The identity matrix, inverse matrices and the determinant condition for their existence, the formula for the inverse of a 2 by 2, writing a linear system as a matrix equation, and solving it with a single multiplication.
Subject: Algebra 2 · 67 slides · symbolic lesson
Open the interactive version of this deck
Title
Algebra 2 · Lesson 3.8
One matrix undoes another, and the determinant decides whether it can
Objectives
Chapter 3 has spent seven lessons taking systems apart. You have graphed them, substituted, eliminated, stacked the numbers into matrices, added and multiplied those matrices, and used determinants to run Cramer's rule. This lesson puts all of it into a single line of arithmetic.
The idea behind it is one you have had since primary school. To undo multiplying by 3 you multiply by one third. This lesson builds the matrix that plays the part of one third, and finds out exactly when it exists.
Keep one question running through every slide: what would this look like if the matrices were ordinary numbers? Almost everything here is the number version wearing a bigger coat, and the two places where the coat does not fit are the two places worth your attention.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 210 — Before, Now, Why and the Key Vocabulary
Warm-up
Both halves of this warm-up return within three slides, so do them properly on paper rather than nodding at them. The first is the idea this whole lesson copies; the second is the test that decides whether the copy works.
\[ 3x = 21 \]
Discussion prompt
Solve the equation above without dividing: multiply both sides by something instead, and say what you multiplied by. Then find the determinant of the matrix with top row 2 and negative 3 and bottom row 1 and 4.
Hold on to the shape of that first answer. You reached for a multiplier that turns the coefficient into 1, and then leaned on the fact that 1 changes nothing. Those are the two ingredients of the whole lesson, and Part 1 builds the matrix version of the second one.
If the determinant came out as anything other than 11, check the signs before moving on. Two times four is eight, and subtracting negative three times one adds three rather than taking three away.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.7, p. 203 — the determinant of a 2 by 2 matrix
Section
Part 1
Concept
Every time you have solved an equation by dividing, you were really multiplying by a reciprocal. The number 3 has the partner one third, and the two of them multiply to 1. That partnership is what lets you get back to where you started.
\[ 3 \cdot \tfrac13 = 1 \qquad \text{and} \qquad 1 \cdot 5 = 5 \]
Figure (svg): A flow diagram. On the top row, 5 becomes 15 after multiplying by 3, and 15 becomes 5 again after multiplying by one third. On the bottom row, 5 becomes 0 after multiplying by 0, and a crossed-out arrow shows that no multiplier brings the 5 back.
The bottom row of the picture is the exception, and it is worth staring at. Multiplying by zero throws information away: every number becomes the same zero, so nothing you multiply by afterwards can tell you which number you began with.
So arithmetic already has a rule of exactly the shape this lesson needs. A number has a reciprocal when it is not zero, and no reciprocal when it is. Matrices will turn out to have the same rule with one quantity in place of the number, and that quantity is the determinant.
Before any of that can happen, though, the matrix world needs its own version of the number 1. Without something that plays the part of 1, there is nothing for a pair of matrices to multiply to.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 210 — the identity matrix introduced as the matrix version of 1
Concept
Here is the matrix that behaves like the number 1: it has ones down the main diagonal, running from the top left corner to the bottom right, and zeros in every other position.
identity matrix — The n by n matrix with 1s on the main diagonal and 0s elsewhere, written I. For any n by n matrix A, both A I and I A give back A.
Figure (svg): The 2 by 2 identity matrix and the 3 by 3 identity matrix drawn side by side, with the ones on the main diagonal in yellow and the zeros in grey, and a dashed line running down each main diagonal.
Notice that the identity comes in sizes. There is a 2 by 2 one, a 3 by 3 one and so on, and the one you need is always the size that makes the multiplication legal. The book writes all of them with the same letter, and which size is meant is read off the matrix beside it.
\[ I = \begin{bmatrix}1 & 0\\0 & 1\end{bmatrix} \qquad I = \begin{bmatrix}1 & 0 & 0\\0 & 1 & 0\\0 & 0 & 1\end{bmatrix} \]
The identity is always square. That is not a fussy restriction: a matrix that is to leave A unchanged from either side has to have the same number of rows as A has columns and the same number of columns as A has rows, and only a square identity can do both.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 210 — the definition of the identity matrix
Worked example
A definition that claims something should be tested rather than believed. Take an ordinary 2 by 2 matrix and multiply it by the identity, using the row-times-column rule from Lesson 3.6 exactly as you did there.
\[ A = \begin{bmatrix}6 & -1\\4 & 3\end{bmatrix}, \qquad I = \begin{bmatrix}1 & 0\\0 & 1\end{bmatrix} \]
Take row 1 of A across column 1 of I
Why: the column of I holds 1 and 0, so the first number of the row survives and the second is silenced.
\[ 6(1) + (-1)(0) = 6 \]
Figure (svg): Matrix A with 6 and minus 1 on its top row shaded, times the 2 by 2 identity with its first column shaded, and the product with the entry 6 ringed. Beside them the arithmetic reads 6 times 1 plus minus 1 times 0 equals 6.
That is the whole trick, drawn. A column of the identity is one 1 among zeros, so it reaches into the row and picks out a single number. Each column picks a different one, and the row is rebuilt in its original order.
Do the same for the other three entries
Why: row 1 against column 2, then row 2 against each column in turn.
\[ A I = \begin{bmatrix}6(1) + (-1)(0) & 6(0) + (-1)(1)\\4(1) + 3(0) & 4(0) + 3(1)\end{bmatrix} = \begin{bmatrix}6 & -1\\4 & 3\end{bmatrix} \]
Now multiply in the other order
Why: Lesson 3.6 warned that order changes a matrix product, so the second order has to be checked rather than assumed.
\[ I A = \begin{bmatrix}1(6) + 0(4) & 1(-1) + 0(3)\\0(6) + 1(4) & 0(-1) + 1(3)\end{bmatrix} = \begin{bmatrix}6 & -1\\4 & 3\end{bmatrix} \]
Check both products against A
Why: both came back as 6 and negative 1 on the top row and 4 and 3 on the bottom, which is A itself, so the identity leaves A alone from either side.
\[ A I = A = I A \]
The identity is one of the very few matrices that commutes with everything, and that is exactly why it is the right target. When you undo a multiplication you want to land on something that will then get out of the way.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 210 — A I equals A and I A equals A
Prediction
The identity is square, but the matrix it multiplies need not be. Think about the dimensions first, using the rule from Lesson 3.6 that the inner numbers must agree, and then about what the answer holds.
\[ I\,B = \begin{bmatrix}1 & 0\\0 & 1\end{bmatrix}\begin{bmatrix}5 & -2 & 7\\0 & 3 & 1\end{bmatrix} \]
Predict first
What is the product of the 2 by 2 identity with that 2 by 3 matrix?
Correct: The same 2 by 3 matrix, unchanged
Why: The inner dimensions are 2 and 2, so the product is defined and comes out 2 by 3. Each column of the answer is built from one column of B, and the row of the identity picks out one number from it, so every entry survives exactly where it was: 5 and negative 2 and 7 on the top row, 0 and 3 and 1 underneath.
The dimension part of that answer is worth saying out loud. A 2 by 2 times a 2 by 3 gives a 2 by 3, so the shape of the answer is fixed before any arithmetic happens, and it is the shape of the matrix that was not the identity.
This is why the size of the identity is decided by the company it keeps. Put the 3 by 3 identity on the right of that same B and it works too, because B has three columns. Put the wrong size on either side and there is no product at all.
From here on, the phrase leaves it unchanged is the thing to listen for. Every solve in this lesson ends by producing the identity next to the unknown, precisely so that the identity can be rubbed out and the unknown left standing alone.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.6, p. 195 — the dimension rule for a matrix product
Section
Part 2
Concept
Now the definition the lesson is named after. Two square matrices are inverses when multiplying them gives the identity, and the book insists on trying it both ways round.
inverse matrices — Two n by n matrices A and B are inverses of each other if A B equals I and B A equals I. The inverse of A is written with a raised negative one after it, and A has an inverse if and only if the determinant of A is not zero.
\[ A A^{-1} = I = A^{-1}A \]
Read that line from the middle outwards. Whichever side the inverse is written on, the product collapses to the matrix that changes nothing, which is what undoing has meant since the first slide of this lesson.
The second half of the definition is the part that will shape everything after Part 3. A matrix has an inverse exactly when its determinant is not zero, and a determinant of zero is the matrix version of trying to undo a multiplication by zero.
One piece of vocabulary care: a matrix with no inverse is not a broken matrix. It is a perfectly good matrix that happens to throw information away, in the same way that multiplying a number by zero is a perfectly good thing to do and simply cannot be reversed.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 210 — the definition of inverse matrices and the determinant condition
Worked example
Part 3 will show you how to find the inverse. Before that, take on trust that these two matrices are supposed to be inverses and test the claim, because testing is a skill you will use after every inverse you ever compute.
\[ A = \begin{bmatrix}3 & 8\\2 & 5\end{bmatrix}, \qquad B = \begin{bmatrix}-5 & 8\\2 & -3\end{bmatrix} \]
Multiply A by B, entry by entry
Why: row 1 of A is 3 and 8; take it across each column of B in turn.
\[ A B = \begin{bmatrix}3(-5) + 8(2) & 3(8) + 8(-3)\\2(-5) + 5(2) & 2(8) + 5(-3)\end{bmatrix} \]
Work out the four entries
Why: watch the cancellations: negative 15 and 16 leave 1, while 24 and negative 24 leave nothing at all.
\[ A B = \begin{bmatrix}1 & 0\\0 & 1\end{bmatrix} \]
Those cancellations are not luck. The off-diagonal entries are built to destroy each other, and the diagonal entries are built to leave the determinant, which the recipe in Part 3 has already divided out.
Watch where each entry of the answer comes from as you work. The 1 in the top left corner is the top row of the first matrix meeting the first column of the second, and the 0 beside it is that same row meeting the other column. Every entry of a product has an address like that, exactly as in Lesson 3.6.
