x\cdot x^2 = 0 \]
Simplify
Why: Collect the cubes.
\[ x^6 - 2x^3 = 0 \]
Factor
Why: Take out x cubed.
\[ x^3\left(x^3 - 2\right) = 0 \]
Solve
Why: Two candidates.
\[ x = 0 \quad\text{or}\quad x = \sqrt[3]{2} \approx 1.260 \]
Find y from y equal to x squared
Why: The second candidate.
\[ y = \left(\sqrt[3]{2}\right)^2 = 2^{2/3} \approx 1.587 \]
Discard the origin
Why: There the denominator is zero as well, so it is not a horizontal tangent; the next slide looks at it.
\[ (0, 0): \quad \frac{0}{0} \]
Use the symmetry for the vertical tangent
Why: Swapping x and y leaves the equation unchanged, so the swapped point has the swapped behaviour.
\[ \left(2^{2/3}, 2^{1/3}\right) \approx (1.587, 1.260) \]
Figure (svg): A close-up of the folium's loop between 0 and 2 on both axes. A horizontal tangent segment sits at the loop's highest point, about (1.26, 1.59), and a vertical tangent segment at its rightmost point, about (1.59, 1.26). The dotted parabola y equals x squared passes through the highest point.
Check the denominator at the top point
Why: It must not vanish, or the slope would not be zero there.
\[ 3y^2 - 3x = 3\cdot 2^{4/3} - 3\cdot 2^{1/3} = 3\cdot 2^{1/3} \approx 3.780 \ne 0 \]
OpenStax Calculus Volume 1, §3.8 Implicit Differentiation §3.8, pp. 272-273 — the folium of Example 3.72
The loop of the folium must have a highest point, where the tangent is flat. So set the numerator of the slope formula to zero. That gives y equal to x squared, a parabola, not yet a point.
Solve it together with the curve's equation by substituting. The cubes collapse to x to the sixth minus two x cubed, which factors. The root x equal to zero is the origin, where the denominator also vanishes, so it is not a horizontal tangent. The other root is the cube root of two, about 1.260, and y is its square, about 1.587.
The folium's equation does not change if you swap x and y, so the swapped point is where the denominator vanishes and the tangent is vertical: the rightmost point of the loop. The dotted parabola in the figure passes through the top point, confirming the substitution. The check makes sure the denominator is not zero at the top point, since otherwise the slope would not really be zero.
Anomaly
Figure (svg): A close-up of the folium near the origin, from minus 0.6 to 0.6 on both axes. Two branches pass through the origin: one leaves along the x-axis, the other along the y-axis, and both axes are highlighted there as tangent lines.
\[ \left.\frac{3y - 3x^2}{3y^2 - 3x}\right|_{(0,0)} = \frac{0}{0} \]
Predict first
The slope formula gives zero over zero at the origin. What does the curve actually do there?
Correct: Two branches cross, one horizontal and one vertical
Why: Near the origin one branch behaves like y ≈ x²/3, leaving along the x-axis, and the other like x ≈ y²/3, leaving along the y-axis. Near the origin the curve is not the graph of any single function, so the method's assumption fails, and the formula signals it with zero over zero.
Commit to an answer before revealing. The formula gives zero over zero at the origin, which on its own tells you nothing.
The close-up figure shows what is happening. Two branches of the folium pass through the origin. One leaves horizontally, hugging the x-axis like a small parabola; the other leaves vertically, hugging the y-axis. So there are two tangent lines through one point, the two axes themselves.
No single slope could possibly describe that, and near the origin the curve is not the graph of any one function of x. The promise behind implicit differentiation fails, and the formula reports it the only way it can, with zero over zero. When you meet this in an exercise, look at the curve before trusting any number.
Worked example
Find all points on the graph at which the tangent line is vertical.
\[ y^3 - 27y = x^2 - 90 \]
Differentiate
Why: Chain-rule factors on both y terms.
\[ 3y^2\frac{dy}{dx} - 27\frac{dy}{dx} = 2x \]
Factor and divide
Why: The slope formula.
\[ \frac{dy}{dx} = \frac{2x}{3y^2 - 27} \]
Set the denominator to zero
Why: A vertical tangent needs an upright slope.
\[ 3y^2 - 27 = 0 \iff y = \pm 3 \]
Find x when y is 3
Why: Substitute into the curve.
\[ 27 - 81 = x^2 - 90 \iff x^2 = 36 \iff x = \pm 6 \]
Find x when y is minus 3
Why: Substitute again.
\[ -27 + 81 = x^2 - 90 \iff x^2 = 144 \iff x = \pm 12 \]
Figure (svg): The curve y cubed minus 27y equals x squared minus 90, symmetric about the y-axis: it opens upward from a low point near (0, minus 6.4), bulges out to x equals plus or minus 12 at height minus 3, pinches in to x equals plus or minus 6 at height 3, then widens again. Vertical tangent segments mark the four points (plus or minus 6, 3) and (plus or minus 12, minus 3).
