11.9 The Dot Product and Projection

A product of two vectors returning a number. Defines it componentwise, establishes its algebraic properties, then proves via the Law of Cosines that it equals the product of the magnitudes times the cosine of the angle between them. From that follow the angle formula, the test for perpendicularity, the projection of one vector onto another with its unique parallel-plus-perpendicular decomposition, and the definition of work as a dot product of force and displacement.

Subject: Trigonometry · 70 slides · symbolic lesson

Open the interactive version of this deck

What this lesson covers

The lesson, slide by slide

1. The Dot Product and Projection

Title

Trigonometry · Section 11.9

One number that knows the angle between two vectors

2. What this lesson gives you

Objectives

Section 11.8 added vectors, subtracted them and scaled them. What it never did was multiply two vectors together. This lesson defines a product of two vectors whose answer is an ordinary number, and then spends the rest of its time showing how much geometry that one number carries.

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, pp. 1034-1043 — the section these objectives come from

Everything you have done with vectors so far has kept them as vectors. You added them and got a vector, you scaled them and got a vector. This lesson introduces an operation that takes two vectors and hands back a plain number, and at first that sounds like a step backwards.

It is the opposite. Because the answer is a number, it can be compared with zero, divided by other numbers and dropped into the middle of ordinary algebra. And the number is not arbitrary: it turns out to equal the two lengths multiplied together times the cosine of the angle between the vectors. That one identity is the engine of the whole section.

From it you get a formula for the angle between any two vectors, a one-line test for perpendicularity, a way to split a vector into a piece along another vector and a piece across it, and the physicist's definition of work. Five objectives, one definition.

3. What could it mean to multiply two vectors?

Warm-up

Discussion prompt

You can add two vectors and get a vector, and you can scale a vector by a number. Before reading on, write down what you think a product of two vectors ought to produce, and one geometric question you would want such a product to answer.

Write something down before you read the answer. Even a wrong guess makes the definition stick, because you will remember what it is not.

The natural first guess is to multiply the matching components and keep the result as a vector. Nothing goes wrong algebraically, but nothing useful happens either. That operation depends entirely on which directions you happened to call the axes, so it cannot be telling you anything about the actual geometry of the two arrows.

The definition you are about to meet does something small and strange: it multiplies the matching components and then throws the vector shape away by adding them. Collapsing two numbers into one looks like losing information, and it is, but what survives is exactly the geometric part: how much the two vectors agree in direction. Keep that phrase in mind, because it is the answer to every question in this lesson.

4. A product of two vectors

Section

Part 1

5. Multiply matching components, then add

Concept

Two vectors go in; one number comes out. That is the whole definition.

\[ \vec{v}\cdot\vec{w} = \langle v_1,v_2\rangle\cdot\langle w_1,w_2\rangle = v_1w_1 + v_2w_2 \]

dot product — For two vectors in component form, the sum of the products of corresponding components. Because the answer is a scalar rather than a vector, it is also called the scalar product.

The raised dot is part of the notation and cannot be dropped. Writing two vectors side by side with no symbol means something else entirely, and writing a number beside a vector means scalar multiplication.

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, p. 1034 — Definition 11.11

The recipe is two multiplications and one addition, and you should be able to do it in your head for small vectors. First times first, second times second, add.

Notice how little the definition looks like the geometry it is about to produce. There is no angle in it, no cosine, no picture. That is typical of a good definition: it is cheap to compute, and the theorems do the work of connecting it to meaning.

Be strict about the raised dot. Writing two vectors next to each other with nothing between them is not an operation that exists here, and a number written beside a vector already means something else, namely scalar multiplication. The dot is the only way to say what you mean.

6. Reading the dot product definition

Notation

Annotate

On: \( \vec{v}\cdot\vec{w} = v_1w_1 + v_2w_2 \)

  • A raised dot, never omitted and never a cross. The cross product is a different operation, and it does not exist for plane vectors.
  • Matching components only: first with first, second with second. Mixing them up (v₁w₂) gives a number with no geometric meaning.
  • The products are ADDED. Keeping them separate would give a vector back and would lose everything useful about the operation.
  • A scalar. It can be positive, zero or negative, and it can be compared with other numbers, which is what makes it so useful.

Step through the annotations and check your own habits against them. The most common slip in the first week is mixing the components: taking the first of one vector with the second of the other. Matching means first with first.

The second thing to internalise is the plus sign. If you kept the two products separate you would have a vector, and that operation, as the warm-up said, carries no geometric meaning. The addition is what collapses the pair into something the geometry can use.

Finally, keep the output type in your head at all times. Almost every error later in this lesson, including the meaningless triple product and the wrong projection formula, comes from losing track of whether the thing in front of you is a number or a vector.

7. The book's first dot product, and a second look at length

Worked example

Compute the dot product of the two vectors below, and then dot the first one with itself.

\[ \vec{v} = \langle 3,4\rangle, \qquad \vec{w} = \langle 1,-2\rangle \]

Multiply the matching components

Why: Three times one, and four times minus two.

\[ (3)(1) = 3, \qquad (4)(-2) = -8 \]

Add the two products

Why: That sum is the dot product.

\[ \vec{v}\cdot\vec{w} = 3 + (-8) = -5 \]

Now dot v with itself

Why: Same rule, with w replaced by v.

\[ \vec{v}\cdot\vec{v} = (3)(3) + (4)(4) = 9 + 16 = 25 \]

Check the second answer against the magnitude

Why: The magnitude of v from Section 11.8 is the square root of nine plus sixteen, which is 5, and 5 squared is 25: the self-dot returns the square of the length.

\[ \|\vec{v}\| = \sqrt{3^2+4^2} = 5, \qquad \|\vec{v}\|^2 = 25 = \vec{v}\cdot\vec{v} \]

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, p. 1034 — the example following Definition 11.11

This is the arithmetic the book does immediately after stating the definition, and it is worth doing slowly once so the pattern becomes automatic. Three times one is three, four times minus two is minus eight, and the sum is minus five.

Pay attention to the sign. A negative dot product is perfectly normal and it is about to acquire a meaning: it will tell you that the angle between these two vectors is obtuse. If you sketch them you will see that one points up and to the right while the other points down and to the right, so they are indeed more than a quarter turn apart.

The second half of the slide is the first surprise of the lesson. Dotting a vector with itself gave twenty-five, which is the square of its length. That is not a coincidence about this particular vector, and the next slide shows why it can never be one.

8. A vector dotted with itself is its length squared

Concept

That last check was not a coincidence. Dotting a vector with itself squares each component and adds, which is exactly what is under the square root in the magnitude formula.

\[ \vec{v}\cdot\vec{v} = v_1v_1 + v_2v_2 = v_1^2 + v_2^2 \]

\[ \|\vec{v}\| = \sqrt{v_1^2+v_2^2} \quad\Longrightarrow\quad \|\vec{v}\|^2 = v_1^2+v_2^2 \]

\[ \vec{v}\cdot\vec{v} = \|\vec{v}\|^2 \]

So the dot product already contains the magnitude. Any time a squared length appears in an argument, it can be rewritten as a dot product and then handled with algebra, and that is precisely the trick the next few slides run on.

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, p. 1035 — Theorem 11.22, relation to magnitude

Look at what dotting a vector with itself actually computes: the first component times itself plus the second component times itself. That is the sum of two squares, and the sum of those same two squares is what sits under the radical in the magnitude formula from Section 11.8.

So the magnitude and the dot product are not two separate ideas; the dot product already contains the magnitude. This is worth more than it looks. It means that any argument involving a squared length can be turned into an argument about dot products, and dot products obey the ordinary rules of algebra.

That translation is exactly what happens two slides from now, in the identity that opens the door to the geometry. Whenever you see a squared magnitude in this section, your first instinct should be to rewrite it as a self-dot.

9. Four properties, and one that is missing

Concept

Theorem 11.22 — The dot product is commutative; it distributes over vector addition; a scalar may be moved freely across it; and a vector dotted with itself is the square of its magnitude.

propertywhat it says
commutativev · w = w · v
distributiveu · (v + w) = u · v + u · w
scalar(k v) · w = k(v · w) = v · (k w)
magnitudev · v = ‖v‖²

There is no associative property on that list, and there cannot be one. An associative law would compare two ways of bracketing a product of three vectors, and neither bracketing is even a legal expression here.

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, pp. 1034-1035 — Theorem 11.22

Three of these four say that the dot product behaves like ordinary multiplication: you can swap the order, you can distribute over a sum, and you can slide a numerical factor around freely. That is why the algebra in the next worked example will feel completely familiar.

The fourth property is the one that is special to vectors, and it is the one you just proved for yourself: a vector dotted with itself is the square of its length.

The missing property matters as much as the present ones. There is no associative law, and the reason is not that it fails but that it cannot even be stated. Bracketing a product of three vectors either way produces a number dotted with a vector, and that is not a legal expression. Getting used to checking types before checking values will save you repeatedly.

10. Proving two of the properties from the definition

Worked example

Every one of these proofs is the same move: unfold the definition, use ordinary arithmetic, fold the definition back up.

Expand the commutative claim

Why: Write both sides in components.

\[ \vec{v}\cdot\vec{w} = v_1w_1 + v_2w_2 \]

Swap each product

Why: Real number multiplication is commutative.

\[ v_1w_1 + v_2w_2 = w_1v_1 + w_2v_2 \]

Fold the definition back up

Why: That is the dot product taken the other way round.

\[ w_1v_1 + w_2v_2 = \vec{w}\cdot\vec{v} \]

Now expand the scalar claim

Why: Scale v first, then dot.

\[ (k\vec{v})\cdot\vec{w} = \langle kv_1, kv_2\rangle\cdot\langle w_1,w_2\rangle = (kv_1)w_1 + (kv_2)w_2 \]

Factor the k out

Why: Associativity and distributivity of real numbers.

\[ k(v_1w_1) + k(v_2w_2) = k(v_1w_1+v_2w_2) = k(\vec{v}\cdot\vec{w}) \]

Check it on numbers

Why: Tripling the first vector triples the dot product, exactly as the property claims.

\[ \langle 3,4\rangle\cdot\langle 1,-2\rangle = -5, \qquad \langle 9,12\rangle\cdot\langle 1,-2\rangle = 9-24 = -15 = 3(-5) \]

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, pp. 1034-1035 — the proofs of Theorem 11.22

Both proofs have the same three-beat shape, and once you see it you can reproduce any of the four properties on demand. Unfold the definition into components, use a fact about ordinary real numbers, then fold the definition back up.

For commutativity the real-number fact is that multiplication of numbers can be done in either order. For the scalar property it is that you can factor a common multiplier out of a sum. Nothing about vectors is used at all, which is precisely why the dot product inherits so much of the behaviour of ordinary multiplication.

The numerical check at the end is the kind of thing worth doing whenever a property feels abstract. Tripling one vector tripled the answer, from minus five to minus fifteen, exactly as the property claims it must.

11. Finish the dot product

Fill the middle

Fill in the blanks

\langle -2,7\rangle\cdot\langle 5,3\rangle = -10 + 21 = 11

Why: Minus two times five is minus ten, and seven times three is twenty-one; the two products add to eleven. The answer is positive, which by the theorem coming next means the angle between these two vectors is acute.

Fill in both blanks before revealing. The first is one multiplication, the second is one addition, and the whole thing should take you about five seconds by the end of this lesson.

Then look at the sign of your answer. It came out positive, and by the time you reach the geometric theorem you will read that as a statement about the two vectors: they are less than a quarter turn apart. Getting into the habit of reading the sign now will make that theorem feel like a confirmation rather than a surprise.

12. Scalar or vector?

Sorting

Sort into buckets

Each expression below is legal. Sort it by the kind of object it produces.

A number
v · w; v · v; ‖v‖
A vector
3v; (v · w) w; v + w
sca
A dot product is a sum of products of numbers, so v · w and v · v are numbers; a magnitude is a length, also a number. v · v is the square of ‖v‖.
vec
Scaling a vector or adding two vectors returns a vector. In (v · w) w the bracket is a number, so the whole expression is that number times w: a vector parallel to w.