Now multiply the other way round
Why: B first, then A, because the definition asks for both products and matrix multiplication is not commutative.
\[ B A = \begin{bmatrix}-5(3) + 8(2) & -5(8) + 8(5)\\2(3) + (-3)(2) & 2(8) + (-3)(5)\end{bmatrix} \]
Verify that the second product is also the identity
Why: negative 15 plus 16 is 1, negative 40 plus 40 is 0, 6 minus 6 is 0, and 16 minus 15 is 1, so both orders give the identity and the two matrices are inverses.
\[ B A = \begin{bmatrix}1 & 0\\0 & 1\end{bmatrix} \]
This is the check to run every single time. It costs eight small multiplications and it catches every sign slip and every swap made in the wrong direction, which between them are the two mistakes this topic generates.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 210 — Example 1 and the Check Inverses note
Picture it
Here is what the definition means when a matrix is actually put to work. Multiply the pair of numbers 1 and 1, written as a column, by the matrix A from the previous slide, and see where it lands.
\[ \begin{bmatrix}3 & 8\\2 & 5\end{bmatrix}\begin{bmatrix}1\\1\end{bmatrix} = \begin{bmatrix}11\\7\end{bmatrix} \]
Figure (svg): A coordinate plane with the point (1, 1) marked, an arrow taking it to (11, 7) labelled A, and a second arrow bringing it back to (1, 1) labelled the inverse of A.
Now send the answer back through the inverse. Three times negative 5 plus eight times 2 is the arithmetic that built the identity a moment ago, and the same cancellations bring the point home.
\[ \begin{bmatrix}-5 & 8\\2 & -3\end{bmatrix}\begin{bmatrix}11\\7\end{bmatrix} = \begin{bmatrix}-55 + 56\\22 - 21\end{bmatrix} = \begin{bmatrix}1\\1\end{bmatrix} \]
This is worth keeping as your mental picture of the word inverse. A matrix moves points about; its inverse is the instruction that brings every one of them back. When no such instruction exists, it is because the matrix has sent two different points to the same place and the return trip would not know which one to choose.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 210 — the inverse as the matrix undoing A
Notation
The notation looks like an exponent and is not one, which causes more trouble here than the arithmetic does. Take it apart before using it.
Annotate
On: \( A A^{-1} = I = A^{-1} A \)
One more habit of the notation: the inverse only ever belongs to a square matrix. A matrix with more columns than rows cannot have one, because no single matrix could give the identity in both orders.
When you write your own work, keep the raised negative one tight against the letter it belongs to. A stray one written on the line turns your inverse into a subtraction, and it is a surprisingly common way to lose a mark on an otherwise perfect answer.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 210 — the symbol for the inverse of A
Trap
Here is the instinct almost everybody has when they first meet a matrix equation. You want X on its own, and A is multiplying it, so you divide both sides by A.
\[ A X = B \;\Longrightarrow\; X = \dfrac{B}{A} \]
The line looks reasonable because it is exactly what you would write if A, X and B were numbers.
It cannot be evaluated, though. There is no operation that divides one matrix by another, so the right-hand side names nothing, and no arithmetic can be started from it.
Multiply instead, by the inverse, and keep track of the side you multiply on.
\[ A^{-1}(A X) = A^{-1}B \]
\[ X = A^{-1}B \]
The two lines are as short as the wrong one, so nothing has been lost. What has been gained is that every symbol on the right names a matrix you can actually build, given A and B.
It is also worth noticing why arithmetic can afford the shortcut and matrices cannot. With numbers, dividing by 3 and multiplying by one third are the same operation, so the fraction hides a multiplication. With matrices only the multiplication exists, so the fraction has nothing to hide.
Sorting
Six pairs, and the only way to be sure is to multiply. For each one, work out the product of the two matrices, and sort by whether it comes out as the identity or as something else. Any single entry that is wrong settles it.
Each matrix is written on one line here, with a semicolon separating the top row from the bottom one. So an item reading 3, 8 then a semicolon then 2, 5 is the matrix whose top row is 3 and 8 and whose bottom row is 2 and 5.
Sort into buckets
Multiply each pair and sort by whether the product is the identity.
The pair that multiplies to negative one on each diagonal entry is the one to study. Everything about it is right except a single sign, and it comes from a determinant of negative 1, which is easy to read as though it were 1.
The pair giving 6 down the diagonal is the same fault on a larger scale, and it is the error the textbook singles out in its own exercise set. You will meet it again in a few slides with the working shown.
One consolation as you do this: a wrong inverse always announces itself. The identity is such a distinctive answer that any slip at all produces something that obviously is not it, which is why the check is so much more useful here than in most topics.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 214 — Exercises 3 to 12
Socratic
Definitions usually contain no spare parts, so a condition stated twice deserves a question. Argue this one out rather than accepting it.
Discussion prompt
Lesson 3.6 showed that A B and B A are usually different matrices. Explain, using that fact, why the definition of inverses asks for both products to be the identity, and say what you would conclude if a classmate showed you one product only.
There is a second, quieter reason to do both. Writing out the two products forces you to see the inverse acting from each side, and that is exactly the distinction that decides how you solve a matrix equation later in this lesson.
Keep the phrase in both orders in your working vocabulary. When a question asks you to show that two matrices are inverses, it is asking for two products, and an answer with one of them is an answer with half the marks.
Notice how unusual this makes the inverse. Almost no pair of matrices you have met in this chapter gives the same answer in both orders, and the pairs that do are special: a matrix with the identity, a matrix with a multiple of the identity, and a matrix with its own inverse. That short list is worth remembering.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 210 — the Check Inverses note beside Example 1
Section
Part 3
Concept
The formula for the inverse of a 2 by 2 is usually handed over to be memorised. It is better earned, and earning it takes a single product you are already able to do.
Start with a general matrix, and try the obvious repair: swap the two entries on the main diagonal, and change the sign of the other two. Then multiply the two matrices together and see what turns up.
\[ \begin{bmatrix}a & b\\c & d\end{bmatrix}\begin{bmatrix}d & -b\\-c & a\end{bmatrix} = \begin{bmatrix}ad - bc & -ab + ba\\cd - dc & -cb + da\end{bmatrix} \]
Look at the two entries off the diagonal. Each one is a product minus itself, so both are zero, whatever the letters stand for. That is what the sign changes were for.
\[ = \begin{bmatrix}ad - cb & 0\\0 & ad - cb\end{bmatrix} = (ad - cb)\begin{bmatrix}1 & 0\\0 & 1\end{bmatrix} \]
So the swapped, negated matrix does not give the identity; it gives the determinant times the identity. One division fixes that, and it is the only thing left to do. The recipe is not a rule to trust, it is this line rearranged.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 210 — the Key Concept box, The Inverse of a 2 by 2 Matrix
Notation
Here is the book's version of the line you just derived. Four things are happening at once, so take them one at a time before using it on numbers.
Annotate
On: \( A^{-1} = \dfrac{1}{ad - cb}\begin{bmatrix}d & -b\\-c & a\end{bmatrix} \)
The book writes the determinant as ad minus cb, which is the same quantity as the more usual ad minus bc, because multiplication does not care about order. Use whichever form you remember, as long as the main diagonal comes first.
Figure (svg): Matrix A with entries a, b, c, d on the left, and on the right the same four letters rearranged: d and a have changed places along the main diagonal, b and c have picked up minus signs, and the whole matrix is divided by ad minus cb.
Say the recipe as three words while you carry it out: swap, negate, divide. Swapping and negating touch different pairs of entries, and mixing up which pair gets which treatment is the single most common error in this lesson.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 210 — the Key Concept box
Worked example
The book's first example, worked at the pace you should work yours. Notice how short it is once the recipe is in the right order.
\[ A = \begin{bmatrix}3 & 8\\2 & 5\end{bmatrix} \]
Find the determinant first
Why: if it turns out to be zero there is no inverse and no point in going on, so this is the move that decides whether the rest happens.
\[ \det A = 3(5) - 8(2) = 15 - 16 = -1 \]
A determinant of negative 1 is a gift: dividing by negative 1 is just changing every sign. Do not let that make you careless, though, because a sign change applied to only half the entries is exactly how this example goes wrong.
Swap the main diagonal and negate the other pair
Why: the 3 and the 5 change places, the 8 and the 2 stay where they are and change sign.
\[ \begin{bmatrix}5 & -8\\-2 & 3\end{bmatrix} \]
Divide every entry by the determinant
Why: one over negative 1 is negative 1, so this scalar multiplication flips all four signs.
\[ A^{-1} = \dfrac{1}{-1}\begin{bmatrix}5 & -8\\-2 & 3\end{bmatrix} = \begin{bmatrix}-5 & 8\\2 & -3\end{bmatrix} \]
Compare the first and last matrices for a moment. Nothing about the answer looks like the matrix it came from, which is why nobody can sensibly guess an inverse. The recipe is short enough to run and the check is short enough to trust, and between them they make guessing unnecessary.
Check by multiplying A by the answer
Why: 3 times negative 5 plus 8 times 2 is 1, and 3 times 8 plus 8 times negative 3 is 0, and the bottom row gives 0 and 1, so the product is the identity.
\[ \begin{bmatrix}3 & 8\\2 & 5\end{bmatrix}\begin{bmatrix}-5 & 8\\2 & -3\end{bmatrix} = \begin{bmatrix}1 & 0\\0 & 1\end{bmatrix} \]
That check is the same one you ran two slides ago, and it is the reason those two matrices were worth trusting then. Every inverse you find from here on gets this treatment before it is used for anything.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 210 — Example 1
Worked example
The first checkpoint after Example 1. This one has a determinant that is not 1 or negative 1, so the fractions stay in the answer, and that is perfectly normal rather than a sign of a mistake.
\[ A = \begin{bmatrix}6 & 1\\2 & 4\end{bmatrix} \]
Find the determinant
Why: main diagonal product first, then subtract the other diagonal product.
\[ \det A = 6(4) - 1(2) = 24 - 2 = 22 \]
Swap the main diagonal and negate the other pair
Why: the 6 and the 4 exchange places; the 1 and the 2 stay and take minus signs.
\[ \begin{bmatrix}4 & -1\\-2 & 6\end{bmatrix} \]
Divide by the determinant
Why: keep the one over 22 outside the brackets rather than turning four entries into four fractions; it is tidier and much easier to check.
\[ A^{-1} = \dfrac{1}{22}\begin{bmatrix}4 & -1\\-2 & 6\end{bmatrix} \]
If you do want the entries written out, they are two elevenths and negative one twenty-second on the top, negative one eleventh and three elevenths underneath. Both forms are correct, and the factored one is far kinder to whoever checks your work, including you.