Check the numerator is not zero
Why: At all four points the numerator 2x is nonzero, so none is a zero-over-zero case; the four points (6, 3), (minus 6, 3), (12, minus 3) and (minus 12, minus 3) are genuine vertical tangents.
\[ 2x = \pm 12 \text{ or } \pm 24 \ne 0 \]
OpenStax Calculus Volume 1, §3.8 Implicit Differentiation §3.8, p. 274 — Exercise 318
This exercise asks only about vertical tangents, so you only need the denominator of the slope formula. Differentiate, factor out dy by dx, and the slope is two x over three y squared minus twenty-seven.
The denominator vanishes when y squared is nine, so y is three or minus three. Each of those heights, substituted back into the curve's equation, gives two values of x: plus or minus six at height three, and plus or minus twelve at height minus three.
The figure shows why. The curve is symmetric about the y-axis and pinches in at height three and bulges out at height minus three; at each pinch and bulge it is momentarily vertical. The check confirms the numerator is nonzero at all four points, so none is a zero-over-zero case.
Sorting
\[ 4x^2 + 25y^2 = 100, \qquad \frac{dy}{dx} = -\frac{4x}{25y} \]
Sort into buckets
Each point lies on the rocket's ellipse. Sort it by the kind of tangent line there.
Use the slope formula under the equation. For each point, work out the numerator four x and the denominator twenty-five y, and ask which, if either, is zero.
The top and bottom of the ellipse have x equal to zero, so the numerator is zero and the tangent is flat. The two ends have y equal to zero, so the denominator is zero and the tangent is upright.
The last two points are ordinary. The rocket's firing point has slope minus three tenths, and the opposite point, reflected through the centre, has exactly the same slope, because an ellipse is symmetric under that reflection and parallel tangents appear at opposite points.
Section
Part 6
Concept
The first derivative is itself an expression in x and y. Differentiate it with respect to x, remembering that y is still a function of x.
\[ \frac{d^2y}{dx^2} = \frac{d}{dx}\left(\frac{dy}{dx}\right) \]
A dy by dx will appear inside the result. Replace it with the first-derivative formula, then use the original equation to simplify if you can.
OpenStax Calculus Volume 1, §3.8 Implicit Differentiation §3.8, p. 271 — Example 3.70
The second derivative is the derivative of the first derivative, as always. What is new is that the first derivative is a formula in x and y, and y still depends on x. So differentiating it produces a dy by dx, just as the original equation did.
That leaves you with an expression containing dy by dx, which you already know. Substitute the formula for it. The result is a second derivative in terms of x and y only.
The last step is optional but often dramatic: the original equation may let you replace a combination of x and y by a constant. For the circle, x squared plus y squared becomes twenty-five, and the second derivative shrinks to a single term.
Worked example
\[ x^2 + y^2 = 25, \qquad \frac{dy}{dx} = -\frac{x}{y} \]
Differentiate both sides of the first-derivative formula
Why: With respect to x. The book's margin says with respect to y here, a misprint: every derivative in this example is with respect to x.
\[ \frac{d^2y}{dx^2} = \frac{d}{dx}\left(-\frac{x}{y}\right) \]
Apply the quotient rule
Why: The top x has derivative 1; the bottom y has derivative dy/dx.
\[ \frac{d^2y}{dx^2} = -\frac{1\cdot y - x\frac{dy}{dx}}{y^2} \]
Distribute the minus sign
Why: Into the numerator.
\[ \frac{d^2y}{dx^2} = \frac{-y + x\frac{dy}{dx}}{y^2} \]
Substitute the first derivative
Why: Replace dy/dx by minus x over y.
\[ \frac{d^2y}{dx^2} = \frac{-y + x\left(-\frac{x}{y}\right)}{y^2} \]
Clear the small fraction
Why: Multiply top and bottom by y.
\[ \frac{d^2y}{dx^2} = \frac{-y^2 - x^2}{y^3} \]
Use the original equation
Why: x squared plus y squared is 25.