Sort these by output type, not by difficulty. Every expression here is legal, so the only question is what kind of object comes out the other end.

The one that catches people is the third: a dot product in brackets, followed by a vector. Because the bracket is a number, the whole expression is a number times a vector, which is scalar multiplication, so the answer is a vector, and specifically one pointing along w. Remember this shape, because the projection formula at the heart of Part 4 has exactly it.

Tracking types like this is not pedantry. It is the fastest way to catch a wrong formula: if you have written something that claims to be a vector but computes to a number, the formula is wrong before you check a single digit.

13. Trap: dotting three vectors

Trap

The trap

A line that looks harmless until you read the types:

\[ \vec{u}\cdot\vec{v}\cdot\vec{w} \]

Meaningless. The first dot product produces a number, and a number cannot be dotted with a vector.

The fix

Work left to right. The bracket below is a scalar, and a scalar times a vector is scalar multiplication, written with no symbol at all.

\[ (\vec{u}\cdot\vec{v})\,\vec{w} \]

That is a legal expression, and it is a vector parallel to w. The projection formula later in this lesson has exactly this shape.

This is the type error the missing associative property predicted, and it shows up on homework constantly, usually in the disguised form of someone trying to cancel a vector from both sides of an equation.

Read it left to right. The first dot product returns a number. The expression then asks you to dot that number with a third vector, and the definition has nothing to say about that: it only accepts two vectors.

The fix is not to add brackets differently but to notice that the legal expression, a bracketed dot product multiplying a vector, is a different and very useful thing. It is a vector pointing along the last factor, and you will meet it as the projection in Part 4.

14. Example 11.9.1: expanding a squared difference

Worked example

Prove the identity below. It is the bridge to every geometric fact in this lesson.

\[ \|\vec{v}-\vec{w}\|^2 = \|\vec{v}\|^2 - 2(\vec{v}\cdot\vec{w}) + \|\vec{w}\|^2 \]

Rewrite the left side as a self-dot

Why: A squared magnitude is a vector dotted with itself.

\[ \|\vec{v}-\vec{w}\|^2 = (\vec{v}-\vec{w})\cdot(\vec{v}-\vec{w}) \]

Distribute over the first factor

Why: The distributive property, taken on the left.

\[ = \vec{v}\cdot(\vec{v}-\vec{w}) - \vec{w}\cdot(\vec{v}-\vec{w}) \]

Distribute again over the second factor

Why: Four terms, exactly as with real numbers.

\[ = \vec{v}\cdot\vec{v} - \vec{v}\cdot\vec{w} - \vec{w}\cdot\vec{v} + \vec{w}\cdot\vec{w} \]

Combine the middle two

Why: They are equal, by commutativity.

\[ = \vec{v}\cdot\vec{v} - 2(\vec{v}\cdot\vec{w}) + \vec{w}\cdot\vec{w} \]

Rewrite the outer self-dots as magnitudes

Why: The fourth property, used twice.

\[ = \|\vec{v}\|^2 - 2(\vec{v}\cdot\vec{w}) + \|\vec{w}\|^2 \]

Check it on the pair from earlier

Why: With v and w as before, the left side uses the difference vector and the right side uses the dot product minus five, and both come out as 45.

\[ \vec{v}-\vec{w} = \langle 2,6\rangle: \; 4+36 = 40; \qquad 25 - 2(-5) + 5 = 40 \]

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, p. 1035 — Example 11.9.1

The book works this one out in painstaking detail, one property per line, and the point of the pedantry is to convince you that nothing new is being used. Every step is one of the four properties.

Once you have seen it, the shortcut is to notice that the result looks exactly like the expansion of a difference of two real numbers squared. That is not an accident: the expansion of a squared difference only ever uses commutativity and distributivity, and the dot product has both.

The reason this identity gets its own example is that it is the bridge to the geometry. It expresses the squared length of the third side of a triangle in terms of dot products, and the Law of Cosines from Section 11.3 expresses the same length in terms of an angle. Setting those two expressions equal is the whole proof of the main theorem, and it happens three slides from now.

15. What the dot product knows

Section

Part 2

16. The angle between two vectors

Concept

Figure (svg): Three panels, each showing two vectors drawn from the same starting point: on the left they lie along the same ray so the angle is zero, in the middle they open out with an arc marking the angle theta, and on the right they point in exactly opposite directions so the angle is pi.

The angle between two vectors is measured with their tails together, and it is never more than a straight angle. Those three pictures are the three cases the proof of the theorem has to handle.

Draw the two vectors with the same starting point. The angle between them is the one the two rays cut out, and it is always taken between zero and a straight angle, never reflex and never negative.

\[ 0 \le \theta \le \pi \]

That restriction is what makes the angle unique: a pair of rays cuts the plane into two angles that add to a full turn, and this convention always takes the smaller one.

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, p. 1035 — the definition of the angle between two vectors

The definition needs care because two rays from a point cut out two angles, not one, and they add to a full turn. The convention takes the smaller one, so the angle between two vectors never exceeds a straight angle and is never negative.

That choice is not arbitrary. It is made so that the angle is determined by the pair of vectors alone, with no orientation to worry about, and so that swapping the two vectors does not change it. It also matches the range of the arccosine exactly, which is what makes the angle formula later so painless.

Look at the three panels. The outer two are the degenerate cases, where the vectors line up along a single line, and they will need separate treatment in the proof because there is no triangle to apply the Law of Cosines to. The middle panel is the general case.

17. The triangle behind the theorem

Concept

Figure (svg): A triangle drawn from two vectors with a common tail: v runs up to the left vertex, w runs along the base to the right vertex, and the third side joins the tip of w to the tip of v. The three sides are labelled with the magnitude of v, the magnitude of w, and the magnitude of v minus w, with the angle theta marked at the common tail.

Place the two vectors tail to tail and the vector from the tip of w to the tip of v closes the triangle. That third side is v minus w, and its length can be computed two different ways.

Two vectors from a common point, and the vector joining their tips, make a triangle. Its three side lengths are the two magnitudes and the magnitude of the difference, and the angle between the vectors sits between the two known sides.

\[ \vec{w} + (\vec{v}-\vec{w}) = \vec{v} \]

Now there are two completely different expressions for the third side squared: the Law of Cosines from Section 11.3, and the algebraic identity just proved. Setting them equal is the whole proof.

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, p. 1036 — the triangle in the proof of Theorem 11.23

Take a moment with this picture; the next slide is nothing but algebra applied to it. Two vectors from a common point, plus the arrow joining their tips, make a genuine triangle.

The key observation is which vector that third side is. Going along w and then along the third side gets you to the tip of v, so the third side must be the difference of v and w. Its length is therefore the magnitude of that difference, the very quantity the previous example expanded.

You now have two expressions for the same number. One comes from trigonometry, the Law of Cosines, and contains an angle. The other comes from algebra and contains a dot product. Two formulas for the same thing always yield an identity, and here the identity is the most important theorem of the section.

18. Proving Theorem 11.23 from the Law of Cosines

Worked example

Take the angle strictly between zero and a straight angle, so that the three points really do form a triangle.

Apply the Law of Cosines to that triangle

Why: Two sides and the included angle give the third side.

\[ \|\vec{v}-\vec{w}\|^2 = \|\vec{v}\|^2 + \|\vec{w}\|^2 - 2\|\vec{v}\|\,\|\vec{w}\|\cos(\theta) \]

Write down the algebraic identity for the same length

Why: Example 11.9.1, unchanged.

\[ \|\vec{v}-\vec{w}\|^2 = \|\vec{v}\|^2 - 2(\vec{v}\cdot\vec{w}) + \|\vec{w}\|^2 \]

Equate the two right sides

Why: They describe the same number.

\[ \|\vec{v}\|^2 + \|\vec{w}\|^2 - 2\|\vec{v}\|\,\|\vec{w}\|\cos(\theta) = \|\vec{v}\|^2 - 2(\vec{v}\cdot\vec{w}) + \|\vec{w}\|^2 \]

Cancel the two squared magnitudes

Why: They appear on both sides.

\[ -2\|\vec{v}\|\,\|\vec{w}\|\cos(\theta) = -2(\vec{v}\cdot\vec{w}) \]

Divide by minus two

Why: And the theorem is proved.

\[ \vec{v}\cdot\vec{w} = \|\vec{v}\|\,\|\vec{w}\|\cos(\theta) \]

Check the two excluded cases

Why: At an angle of zero the vectors are positive multiples of each other and the dot product is the product of the lengths, matching a cosine of 1; at a straight angle the multiple is negative and the dot product is minus that product, matching a cosine of minus 1.

\[ \cos(0) = 1, \qquad \cos(\pi) = -1 \]

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, p. 1036 — Theorem 11.23 and its proof

Follow the cancellation carefully; the whole proof is four lines and it is very much worth being able to reconstruct. Both expressions for the squared third side contain the squared magnitude of v and the squared magnitude of w, so those four terms vanish when you set the expressions equal.

What is left is an equation between two things that both carry a factor of minus two. Divide it out and the theorem is standing there: the dot product equals the product of the two lengths times the cosine of the angle.

The last step handles the two degenerate cases separately, because with the vectors along a single line there is no triangle and the Law of Cosines has nothing to work on. The book does these first, by writing one vector as a positive or negative multiple of the other. Both come out agreeing with the formula, which is why the theorem can be stated without any exceptions.

19. Reading the geometric formula

Notation

Annotate

On: \( \vec{v}\cdot\vec{w} = \|\vec{v}\|\,\|\vec{w}\|\cos(\theta) \)

  • Computed from components, with no picture and no angle: multiply, multiply, add.
  • Computed from the geometry, with no components at all: two lengths and the angle between them.
  • Both are positive, always. So they scale the answer but can never change its sign.
  • The only factor that can be negative, so the sign of the whole dot product is the sign of the cosine — that is, it is decided by the angle alone.
  • The theorem needs both vectors nonzero: the zero vector has no direction, so there is no angle to speak of.

This is the single most important line in the section, so read the annotations slowly. The two sides of the equation are computed by completely different means: components on the left, geometry on the right. That is what makes it powerful, because you can always compute the left side and thereby learn something about the right.

The observation to carry away is the one about signs. Both magnitudes are lengths, so both are positive, and they can never flip the sign of the answer. The cosine is the only factor that can be negative, and the cosine's sign is decided entirely by the angle.

So the sign of a dot product is a direct statement about the angle, available without dividing by anything or taking any inverse function. That single bit of information turns out to be enough for a surprising amount of work, including the entire perpendicularity test in Part 3.

20. The sign of the dot product turns with the angle

Picture it

Figure (svg): A fan of five vectors of equal length drawn from one point, at 25, 60, 90, 130 and 165 degrees to a fixed horizontal vector w. Each is labelled with the dot product it makes with w: plus 10.9, plus 6.0, 0, minus 7.7 and minus 11.6, coloured green for positive, yellow for zero and red for negative.

Both lengths stay the same the whole way round, so every change in the number comes from the angle alone. The number crosses zero exactly at the right angle.

Every arrow in the fan has the same length and meets the same partner, so the only thing changing is the angle. Below a right angle the number is positive, at a right angle it is exactly zero, and beyond a right angle it is negative.

Everything in the fan has been held fixed except the angle. All five arrows have length three, and they all meet the same partner of length four, so the product of the lengths is twelve in every case and every change in the number comes from the cosine alone.

Follow the numbers as the arrow swings anticlockwise. At twenty-five degrees you get almost eleven, close to the maximum of twelve. At sixty degrees it has fallen to six, half the maximum, because the cosine of sixty degrees is one half. At ninety it is exactly zero, and past that it goes negative and keeps falling.