Check by multiplying, keeping the fraction outside
Why: the four products inside come to 22, 0, 0 and 22, and one twenty-second of that is the identity.
\[ \dfrac{1}{22}\begin{bmatrix}6 & 1\\2 & 4\end{bmatrix}\begin{bmatrix}4 & -1\\-2 & 6\end{bmatrix} = \dfrac{1}{22}\begin{bmatrix}22 & 0\\0 & 22\end{bmatrix} = I \]
Notice what the check looks like when the fraction is kept outside: the matrix product comes out as the determinant times the identity, exactly as the derivation promised, and the fraction then cancels it.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 211 — Guided Practice 1
Fill the middle
Second checkpoint, and this time you supply the two pieces the recipe generates. Work out the determinant first, then decide what happens to the entry in the top right corner.
\[ A = \begin{bmatrix}-1 & 5\\-4 & 8\end{bmatrix} \]
Fill in the blanks
A^12 = \dfrac-5___}\begin___8 & ___\\4 & -1\end___
Why: The determinant is negative 1 times 8 minus 5 times negative 4, which is negative 8 plus 20, so 12. The entry in the top right is the old 5 with its sign changed, giving negative 5, and the bottom left is the old negative 4 with its sign changed, giving 4. Check it: the product of the original matrix with the bracket is 12 down the diagonal and 0 elsewhere, and one twelfth of that is the identity.
The bottom left entry is already filled in for you as 4, and it is worth asking where the sign went. The original entry there was negative 4, so changing its sign gives positive 4. Negating does not mean writing a minus sign, it means swapping the sign for the other one.
The two diagonal entries, 8 and negative 1, have simply traded places. If you find yourself wondering whether they also change sign, look back at the derivation: only the off-diagonal pair needed sign changes to make the off-diagonal products cancel.
A determinant of 12 leaves four fractions in the answer. Leave them as one twelfth outside the bracket unless a question demands otherwise, and you will have less arithmetic and fewer chances to slip.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 211 — Guided Practice 2
Worked example
The third checkpoint is the sign test. All four entries start negative, so every step involves a negative times a negative, and the answer is less obvious than it looks.
\[ A = \begin{bmatrix}-3 & -4\\-1 & -2\end{bmatrix} \]
Find the determinant
Why: negative 3 times negative 2 is positive 6, and negative 4 times negative 1 is positive 4, so the determinant is 6 minus 4.
\[ \det A = (-3)(-2) - (-4)(-1) = 6 - 4 = 2 \]
A positive determinant from an all-negative matrix surprises people every year. Both diagonal products are positive here, and the subtraction between them is what leaves a small positive number.
Swap the main diagonal, negate the other pair
Why: negative 3 and negative 2 change places; negative 4 and negative 1 stay put and become positive 4 and positive 1.
\[ \begin{bmatrix}-2 & 4\\1 & -3\end{bmatrix} \]
Divide by the determinant
Why: one half multiplies all four entries.
\[ A^{-1} = \dfrac{1}{2}\begin{bmatrix}-2 & 4\\1 & -3\end{bmatrix} = \begin{bmatrix}-1 & 2\\\tfrac12 & -\tfrac32\end{bmatrix} \]
Both forms of that answer are shown so you can compare them. Halving a whole number sometimes gives a whole number and sometimes a fraction, so an answer with a mixture of the two is not a sign that something has gone wrong.
Verify by multiplying the original by the answer
Why: the top row gives negative 3 times negative 2 plus negative 4 times 1, which is 2, and negative 3 times 4 plus negative 4 times negative 3, which is 0; the bottom row gives 0 and 2, and one half of that is the identity.
\[ \dfrac{1}{2}\begin{bmatrix}-3 & -4\\-1 & -2\end{bmatrix}\begin{bmatrix}-2 & 4\\1 & -3\end{bmatrix} = \dfrac{1}{2}\begin{bmatrix}2 & 0\\0 & 2\end{bmatrix} = I \]
Two of the entries in the answer are negative and two are positive, which is not a pattern you could have guessed from the all-negative start. Guessing is exactly what the check is there to protect you from.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 211 — Guided Practice 3
Error analysis
This is the textbook's own error-analysis exercise, and the mistake in it is the most popular one in the topic. Everything except the last move is right, which is what makes it worth studying.
Annotate
On: \( \begin{bmatrix}2 & 4\\1 & 5\end{bmatrix}^{-1} = 6\begin{bmatrix}5 & -4\\-1 & 2\end{bmatrix} \)
\[ \begin{bmatrix}2 & 4\\1 & 5\end{bmatrix}\begin{bmatrix}30 & -24\\-6 & 12\end{bmatrix} = \begin{bmatrix}36 & 0\\0 & 36\end{bmatrix} \]
The line above is the check catching the error. The wrong answer produces 36 down the diagonal rather than 1, and 36 is 6 times 6: one factor of 6 from multiplying instead of dividing, and one more from failing to divide at all.
\[ A^{-1} = \dfrac{1}{6}\begin{bmatrix}5 & -4\\-1 & 2\end{bmatrix} \]
Why is the mistake so tempting? Because the determinant is the number you just worked out, and it is sitting in your head asking to be used. The formula wants its reciprocal, and the moment to remember that is when you write the fraction bar, not afterwards.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 214 — Exercise 11
Trap
Finding the inverse of the matrix with rows 3, 8 and 2, 5, here is a version that gets the two halves of the recipe the wrong way round.
\[ A = \begin{bmatrix}3 & 8\\2 & 5\end{bmatrix} \;\Longrightarrow\; \begin{bmatrix}-3 & 8\\2 & -5\end{bmatrix} \]
The diagonal entries have been negated and the other two left alone, which is the mirror image of what the formula asks.
It is an easy slip to make, because the words swap and negate are both being applied to a diagonal, and it is not obvious from the formula alone which pair gets which.
Multiply it out and the error announces itself at once.
\[ \begin{bmatrix}3 & 8\\2 & 5\end{bmatrix}\begin{bmatrix}-3 & 8\\2 & -5\end{bmatrix} = \begin{bmatrix}7 & -16\\4 & -9\end{bmatrix} \]
Nothing about that is the identity, so the proposed inverse is not one.
\[ A^{-1} = \begin{bmatrix}-5 & 8\\2 & -3\end{bmatrix} \]
The derivation is what settles which pair is which. The sign changes exist to make the off-diagonal products cancel, and only the off-diagonal entries appear in those products, so only they change sign.
Elimination
A multiple-choice question from the book, and a good one, because three of its four answers are mistakes people actually make. Work out the determinant first, then eliminate rather than compute all four.
\[ A = \begin{bmatrix}10 & -3\\3 & -1\end{bmatrix} \]
Eliminate the wrong options
Which matrix is the inverse of the one above?
Survives elimination: c
Why: The determinant is 10 times negative 1 minus negative 3 times 3, which is negative 10 plus 9, so negative 1. Swapping the diagonal gives negative 1 and 10, negating the other pair gives 3 and negative 3, and dividing by negative 1 flips every sign, leaving 1 and negative 3 on the top row with 3 and negative 10 underneath. Multiply it by A to confirm: 10 minus 9 is 1 on the diagonal, and the off-diagonal entries cancel to 0.
The surviving matrix and the undivided one that keeps 10 in its bottom right corner differ only by a sign on every entry, and that is the determinant of negative 1 doing its job. Whenever the determinant is negative, expect the final division to change every sign in the bracket.
The candidate that simply repeats the original matrix is worth a thought rather than a laugh. Some matrices genuinely are their own inverses, the identity among them, and the only way to know is to multiply and look for the identity.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 214 — Exercise 12
Concept
Everything so far has assumed the determinant was some number you could divide by. Now look at what the formula says when it is zero.
\[ A = \begin{bmatrix}2 & 3\\4 & 6\end{bmatrix}, \qquad \det A = 2(6) - 3(4) = 12 - 12 = 0 \]
The recipe asks for one over the determinant, and one over zero does not name a number. So the process does not produce a wrong inverse here; it produces nothing at all, and the book's statement that a matrix has an inverse if and only if its determinant is not zero is exactly this observation.
Compare it with the first slide of the lesson. Multiplying a number by zero destroys the number, so the trip back does not exist. A matrix with determinant zero destroys information in the same way, and its determinant is how you detect that before wasting any effort.
Look at the entries of this particular matrix and you can see the damage. The second row is exactly twice the first, so the matrix carries only one genuine piece of information dressed up as two.
\[ \begin{bmatrix}4 & 6\end{bmatrix} = 2\begin{bmatrix}2 & 3\end{bmatrix} \]
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 210 — a matrix has an inverse if and only if its determinant is not zero
Counterexample
Saying the formula breaks down is not quite the same as saying no inverse exists. Perhaps some other matrix, found some other way, would do the job. Settle it with the definition rather than the recipe.
Discussion prompt
Suppose some matrix B satisfied the equation given below, with the matrix whose rows are 2, 3 and 4, 6. Multiply out the top left and bottom left entries of that product in terms of the entries of B, and explain why the two answers cannot both be right.
Figure (svg): Two parallel lines on a coordinate plane, one through 2 on the vertical axis and one through 4, both falling at the same rate, with the note that they never meet.
The picture is the same fact drawn as lines. Any system whose coefficient matrix is this one has two equations that describe lines of identical steepness, so they either never meet or lie on top of each other, and there is no single crossing to find.
That is the bridge back to Lesson 3.1, where you sorted systems into one solution, no solution and infinitely many. The determinant of the coefficient matrix is the arithmetic that tells the first case from the other two, without a graph.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 214 — Exercise 41, write a 2 by 2 matrix that has no inverse
Tweak it
Zero is a hard boundary, but the approach to it is worth seeing. The matrix below has 4 in its top right corner and 3 in its bottom right, the slider sets its bottom left entry, and the horizontal axis of the graph is the entry in its top left corner.