\[ \frac{d^2y}{dx^2} = -\frac{x^2 + y^2}{y^3} = -\frac{25}{y^3} \]
Check against the explicit upper half at (3, 4)
Why: Differentiating the root twice gives minus 25 over the three-halves power of 25 minus x squared. At x equal to 3 both routes give minus 25 over 64.
\[ -\frac{25}{4^3} = -0.390625 = -\frac{25}{(25 - 9)^{3/2}} \]
OpenStax Calculus Volume 1, §3.8 Implicit Differentiation §3.8, p. 271 — Example 3.70
Start from minus x over y and differentiate it with respect to x using the quotient rule. The top, x, has derivative one. The bottom, y, has derivative dy by dx, and that is where the chain rule enters. The book's margin note says with respect to y at this step, which is a misprint; everything here is differentiated with respect to x.
Distribute the minus sign, then replace dy by dx by minus x over y. The result has a fraction inside a fraction, so multiply top and bottom by y to clear it. You reach minus y squared minus x squared, over y cubed.
Now use the circle's equation: x squared plus y squared is twenty-five, so the second derivative is minus twenty-five over y cubed. The check differentiates the explicit upper half twice and evaluates at x equal to three. Both routes give minus twenty-five over sixty-four, about minus 0.3906.
Picture it
Figure (svg): The circle of radius 5 with its upper half drawn in red, labelled y positive so the second derivative is negative, concave down, and its lower half in green, labelled y negative so the second derivative is positive, concave up.
\[ \frac{d^2y}{dx^2} = -\frac{25}{y^3}: \quad y > 0 \Rightarrow \text{negative}, \quad y < 0 \Rightarrow \text{positive} \]
The second derivative's sign depends only on which half of the circle you are on. The upper half bends down, like a cap; the lower half bends up, like a cup.
The second derivative tells you which way a curve bends. A negative second derivative means concave down, bending like a cap; a positive one means concave up, bending like a cup.
For the circle the second derivative is minus twenty-five over y cubed. Cubing keeps the sign, so on the upper half, where y is positive, the second derivative is negative, and on the lower half it is positive. The figure colours the two halves accordingly.
This matches what you see, and that is the point: an abstract formula has turned into a statement about the shape. It also shows the second derivative growing without bound as y approaches zero, at the sides of the circle, where the curve turns vertical and bends ever more sharply as a graph of x.
Worked example
Continue Checkpoint 3.49 one derivative further.
\[ x^2 - y^2 = 16, \qquad \frac{dy}{dx} = \frac{x}{y} \]
Differentiate with the quotient rule
Why: The bottom y has derivative dy/dx.
\[ \frac{d^2y}{dx^2} = \frac{1\cdot y - x\frac{dy}{dx}}{y^2} \]
Substitute the first derivative
Why: Replace dy/dx by x over y.
\[ \frac{d^2y}{dx^2} = \frac{y - x\cdot\frac{x}{y}}{y^2} \]
Clear the small fraction
Why: Multiply top and bottom by y.
\[ \frac{d^2y}{dx^2} = \frac{y^2 - x^2}{y^3} \]
Use the original equation
Why: y squared minus x squared is minus 16.
\[ \frac{d^2y}{dx^2} = -\frac{16}{y^3} \]
Evaluate at (5, 3)
Why: Three cubed is 27.
\[ \left.\frac{d^2y}{dx^2}\right|_{(5,3)} = -\frac{16}{27} \approx -0.5926 \]
Check with the explicit upper branch
Why: On the upper right branch y is the root of x squared minus 16; differentiating twice gives minus 16 over the three-halves power, the same number.
\[ y = \sqrt{x^2-16} \Rightarrow y'' = -\frac{16}{(x^2-16)^{3/2}} = -\frac{16}{9^{3/2}} = -\frac{16}{27} \]
OpenStax Calculus Volume 1, §3.8 Implicit Differentiation §3.8, p. 274 — Checkpoint 3.49, continued
This follows exactly the pattern of Example 3.70. Differentiate x over y with the quotient rule, remembering that the derivative of y is dy by dx, then substitute x over y for it.
After clearing the fraction you reach y squared minus x squared over y cubed, and the hyperbola's equation says y squared minus x squared is minus sixteen. So the second derivative is minus sixteen over y cubed, which at (5, 3) is minus sixteen twenty-sevenths.
The negative sign says the upper right branch is concave down near (5, 3), bending away from its steep tangent towards the gentler asymptote. The check differentiates the explicit branch, the square root of x squared minus sixteen, twice, and gets the same number, which is reassuring evidence that each substitution was right.