Notice how quickly the value falls near the right angle and how slowly it changes near zero degrees. That is the shape of a cosine, and it means a dot product is a rather insensitive way of detecting small angles but a very sharp way of detecting a right angle.

21. The same story as a cosine curve

Picture it

Figure (svg): The graph of twelve times the cosine of theta for theta from 0 to 180 degrees, shaded green above the axis from 0 to 90 degrees and red below it from 90 to 180 degrees, crossing zero at 90 degrees.

The dot product of two fixed lengths is a cosine curve in the angle. It is largest when the vectors agree, zero when they are perpendicular, and most negative when they oppose.

Plot the dot product against the angle and you get a cosine curve scaled by the two lengths. It starts at the product of the lengths, falls through zero at a quarter turn, and bottoms out at minus that product when the vectors oppose each other.

This is the fan again, plotted rather than drawn. The horizontal axis is the angle and the vertical axis is the dot product, with the two lengths fixed at three and four.

Three features are worth naming. At an angle of zero the value is the product of the lengths, which is as large as a dot product of these two vectors can ever be. At a straight angle it is minus that product, as small as it can ever be. And in between it passes through zero exactly once.

That single crossing is the reason a zero dot product pins the angle exactly. Any other value of the dot product leaves the angle ambiguous until you also know the lengths, but zero does not, because zero can only happen at one place on this curve.

22. Turn the angle and watch the dot product

Tweak it

Parameter explorer

The horizontal axis is the angle in degrees, from 0 to 180. Set both lengths, then read off the angle where the curve crosses zero, and check whether it moves when you change the lengths.

\[ \vec{v}\cdot\vec{w} = ({a})({b})\cos\theta \]

  • a — from 1 to 5: length of v
  • b — from 1 to 5: length of w

Move the two length sliders and watch what happens to the curve. It stretches and shrinks vertically, sometimes dramatically, and the peak value is always the product of the two lengths you have set.

Now watch the crossing point. It does not move. No matter how you set the lengths, the curve passes through zero at ninety degrees and nowhere else, because a product is zero exactly when one of its factors is, and the only factor that can vanish is the cosine.

That is the whole content of the perpendicularity theorem in Part 3, discovered with a slider instead of a proof. It is also why you can read a right angle off a dot product without ever computing a magnitude: the lengths simply do not matter to that one question.

23. Solving for the angle itself

Concept

The geometric formula has only one unknown once the components are known. Divide by the two lengths, which are nonzero, and the cosine is exposed.

\[ \cos(\theta) = \frac{\vec{v}\cdot\vec{w}}{\|\vec{v}\|\,\|\vec{w}\|} \]

\[ \theta = \arccos\left(\frac{\vec{v}\cdot\vec{w}}{\|\vec{v}\|\,\|\vec{w}\|}\right) = \arccos(\hat{v}\cdot\hat{w}) \]

Theorem 11.24 — The angle between two nonzero vectors is the arccosine of their dot product divided by the product of their magnitudes, which is the same as the arccosine of the dot product of their unit vectors.

No case analysis is needed. The angle was defined to lie between zero and a straight angle, and that is exactly the range of the arccosine, so the formula returns the right angle every time.

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, p. 1036 — Theorem 11.24

The rearrangement is pure algebra: the two magnitudes are nonzero numbers, so you may divide by them, and that isolates the cosine. Taking the arccosine of both sides then gives the angle.

The reason this works without any case analysis is the convention set earlier, that the angle lies between zero and a straight angle. That interval is exactly the range of the arccosine function, so the arccosine always returns the angle you actually want. Compare that with solving a cosine equation in Chapter 10, where you had to hunt for a second solution in another quadrant; here there is nothing to hunt for.

The second form of the formula is worth noticing too. Dividing each vector by its own length turns it into a unit vector, so the angle between two vectors is just the arccosine of the dot product of their directions. Lengths are irrelevant to angle, and that formula says so explicitly.

24. Example 11.9.2(1): an angle of one hundred fifty degrees

Worked example

Find the angle between these two vectors.

\[ \vec{v} = \langle 3,-3\sqrt{3}\rangle, \qquad \vec{w} = \langle -\sqrt{3},1\rangle \]

Compute the dot product

Why: Two products, then add.

\[ \vec{v}\cdot\vec{w} = (3)(-\sqrt{3}) + (-3\sqrt{3})(1) = -3\sqrt{3}-3\sqrt{3} = -6\sqrt{3} \]

Compute the two magnitudes

Why: Nine plus twenty-seven is thirty-six; three plus one is four.

\[ \|\vec{v}\| = \sqrt{9+27} = 6, \qquad \|\vec{w}\| = \sqrt{3+1} = 2 \]

Divide

Why: Twelve in the denominator, and the radical survives.

\[ \cos(\theta) = \frac{-6\sqrt{3}}{12} = -\frac{\sqrt{3}}{2} \]

Take the arccosine

Why: A common value from the unit circle, in the second quadrant.

\[ \theta = \arccos\left(-\frac{\sqrt{3}}{2}\right) = \frac{5\pi}{6} = 150^{\circ} \]

Check the sign against the picture

Why: The dot product came out negative, so the angle had to be obtuse, and one hundred fifty degrees is obtuse. As a second check, v points into the fourth quadrant at minus sixty degrees and w into the second at one hundred fifty degrees, a gap of two hundred ten degrees, whose smaller companion is one hundred fifty.

\[ 360^{\circ} - 210^{\circ} = 150^{\circ} \]

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, p. 1037 — Example 11.9.2, part 1

Work through the radicals carefully rather than reaching for a calculator, because this problem is built so that everything simplifies. The dot product gathers two copies of minus three root three, giving minus six root three, and both magnitudes come out as whole numbers.

When the quotient turns out to be minus the square root of three over two, you should recognise it immediately from the unit circle in Chapter 10. It is a reference value, and since the arccosine returns angles between zero and a straight angle, the answer is the second-quadrant one: one hundred fifty degrees.

The check at the end is a good habit. The sign of the dot product predicted an obtuse angle before any dividing happened, so an answer under ninety degrees would have been wrong on inspection. As a second confirmation, the two vectors point at minus sixty and one hundred fifty degrees from the positive horizontal, and the gap between those directions, taken the short way, is one hundred fifty degrees.

25. Example 11.9.2(3): an angle that is not a common one

Worked example

The same procedure, on a pair whose answer is not a familiar angle.

\[ \vec{v} = \langle 3,-4\rangle, \qquad \vec{w} = \langle 2,1\rangle \]

Compute the dot product

Why: Six and minus four.

\[ \vec{v}\cdot\vec{w} = (3)(2) + (-4)(1) = 6-4 = 2 \]

Compute the magnitudes

Why: A three-four-five triangle, and the root of five.

\[ \|\vec{v}\| = \sqrt{9+16} = 5, \qquad \|\vec{w}\| = \sqrt{4+1} = \sqrt{5} \]

Divide and rationalise

Why: Multiply above and below by the root of five.

\[ \cos(\theta) = \frac{2}{5\sqrt{5}} = \frac{2\sqrt{5}}{25} \]

Leave the answer as an arccosine

Why: That number is not the cosine of any common angle, so the exact answer is the arccosine itself.

\[ \theta = \arccos\left(\frac{2\sqrt{5}}{25}\right) \]

Check what it is worth as a decimal

Why: The fraction is about 0.1789, just above zero, so the angle should be a little under a right angle, and it is: about 79.7 degrees.

\[ \frac{2\sqrt{5}}{25} \approx 0.1789 \;\Longrightarrow\; \theta \approx 79.7^{\circ} \]

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, p. 1037 — Example 11.9.2, part 3

This is the ordinary case, the one most exercises produce. The arithmetic is easy, the quotient simplifies a little when you rationalise, and then it stops: two root five over twenty-five is not the cosine of any angle you know.

Resist the urge to reach for a decimal and call that the exact answer. The exact answer is the arccosine expression itself, and that is what the book writes down. A decimal approximation is a different kind of answer, useful for checking and for reporting a result, and you should label it as approximate when you give it.

The final line is the check, and notice what it checks. The quotient is a small positive number, only a little above zero, so the angle must be a little below ninety degrees. Getting about seventy-nine point seven confirms both the arithmetic and your reading of the sign.

26. What the sign alone tells you

Sorting

Sort into buckets

Each pair of vectors is nonzero, and only the dot product is given. Sort by the angle it forces.

Less than a right angle
v · w = 12; v · w = 0.001; v · w = ‖v‖‖w‖
Exactly a right angle
v · w = 0
More than a right angle
v · w = −7; v · w = −0.5
acute
A positive dot product forces a positive cosine, and that means an angle under ninety degrees. How positive does not matter: 0.001 and 12 are both simply acute. The last one is the extreme case, cosine exactly 1, an angle of zero.
right
Zero is the only value that pins the angle exactly, because the cosine vanishes only at ninety degrees on this interval.
obtuse
A negative dot product forces a negative cosine, so the angle exceeds ninety degrees, whether the number is −0.5 or −7.

Sort these before reading the rationales. You are given no components and no lengths, only the dot product, and the claim is that this is already enough to classify the angle.

The two that tempt people are the small ones. A dot product of one thousandth feels like it ought to mean nearly perpendicular, and a dot product of minus one half feels barely negative. But the classification does not care how big the number is, only which side of zero it sits on, because the magnitudes are always positive and cannot change the sign.

The last item is the extreme case. If the dot product equals the product of the two lengths then the cosine must be exactly one, so the angle is zero and the vectors point the same way. That is the largest a dot product can ever be for a given pair of lengths.

27. Trap: a large dot product means a small angle

Trap

The trap

Two pairs of vectors are compared:

\[ \vec{a}\cdot\vec{b} = 100 \]

\[ \vec{c}\cdot\vec{d} = 2 \]

Wrong to conclude that the first pair is more nearly aligned.

The fix

The size of a dot product mixes up two different things: the lengths and the angle. Long vectors give big numbers whatever the angle.

\[ (100)(1)\cos(0^{\circ}) = 100 \]

\[ (1000)(1)\cos(84.26^{\circ}) \approx 100 \]

Only after dividing by both lengths does the number say anything about direction. That quotient is the cosine, and it never leaves the interval from minus one to one.

This is the natural misreading of the previous slide, and it is worth naming explicitly. The sign of a dot product tells you about the angle. The size of a dot product does not, because it mixes the angle together with both lengths.

The two readings shown on the right both produce the same number from wildly different geometry: one pair perfectly aligned but short, another pair nearly perpendicular but very long. Nothing in the number distinguishes them.

What does carry pure direction information is the dot product divided by both magnitudes, because that quotient is exactly the cosine and is trapped between minus one and one. If you ever want to compare how aligned two different pairs of vectors are, compare those quotients, never the raw dot products.

28. Find the error: the magnitudes were left out

Error analysis

A student computes the angle between the vectors of Example 11.9.2(3) as follows.

Annotate

On: \( \vec{v}\cdot\vec{w} = 2 \;\Longrightarrow\; \theta = \arccos(2) \)

  • This part is right: three times two plus minus four times one really is 2.
  • The theorem says the dot product equals the product of the magnitudes TIMES the cosine. The cosine is the dot product divided by 5 and by the root of 5, not the dot product itself.
  • Undefined. The arccosine only accepts inputs from minus one to one, and getting a number outside that range is the alarm bell that a division was skipped.
  • Divide by both magnitudes first: 2 over 5 root 5, about 0.1789, which is safely inside the domain and gives about 79.7 degrees.

Try to spot the problem before revealing the annotations. The arithmetic in the first half is correct, so the error is not a slip in the multiplication.

The theorem says the dot product is the product of the magnitudes times the cosine, and this student has treated it as though the dot product were the cosine. The magnitudes have simply gone missing, and they matter: dividing by five and by the square root of five changes two into about one hundred seventy-nine thousandths.

There is a built-in alarm for this mistake, and you should learn to hear it. The arccosine only accepts numbers between minus one and one. The moment you find yourself asking a calculator for the arccosine of something bigger than one, you have skipped a division, and no amount of rechecking the multiplication will find it.