What is plotted is the top left entry of the inverse. Watch what that entry does as the top left entry of the matrix creeps towards the value that makes the determinant zero.
Parameter explorer
Slide the bottom-left entry, then read off what happens to the plotted entry of the inverse as the top-left entry approaches the value that makes the determinant zero.
\[ \dfrac{3}{3x - 4({k})} \]
The gap in the curve is not a drawing error. At the value where the determinant is zero there is no inverse, so there is no entry to plot, and the curve has to break.
Moving the slider moves the gap, because the determinant depends on all four entries. That is the point of the widget: being close to having no inverse is not a property of one number in the matrix, it is a property of the combination the determinant measures.
There is a practical warning in the steepness. A matrix whose determinant is small but not zero does have an inverse, and that inverse has enormous entries, so small errors in the original numbers become large errors in the answer. Later courses spend whole chapters on that fact.
Comparison
Five matrices. Work out each determinant, then say whether the matrix has an inverse. The rule is the whole content of this part, and it takes one subtraction to apply.
Comparison matrix
| matrix | determinant | has an inverse? |
|---|---|---|
| [3, 8; 2, 5] | -1 | yes |
| [6, 1; 2, 4] | 22 | yes |
| [2, 3; 4, 6] | 0 | no |
| [−24, 60; −6, 30] | -360 | yes |
| [1, 2; 2, 4] | 0 | no |
The two rows with a determinant of zero are worth comparing with each other. In both of them the second row is a multiple of the first, and that is always what a zero determinant means for a 2 by 2 matrix: one row is the other one rescaled.
The row with a determinant of negative 360 is a reminder that large numbers are not the difficulty here. Its entries are the biggest on the table and it has a perfectly good inverse, whose entries are correspondingly tiny.
Notice also that nothing in the verdict column needed the inverse itself. You can answer the question does this have an inverse with one subtraction, long before doing any of the work of finding it, and on a long exercise set that is worth real time.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 214 — Exercises 3 to 10 and Exercise 41
Matching
Four matrices from the exercise set, four inverses, shuffled. You can do this without finishing any of the calculations: the determinant tells you what kind of numbers to expect, and one entry settles each pairing.
Match the pairs
Why: The determinants are negative 1, 1, negative 3 and negative 4 in that order. A determinant of 1 needs no division, so that matrix pairs with the swapped and negated version of itself with no fractions in it. A determinant of negative 1 flips every sign, giving the all-negative partner. The determinant of negative 3 is the only one producing thirds, and the determinant of negative 4 is the only one producing quarters, so the two fractional answers sort themselves.
Reading the denominators is the fastest route, and it is a genuine strategy rather than a trick for this question. The determinant is the only thing the recipe divides by, so it decides what denominators can appear in the answer.
The pairing with no fractions at all is worth noticing. A determinant of 1 means the third step of the recipe changes nothing, so swap and negate is the entire calculation, which is why examples are so often built with a determinant of 1.
If two of these had shared a determinant you would have had to look at an entry instead. The top left entry of any inverse is the old bottom right entry divided by the determinant, and that single number separates almost any pair.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 214 — Exercises 3, 4, 6 and 7
Section
Part 4
Concept
Here is the shape of problem the inverse was built for. A known matrix multiplies an unknown matrix, and you want the unknown on its own.
\[ A X = B \]
With numbers you would divide by A. With matrices there is no division, so you multiply by the inverse instead, and now the side you multiply on genuinely matters.
Figure (svg): Two rows. The top row shows the inverse of A written immediately to the left of A in the product, with the two cancelling to the identity and a green tick. The bottom row shows the inverse written on the right of the product, where A and X sit between it and A, with a red cross.
In the product A X, the matrix A sits immediately to the left of X. To cancel it you must place the inverse immediately to the left of A, so that the two of them meet with nothing in between. Written on the right, the inverse is separated from A by the whole of X, and there is no rule that lets you slide past it.
So the instruction is: multiply each side of the equation on the left by the inverse of A. Both sides, because it is an equation, and on the left on both sides, because otherwise you are performing two different operations and the equality is lost.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 211 — multiply both sides of the equation by the inverse on the left
Worked example
The book's second example. X here is not a pair of numbers but a whole 2 by 2 matrix, which changes nothing about the method and is worth seeing once so that the method does not feel tied to systems.
\[ \begin{bmatrix}2 & -7\\-1 & 4\end{bmatrix}X = \begin{bmatrix}-21 & 3\\12 & -2\end{bmatrix} \]
Find the determinant of the matrix on the left
Why: 8 minus 7, so 1, which is the friendliest determinant there is: the division at the end will do nothing.
\[ \det A = 2(4) - (-7)(-1) = 8 - 7 = 1 \]
Build the inverse
Why: swap the 2 and the 4, change the signs of negative 7 and negative 1, and divide by 1.
\[ A^{-1} = \dfrac{1}{1}\begin{bmatrix}4 & 7\\1 & 2\end{bmatrix} = \begin{bmatrix}4 & 7\\1 & 2\end{bmatrix} \]
Both off-diagonal entries started negative, so negating them made them positive. That is the step to slow down on when the original matrix has minus signs already in it.
Multiply both sides on the left by the inverse
Why: the left-hand side becomes the inverse times A times X, and the first two collapse to the identity.
\[ \begin{bmatrix}4 & 7\\1 & 2\end{bmatrix}\begin{bmatrix}2 & -7\\-1 & 4\end{bmatrix}X = \begin{bmatrix}4 & 7\\1 & 2\end{bmatrix}\begin{bmatrix}-21 & 3\\12 & -2\end{bmatrix} \]
Simplify each side
Why: the identity on the left leaves X alone; the right-hand side is one ordinary matrix product, four entries of row-times-column work.
\[ I X = \begin{bmatrix}4(-21) + 7(12) & 4(3) + 7(-2)\\1(-21) + 2(12) & 1(3) + 2(-2)\end{bmatrix} = \begin{bmatrix}0 & -2\\3 & -1\end{bmatrix} \]
Look at what happened to the unknown. It was never taken apart, never split into four separate numbers and never solved for one entry at a time; the whole 2 by 2 matrix came out of a single product. That is what makes this method worth learning rather than treating each entry as its own little equation.
Check by putting the answer back into the original equation
Why: A times the answer has to give B: 2 times 0 minus 7 times 3 is negative 21, 2 times negative 2 minus 7 times negative 1 is 3, and the bottom row gives 12 and negative 2.
\[ \begin{bmatrix}2 & -7\\-1 & 4\end{bmatrix}\begin{bmatrix}0 & -2\\3 & -1\end{bmatrix} = \begin{bmatrix}-21 & 3\\12 & -2\end{bmatrix} \]
That is B exactly, so X is right. Checking by substituting into the original equation is better than checking your inverse alone, because it tests the multiplication as well.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 211 — Example 2
Trap
Same equation, and an answer built by putting the inverse on the right of both sides.
\[ A X A^{-1} = B A^{-1} \]
The left-hand side has not simplified at all. The inverse and A are separated by X, and there is no rule permitting you to move X out of the way.
Worse, the right-hand side has genuinely changed, so you now have a true but useless equation, in which the unknown is buried in the middle of a product.
Put the inverse on the left of both sides instead.
\[ A^{-1}A X = A^{-1}B \]
\[ X = A^{-1}B \]
Now the two matrices that cancel are next to each other, the identity appears, and it steps aside to leave X.
The arithmetic difference is real, not cosmetic. In Example 2 the correct product gives the top row 0 and negative 2, while multiplying on the right gives an entirely different matrix, and only one of them satisfies the original equation.
Worked example
The checkpoint after Example 2. The method does not change, but the determinant is negative 24 this time, so fractions appear and the arithmetic needs more care.
\[ \begin{bmatrix}-4 & 1\\0 & 6\end{bmatrix}X = \begin{bmatrix}8 & 9\\24 & 6\end{bmatrix} \]
Find the determinant
Why: negative 4 times 6 is negative 24, and 1 times 0 is nothing to subtract.
\[ \det A = (-4)(6) - (1)(0) = -24 \]
Build the inverse
Why: swap negative 4 and 6, negate the 1 and the 0, then divide by negative 24.
\[ A^{-1} = \dfrac{1}{-24}\begin{bmatrix}6 & -1\\0 & -4\end{bmatrix} \]
A zero in the matrix stays a zero when negated, which feels like a non-event and is worth a moment anyway. It is the reason this inverse has a zero in the same corner as the original.
Multiply the constant matrix on the left by the inverse
Why: keep the fraction outside and multiply the two brackets first.
\[ X = \dfrac{1}{-24}\begin{bmatrix}6 & -1\\0 & -4\end{bmatrix}\begin{bmatrix}8 & 9\\24 & 6\end{bmatrix} = \dfrac{1}{-24}\begin{bmatrix}24 & 48\\-96 & -24\end{bmatrix} \]
Divide every entry by negative 24
Why: 24 over negative 24 is negative 1, 48 over negative 24 is negative 2, and so on.
\[ X = \begin{bmatrix}-1 & -2\\4 & 1\end{bmatrix} \]
Every one of those four divisions came out whole, which is the usual sign in a textbook question that the earlier steps were right. If three of your entries divide neatly and one does not, suspect the entry rather than the determinant, and recheck that single row-times-column calculation.
Check in the original equation
Why: negative 4 times negative 1 plus 1 times 4 is 8, negative 4 times negative 2 plus 1 times 1 is 9, and the bottom row gives 6 times 4, which is 24, and 6 times 1, which is 6.
\[ \begin{bmatrix}-4 & 1\\0 & 6\end{bmatrix}\begin{bmatrix}-1 & -2\\4 & 1\end{bmatrix} = \begin{bmatrix}8 & 9\\24 & 6\end{bmatrix} \]
Keeping the fraction outside until the very end paid off here. Every number inside the bracket was a whole number, and the only division happened once, at the last line.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 211 — Guided Practice 4
Faded example
Here the determinant is 1 again, so the inverse costs nothing but a swap and two sign changes. The first two lines are done for you; supply the two missing entries of the product.
\[ \begin{bmatrix}1 & 1\\4 & 5\end{bmatrix}X = \begin{bmatrix}2 & 3\\-1 & 6\end{bmatrix}, \qquad A^{-1} = \begin{bmatrix}5 & -1\\-4 & 1\end{bmatrix} \]
Fill in the blanks
X = \begin115 & -1\\-4 & 1\end-6\begin___2 & 3\\-1 & 6\end___ = \begin______ & 9\\-9 & ___\end___
Why: The top left entry is 5 times 2 plus negative 1 times negative 1, which is 10 plus 1, so 11. The bottom right is negative 4 times 3 plus 1 times 6, which is negative 12 plus 6, so negative 6. Check the whole thing in the original equation: the first matrix times the answer gives 11 minus 9, which is 2, and 9 plus 6, which is 3, on the top row, and 44 minus 45, which is negative 1, and 36 plus 30, which is 6, underneath.