Trap
Differentiating the circle's slope formula:
\[ \frac{d}{dx}\left(-\frac{x}{y}\right) = -\frac{1}{y} \]
Wrong. That treats y as a fixed number.
y still changes with x, so the quotient rule is needed and a dy by dx appears, to be replaced by minus x over y.
\[ -\frac{y - x\,y'}{y^2} \]
\[ = -\frac{25}{y^3} \]
This slip happens because the first-derivative formula looks like an ordinary fraction in two letters, and it is easy to forget that one of them is a hidden function of x.
If y really were constant, minus x over y would indeed have derivative minus one over y. But y changes as you move along the circle, and ignoring that change gives an answer that is wrong at almost every point: at (3, 4) it gives minus one quarter instead of the correct minus twenty-five over sixty-four.
Use the quotient rule, let dy by dx appear, and then substitute the first-derivative formula for it. Every derivative in this section, first or second, treats y as a function of x.
Comparison
Comparison matrix
| curve | dy/dx | d²y/dx² |
|---|---|---|
| x² + y² = 25 | −x/y | −25/y³ |
| x² − y² = 16 | x/y | −16/y³ |
| 4x² + 25y² = 100 | −4x/(25y) | −16/(25y³) |
Each second derivative is a constant over y cubed, because in each case the original equation absorbs the x squared terms. At (3, 8/5) on the ellipse the entry gives minus 16 over 25 times 4.096, which is minus 0.15625.
Fill in each blank by repeating the moves you have just practised. The circle's second derivative comes from Example 3.70. The hyperbola's first derivative comes from Checkpoint 3.49. The ellipse's first derivative comes from Example 3.73.
Notice the family resemblance once the table is complete. Each first derivative is a multiple of x over y, and each second derivative is a constant over y cubed. That is because in each case the original equation lets you replace the combination of x squared and y squared by a constant.
The ellipse's second derivative, minus sixteen over twenty-five y cubed, was found the same way. At the rocket's firing point it gives minus 0.15625, negative, so the upper half of the ellipse bends downward there, as the picture of the orbit suggests.
Real world
\[ PV^{1.4} = C \]
Discussion prompt
When air is compressed so quickly that no heat escapes, its pressure P and volume V satisfy the equation above, with C a constant. Treat P as a function of V and differentiate implicitly to find how fast pressure changes with volume. Then evaluate it when P is 100 kilopascals and V is 2 litres.
Write your derivative before revealing. The letters are different, but the structure is exactly the one you have practised: an equation linking two quantities, one of which you treat as a function of the other.
Here volume plays the part of x and pressure the part of y. The left side is a product, so the product rule gives the derivative of P times V to the 1.4, plus P times the derivative of V to the 1.4. Solving gives minus 1.4 P over V, and the constant C has vanished entirely.
At 100 kilopascals and 2 litres the rate is minus seventy kilopascals per litre. The minus sign is the physics: squeezing the gas raises its pressure. In the next section, related rates, you will differentiate equations like this with respect to time instead, and the same chain-rule factor appears on every quantity.
Section
Part 7
Pattern
Figure (svg): A flow diagram. An equation in x and y goes to: differentiate every term, with a dy/dx on every y; then gather the dy/dx terms; then factor and divide. From the resulting slope formula three arrows lead to: substitute a point for a tangent line, set the numerator or denominator to zero for horizontal or vertical tangents, and differentiate again for the second derivative.
The diagram shows the method as a pipeline. An equation goes in; one pass of differentiation followed by some algebra produces a slope formula, usually a fraction in x and y.
Everything else in the section branches from that formula. Substitute a point for a tangent or normal line. Set the numerator or the denominator to zero for horizontal or vertical tangents. Differentiate once more, and substitute, for the second derivative.
The list is the checklist to run through on any problem. The two habits that prevent most errors are scanning every term for a y before differentiating, and substituting numbers only at the very end.
Check
Check your understanding
What is the derivative of x y² with respect to x?
Answer: A
Why: Product rule: the derivative of x times y squared, plus x times the derivative of y squared. The second piece is x times 2y times dy/dx, so the answer is y squared plus 2xy dy/dx.
The term is a product, x times y squared, so the product rule gives two pieces. The first is the derivative of x times y squared, which is simply y squared.
The second piece is x times the derivative of y squared, which is two y dy by dx. Together that is y squared plus two x y dy by dx. Each distractor drops exactly one of these ingredients: a whole piece of the product rule, the chain-rule factor, or the x.
Check
Check your understanding
What is the slope of the tangent to x² + xy + y² = 7 at the point (2, 1)?