29. Finish the angle calculation

Faded example

The vectors are the three-four vector and the five-twelve vector, two of the most familiar right triangles there are.

Fill in the blanks

\cos\theta = \frac1365})} = \frac______}

Why: The dot product is fifteen plus forty-eight, which is sixty-three. The magnitudes are five and thirteen, so the denominator is sixty-five and the cosine is sixty-three over sixty-five, giving an angle of about 14.25 degrees: the two vectors very nearly agree.

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, p. 1043 — Exercise 8

Fill in the two blanks and then compute the angle yourself. Both magnitudes here come from right triangles you already know by heart, the three-four-five and the five-twelve-thirteen.

The answer, sixty-three over sixty-five, is very close to one, and you should read that before you touch a calculator: a cosine near one means an angle near zero, so these two vectors very nearly point the same way. Working it out gives about fourteen and a quarter degrees, which matches.

Getting into the habit of predicting the angle from the quotient before evaluating it is what stops a mistyped calculator entry from going unnoticed.

30. Estimate the angle before computing it

Estimation

\[ \vec{v} = \langle 1,17\rangle, \qquad \vec{w} = \langle -1,0\rangle \]

Predict first

One vector points almost straight up and the other points straight left. Estimate the angle between them before doing any arithmetic.

  • About 5 degrees
  • About 87 degrees
  • About 93 degrees
  • About 175 degrees

Correct: About 93 degrees

Why: Straight up and straight left are exactly ninety degrees apart. The first vector leans very slightly to the right of straight up, which opens the angle a little rather than closing it, so the answer must be just over ninety. The dot product is minus one, a small negative number, confirming a barely obtuse angle: about 93.4 degrees.

\[ \vec{v}\cdot\vec{w} = -1 \;\Longrightarrow\; \cos\theta = \frac{-1}{\sqrt{290}} \approx -0.0587 \;\Longrightarrow\; \theta \approx 93.4^{\circ} \]

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, p. 1043 — Exercise 7

Commit to one of the four options before you look. The point of the exercise is that a rough picture beats arithmetic for catching errors, and this pair is easy to picture: one vector is almost straight up, the other is exactly straight left.

Straight up and straight left are ninety degrees apart. Now ask which way the first vector's slight lean takes it. It leans a little to the right of straight up, which is away from the leftward vector, so the angle opens rather than closes and the answer must be just over ninety degrees.

The arithmetic confirms it. The dot product is minus one, small but negative, and negative means obtuse. The exact value comes out at about ninety-three and a half degrees, which is exactly the small nudge past a right angle the picture predicted.

31. Zero means perpendicular

Section

Part 3

32. The dot product detects a right angle

Concept

Figure (svg): Two vectors drawn from the origin on a grid: one to the point 2 comma 2 and one to the point 5 comma minus 5, with a small square at the origin marking the right angle between them and the dot product written as ten minus ten equals zero.

Ten and minus ten cancel, and the picture says why: turning the first vector clockwise by a quarter turn lands it exactly along the second.

Of all the values a dot product can take, one is special. Zero pins the angle exactly, because the cosine vanishes at exactly one place between zero and a straight angle.

Theorem 11.25 — Two nonzero vectors are perpendicular, written with the perpendicular sign, exactly when their dot product is zero. Perpendicular vectors are also called orthogonal.

\[ \vec{v}\perp\vec{w} \iff \vec{v}\cdot\vec{w} = 0 \]

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, p. 1037 — Theorem 11.25

Everything that has been said about the sign now pays off in a single sharp statement. Zero is the only value of the dot product that pins the angle exactly, and it pins it at a right angle.

Look at the drawn pair. One vector goes two right and two up, the other five right and five down. The products of the matching components are ten and minus ten, which destroy each other. The picture shows the same fact geometrically: turning the first vector a quarter turn clockwise lands it along the second.

Two things to keep straight in the wording. The theorem needs both vectors to be nonzero, since the zero vector has no direction and no angle. And orthogonal is simply another word for perpendicular; both will be used from here on, and the book prefers orthogonal.

33. Example 11.9.2(2): the dot product vanishes

Worked example

Find the angle between these two vectors.

\[ \vec{v} = \langle 2,2\rangle, \qquad \vec{w} = \langle 5,-5\rangle \]

Compute the dot product

Why: Ten and minus ten.

\[ \vec{v}\cdot\vec{w} = (2)(5) + (2)(-5) = 10 - 10 = 0 \]

Divide by the magnitudes

Why: Whatever they are, zero divided by a positive number is zero, so they never need to be computed.

\[ \cos(\theta) = \frac{0}{\|\vec{v}\|\,\|\vec{w}\|} = 0 \]

Take the arccosine

Why: The arccosine of zero.

\[ \theta = \arccos(0) = \frac{\pi}{2} = 90^{\circ} \]

Check against the components

Why: The first vector has equal components, so it runs along the line of slope one; the second has opposite components, so it runs along the line of slope minus one. Those slopes multiply to minus one, so the lines are perpendicular.

\[ (1)(-1) = -1 \]

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, p. 1037 — Example 11.9.2, part 2

This is the third part of the book's example, and it is much quicker than the other two because the dot product comes out zero at the first step.

Notice what you never had to do. Neither magnitude was computed, because zero divided by any positive number is zero, whatever that number is. Students often grind out both square roots here out of habit; the moment your dot product is zero, you are finished.

The check offers a second route to the same conclusion, through Chapter 2. The first vector has equal components so it lies along a line of slope one, and the second has opposite components so it lies along a line of slope minus one. Those slopes multiply to minus one, which is the perpendicular-lines test, and the next-but-one slide proves that test with vectors.

34. Proving the test in both directions

Worked example

An if-and-only-if claim needs both implications. Each one is a single application of a theorem already proved.

Assume the vectors are perpendicular

Why: Then the angle is a quarter turn by definition.

\[ \theta = \frac{\pi}{2} \]

Substitute into the geometric formula

Why: The cosine of a quarter turn is zero, so the whole product collapses.

\[ \vec{v}\cdot\vec{w} = \|\vec{v}\|\,\|\vec{w}\|\cos\left(\frac{\pi}{2}\right) = 0 \]

Now assume the dot product is zero

Why: Feed it into the angle formula from Theorem 11.24.

\[ \theta = \arccos\left(\frac{0}{\|\vec{v}\|\,\|\vec{w}\|}\right) = \arccos(0) \]

Read off the angle

Why: Which is a quarter turn, so the vectors are perpendicular.

\[ \theta = \frac{\pi}{2} \;\Longrightarrow\; \vec{v}\perp\vec{w} \]

Check where the nonzero assumption was used

Why: Both directions divide by the two magnitudes, which is only legal because neither vector is the zero vector. The zero vector has no direction, so it is neither perpendicular nor parallel to anything.

\[ \|\vec{v}\| \ne 0, \qquad \|\vec{w}\| \ne 0 \]

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, p. 1037 — the proof of Theorem 11.25

An if-and-only-if statement is really two statements, and both need proving. Get into the habit of asking which direction you are in, because on an exam a proof of only one half earns only half the marks.

Neither half is hard here. Going forwards, perpendicular means the angle is a quarter turn, and the cosine of a quarter turn is zero, which kills the whole product. Going backwards, a dot product of zero makes the angle formula return the arccosine of zero, which is a quarter turn.

The last step points at where the hypothesis was used. Both directions divide by the two magnitudes, and that division is only legal because both vectors are nonzero. This is why the theorem says nonzero, and why the zero vector is neither perpendicular nor parallel to anything: those words simply do not apply to it.

35. Example 11.9.3: perpendicular slopes, proved with vectors

Worked example

Figure (svg): Two dashed lines through the origin, one of slope 2 and one of slope minus one half, with their direction vectors drawn from the origin to the points 1 comma 2 and 1 comma minus one half, and a small square marking the right angle between those two direction vectors.

Every line carries a direction vector whose second entry is its slope. Two lines meet at a right angle exactly when those two direction vectors do.

Prove that two lines are perpendicular exactly when the product of their slopes is minus one.

Find two points on the first line

Why: Substitute x equal to zero and x equal to one into its equation.

\[ P(0,b_1), \qquad Q(1, m_1+b_1) \]

Build a direction vector from them

Why: Subtract the coordinates; the intercept cancels.

\[ \vec{v}_1 = \overrightarrow{PQ} = \langle 1-0,\; (m_1+b_1)-b_1\rangle = \langle 1,m_1\rangle \]

Do the same for the second line

Why: Identical algebra with the second slope.

\[ \vec{v}_2 = \langle 1,m_2\rangle \]

Dot the two direction vectors

Why: One times one, plus the product of the slopes.

\[ \vec{v}_1\cdot\vec{v}_2 = (1)(1) + m_1m_2 = 1 + m_1m_2 \]

Set it to zero

Why: The lines are perpendicular exactly when their directions are, and that is exactly when this dot product vanishes.

\[ 1 + m_1m_2 = 0 \iff m_1m_2 = -1 \]

Check it on the picture

Why: The drawn lines have slopes two and minus one half, whose product is minus one, and their direction vectors dot to one minus one, which is zero.

\[ \langle 1,2\rangle\cdot\langle 1,-\tfrac12\rangle = 1 - 1 = 0 \]

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, pp. 1037-1038 — Example 11.9.3

You have known the perpendicular slopes rule since Chapter 2, probably as a fact to memorise. Here is a proof of it that takes four lines, and the reason it is so short is that the dot product does all the geometry for you.

The clever step is the first one: turn each line into a vector. Substituting two convenient values of x gives two points on the line, and subtracting them gives a direction vector. Watch the intercept cancel in that subtraction, which is why the answer depends on the slope alone; parallel lines have the same direction vector regardless of where they sit.

After that it is one dot product. Two direction vectors, each of the form one and a slope, dot to one plus the product of the slopes. Setting that to zero is the same as saying the product of the slopes is minus one, and the rule from Chapter 2 falls out. Any perpendicular pair you can think of will satisfy it; check it on the drawn lines with slopes two and minus one half.

36. Rotate a vector by a quarter turn

Prediction

Take any nonzero vector and build a new one by swapping the components and changing the sign of the first.

\[ \vec{v} = \langle a,b\rangle \;\longmapsto\; \vec{v}^{\,\perp} = \langle -b,a\rangle \]

Predict first

What is the dot product of a vector with that new vector?

  • Always zero
  • Always the square of the length
  • Twice the product of the components
  • It depends on the vector

Correct: Always zero

Why: The dot product is a times minus b plus b times a, which cancels to zero for every a and b. So the swap-and-negate move always produces a perpendicular vector, and that is the quickest way to write down a vector at right angles to a given one.

\[ \langle a,b\rangle\cdot\langle -b,a\rangle = -ab + ab = 0 \]

Predict before you compute. Take a vector, swap its two components and change the sign of the new first one, and ask what the dot product with the original comes to.

The algebra is one line and the two terms cancel for every choice of components. So the answer is always zero, which by the theorem means the new vector is always perpendicular to the old one. Geometrically the swap-and-negate move is a quarter turn anticlockwise.

This is genuinely useful. Whenever a problem asks you for a vector perpendicular to a given one, you do not need to solve anything: swap the components and negate one of them. It also explains the earlier example, where the two-two vector and the five-minus-five vector were perpendicular; the second is a rotated and stretched copy of the first.

37. Zero product, and neither factor is zero

Counterexample

Discussion prompt

For real numbers, a product of zero forces one of the factors to be zero. Give a pair of nonzero vectors whose dot product is zero, and say what that tells you about cancelling in dot product equations.

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, p. 1037 — the footnote on the zero product property

Write your example down before revealing. Any perpendicular pair works, and the simplest is the pair of unit vectors along the two axes.

The consequence is worth stating loudly, because it breaks a habit built over years of ordinary algebra. With real numbers, a product of zero forces a factor to be zero, and that is what lets you cancel. With dot products you may never cancel a common vector from both sides of an equation.