The determinant here is 1 times 5 minus 1 times 4, which is 1, which is why the given inverse has no fraction in front of it. Confirm that for yourself before using it; an inverse handed to you is exactly the kind of thing worth two seconds of suspicion.
Notice that the unknown X came out with no fractions either. That is not a coincidence: when the determinant is 1 and everything else is whole, nothing in the calculation can introduce a fraction.
If you want the full check, multiply the original matrix by your answer and look for the constant matrix. It is the same two-line habit as every other example in this lesson, and it costs less than redoing the question.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 214 — Exercise 13
Section
Part 5
Concept
Now the payoff. A linear system is nothing but coefficients, letters and constants, and those three things can be separated into three matrices whose product rebuilds the system exactly.
Figure (svg): The system minus x plus two y equals 4 and three x plus four y equals minus 2 on the left, and on the right the same numbers as a 2 by 2 coefficient matrix times a 2 by 1 matrix of variables, equal to a 2 by 1 matrix of constants, with the dimensions written underneath.
Check that the picture really does rebuild the system before accepting it. The top row of the coefficient matrix is negative 1 and 2; taken across the column holding x and y it gives negative x plus two y, and the top entry of the constant matrix is 4. That is the first equation, entry for entry.
\[ A X = B \]
Three letters, and each one names a whole matrix now. A holds the coefficients in the order the equations are written, X holds the variables in a single column, and B holds the numbers from the right-hand sides.
The dimensions are what make it work. A 2 by 2 times a 2 by 1 gives a 2 by 1, which is the shape of B, so the equation is consistent as well as convenient.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 212 — Step 1 of the Key Concept box
Notation
The book gives each of the three a name, and questions use those names, so learn them here rather than guessing later.
Annotate
On: \( \begin{bmatrix}a & b\\c & d\end{bmatrix}\begin{bmatrix}x\\y\end{bmatrix} = \begin{bmatrix}e\\f\end{bmatrix} \)
The book's own vocabulary exercise asks you to point at the matrix of variables and the matrix of constants in an equation like this one, so the names are examinable in their own right.
One more habit worth building: write the variables down the column in alphabetical order and keep that order in every row of A. Most wrong answers in this part are a right calculation on a coefficient matrix whose columns were filled in inconsistently.
There is a reason the constants sit in a column rather than a row. The product of a 2 by 2 matrix with a 2 by 1 column is a 2 by 1 column, and a row would make the dimensions disagree, so the shape of B is forced by the shape of the answer you want.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 214 — Exercise 1, identify the matrix of variables and the matrix of constants
Step zero
A system does not always arrive in the shape the matrix needs. Deciding what to do about that is a step before the method starts, and skipping it is a reliable way to get a clean wrong answer.
\[ y = 2x - 7 \qquad \text{and} \qquad 3x = 16 - 4y \]
Discussion prompt
Rewrite both equations so that the x term and the y term are on the left in that order and the constant is on the right. Then write the coefficient matrix and the matrix of constants, and say what would go wrong if you built the matrix from the equations as they stand.
Two details to watch as you rearrange. A term that moves across the equals sign changes sign, and a variable with no visible number in front of it has a coefficient of 1, or negative 1 when a minus sign is attached.
The habit is cheap and it protects everything downstream. Once the coefficient matrix is wrong, the determinant is wrong, the inverse is wrong, and the answer is wrong, with no stage at which the arithmetic itself will complain.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 212 — the system written in standard form before Step 1
Concept
The book puts the whole method in a Key Concept box, and it really is only three steps once the pieces of this lesson are in place.
Figure (svg): Four stacked boxes. The first reads A X equals B, the second shows the inverse of A written in front of both sides, the third shows the left side collapsing to I X, and the fourth reads X equals the inverse of A times B. A reason is written beside each arrow.
The middle arrow of the diagram is the only place any thought is required. Everything above it is copying numbers into boxes, and everything below it is one matrix product, so the mathematics of the lesson is concentrated in a single move.
\[ X = A^{-1}B \]
Notice what this method gives you that substitution and elimination did not. The inverse depends only on the coefficients, so once you have it you can solve the same system with any right-hand side at all for the cost of one multiplication, which is exactly what the gift-basket problem later in this deck needs.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 212 — the Key Concept box, Using an Inverse Matrix to Solve a Linear System
Worked example
The book's own worked system, done in full. Follow the three steps in order and notice how little of it is thinking and how much is copying.
\[ 2x - 3y = 19 \qquad \text{and} \qquad x + 4y = -7 \]
Write the system as a matrix equation
Why: coefficients into A in the order the variables appear, unknowns into a column, right-hand sides into another column.
\[ \begin{bmatrix}2 & -3\\1 & 4\end{bmatrix}\begin{bmatrix}x\\y\end{bmatrix} = \begin{bmatrix}19\\-7\end{bmatrix} \]
Find the determinant of the coefficient matrix
Why: 2 times 4 is 8, and subtracting negative 3 times 1 adds 3, so it is not zero and the method will work.
\[ \det A = 2(4) - (-3)(1) = 8 + 3 = 11 \]
Build the inverse
Why: swap the 2 and the 4, change the signs of negative 3 and 1, and divide by 11.
\[ A^{-1} = \dfrac{1}{11}\begin{bmatrix}4 & 3\\-1 & 2\end{bmatrix} \]
Multiply the matrix of constants on the left
Why: one row-times-column calculation for each unknown, with the fraction kept outside until the end.
\[ X = \dfrac{1}{11}\begin{bmatrix}4 & 3\\-1 & 2\end{bmatrix}\begin{bmatrix}19\\-7\end{bmatrix} = \dfrac{1}{11}\begin{bmatrix}76 - 21\\-19 - 14\end{bmatrix} = \dfrac{1}{11}\begin{bmatrix}55\\-33\end{bmatrix} \]
Read the answer out of the column
Why: 55 over 11 is 5 and negative 33 over 11 is negative 3, and the column holds x above y.
\[ \begin{bmatrix}x\\y\end{bmatrix} = \begin{bmatrix}5\\-3\end{bmatrix} \]
Figure (svg): Two lines on a coordinate plane crossing at the point 5, minus 3, which is marked with a dot and dashed guide lines down to the axes.
The crossing in the picture is the answer, and it is the same answer graphing would have given you in Lesson 3.1 with far more drawing and far less certainty. The matrix method never plots a point; it computes one.
It is worth noticing where the two given numbers, 19 and negative 7, actually entered the calculation. They appeared only at the very last multiplication: the determinant and the inverse were built from the coefficients alone. Change the right-hand sides and everything above that final line still stands.
Check both original equations
Why: twice 5 is 10, and subtracting three lots of negative 3 adds 9, giving 19; and 5 plus four lots of negative 3 is 5 minus 12, which is negative 7.
\[ 2(5) - 3(-3) = 10 + 9 = 19 \qquad 5 + 4(-3) = 5 - 12 = -7 \]
Both equations, not just the first. A wrong answer will often satisfy one of them by accident, and the second equation is what catches it.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 212 — Example 4
Worked example
The checkpoint straight after Example 4. Work it yourself first and then read on, because the value of this one is entirely in whether you can run the method without being led.
\[ 4x + y = 10 \qquad \text{and} \qquad 3x + 5y = -1 \]
Write the matrix equation
Why: the coefficients of x go in the first column, the coefficients of y in the second.
\[ \begin{bmatrix}4 & 1\\3 & 5\end{bmatrix}\begin{bmatrix}x\\y\end{bmatrix} = \begin{bmatrix}10\\-1\end{bmatrix} \]
Find the determinant
Why: 20 minus 3, which is 17: not zero, so there is exactly one solution to find.
\[ \det A = 4(5) - 1(3) = 20 - 3 = 17 \]
Build the inverse
Why: swap the 4 and the 5, negate the 1 and the 3, divide by 17.
\[ A^{-1} = \dfrac{1}{17}\begin{bmatrix}5 & -1\\-3 & 4\end{bmatrix} \]
Multiply the constants by the inverse on the left
Why: 5 times 10 plus negative 1 times negative 1 on the top, and negative 3 times 10 plus 4 times negative 1 underneath.
\[ X = \dfrac{1}{17}\begin{bmatrix}50 + 1\\-30 - 4\end{bmatrix} = \dfrac{1}{17}\begin{bmatrix}51\\-34\end{bmatrix} = \begin{bmatrix}3\\-2\end{bmatrix} \]
The moment when 51 and negative 34 both turn out to be multiples of 17 is the usual sign that the work is right. Textbook systems are built to come out whole, so an answer that does not divide neatly is worth rechecking before you write it down.
Check in both equations
Why: four threes plus negative 2 is 12 minus 2, which is 10; and three threes plus five lots of negative 2 is 9 minus 10, which is negative 1.
\[ 4(3) + (-2) = 10 \qquad 3(3) + 5(-2) = -1 \]
Both hold, so the solution is the pair 3 and negative 2. Writing the answer as a coordinate pair is fine, as long as you keep x first, which is the order the column gave you.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 213 — Guided Practice 8
Worked example
From the exercise set, and pleasant to work because every coefficient is positive. The answer is not, which is a useful reminder that the signs in the question tell you very little about the signs in the solution.
\[ 5x + 7y = 20 \qquad \text{and} \qquad 3x + 5y = 16 \]
Write the matrix equation
Why: coefficients, variables, constants, in that order.
\[ \begin{bmatrix}5 & 7\\3 & 5\end{bmatrix}\begin{bmatrix}x\\y\end{bmatrix} = \begin{bmatrix}20\\16\end{bmatrix} \]
Find the determinant
Why: 25 minus 21, which is 4.
\[ \det A = 5(5) - 7(3) = 25 - 21 = 4 \]
Build the inverse
Why: swap the two 5s, which changes nothing visible, negate the 7 and the 3, and divide by 4.
\[ A^{-1} = \dfrac{1}{4}\begin{bmatrix}5 & -7\\-3 & 5\end{bmatrix} \]
The two 5s swapping places and looking identical afterwards catches people out. Carry out the swap anyway, as a move; if you start deciding which swaps are worth doing, one day you will skip one that matters.