Answer: A
Why: Differentiating gives 2x + y + x dy/dx + 2y dy/dx = 0, so dy/dx = −(2x + y)/(x + 2y). The point is on the curve (4 + 2 + 1 = 7), and substituting gives −5/4.
Differentiate first. The x y term needs the product rule, giving y plus x dy by dx, and y squared gives two y dy by dx. Gathering the derivative terms gives dy by dx times x plus two y, equal to minus two x minus y.
Check the point is on the curve: four plus two plus one is seven. Substituting gives minus five over four. The most informative distractor is minus five halves, which is what you get if the product rule on x y loses its second piece.
Check
Check your understanding
For y³ + x³ − 3xy = 0, with dy/dx = (3y − 3x²)/(3y² − 3x), where is the tangent vertical (other than the origin)?
Answer: A
Why: Vertical means the denominator is zero: y² = x. Substituting x = y² into the curve gives y⁶ − 2y³ = 0, so y = 2^(1/3) and x = 2^(2/3). The numerator is not zero there.
A vertical tangent needs the denominator of the slope formula to vanish while the numerator does not. The denominator is three y squared minus three x, which is zero when x equals y squared.
Substituting x equal to y squared into the curve gives y to the sixth minus two y cubed equal to zero, so y is the cube root of two and x is its square. That point, about (1.59, 1.26), is the rightmost point of the loop. Its mirror image, (1.26, 1.59), is where the numerator vanishes, the flat top of the loop, so be careful which zero you set.
Explain it to yourself
Discussion prompt
In two or three sentences, explain to a classmate why the derivative of x squared is 2x but the derivative of y squared is 2y times dy by dx, when both are differentiated with respect to x.
Write your explanation before revealing, in words a classmate who missed the lesson would follow. Being able to explain the factor is the surest sign you will not forget it under exam pressure.
The key idea is what differentiating with respect to x means: how fast something changes as x changes. For x squared, the thing inside is x itself, changing at rate one, so the power rule is the whole story. For y squared, the thing inside is y, which changes at its own rate, dy by dx, as x changes. The chain rule multiplies the rate of the outer squaring, two y, by that inner rate.
Exit ticket
\[ xy^2 + y = 6 \quad \text{at} \quad (1, 2) \]
Discussion prompt
Check the point lies on the curve, find dy by dx implicitly, and write the equation of the tangent line at the point.
This combines the main skills in one problem: checking a point, differentiating a product implicitly, solving for the derivative, and writing a tangent line.
The point checks, since four plus two is six. Differentiating gives y squared plus two x y dy by dx plus dy by dx equal to zero, so dy by dx is minus y squared over two x y plus one. At (1, 2) that is minus four fifths, and the tangent is y equals minus four fifths x plus fourteen fifths.
The alternative check is a nice trick. This equation is easy to solve for x instead of y, so you can find dx by dy directly and take its reciprocal. Getting minus four fifths both ways is strong evidence that the implicit work is right.
Recap
| goal | what you do | example |
|---|---|---|
| dy/dx from an equation | differentiate both sides, y′ on every y term, gather, factor, divide | x² + y² = 25 gives −x/y |
| tangent line | check the point, substitute it, point-slope form | y = 3x/4 − 25/4 at (3, −4) |
| normal line | slope −1/m through the same point | y = −x + 2 at (1, 1) on Exercise 317 |
| horizontal / vertical tangents | numerator = 0 / denominator = 0, together with the curve | folium: (1.26, 1.59) flat, (1.59, 1.26) upright |
| second derivative | differentiate y′ again, substitute y′, simplify with the equation | circle: −25/y³ |
\[ \frac{d}{dx}\left(y^n\right) = n\,y^{n-1}\frac{dy}{dx} \]
Next, related rates differentiate an equation with respect to time instead of x, and every quantity picks up the same kind of factor.
Stewart, Calculus: Early Transcendentals 8e, §3.5 Implicit Differentiation §3.5, pp. 208-217 — the same material in Stewart
One idea runs through the whole section: every y is a function of x, so every y term differentiates with a factor of dy by dx. After that, finding the derivative is algebra, and the answer usually involves both x and y.
The table lists what you can do with the slope formula: tangent lines, normal lines, horizontal and vertical tangents, and second derivatives. In every case, check the point is on the curve and substitute numbers last.
The next section, related rates, uses exactly this skill with time as the independent variable. When two quantities are linked by an equation and both change with time, differentiating the equation with respect to time links their rates, with the same chain-rule factor on every term.
Want this taught 1-on-1? Alexander tutors Calculus I — $55/session, free consultation.