Think about why not. If two dot products with the same first vector are equal, all you learn is that the difference of the other two vectors is perpendicular to it, and perpendicular is a long way from zero. There is a whole line of vectors that produce the same dot product, and cancelling would be claiming there is only one.

38. Perpendicular, parallel, or neither?

Discrimination

Sort into buckets

Sort each pair. One dot product settles perpendicularity; for parallel, look for a scalar multiple.

Perpendicular
(2, 2) and (5, −5); (−2, 1) and (3, 6)
Parallel
(3, 4) and (−6, −8); (1, 2) and (3, 6)
Neither
(3, −4) and (2, 1); (1, 17) and (−1, 0)
perp
Both dot products come out zero: 10 − 10 for the first pair, and −6 + 6 for the fourth. Zero is the whole test.
par
Each second vector is a scalar multiple of the first: −2 times for the second pair, which reverses it, and 3 times for the fifth, which does not. Parallel pairs have dot products equal to plus or minus the product of the lengths.
nei
The dot products are 2 and −1, neither zero nor extreme, so these pairs sit at some ordinary angle: about 79.7 degrees and about 93.4 degrees.

Two different tests are in play here and you should apply them in the cheaper order. First compute the dot product: if it is zero you are done, the pair is perpendicular. Only if it is not zero do you need to check whether one vector is a scalar multiple of the other.

The parallel cases are worth inspecting. In one of them the multiplier is negative, so the vectors point in opposite directions along the same line; in the other it is positive. Both count as parallel, and both have dot products of the largest possible size for their lengths, one positive and one negative.

The pairs in the third bucket are the ordinary ones: dot products of two and minus one, close to neither extreme. Those are the pairs where you would actually have to compute the angle, and you already have: about eighty degrees and about ninety-three degrees, from earlier slides.

39. The shadow: projection

Section

Part 4

40. The shadow one vector casts on another

Concept

Figure (svg): A vector v leaning up and to the right, and a vector w below it. A dashed line drops from the tip of v straight down onto the line of w, meeting it at a right angle at a point R. The piece of w from the origin to R is drawn as a thick bright bar labelled the shadow p, and two fainter dashed rays show other points of v dropping onto the same bar.

Shine a light straight down onto the line of w and v casts a shadow on it. That shadow, from the tail to the foot of the perpendicular, is the projection of v onto w.

Shine a light perpendicular to w and the vector v casts a shadow along the line of w. Drop a perpendicular from the tip of v to the line of w, call the foot R, and the vector from the tail to R is the shadow.

orthogonal projection — The vector from the common tail to the foot of the perpendicular dropped from the tip of v onto the line of w. It points along w and measures how much of v lies in the direction of w.

It is a vector, not a number, and it always lies along the line of w. So it is determined by just two things: its direction, which is w's direction, and its length, which the right triangle in the picture gives away.

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, p. 1038 — the construction of the orthogonal projection

Take a moment with this picture, because everything in Part 4 is a comment on it. Imagine light falling perpendicular to the line of w. The vector v blocks some of it, and the shadow that lands on the line of w is the projection.

The construction is precise: drop a perpendicular from the tip of v onto the line containing w, and call the foot of that perpendicular R. The projection is the vector from the shared tail to R. The faint dashed rays in the picture show other points of v dropping to their own points of the shadow, which is what makes the word shadow more than a metaphor.

Two things are worth fixing now. The projection is a vector, not a number, and it always lies along the line of w. That means it is completely determined by two pieces of information, a direction, which is just w's direction, and a length, which the right triangle in the picture is about to give you.

41. Deriving the length of the shadow

Worked example

Figure (svg): A right triangle with the horizontal leg from O to R, the vertical leg from R up to T, and the hypotenuse from O to T. The hypotenuse is labelled the magnitude of v, the horizontal leg the magnitude of p, the angle at O is theta, and a small square marks the right angle at R.

Right-triangle trigonometry from Chapter 10 is all the projection formula needs: the shadow is the adjacent side, v is the hypotenuse.

Work inside the right triangle made by the tail, the foot of the perpendicular and the tip of v, taking the angle acute for now.

Read the cosine off the right triangle

Why: Adjacent over hypotenuse, from Chapter 10.

\[ \cos(\theta) = \frac{\|\vec{p}\|}{\|\vec{v}\|} \]

Solve for the length of the shadow

Why: Multiply through by the hypotenuse.

\[ \|\vec{p}\| = \|\vec{v}\|\cos(\theta) \]

Replace the cosine using Theorem 11.23

Why: The geometric formula, rearranged.

\[ \|\vec{v}\|\cos(\theta) = \frac{\|\vec{v}\|\,\|\vec{w}\|\cos(\theta)}{\|\vec{w}\|} = \frac{\vec{v}\cdot\vec{w}}{\|\vec{w}\|} \]

Recognise the unit vector

Why: Dividing w by its own length is exactly the unit vector of Section 11.8.

\[ \frac{\vec{v}\cdot\vec{w}}{\|\vec{w}\|} = \vec{v}\cdot\left(\frac{1}{\|\vec{w}\|}\vec{w}\right) = \vec{v}\cdot\hat{w} \]

Attach the direction

Why: A vector is its length times its unit direction, and the shadow points along w.

\[ \vec{p} = \|\vec{p}\|\,\hat{p} = (\vec{v}\cdot\hat{w})\,\hat{w} \]

Check the extreme case

Why: If v already points along w then the angle is zero, the cosine is 1, and the shadow has the full length of v, which is exactly what the picture would show with the perpendicular dropping onto the tip itself.

\[ \theta = 0 \;\Longrightarrow\; \|\vec{p}\| = \|\vec{v}\| \]

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, p. 1038 — the derivation before Definition 11.12

Strip the picture down to its right triangle and the derivation is Chapter 10 trigonometry. The shadow is the side adjacent to the angle, v is the hypotenuse, and the cosine is adjacent over hypotenuse.

The rest is rewriting. The length of the shadow is the length of v times the cosine, but you usually do not know the angle, so you replace the cosine using the geometric theorem. Doing that introduces the dot product on top and the length of w underneath, and dividing w by its own length is exactly the unit vector from Section 11.8.

The final assembly uses a principle from Section 11.8 as well: any vector equals its length times its own unit direction. The length is the number you just found and the direction is w's direction, so multiplying them gives the projection. Notice the unit vector of w appears twice for two different reasons, once inside the dot product to supply the length and once outside to supply the direction.

42. The obtuse case gives the same formula

Concept

Figure (svg): An obtuse pair: w points right and slightly up, v points up and to the left. The perpendicular dropped from the tip of v meets the line of w behind the origin, so the shadow bar runs backwards from the origin, drawn in red. The angle theta between the vectors is obtuse and its supplement theta prime is marked on the other side.

When the angle is obtuse the foot of the perpendicular lands behind the tail, so the shadow points backwards along w. One formula still covers it, because the cosine has already turned negative.

If the angle is obtuse the foot of the perpendicular lands behind the tail, so the shadow points the opposite way along w. Two sign changes then cancel.

\[ \hat{p} = -\hat{w}, \qquad \|\vec{p}\| = \|\vec{v}\|\cos(\theta') = -\|\vec{v}\|\cos(\theta) \]

\[ \vec{p} = \|\vec{p}\|\,\hat{p} = \left(-(\vec{v}\cdot\hat{w})\right)(-\hat{w}) = (\vec{v}\cdot\hat{w})\,\hat{w} \]

The length is measured with the supplementary angle inside the triangle, and the direction is reversed. Both facts carry a minus sign, and multiplying them together restores the formula unchanged. At exactly a right angle the shadow collapses to the zero vector, which the formula also gives, since the dot product is then zero.

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, pp. 1038-1039 — the obtuse and right-angle cases

The derivation you just followed quietly assumed the angle was acute, because it used a triangle with the foot of the perpendicular in front of the tail. When the angle is obtuse the foot lands behind the tail, and both the direction and the length calculation change.

Track the two minus signs. The direction is now the reverse of w's direction, which is one minus sign. And the length comes from the triangle's own angle, which is the supplement of the angle between the vectors, so the cosine picks up a minus sign too. Multiplying two negatives gives a positive, and the formula survives untouched.

This is why the definition can be stated without cases. It is also why the scalar part of the projection is allowed to be negative: a negative value is the formula's way of saying the shadow points backwards along w. At exactly a right angle the shadow collapses to a single point, the zero vector, and the formula gives that too.

43. Reading the projection formula

Notation

Annotate

On: \( \operatorname{proj}_{\vec{w}}(\vec{v}) = (\vec{v}\cdot\hat{w})\,\hat{w} \)

  • w is what you project ONTO; v is what casts the shadow. Swapping them gives a different vector, pointing along v instead.
  • A number: the scalar projection, how much of v lies in w's direction. It is negative when the angle is obtuse.
  • A unit vector supplying the direction. Number times direction is the vector, which is why ŵ appears twice and w itself not at all.
  • A vector parallel to w, and always a scalar multiple of it, however long or short w happens to be.

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, p. 1039 — Definition 11.12

The single most common error with this formula is getting the two vectors the wrong way round, so start with the annotation about the subscript. The vector in the subscript is the one you project onto, the one whose line the shadow lands on. The vector in the brackets is the one casting the shadow.

The two parts of the formula do two different jobs. The bracket is a number, called the scalar projection, and it measures how much of v runs in w's direction, signed. The second factor is a unit vector, and its only job is to supply the direction.

Notice that w itself never appears, only its unit vector. That is the formula saying, quite explicitly, that the length of w is irrelevant. Doubling w does not change the shadow at all, because it does not change the line the shadow lands on.

44. Three formulas, one vector

Concept

Normalising w is a nuisance in practice, because it introduces a square root that usually cancels later. Two rearrangements avoid it entirely.

\[ \operatorname{proj}_{\vec{w}}(\vec{v}) = (\vec{v}\cdot\hat{w})\,\hat{w} = \left(\frac{\vec{v}\cdot\vec{w}}{\|\vec{w}\|^2}\right)\vec{w} \]

\[ \left(\frac{\vec{v}\cdot\vec{w}}{\|\vec{w}\|^2}\right)\vec{w} = \left(\frac{\vec{v}\cdot\vec{w}}{\vec{w}\cdot\vec{w}}\right)\vec{w} \]

Each step just replaces the unit vector by w divided by its length, twice, and then rewrites the squared length as a self-dot. The last form is the one to compute with: two dot products and a division, and no radicals anywhere.

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, p. 1039 — Theorem 11.26

The theoretically clean formula is a nuisance to compute with, because normalising w means dividing by a square root, and that square root nearly always cancels a few lines later. The two rearrangements do the cancelling in advance.

Watch where the two copies of the unit vector go. Each one contributes a division by the length of w, so together they contribute a division by the length squared, and the two bare copies of w recombine into one. Then the squared length becomes a self-dot, using the property from Part 1.

The last form is the one to use in practice, and you should commit it to memory: two dot products, a division, and a scalar multiplication. No radicals appear anywhere, which means exact answers stay exact and fractions stay tidy.

45. Example 11.9.4: projecting one-eight onto minus-one-two

Worked example

Figure (svg): A grid showing the vector v to the point 1 comma 8, the short vector w to the point minus 1 comma 2, the projection p drawn as a thick bar from the origin to the point minus 3 comma 6 along the line of w, and the leftover q drawn from the tip of p to the tip of v, with a small square marking the right angle at the tip of p.