Multiply by the matrix of constants
Why: 100 minus 112 on the top, negative 60 plus 80 underneath.
\[ X = \dfrac{1}{4}\begin{bmatrix}5(20) - 7(16)\\-3(20) + 5(16)\end{bmatrix} = \dfrac{1}{4}\begin{bmatrix}-12\\20\end{bmatrix} = \begin{bmatrix}-3\\5\end{bmatrix} \]
The minus signs in this calculation all came from the inverse rather than from the question. That is where negative answers to friendly-looking systems are born, and it is another reason to build the inverse carefully before rushing into the multiplication.
Check both equations
Why: five lots of negative 3 is negative 15, and seven fives is 35, which leaves 20; three lots of negative 3 is negative 9, and five fives is 25, which leaves 16.
\[ 5(-3) + 7(5) = 20 \qquad 3(-3) + 5(5) = 16 \]
A positive system with a negative solution is completely ordinary. The coefficients describe the directions of the two lines; where they happen to cross is a separate question.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 215 — Exercise 33
Prediction
Before working it out, look hard at the two equations. Something about the second one should feel familiar from Part 3, and from Lesson 3.1 before that.
\[ 2x - y = -6 \qquad \text{and} \qquad 6x - 3y = -18 \]
Predict first
What happens when you try to solve this system with an inverse matrix?
Correct: The determinant is zero, so there is no inverse and the method stops
Why: The determinant is 2 times negative 3 minus negative 1 times 6, which is negative 6 plus 6, so zero. Every entry of the second equation is three times the matching entry of the first, so the two equations say the same thing twice and the coefficient matrix has one row that is a multiple of the other. With no inverse there is no single answer to compute.
Notice how quickly the determinant told you. One subtraction, done before any inverse was attempted, ruled out the entire method, which is exactly why the book puts finding the determinant at the front of the recipe.
The book's own note beside Example 4 says the same thing in words: this method works when A has an inverse, and when it does not the system has either no solution or infinitely many. Which of the two it is takes a second look at the constants.
Here the constants agree with the coefficients: negative 18 really is three times negative 6. So the two equations are the same equation, and every point on that line solves the system.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 213 — Guided Practice 9 and the Solve Systems note on page 212
Intuition
This is the thread that ties the chapter together, so it is worth stating plainly. In Lesson 3.1 you sorted every system into three cases by drawing it. The determinant sorts them by arithmetic instead.
Figure (svg): A single line drawn twice on a coordinate plane, once thick in blue and once dashed in orange on top of it, showing that the two equations of the system describe the same line.
The picture above is the checkpoint you just predicted. Both equations draw the same line, so there is no single crossing and the inverse method has nothing to hand back: every point on the line is a solution, and a matrix product can only ever produce one answer.
| determinant of A | what the lines do | what the system has |
|---|---|---|
| not zero | cross at exactly one point | one solution, which the inverse finds |
| zero, constants matching | lie on top of each other | infinitely many solutions |
| zero, constants not matching | run parallel, never meeting | no solution |
So a determinant of zero never means you have made a mistake. It means the question was one of the other two kinds, and the tools for those are the ones from Lessons 3.1 and 3.2, where you compared slopes and intercepts.
It also explains a detail of Lesson 3.7 you may have wondered about. Cramer's rule puts the determinant underneath both fractions, so it fails in exactly the same cases, for exactly the same reason: the same zero, in the same place.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 212 — the Solve Systems note
Comparison
Fill in the determinant of each coefficient matrix and then the verdict. Work strictly left to right, because the determinant is what you are allowed to use to decide the last column.
Comparison matrix
| system | coefficient matrix | determinant | what it has |
|---|---|---|---|
| 3x − y = −5 and −4x + 2y = 8 | [3, −1; −4, 2] | 2 | one solution |
| 2x − y = −6 and 6x − 3y = −18 | [2, −1; 6, −3] | 0 | infinitely many |
| x + 2y = 1 and 2x + 4y = 9 | [1, 2; 2, 4] | 0 | none |
The first row is the one to do properly, because it is also Guided Practice 10. Its determinant is 6 minus 4, so 2, and running the method through gives x equal to negative 1 and y equal to 2, which you can check in both equations in a few seconds.
The last two rows have the same determinant and different verdicts, which is the whole point of putting them side by side. A zero determinant tells you there is no single answer; it does not tell you which of the other two cases you are in.
To tell those two apart, compare the constants with the coefficients. In the middle row, multiplying the first equation by 3 reproduces the second one exactly, constants included. In the last row, multiplying by 2 gives 2 on the right, not 9, so the two equations contradict each other and nothing can satisfy both.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 213 — Guided Practice 9 and 10
Ranking
Five moves, deliberately scrambled. Two of them can be done in either order in principle, but only one order is sensible, and saying why is the point of the exercise.
Put in order
Put the five moves of the inverse-matrix method into the order you would carry them out.
Why: Standard form comes first, because the coefficient matrix cannot be written correctly from equations that are not arranged. The three matrices come next. The determinant is then worked out before any inverse is attempted, since a determinant of zero means there is nothing to build. Only then is the inverse worth the effort, and the substitution at the end is what turns a plausible answer into a confirmed one.
The determinant sitting before the inverse rather than inside it is the ordering worth defending. Finding the determinant is one subtraction; finding the inverse is a swap, two sign changes and four divisions. Doing the cheap test first is what stops you from doing the expensive work on a system that has no answer.
The last move is the one most often dropped under time pressure. It costs two substitutions and it is the only step that can tell you whether the previous four were done correctly, which makes it the best value move in the list.
If you found yourself wanting to put the check second, that instinct is not silly. You can indeed check the inverse as soon as you have it, by multiplying it by A and looking for the identity, and the strongest answers do both checks.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 212 — the three steps of the Key Concept box
Section
Part 6
Concept
Everything so far has been 2 by 2, and there is a reason. The swap-negate-divide recipe only works at that size, and the inverse of a 3 by 3 matrix is, in the book's own words, difficult to compute by hand.
So for three equations in three unknowns you enter the coefficient matrix into a graphing calculator and ask it for the inverse. That is not a retreat from the mathematics: the definition, the determinant condition and the three-step method are all unchanged, and only the arithmetic has been handed over.
\[ A^{-1}A = I = A A^{-1} \]
What does not get handed over is the checking. A calculator will happily return an inverse built from a matrix you typed in wrongly, and the only thing standing between that and a wrong answer is you multiplying the two matrices together and looking for the identity.
The book asks for exactly that after every calculator inverse in the exercise set, and the next slide does it in full so you can see how much work it really is.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 211 — Inverse of a 3 by 3 matrix
Worked example
The book enters this matrix, reads off the inverse, and then uses the calculator to confirm it. Confirm it yourself instead, because doing so once tells you what the calculator is actually claiming.
\[ A = \begin{bmatrix}2 & 1 & -2\\5 & 3 & 0\\4 & 3 & 8\end{bmatrix}, \qquad A^{-1} = \begin{bmatrix}12 & -7 & 3\\-20 & 12 & -5\\1.5 & -1 & 0.5\end{bmatrix} \]
The halves in the bottom row are worth explaining before you start. The determinant of this matrix is 2, so every entry of the inverse is something divided by 2, and an odd numerator leaves a half behind.
Take row 1 of A across column 1 of the inverse
Why: the same row-times-column rule as always, now with three products to add instead of two.
\[ 2(12) + 1(-20) + (-2)(1.5) = 24 - 20 - 3 = 1 \]
Take row 1 across the other two columns
Why: these should give zeros, since they are off the diagonal of the answer.
\[ 2(-7) + 1(12) + (-2)(-1) = -14 + 12 + 2 = 0 \]
\[ 2(3) + 1(-5) + (-2)(0.5) = 6 - 5 - 1 = 0 \]
One row of the product is already finished: 1, then 0, then 0, which is the top row of the 3 by 3 identity. Two rows of work remain, and they behave the same way.
Do rows 2 and 3 the same way
Why: row 2 gives 60 minus 60, then negative 35 plus 36, then 15 minus 15; row 3 gives 48 minus 60 plus 12, then negative 28 plus 36 minus 8, then 12 minus 15 plus 4.
\[ \begin{bmatrix}0 & 1 & 0\end{bmatrix} \quad \text{and} \quad \begin{bmatrix}0 & 0 & 1\end{bmatrix} \]
Verify that the nine entries form the identity
Why: ones down the main diagonal and zeros everywhere else, so the calculator's matrix really is the inverse of A.
\[ A A^{-1} = \begin{bmatrix}1 & 0 & 0\\0 & 1 & 0\\0 & 0 & 1\end{bmatrix} \]
Nine dot products is more work than a 2 by 2 check, which is exactly why people skip it. Do the first row by hand and let the calculator do the rest, and you will still catch a mistyped entry.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 211 — Example 3
Worked example
Now a problem that is worth the machinery. A company sells three gift baskets, each holding some combination of movie passes, popcorn and DVDs, and you are told only the three total prices.
Figure (svg): Three gift baskets drawn as boxes. The basic basket holds two movie passes and one bag of popcorn and costs 15 dollars 50, the medium holds two passes, two popcorn and one DVD and costs 37 dollars, and the super holds four passes, three popcorn and two DVDs and costs 72 dollars 50.
Name the unknowns and write one equation per basket
Why: let m be the cost of a movie pass, p a packet of popcorn and d a DVD; each basket's contents become the coefficients and its price becomes the constant.
\[ 2m + p = 15.50, \quad 2m + 2p + d = 37.00, \quad 4m + 3p + 2d = 72.50 \]
Notice the first equation. The basic basket has no DVD in it, so the coefficient of d there is 0, and that zero has to be written into the matrix. A gap would push the other coefficients into the wrong columns.
Write the system as a matrix equation
Why: three rows of coefficients, a column of unknowns, a column of prices.
\[ \begin{bmatrix}2 & 1 & 0\\2 & 2 & 1\\4 & 3 & 2\end{bmatrix}\begin{bmatrix}m\\p\\d\end{bmatrix} = \begin{bmatrix}15.50\\37.00\\72.50\end{bmatrix} \]
Get the inverse of the coefficient matrix from a calculator
Why: its determinant is 2, so the entries come out in halves and whole numbers.
\[ A^{-1} = \begin{bmatrix}0.5 & -1 & 0.5\\0 & 2 & -1\\-1 & -1 & 1\end{bmatrix} \]
Multiply the column of prices on the left by the inverse
Why: one row-times-column calculation for each unknown, three products added each time.
\[ X = \begin{bmatrix}0.5(15.5) - 1(37) + 0.5(72.5)\\0(15.5) + 2(37) - 1(72.5)\\-1(15.5) - 1(37) + 1(72.5)\end{bmatrix} = \begin{bmatrix}7\\1.5\\20\end{bmatrix} \]
Nothing in that column was solved for separately. Each price is one row of the inverse taken across the column of totals, so the three answers arrive together in a single multiplication, which is what three equations in three unknowns would otherwise cost you a page of elimination to produce.
Check all three prices
Why: two passes and a popcorn is 14 plus 1.50, which is 15.50; two passes, two popcorn and a DVD is 14 plus 3 plus 20, which is 37; four passes, three popcorn and two DVDs is 28 plus 4.50 plus 40, which is 72.50.
\[ 2(7) + 1.5 = 15.5, \quad 2(7) + 2(1.5) + 20 = 37, \quad 4(7) + 3(1.5) + 2(20) = 72.5 \]
So a movie pass costs 7 dollars, a packet of popcorn 1 dollar 50 and a DVD 20 dollars. Three prices that were never listed anywhere, recovered from three totals.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 213 — Example 5
Real world
Here is where the inverse earns its keep. The company changes its prices; the contents of the baskets do not change. Ask yourself which part of the calculation you have to redo before you look at the answer.
\[ \begin{bmatrix}2 & 1 & 0\\2 & 2 & 1\\4 & 3 & 2\end{bmatrix}\begin{bmatrix}m\\p\\d\end{bmatrix} = \begin{bmatrix}17\\35\\69\end{bmatrix} \]
Discussion prompt
The basic basket now costs 17 dollars, the medium 35 dollars and the super 69 dollars. Find the new cost of each item, and say which parts of the Example 5 working you were able to reuse without repeating.
This is the structural advantage of the inverse over every other method in the chapter. Substitution, elimination and Cramer's rule all start again from the beginning when the right-hand side changes, because they mix the constants into the working from the first line.
The inverse keeps them separate. It is built from the coefficients alone, so it is a reusable object: one calculation that answers this question for every possible set of prices.
Businesses really do use it this way. A supplier who reprices weekly has the same coefficient matrix all year, and each new price list costs one multiplication rather than a fresh solve.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 213 — Guided Practice 11
Worked example
A word problem with two unknowns, so you are back to a 2 by 2 and the swap-negate-divide recipe. The hardest part is deciding what the two equations are.
A pilot has 200 hours of flight time altogether, split between single-engine and twin-engine aircraft. Single-engine time costs 60 dollars an hour, twin-engine time costs 240 dollars an hour, and 21,000 dollars has been spent in total.
Write one equation for hours and one for money
Why: let s be the single-engine hours and t the twin-engine hours; hours add to hours, and dollars add to dollars.
\[ s + t = 200 \qquad \text{and} \qquad 60s + 240t = 21\,000 \]
Keeping the two equations in different units is what stops them from being the same equation twice. One counts time, the other counts money, and it is the different rates that make the second one carry new information.
Write the matrix equation and find the determinant
Why: the coefficients of s and t in each equation, then 1 times 240 minus 1 times 60.
\[ \begin{bmatrix}1 & 1\\60 & 240\end{bmatrix}\begin{bmatrix}s\\t\end{bmatrix} = \begin{bmatrix}200\\21\,000\end{bmatrix}, \quad \det A = 240 - 60 = 180 \]
Build the inverse
Why: swap the 1 and the 240, negate the other 1 and the 60, and divide by 180.
\[ A^{-1} = \dfrac{1}{180}\begin{bmatrix}240 & -1\\-60 & 1\end{bmatrix} \]
Figure (svg): Two lines on a graph of twin-engine hours against single-engine hours: the total-hours line falling from 200, and the total-cost line falling more gently, crossing at 150 single-engine hours and 50 twin-engine hours.
The two lines in the picture are the two equations, and the dot where they meet is the answer the inverse is about to produce. Reading it off a graph would be guesswork at this scale; the arithmetic is exact.
Multiply the constants by the inverse on the left
Why: 240 times 200 is 48,000, less 21,000; and negative 60 times 200 is negative 12,000, plus 21,000.
\[ X = \dfrac{1}{180}\begin{bmatrix}48\,000 - 21\,000\\-12\,000 + 21\,000\end{bmatrix} = \dfrac{1}{180}\begin{bmatrix}27\,000\\9000\end{bmatrix} = \begin{bmatrix}150\\50\end{bmatrix} \]
Check both facts about the pilot
Why: 150 hours plus 50 hours is the 200 hours of flight time, and 60 times 150 is 9000 dollars while 240 times 50 is 12,000 dollars, which together are the 21,000 spent.
\[ 150 + 50 = 200 \qquad 60(150) + 240(50) = 9000 + 12\,000 = 21\,000 \]
Both totals come out exactly, so the pilot has flown 150 hours single-engine and 50 hours twin-engine. Checking the money as well as the hours is what confirms the expensive coefficient was used on the right unknown.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 215 — Exercise 43
Estimation
Dirk Nowitzki took 976 shots in the 2003 to 2004 season and scored 1680 points from them, with three-point field goals, two-point field goals and one-point free throws. He made 135 more two-pointers than free throws.
\[ t + w + f = 976, \quad 3t + 2w + f = 1680, \quad w - f = 135 \]
Predict first
Before solving, which of the three kinds of shot did he make most of?
Correct: Two-point field goals, by a clear margin
Why: Divide the points by the shots: 1680 over 976 is about 1.72 points per shot, which is much nearer 2 than 1 or 3, so the two-pointers must dominate the count. Solving confirms it: 99 three-pointers, 506 two-pointers and 371 free throws, and the two-pointers are more than half of all his shots.
The estimate is worth making a habit. An average of 1.72 points per shot cannot be produced by a pile of free throws or a pile of three-pointers, so the middle option has to carry most of the weight, and any answer that says otherwise is wrong before you check it.
\[ \begin{bmatrix}1 & 1 & 1\\3 & 2 & 1\\0 & 1 & -1\end{bmatrix}\begin{bmatrix}t\\w\\f\end{bmatrix} = \begin{bmatrix}976\\1680\\135\end{bmatrix} \]
Look at the third row of that coefficient matrix. The sentence about 135 more two-pointers than free throws contains no three-pointers at all, so the first coefficient is 0, and the comparison becomes a subtraction: two-pointers minus free throws.
Check the solution against all three facts. The three counts add to 976; three 99s plus two 506s plus 371 is 297 plus 1012 plus 371, which is 1680; and 506 minus 371 is 135. All three hold, so the inverse did its job.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 215 — Exercise 44
Worked example
This is the problem the book names on its opening page as the reason for the lesson. A cooking class wants to use up exactly 8 cups of buttermilk and 11 eggs by baking batches of rolls and batches of muffins.
A batch of rolls uses 2 cups of buttermilk and 3 eggs. A batch of muffins uses 1 cup and 1 egg. Nothing may be left over, which is what makes it a system rather than an inequality.
Write one equation per ingredient
Why: let r be the number of batches of rolls and m the batches of muffins; buttermilk is counted in one equation and eggs in the other.
\[ 2r + m = 8 \qquad \text{and} \qquad 3r + m = 11 \]
Notice that the equations are organised by ingredient rather than by recipe. It is tempting to write one equation per recipe, but a recipe is not a total, and only totals give equations.
Write the matrix equation and take the determinant
Why: 2 times 1 minus 1 times 3, which is negative 1.
\[ \begin{bmatrix}2 & 1\\3 & 1\end{bmatrix}\begin{bmatrix}r\\m\end{bmatrix} = \begin{bmatrix}8\\11\end{bmatrix}, \qquad \det A = 2 - 3 = -1 \]
Build the inverse
Why: swap the two 1s, negate the 1 and the 3, then divide by negative 1, which flips every sign.
\[ A^{-1} = \dfrac{1}{-1}\begin{bmatrix}1 & -1\\-3 & 2\end{bmatrix} = \begin{bmatrix}-1 & 1\\3 & -2\end{bmatrix} \]
Multiply the constants on the left by the inverse
Why: negative 8 plus 11 on the top, and 24 minus 22 underneath.
\[ X = \begin{bmatrix}-1 & 1\\3 & -2\end{bmatrix}\begin{bmatrix}8\\11\end{bmatrix} = \begin{bmatrix}3\\2\end{bmatrix} \]
Check both ingredients are used up exactly
Why: three batches of rolls take 6 cups and 9 eggs, two batches of muffins take 2 cups and 2 eggs, which is 8 cups and 11 eggs with nothing left.
\[ 2(3) + 2 = 8 \qquad 3(3) + 2 = 11 \]
Both answers also have to be whole numbers to make sense, and they are. A system like this can perfectly well produce two and a half batches, and then the honest answer is that the ingredients cannot be used up exactly.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 216 — Exercise 45
Picture it
One last use of the inverse, and a different kind. The columns of the matrix T below are the three corners of a triangle, and A is a transformation matrix: multiplying by it moves every corner somewhere new.
\[ A = \begin{bmatrix}0 & 1\\-1 & 0\end{bmatrix}, \qquad T = \begin{bmatrix}1 & 3 & 5\\1 & 4 & 2\end{bmatrix} \]
\[ A T = \begin{bmatrix}1 & 4 & 2\\-1 & -3 & -5\end{bmatrix}, \qquad A(A T) = \begin{bmatrix}-1 & -3 & -5\\-1 & -4 & -2\end{bmatrix} \]
Figure (svg): Three congruent triangles on a coordinate plane: the original with vertices at 1 comma 1, 3 comma 4 and 5 comma 2, a second one a quarter turn clockwise from it, and a third a half turn from the original.