The projection runs three times as far as w itself, and the leftover from its tip to the tip of v meets it at a right angle. That right angle is what the check confirms in numbers.

\[ \vec{v} = \langle 1,8\rangle, \qquad \vec{w} = \langle -1,2\rangle \]

Dot v with w

Why: Minus one plus sixteen.

\[ \vec{v}\cdot\vec{w} = (1)(-1)+(8)(2) = -1+16 = 15 \]

Dot w with itself

Why: One plus four.

\[ \vec{w}\cdot\vec{w} = (-1)^2+2^2 = 1+4 = 5 \]

Divide to get the scalar

Why: Fifteen over five.

\[ \frac{\vec{v}\cdot\vec{w}}{\vec{w}\cdot\vec{w}} = \frac{15}{5} = 3 \]

Multiply w by that scalar

Why: Three times each component.

\[ \vec{p} = 3\langle -1,2\rangle = \langle -3,6\rangle \]

Check the direction and the drop

Why: The answer is three times w, so it certainly lies along w; and the vector from its tip to the tip of v is four right and two up, which dots with w to minus four plus four, zero, so the drop really is perpendicular.

\[ \langle 4,2\rangle\cdot\langle -1,2\rangle = -4+4 = 0 \]

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, pp. 1039-1040 — Example 11.9.4

This is the book's own example, with its own figure. Work it with the last of the three formulas and it is four short steps: dot the two vectors, dot w with itself, divide, and scale.

The quotient here comes out as the whole number three, which makes the picture easy to read: the projection is three copies of w laid end to end. Do not expect that in general; a fractional multiplier is much more common, as the next exercise shows.

Spend a moment on the check, because it is the check you should perform every single time. The vector from the tip of the projection to the tip of v is the leftover, four right and two up, and dotting it with w gives minus four plus four, which is zero. So the drop really was perpendicular, which is the definition of a projection, and any arithmetic slip in the first four steps would have shown up here as a nonzero answer.

46. The leftover is perpendicular, always

Concept

Figure (svg): A vector v split into two pieces: a thick bar p lying along the horizontal vector w, and an arrow q rising straight up from the tip of p to the tip of v, with a small square marking the right angle between them and the equation v equals p plus q written beside it.

Every vector splits, in exactly one way, into a piece that runs along w and a piece that runs across it. The dot product finds the first piece, and subtraction finds the second.

Subtract the shadow from the original vector and what remains is the part of v that runs across w. That leftover is perpendicular to w every time, and the split can be done in only one way.

\[ \vec{q} = \vec{v} - \vec{p}, \qquad \vec{v} = \vec{p} + \vec{q} \]

Theorem 11.27 — For nonzero vectors there are unique vectors p and q with v equal to their sum, where p is a scalar multiple of w and q dotted with w is zero. They are called the components of v parallel to w and orthogonal to w.

This is how a single vector gets taken apart along directions you choose: a force into the part that pushes a crate forward and the part that presses it into the floor, a velocity into the part along a slope and the part into it.

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, p. 1040 — Theorem 11.27

You have now seen the perpendicular check pass once. The theorem says it always passes, and the picture shows why the split feels inevitable: one piece of v runs along w and the rest runs straight across it.

The word to notice in the theorem is unique. There is exactly one way to write v as a piece parallel to w plus a piece perpendicular to w, which means the projection is not merely a convenient choice; it is forced.

This is one of the most reused ideas in all of applied mathematics. A force on a slope splits into a part that slides the object along and a part that presses it into the surface. A velocity splits into a part along a wall and a part into it. Every time, the parallel part is a projection and the perpendicular part is what is left over.

47. Checking the decomposition in general

Worked example

Show the leftover is always perpendicular to w, whatever the two vectors were.

Write down what the leftover is

Why: The definition of q, with p the projection.

\[ \vec{q}\cdot\vec{w} = (\vec{v}-\vec{p})\cdot\vec{w} = \vec{v}\cdot\vec{w} - \vec{p}\cdot\vec{w} \]

Substitute the projection formula

Why: Using the third form of Theorem 11.26.

\[ \vec{p}\cdot\vec{w} = \left(\left(\frac{\vec{v}\cdot\vec{w}}{\vec{w}\cdot\vec{w}}\right)\vec{w}\right)\cdot\vec{w} \]

Pull the scalar out of the dot product

Why: The scalar property of Theorem 11.22.

\[ \vec{p}\cdot\vec{w} = \left(\frac{\vec{v}\cdot\vec{w}}{\vec{w}\cdot\vec{w}}\right)(\vec{w}\cdot\vec{w}) = \vec{v}\cdot\vec{w} \]

Substitute back

Why: The two terms are identical.

\[ \vec{q}\cdot\vec{w} = \vec{v}\cdot\vec{w} - \vec{v}\cdot\vec{w} = 0 \]

Verify on the previous example

Why: There v was one-eight, p was minus-three-six, so q is four-two, and the general result predicts its dot product with w is zero, which is exactly what was computed.

\[ \vec{q} = \langle 1,8\rangle - \langle -3,6\rangle = \langle 4,2\rangle, \qquad \vec{q}\cdot\vec{w} = 0 \]

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, p. 1041 — the proof of Theorem 11.27

This proof is short and it is the payoff for all four properties from Part 1. Watch how the scalar comes out of the dot product in the middle step, leaving w dotted with w in both the numerator and the denominator of the same expression.

Those two cancel, and the whole thing collapses to the dot product of v and w minus itself, which is zero. Nothing about the particular vectors was used, so the leftover is perpendicular to w always, no matter how strangely the two vectors are placed.

Read that as a licence to use the check freely. When you finish a projection, subtract and dot; a zero means every number in your calculation was almost certainly right, and anything else means you have a mistake to find.

48. Exercise 1: all four quantities for one pair

Worked example

The section's exercises ask for the same four things every time: the dot product, the angle, the projection, and the leftover with its check.

\[ \vec{v} = \langle -2,-7\rangle, \qquad \vec{w} = \langle 5,-9\rangle \]

Compute the dot product

Why: Minus ten plus sixty-three.

\[ \vec{v}\cdot\vec{w} = (-2)(5)+(-7)(-9) = -10+63 = 53 \]

Compute the magnitudes

Why: Four plus forty-nine, and twenty-five plus eighty-one.

\[ \|\vec{v}\| = \sqrt{53}, \qquad \|\vec{w}\| = \sqrt{106} \]

Divide, and simplify the radical

Why: The product of the roots is the root of fifty-three times one hundred six, and one hundred six is twice fifty-three.

\[ \cos\theta = \frac{53}{\sqrt{53}\sqrt{106}} = \frac{53}{53\sqrt{2}} = \frac{1}{\sqrt{2}} \]

Read off the angle

Why: A value from the unit circle.

\[ \theta = 45^{\circ} \]

Compute the projection

Why: Fifty-three over one hundred six is one half.

\[ \vec{p} = \frac{53}{106}\langle 5,-9\rangle = \left\langle \tfrac52, -\tfrac92 \right\rangle \]

Subtract to get the leftover

Why: Component by component.

\[ \vec{q} = \langle -2,-7\rangle - \left\langle \tfrac52,-\tfrac92\right\rangle = \left\langle -\tfrac92, -\tfrac52 \right\rangle \]

Check the leftover is perpendicular

Why: Minus forty-five halves plus forty-five halves is zero, so q really is across w; and the projection came out as exactly half of w, which fits an angle of forty-five degrees on a pair whose lengths differ by a factor of the root of two.

\[ \vec{q}\cdot\vec{w} = -\tfrac{45}{2} + \tfrac{45}{2} = 0 \]

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, p. 1043 — Exercise 1

This is what the first twenty exercises of the section actually ask for, so treat it as the template. Four questions, and each one feeds the next.

The radical simplification in the middle is worth learning to see. The product of the square root of fifty-three and the square root of one hundred six is the square root of their product, and one hundred six is twice fifty-three, so fifty-three comes straight back out of the radical and cancels the numerator. What remains is one over the square root of two, which you know is forty-five degrees.

The projection turns out to be exactly half of w, and you can sanity-check that against the angle. At forty-five degrees the shadow is the length of v times the cosine of forty-five degrees, and since the length of w happens to be the square root of two times the length of v, half of w is exactly right. Then the final dot product comes out zero, as the theorem promised.

49. Find the error: projected the wrong way round

Error analysis

Asked for the projection of v onto w, a student writes this.

Annotate

On: \( \operatorname{proj}_{\vec{w}}(\vec{v}) = \left(\frac{\vec{v}\cdot\vec{w}}{\vec{v}\cdot\vec{v}}\right)\vec{v} \)

  • Correct, and it is the same either way round, because the dot product is commutative.
  • Wrong vector. Dividing by v dotted with v cancels against the wrong length and leaves a formula that is not even parallel to w.
  • This is the real giveaway. The answer must point along the vector being projected ONTO, so w has to be the last factor.
  • This expression is the projection of w onto v, a different vector entirely. On Example 11.9.4 it gives 15 over 65 times the one-eight vector, not the minus-three-six vector.
  • The subscript and the final factor are always the same vector. If they disagree, the formula is wrong.

Look at the formula before revealing the notes and ask one question: which vector does the answer point along? Whatever else is going on, the answer has to be parallel to the vector in the subscript.

It is not. The expression ends with a factor of v, so whatever it computes points along v, and a projection onto w cannot do that. That single observation catches the error without checking a single number.

Diagnosing it more precisely: this student has divided by v dotted with v instead of w dotted with w, which is the formula for projecting w onto v. The memory hook in the last annotation is the one to keep. The vector in the subscript and the vector at the end are always the same one, and if they disagree, something has been swapped.

50. Match each expression to what it is

Matching

Match the pairs

  • a. v · w
  • b. v · v
  • c. (v · w / w · w) w
  • d. v − proj of v onto w
  • e. v · w divided by both magnitudes
  • r1. a number, zero exactly when the vectors are perpendicular
  • r2. the square of the length of v
  • r3. a vector parallel to w
  • r4. a vector perpendicular to w
  • r5. the cosine of the angle between them

Why: Three of the five are numbers and two are vectors, and reading off which is which is half the skill. The projection is always parallel to the vector in the subscript, and the leftover is always perpendicular to it; dividing the dot product by both lengths strips out the sizes and leaves pure direction information.

Match all five before checking. Three of these are numbers and two are vectors, and telling them apart at a glance is the skill being practised.

Two pairs are easy to confuse. The projection and the leftover look similar on the page, but one is parallel to w and the other is perpendicular to it; that is the entire content of the decomposition theorem. The dot product and the dot product divided by both lengths also look similar, but only the second one is trapped between minus one and one and therefore says anything about direction on its own.

If you got any of them wrong, go back to the expression and count the vectors left over after the dot products are taken. No vectors left means a number; one vector left means a vector pointing along it.

51. Projecting onto a vector that points backwards

Prediction

\[ \vec{v} = \langle 3,4\rangle, \qquad \vec{w} = \langle -6,-8\rangle \]

Predict first

The second vector is minus two times the first. What is the projection of the first onto the second?

  • v itself
  • the negative of v
  • the zero vector
  • w itself

Correct: v itself

Why: The dot product is minus fifty and w dotted with itself is one hundred, so the scalar is minus one half, and minus one half times w is the three-four vector, which is v. That is right: v already lies along the line of w, so its shadow on that line is all of v. Projecting onto a reversed vector does not reverse the shadow, because the projection depends on the line, not on which way along it w points.

\[ \frac{-50}{100}\langle -6,-8\rangle = -\tfrac12\langle -6,-8\rangle = \langle 3,4\rangle = \vec{v} \]

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, p. 1043 — Exercise 4

Commit to an option before revealing, and expect the answer to feel slightly wrong at first. The second vector points the opposite way along the same line as the first.

The arithmetic gives a negative scalar, minus one half, and multiplying a backwards-pointing vector by a negative number turns it forwards again. So the projection is v itself.

The geometry explains why that had to happen. The projection only depends on the line that w lies along, not on which direction along that line w points, because the perpendicular from the tip of v lands in the same place either way. Here v is already on that line, so its shadow is all of it. Reversing w changes the sign of the scalar and the direction of w simultaneously, and those two changes cancel, which is the same cancellation you saw in the obtuse case.

52. Put the projection calculation in order

Ranking

Put in order

Order the steps for finding the projection of v onto w and the perpendicular leftover.

  1. Dot v with w
  2. Dot w with itself
  3. Divide the first number by the second
  4. Multiply w by that quotient
  5. Subtract the result from v
  6. Dot the leftover with w and confirm it is zero

Why: Both dot products come first because the quotient needs them, and the quotient must be a number before it can scale w. Subtracting produces the leftover, and dotting that with w is the free check that catches almost every arithmetic slip in the first four steps.

Drag these into order before checking. The dependencies do most of the work: you cannot divide until both dot products exist, and you cannot scale w until the quotient is a number.

The step people leave out is the last one. Dotting the leftover with w and confirming a zero costs about ten seconds and catches nearly every arithmetic error in the preceding five steps, because a slip anywhere upstream produces a nonzero answer here. The book's exercises ask for it explicitly for exactly that reason.

53. Work: the dot product in physics

Section

Part 5

54. Pushing, carrying, dragging

Picture it

Figure (svg): Three panels, each with a crate and a rightward motion arrow. In the first the force arrow points right along the motion and the work is positive; in the second the force arrow points straight up while the crate moves sideways and the work is zero; in the third the force arrow points left against the motion and the work is negative.

The distance travelled is the same in all three. Only the angle between the force and the motion changes, and it is the angle that decides the sign.

In physics, a constant force acting along the direction of motion does work equal to the force times the distance. When the force is not along the motion, only the part of it that is along the motion counts, and that part is exactly what the projection measures.

In everyday speech work means effort. In physics it is a precise quantity, and the three panels show why the angle between the force and the motion is the whole story.

In the first panel the push is along the motion, so all of it counts and the work is the force times the distance, positive. In the third the force opposes the motion, so the work is negative: a brake does negative work on a car. In the middle panel the force is perpendicular to the motion and the work is exactly zero.

That middle case is the one that feels wrong, and it is worth sitting with. Carrying a heavy box across a level room is genuinely tiring, but the tiredness is your muscles burning energy to stay tensed, not work being done on the box. The box gains nothing: it is moving sideways while the force holding it points straight up.

55. Work is a dot product

Concept

Let the force be a vector and let the displacement be the vector from the starting point to the finishing point. The amount of the force lying along the motion is the scalar projection of the force onto the displacement.

\[ \text{force along the motion} = \vec{F}\cdot\widehat{PQ} \]

\[ W = (\vec{F}\cdot\widehat{PQ})\,\|\overrightarrow{PQ}\| = \vec{F}\cdot\left(\|\overrightarrow{PQ}\|\,\widehat{PQ}\right) = \vec{F}\cdot\overrightarrow{PQ} \]

Theorem 11.28 — The work done by a constant force applied along a displacement is the dot product of the force and the displacement, which is also the product of their magnitudes times the cosine of the angle between them.

\[ W = \vec{F}\cdot\overrightarrow{PQ} = \|\vec{F}\|\,\|\overrightarrow{PQ}\|\cos(\theta) \]

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, pp. 1041-1042 — Theorem 11.28

The reasoning is short. Only the part of the force lying along the motion moves the object, and you know how to measure that part: it is the scalar projection of the force onto the displacement.

Multiply that by the distance travelled and a small piece of algebra in the second line turns the whole thing into a single dot product of the force with the displacement. The unit vector in the scalar projection recombines with the distance to rebuild the displacement vector itself.

So the theorem has two forms, and you should hold both. If you are given components, dot them. If you are given two magnitudes and an angle, which is far more common in word problems, multiply the two magnitudes by the cosine. Both give the same number, and the last slide of this part checks one against the other.

56. Example 11.9.5: Taylor pulls her wagon

Worked example

Figure (svg): A wagon on level ground with a handle leaving the wagon at 30 degrees above the horizontal. A force arrow of 10 pounds points along the handle, dashed lines show its horizontal part of about 8.66 pounds and its vertical part of 5 pounds, and an arrow along the ground marks the 50 foot displacement.

Only the part of the pull that lies along the ground moves the wagon forward. The upward part lifts against the axles and contributes nothing to the work.

Taylor exerts a force of 10 pounds to pull her wagon 50 feet over level ground, holding the handle at 30 degrees to the horizontal for the whole journey. Find the work done.

Write the force in components

Why: Magnitude times the cosine and sine of the angle, from Section 11.8.

\[ \vec{F} = 10\langle \cos(30^{\circ}),\sin(30^{\circ})\rangle = \langle 5\sqrt{3},\,5\rangle \]

Write the displacement in components

Why: Fifty feet in the positive direction, with no vertical part.

\[ \overrightarrow{PQ} = 50\,\hat{\imath} = \langle 50,0\rangle \]

Take the dot product

Why: The vertical components multiply to zero and drop out.

\[ W = \langle 5\sqrt{3},5\rangle\cdot\langle 50,0\rangle = 250\sqrt{3}+0 \]

State the answer with its units

Why: Pounds times feet.

\[ W = 250\sqrt{3} \approx 433.01 \text{ foot-pounds} \]

Check it the other way

Why: The second form of the theorem uses the magnitudes and the angle directly, with no components at all, and gives the same number, which also confirms the 30 degrees was used as the angle between the force and the motion.

\[ W = (10)(50)\cos(30^{\circ}) = 500\cdot\frac{\sqrt{3}}{2} = 250\sqrt{3} \]

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, p. 1042 — Example 11.9.5

The book solves this twice, and so does the slide, because the two methods illuminate different things. The component method makes it visible why the vertical part of the pull disappears: it gets multiplied by the zero vertical component of the displacement.

That is the physical content. Five pounds of the ten-pound pull is directed straight upwards, and the wagon never rises, so those five pounds do nothing at all. Only the horizontal part, about eight and two thirds pounds, moves the wagon down the path.

The second method is the one to use on an exam: two magnitudes and a cosine, straight into the formula. The agreement between the two is a real check, not a formality, and it confirms in particular that thirty degrees is the angle between the force and the motion, which is exactly the quantity the formula asks for.

57. Exercise 23: the same wagon, loaded with rocks

Worked example

Taylor fills the wagon with rocks and now needs 13 pounds of force, at 15 degrees to the horizontal, to pull it 25 feet. Find the work, to two decimal places.

Identify the three quantities

Why: Force magnitude, distance, and the angle between them.

\[ \|\vec{F}\| = 13, \qquad \|\overrightarrow{PQ}\| = 25, \qquad \theta = 15^{\circ} \]

Substitute into the theorem

Why: The second form, since no components are given.

\[ W = (13)(25)\cos(15^{\circ}) = 325\cos(15^{\circ}) \]

Evaluate the cosine

Why: A calculator value, kept to six places.

\[ \cos(15^{\circ}) \approx 0.965926 \]

Multiply out

Why: Three hundred twenty-five times that cosine.

\[ W \approx 313.93 \text{ foot-pounds} \]

Check the size against the last example

Why: Less force over half the distance should be far less work, and it is: about 314 foot-pounds against 433. The shallower angle helps a little, since the cosine of 15 degrees is nearer to one than the cosine of 30 degrees. The book's answer key prints 313.92, which truncates rather than rounds the true value of 313.9259.

\[ 325\cos(15^{\circ}) = 313.9259\ldots \;\Longrightarrow\; 313.93 \text{ to two places} \]

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, p. 1043 — Exercise 23

This is the pure plug-in version, and the whole skill is identifying which number is which. The force magnitude, the distance, and the angle between them, in that order.

Before evaluating, predict the size. The force is larger than in the previous example but the distance is halved, so you should expect noticeably less work; the shallower angle claws a little back, since the cosine of fifteen degrees is closer to one. Landing near three hundred fourteen, against four hundred thirty-three, is consistent with all of that.

The book's answer key gives three hundred thirteen point nine two, but the exact value is three hundred thirteen point nine two five nine, which rounds to three hundred thirteen point nine three at two decimal places. The key truncated instead of rounding. Your rounded answer is the correct one, and it is worth knowing that even a careful textbook does this occasionally.

58. Trap: force times distance, whatever the angle

Trap

The trap

Asked for the work in the wagon problem, a student writes:

\[ W = (10)(50) = 500 \]

Wrong. That is the work only if the pull is along the motion.

The fix

The force is applied at an angle, so part of it is lifting rather than pulling. Only the horizontal part moves the wagon forward.

\[ W = (10)(50)\cos(30^{\circ}) \approx 433 \]

The upright part of the pull is 5 pounds, and it does no work at all, because the wagon does not rise. Whenever an angle is mentioned in a work problem, it belongs in the formula.

This is the most common mistake in every work problem, and it happens because the formula force times distance is the one people remember from school. That formula is a special case, valid only when the force points along the motion.

The moment a problem mentions an angle, a handle, a rope held at some tilt, or a ramp, the cosine belongs in the calculation. The pull in the wagon problem is ten pounds, but only about eight and two thirds of those pounds are pulling the wagon forward; the rest is lifting the handle.

One quick self-test before you finish a work problem: if the answer you wrote down does not contain a cosine and the problem mentioned an angle, go back.

59. Exercise 25: a barrel pushed up an incline

Worked example

Figure (svg): A shallow ramp rising to the right at 12.5 degrees with a barrel on it. An arrow along the ramp marks the 10 foot motion, an arrow straight up from the barrel marks the 200 pound force needed to overcome gravity, and the angle between the two, 77.5 degrees, is marked at the barrel.

The ramp tilts by 12.5 degrees, but the force is straight up, so the angle the formula wants is what is left of a right angle: 77.5 degrees.

A 200 pound barrel is moved 10 feet up a ramp that rises at 12.5 degrees. Ignore everything except gravity, so the force needed acts straight upwards. Find the work, to two decimal places.

Identify the direction of the force

Why: Gravity pulls straight down, so overcoming it means pushing straight up: the force makes a right angle with the horizontal.

\[ \vec{F} \text{ points at } 90^{\circ} \]

Identify the direction of the motion

Why: Along the ramp, which is tilted by twelve and a half degrees.

\[ \overrightarrow{PQ} \text{ points at } 12.5^{\circ} \]

Subtract to get the angle between them

Why: This is the step the problem is built around: the angle in the formula is not the angle of the ramp.

\[ \theta = 90^{\circ} - 12.5^{\circ} = 77.5^{\circ} \]

Substitute into the theorem

Why: Force, distance, cosine.

\[ W = (200)(10)\cos(77.5^{\circ}) = 2000\cos(77.5^{\circ}) \]

Evaluate

Why: The cosine of seventy-seven and a half degrees is about 0.216440.

\[ W \approx 432.88 \text{ foot-pounds} \]

Check the answer against a limiting case

Why: Lifting the barrel straight up would put the angle at zero and give the full 2000 foot-pounds, while sliding it along level ground would put the angle at ninety degrees and give nothing. A shallow ramp sits near the second case, and 433 is indeed a small fraction of 2000.

\[ \cos(0^{\circ}) = 1, \qquad \cos(90^{\circ}) = 0, \qquad \cos(77.5^{\circ}) \approx 0.2164 \]

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, p. 1044 — Exercise 25

This exercise exists to punish exactly one assumption, and the book's own hint says so. The angle in the work formula is between the force and the motion, and here the force and the ramp point in different directions.

Read the set-up carefully. You are told to ignore everything but gravity, and gravity acts straight down, so the force you must supply to overcome it acts straight up. The motion, however, is along the ramp. Straight up is ninety degrees from horizontal and the ramp is twelve and a half degrees from horizontal, so the angle between them is what remains: seventy-seven and a half degrees.

The limiting cases at the end tell you the answer is plausible. Lifting the barrel straight up would take the full two thousand foot-pounds; sliding it along level ground, against gravity only, would take nothing. A shallow ramp is much nearer the second case, so an answer of about four hundred thirty out of two thousand is exactly the right kind of size.

60. Which one does no work at all?

Elimination

Eliminate the wrong options

In the technical sense of this section, exactly one of these does zero work. Eliminate the others.