Read the corners off the picture and the transformation names itself. The corner at 1 and 1 has gone to 1 and negative 1, and the corner at 3 and 4 to 4 and negative 3: every point has been turned a quarter turn clockwise about the origin, and multiplying twice turns it a half turn.
Now the question the exercise actually asks. Given only the twice-turned triangle, how do you get the original back? Multiply by the inverse of A, twice, because each multiplication by the inverse undoes one quarter turn.
\[ A^{-1} = \begin{bmatrix}0 & -1\\1 & 0\end{bmatrix}, \qquad A^{-1}A^{-1}\bigl(A(AT)\bigr) = T \]
That is the whole idea of this lesson in a picture. A matrix is an instruction; its inverse is the instruction that puts everything back; and the determinant of this one is 1, which is exactly what you would expect from a turn that neither stretches nor squashes anything.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 216 — Exercise 49
Section
Part 7
Pattern
Everything in this lesson fits into one route, with a single decision point. Follow it from the left and notice that the decision comes before any of the expensive work.
Figure (svg): A flow chart. Write each equation in standard form, build the three matrices, then find the determinant. If it is zero the chart branches to no unique solution; if it is not zero it continues to X equals the inverse of A times B, and then to checking in the original equations.
The order of the third and fifth items is what separates a confident solver from a slow one. The determinant is cheap and the inverse is not, so the cheap test always comes first.
For three unknowns the shape of the route is identical and only the middle changes: the inverse comes from a calculator rather than from the recipe, and everything before and after it stays exactly as it is here.
One habit to carry out of this lesson is the order of the last two items. Substituting into the original equations, rather than into the matrix equation you wrote, is what catches a mistake made at the very first step, when the coefficients were copied across.
Edge cases
Push the method to a case the book does not cover and see where it stops. Three facts about two unknowns sounds like more information, so it ought to be easier rather than impossible.
\[ x + y = 5, \quad 2x - y = 1, \quad x + 2y = 9 \]
Discussion prompt
Write the coefficient matrix for that system and give its dimensions. Then say why the inverse-matrix method cannot be used on it, and what you would do instead.
Notice how the failure differs from a determinant of zero. There, the matrix was square and the system had too little information; here the matrix is not square at all, and the system has more information than two unknowns can absorb.
The moral is that the inverse-matrix method answers one shape of question extremely well: as many equations as unknowns, and a coefficient matrix whose determinant is not zero. Outside that shape it does not give a wrong answer, it gives no answer, which is the safer kind of failure.
If you meet a system with more equations than unknowns in a real setting, three measurements of two quantities for instance, the honest question is usually not what solves all three but what comes closest to all three. That question has an answer, and it is built on this chapter.
Check
One application of the recipe, with a determinant that is negative. Work out the determinant before you look at the options, because it decides the signs of every entry in the answer.
Check your understanding
What is the inverse of the matrix with top row 1 and negative 5 and bottom row negative 1 and 4?
Answer: A
Why: The determinant is 1 times 4 minus negative 5 times negative 1, which is 4 minus 5, so negative 1. Swapping the main diagonal and negating the other pair gives 4 and 5 on the top row with 1 and 1 underneath, and dividing by negative 1 changes all four signs. Multiplying the original by it gives 1 and 0 on the top row and 0 and 1 underneath, so it is the inverse.
The quickest way to eliminate is to multiply the top row of the original by the first column of each candidate and look for a 1. Anything that gives negative 1 has the determinant on the wrong side of the division.
Notice how many of the wrong answers use the correct four numbers. Almost nothing in this topic goes wrong at the multiplication stage; it goes wrong in deciding which entry moves and which entry changes sign.
If you want to be certain, run the check in both orders. It takes eight small products and it is the only thing that distinguishes a right answer from a plausible one.
Check
Four systems. Only one of them defeats the inverse-matrix method, and you can find it with four subtractions and no solving at all.
Check your understanding
Which of these systems cannot be solved with an inverse matrix?
Answer: A
Why: Its coefficient matrix has determinant 1 times 4 minus 2 times 2, which is zero, so that matrix has no inverse. Its second equation has coefficients exactly twice the first one's, while its constant is not twice the first one's, so the two lines are parallel and the system has no solution at all. The other three have determinants of 17, 2 and 11, all safely away from zero.
The test is worth naming precisely, because it is all the question requires: work out the determinant of each coefficient matrix and look for a zero. No inverse needs to be built and no system needs to be solved.
Once you have found it, the follow-up question is which of the two zero cases it is. Doubling the first equation gives a right-hand side of 2 rather than 9, so the equations contradict each other and there is no solution rather than infinitely many.
Watch for the shape of the giveaway in future. A pair of equations where one row of coefficients is a clean multiple of the other is almost always a determinant of zero in disguise.
Check
The book's own multiple-choice system. The fastest honest route is to run the method, and the fastest dishonest route is to substitute the options, which is worth knowing as a fallback when time is short.
\[ 3x - 5y = -26 \qquad \text{and} \qquad -x + 2y = 10 \]
Check your understanding
What is the solution of that system?
Answer: C
Why: The determinant is 3 times 2 minus negative 5 times negative 1, which is 6 minus 5, so 1, and the inverse is simply the swapped and negated matrix: 2 and 5 on the top row, 1 and 3 underneath. Multiplying it by the constants gives 2 times negative 26 plus 5 times 10, which is negative 2, and negative 26 plus 30, which is 4. Both equations check: negative 6 minus 20 is negative 26, and 2 plus 8 is 10.
A determinant of 1 is a gift worth spotting early, because it means the third step of the recipe does nothing at all and the inverse is just the swapped, negated matrix.
The wrong option that satisfies the first equation is the one to learn from. Any single equation has infinitely many solutions, so passing one test proves nothing; it is passing both that identifies the crossing point.
If a question like this appears with the clock running, substitution into both equations settles it in about fifteen seconds. Keep the full method for when you are asked to show working, and use the shortcut when you are asked only for the answer.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 215 — Exercise 34
Explain it to yourself
One idea has run underneath every part of this lesson. Putting it into your own sentences is what turns it from a procedure you can follow into something you can rebuild if you forget it.
Discussion prompt
In three or four sentences, explain how solving a matrix equation by multiplying by an inverse is the same skill as solving a number equation by multiplying by a reciprocal. Then say which number plays the part of the determinant, and why the two failures happen for the same reason.
There is one genuine difference to include if you want a complete answer. Numbers may be multiplied in either order, so the side does not matter, while matrices may not, which is why the inverse has to go on the left of both sides.
If your explanation leaned on the phrase you just divide, rewrite it. Division is the one word that does not survive the trip from numbers to matrices, and the whole lesson exists because it does not.
Test yourself with the case that decides whether you have understood it. If the determinant of a coefficient matrix comes out as zero, what have you learned about the system? Anyone who can answer that in one sentence has the lesson.
Exit ticket
One system, worked from the beginning to a checked answer. Use the full route: standard form, three matrices, determinant, inverse, multiply, check.
\[ 4x - y = 10 \qquad \text{and} \qquad -7x - 2y = -25 \]
Discussion prompt
Solve the system with an inverse matrix, showing the determinant, the inverse and the final multiplication, and confirm your answer in both original equations.
Two places in that working generate most of the errors. The determinant involves subtracting a product of two negatives, which adds rather than subtracts; and the division at the end is by a negative number, so every sign changes.
If your answer came out as negative 3 and negative 2, you have the right numbers and one missing sign change. Go back to the line where you divided by negative 15 and check that all four entries were treated the same way.
Notice that the inverse was worth building even here, where elimination would have been quick. The practice matters more than the speed at this stage, and the method is the one that scales to three unknowns while elimination becomes unpleasant.
McDougal Littell Algebra 2 (Texas Edition), Ch. 3 Linear Systems and Matrices — Lesson 3.8 Use Inverse Matrices to Solve Linear Systems §3.8, p. 215 — Exercise 25
Recap
This lesson gave the chapter its last tool, and it is the one that treats a whole system as a single equation. Everything in it came from one idea: build the matrix that undoes A, then use it the way you have always used a reciprocal.
| the question | what to do | where it fails |
|---|---|---|
| is this matrix the identity? | check for 1s on the main diagonal and 0s elsewhere | it is never anything but square |
| are these two matrices inverses? | multiply them in both orders and look for the identity | one product alone proves only half |
| what is the inverse of a 2 by 2? | swap the main diagonal, negate the other pair, divide by the determinant | a determinant of zero: there is no inverse |
| how do I solve A X = B? | multiply both sides on the left by the inverse of A | multiplying on the right leaves X stranded |
| how do I solve a system? | write it as A X = B in standard form, then X is the inverse of A times B | a zero determinant means no unique solution |
\[ X = A^{-1}B \]
That single line is the lesson. The work is all in building the inverse, and the determinant tells you in one subtraction whether the building is possible.
The thread back through the chapter is worth holding on to. A determinant of zero is the same event you met as parallel or identical lines in Lesson 3.1, as a contradiction in Lesson 3.2, and as a zero denominator in Cramer's rule in Lesson 3.7. Four lessons, one number.
Chapter 4 changes the subject to quadratic equations, where the number that decides how many solutions exist is called the discriminant. When you meet it, notice the shape of the idea: one quantity computed from the coefficients, telling you what kind of answer to expect before you look for it.
OpenStax Algebra and Trigonometry 2e, §11.7 Solving Systems with Inverses §11.7 — the same method, with the inverse of a 3 by 3 found by row reduction
Want this taught 1-on-1? Alexander tutors Algebra 2 — $55/session, free consultation.