  • A. Pulling a sled 20 feet with a rope held at 30 degrees above the horizontal
  • B. Carrying a 20 pound box across a level room at a steady height
  • C. Lowering a box slowly to the floor
  • D. Pushing a crate 5 feet along the floor

Survives elimination: B

Why: Carrying a box horizontally means the supporting force points straight up while the motion is horizontal, so the angle is exactly ninety degrees and the cosine is zero. It is tiring because your muscles burn energy holding the box, but no work in this sense is done on the box: nothing moves in the direction of the force.

Eliminate three and keep one. For each situation, picture the direction of the force and the direction of the motion, then ask what angle they make.

Two of them are easy: the push along the floor has a zero angle and the sled rope has a thirty degree angle, so both do positive work. Lowering a box is the interesting negative case, because your hands push up while the box moves down, making a straight angle and negative work.

The survivor is carrying the box across the room. The supporting force is straight up and the motion is horizontal, so the angle is exactly ninety degrees and the cosine is zero. Remember the distinction from the earlier panel: no work is being done on the box, even though holding it up is tiring.

61. Where the dot product turns up

Real world

Discussion prompt

A solar panel faces a fixed direction, and the sunlight arrives along a fixed direction. Describe how you would use a dot product to decide how much energy the panel collects, and what angle gives the most.

Write your description before revealing. The trick is choosing the right two vectors, which is most of the skill in any applied dot product problem.

For a surface, the natural vector is the one sticking straight out of it, perpendicular to the surface, and for light the natural vector points along the direction of travel with a length recording the intensity. The energy collected is proportional to the dot product of those two, because only the component of the light arriving head-on is captured.

The consequence is familiar: a panel facing straight into the sun collects the most, and a panel edge-on to the sun collects nothing, with a cosine running smoothly between. The same calculation runs in every three-dimensional video game you have ever played, once per surface, to decide how brightly it should be lit.

62. Putting it together

Section

Part 6

63. Pattern: what to do with two vectors

Pattern

Figure (svg): A flow diagram: two vectors feed into a box marked v dot w, and from a vertical spine five labelled arrows lead out to five results: acute right or obtuse from the sign, the angle from the arccosine, perpendicular or not from asking whether it is zero, the projection from multiplying by w over w dot w, and the work done when the vectors are a force and a displacement.

One number, five questions. Which one you are asking decides what you do with the number next.
  1. Compute the dot product first. Multiply matching components and add: one number.
  2. Want the angle? Divide by both magnitudes and take the arccosine. Dividing first is what keeps the input inside the arccosine's domain.
  3. Only want to know if they are perpendicular? Stop at the dot product and ask whether it is zero.
  4. Want the shadow? Divide the dot product by w dotted with itself, then multiply that number by w. Subtract from v for the perpendicular part, and dot it with w to check.
  5. Force and displacement? The dot product is the work, in force units times distance units.

This is the order to work in whenever a problem hands you two vectors, and it runs from the cheapest computation to the most expensive.

Start with the dot product, always, because it is two multiplications and an addition and it may finish the problem. If all that was wanted is whether the vectors are perpendicular, you are done at that point. If the angle is wanted, divide by both magnitudes before taking the arccosine, which is the step that error-analysis slide was about.

For a projection, use the third formula: dot product over w dotted with itself, times w, and subtract from v for the perpendicular piece. Then always finish with the free check. And when the two vectors happen to be a force and a displacement, the dot product you computed in the very first step is already the answer, in force units times distance units.

64. The four questions one dot product answers

Comparison

Comparison matrix

what you want to knowwhat to computewhat the answer is
Are they perpendicular?whether the dot product is zeroyes or no
What is the angle?arccos of the dot product over both magnitudesan angle from 0 to 180 degrees
How much of v lies along w?v dotted with the unit vector of wa number, possibly negative
What is the shadow vector?the dot product over w dotted with w, times wa vector parallel to w
How much work was done?the force dotted with the displacementa number, in foot-pounds

Fill in the blanks before checking. The table is worth copying into your notes, because it is the whole lesson compressed into five rows.

Notice how much comes from a single computation. One dot product answers the perpendicularity question outright, feeds the angle formula, gives the scalar projection once you normalise, and is the work if the vectors happen to be a force and a displacement.

The column that trips people in exams is the last one. Keep track of whether each answer is a number or a vector, and if it is a number, whether it carries units.

65. Check: reading a dot product

Check

Check your understanding

Two nonzero vectors have dot product zero. Which conclusion is justified?

  • A. One of them must be the zero vector
  • B. They point in opposite directions
  • C. They are perpendicular (correct)
  • D. They are parallel

Answer: C

Why: Theorem 11.25: for nonzero vectors, a dot product of zero happens exactly when the angle is ninety degrees. The cosine vanishes at exactly one place between zero and a straight angle, so nothing else can produce a zero.

Why A tempts people
There is no zero product property for the dot product. The vectors (1, 0) and (0, 1) are both nonzero and dot to zero.
Why B tempts people
Opposite directions give a cosine of minus one, so the dot product would be minus the product of the lengths: as negative as it can get, not zero.
Why D tempts people
Parallel vectors have a cosine of plus or minus one, so their dot product has the largest possible size, not the smallest.

This question is about what a zero dot product does and does not prove. Three of the four options describe real situations, but only one of them is what zero means.

The tempting wrong answer is the first, because with ordinary numbers a zero product does force a zero factor. That instinct is wrong here, and the counterexample slide gave you two perfectly ordinary nonzero vectors dotting to zero.

The two parallel options fail for the opposite reason: parallel vectors, in either direction, have cosines of plus or minus one and therefore dot products of the largest possible size for their lengths.

66. Check: from magnitudes to the angle

Check

Check your understanding

Two vectors each have length 10, and their dot product is −50. What is the angle between them?

  • A. 60 degrees
  • B. 120 degrees (correct)
  • C. 150 degrees
  • D. There is not enough information

Answer: B

Why: Divide the dot product by both lengths: −50 over 100 is −0.5, and the arccosine of −0.5 is 120 degrees. The magnitudes and the dot product are all the angle formula needs, so the components are never required.

Why A tempts people
60 degrees has cosine +0.5, which would give a dot product of +50. The sign was dropped.
Why C tempts people
150 degrees has cosine about −0.866, which would give a dot product of about −86.6, not −50.
Why D tempts people
The formula needs only the dot product and the two magnitudes, and all three were given.

Notice what you are given and what you are not. There are no components anywhere in this problem, and none are needed: the angle formula asks only for the dot product and the two magnitudes.

Divide minus fifty by one hundred to get minus one half, then take the arccosine. If you know the unit circle, you should read minus one half straight off as one hundred twenty degrees without reaching for a calculator.

The sixty-degree option is what you get by dropping the minus sign, which is worth guarding against: the negative dot product told you before any arithmetic that the angle had to be obtuse, so any answer under ninety degrees was wrong on inspection.

67. Check: the length of the shadow

Check

Check your understanding

A vector v of length 10 makes a 60 degree angle with w. How long is the projection of v onto w?

  • A. 5 (correct)
  • B. about 8.66
  • C. 10
  • D. 0.5

Answer: A

Why: The length of the shadow is the length of v times the cosine of the angle: 10 times cos 60 degrees, which is 10 times one half, namely 5. Notice the length of w never enters — it sets the direction, not the size of the shadow.

Why B tempts people
That is 10 times the sine of 60 degrees, which is the length of the perpendicular leftover, not the shadow.
Why C tempts people
The full length of v would only be the shadow if the angle were zero, with v already lying along w.
Why D tempts people
That is the cosine on its own. The cosine must still be multiplied by the length of v.

This asks for the length of the projection, not the projection itself, so the answer is a number. The formula is the length of v times the cosine of the angle, straight off the right triangle.

Ten times the cosine of sixty degrees is ten times one half, so the shadow is five units long, exactly half of v. The tempting wrong answer uses the sine instead, which gives the perpendicular leftover rather than the shadow; check which side of the triangle you are naming before you pick a ratio.

Notice that the length of w was never mentioned in the question and is not needed. It determines the direction of the shadow, never its length.

68. Explain the shadow

Explain it to yourself

Discussion prompt

The projection formula divides by w dotted with itself, never by v dotted with itself. Explain in two sentences why the length of w must be divided out, and why the length of v must not be.

Write two sentences before revealing. This is the question that separates having memorised the formula from understanding it.

The length of v belongs in the answer, because a longer v really does cast a longer shadow; that is why the formula never divides by v's length. The length of w does not belong, because stretching w does not move the line it lies along and so cannot move the foot of the perpendicular.

So the formula has to cancel w's length out completely. Counting the factors makes it exact: dividing by w dotted with itself removes two copies of that length, and the final multiplication by w puts exactly one back, leaving the net effect of dividing by the length once, which is precisely what turns w into a direction.

69. Exit ticket

Exit ticket

\[ \vec{v} = \langle 4,3\rangle, \qquad \vec{w} = \langle 6,8\rangle \]

Discussion prompt

Find the dot product, the angle to the nearest tenth of a degree, the projection of v onto w, and the perpendicular leftover with its check. Then say what the number would mean if v were a force in pounds and w a displacement in feet.

This runs the whole lesson on one pair of vectors, so do it without looking back and see which step you have to hunt for.

The numbers are friendly on purpose. The dot product is forty-eight, the magnitudes are five and ten, and the cosine is forty-eight over fifty, which is nought point nine six. A cosine that close to one says immediately that the vectors nearly agree in direction, and the angle confirms it at about sixteen and a third degrees.

The projection scalar is forty-eight over one hundred, so the shadow is just under half of w, and subtracting gives a leftover whose dot product with w is zero. Then the final sentence changes the interpretation without changing a single number: as a force and a displacement, that first dot product of forty-eight is the work, in foot-pounds, and the small angle is why nearly all of the force is doing something useful.

70. Recap

Recap

what you computewhat it needswhat it tells you
v · wthe four componentsa single number, positive for acute, zero for a right angle, negative for obtuse
v · vone vectorthe square of its length
arccos of v · w over both lengthsthe dot product and both magnitudesthe angle, from 0 to 180 degrees
proj of v onto wtwo dot productsthe shadow: a vector along w
v minus the projectionthe projectionthe part of v across w, perpendicular to it
F · PQa force and a displacementthe work done, in force units times distance units

\[ \vec{v}\cdot\vec{w} = \|\vec{v}\|\,\|\vec{w}\|\cos(\theta), \qquad \operatorname{proj}_{\vec{w}}(\vec{v}) = \left(\frac{\vec{v}\cdot\vec{w}}{\vec{w}\cdot\vec{w}}\right)\vec{w} \]

Next, Section 11.10 leaves vectors behind and describes curves by letting a point's two coordinates each be a function of time.

Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection §11.9, pp. 1034-1043 — the whole section

OpenStax Algebra and Trigonometry 2e, §10.8 Vectors §10.8 Vectors — the same material in OpenStax

One definition, six computations. The dot product itself is two multiplications and an addition, and everything on this slide is built from it. If you remember only one line, remember that it equals the product of the two lengths times the cosine of the angle, because every other row follows from that.

The three most common exam errors are all worth a last look. Do not read the size of a dot product as a statement about the angle; divide by both lengths first. Do not put the wrong vector in the projection formula; the subscript and the final factor are always the same one. And do not drop the cosine from a work problem that mentions an angle.

Section 11.10 changes subject entirely, leaving vectors for parametric equations, where a curve is described by letting each coordinate of a moving point be its own function of time.

Sources

  1. Stitz & Zeager, College Trigonometry, Ch. 11 Applications of Trigonometry — §11.9 The Dot Product and Projection — Stitz & Zeager, College Trigonometry, Version 3 Corrected Edition, July 4 2013, pp. 1034-1043
  2. OpenStax Algebra and Trigonometry 2e, §10.8 Vectors

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

Book on Wyzant · Text (657) 465-8108