i i “main”

2007/2/16
page 139

i i

2.4 Elementary Row Operations and Row-Echelon Matrices 139

   
3. Verify that for all values of t, 2 1 3 3
10. A =  4 −1 2  , b =  1 .
(1 − t, 2 + 3t, 3 − 2t) 7 6 3 −5
is a solution to the linear system 11. Consider the m × n homogeneous system of linear
x1 + x2 + x3 = 6, equations
x1 − x2 − 2x3 = −7,
5x1 + x2 − x3 = 4. Ax = 0. (2.3.2)

4. Verify that for all values of s and t,

(s, s − 2t, 2s + 3t, t) (a) If x = [x1 x2 . . . xn ]T and y = [y1 y2 . . . yn ]T

are solutions to (2.3.2), show that
is a solution to the linear system
z=x+y and w = cx
x1 + x2 − x3 + 5x4 = 0,
2x2 − x3 + 7x4 = 0,
are also solutions, where c is an arbitrary scalar.
4x1 + 2x2 − 3x3 + 13x4 = 0.
(b) Is the result of (a) true when x and y are solu-
5. By making a sketch in the xy-plane, prove that the tions to the nonhomogeneous system Ax = b?
following linear system has no solution: Explain.
2x + 3y = 1, For Problems 12–15, write the vector formulation for the
2x + 3y = 2. given system of differential equations.

For Problems 6–8, determine the coefficient matrix, A, the 12. x1
= −4x1 + 3x2 + 4t, x2
= 6x1 − 4x2 + t 2 .
right-hand-side vector, b, and the augmented matrix A# of
the given system. 13. x1
= t 2 x1 − tx2 , x2
= (− sin t)x1 + x2 .
x1 + 2x2 − 3x3 = 1,
14. x1
= e2t x2 , x2
+ (sin t)x1 = 1.
6. 2x1 + 4x2 − 5x3 = 2,
7x1 + 2x2 − x3 = 3.
15. x1
= (− sin t)x2 + x3 + t, x2
= −et x1 + t 2 x3 + t 3 ,
x + y + z − w = 3, x3
= −tx1 + t 2 x2 + 1.
7.
2x + 4y − 3z + 7w = 2.
For Problems 16–17 verify that the given vector function x
x1 + 2x2 − x3 = 0, defines a solution to x
= Ax + b for the given A and b.
8. 2x1 + 3x2 − 2x3 = 0,  4t     
5x1 + 6x2 − 5x3 = 0. e 2 −1 0
16. x(t) = , A = , b(t) = .
−2e4t −2 3 0
For Problems 9–10, write the system of equations with the
given coefficient matrix and right-hand-side vector.    
4e−2t + 2 sin t 1 −4
    17. x(t) = , A = ,
1 −1 2 3 1 3e−2t − cos t −3 2
 
9. A =  1 1 −2 6  , b =  −1 . −2(cos t + sin t)
b(t) = .
3 1 4 2 2 7 sin t + 2 cos t

2.4 Elementary Row Operations and Row-Echelon Matrices

In the next section we will develop methods for solving a system of linear equations.
These methods will consist of reducing a given system of equations to a new system that
has the same solution set as the given system but is easier to solve. In this section we
introduce the requisite mathematical results.

i i

i i
 i i “main”
2007/2/16
page 140

i i

140 CHAPTER 2 Matrices and Systems of Linear Equations

Elementary Row Operations

The first step in deriving systematic procedures for solving a linear system is to determine
what operations can be performed on such a system without altering its solution set.

Example 2.4.1 Consider the system of equations

x1 + 2x2 + 4x3 = 2, (2.4.1)

2x1 − 5x2 + 3x3 = 6, (2.4.2)
4x1 + 6x2 − 7x3 = 8. (2.4.3)

Solution: If we permute (i.e., interchange), say, Equations (2.4.1) and (2.4.2), the
resulting system is
2x1 − 5x2 + 3x3 = 6,
x1 + 2x2 + 4x3 = 2,
4x1 + 6x2 − 7x3 = 8,
which certainly has the same solution set as the original system. Returning to the original
system, if we multiply, say, Equation (2.4.2) by 5, we obtain the system

x1 + 2x2 + 4x3 = 2,
10x1 − 25x2 + 15x3 = 30,
4x1 + 6x2 − 7x3 = 8,

which again has the same solution set as the original system. Finally, if we add, say,
twice Equation (2.4.1) to Equation (2.4.3), we obtain the system

x1 + 2x2 + 4x3 = 2, (2.4.4)

2x1 − 5x2 + 3x3 = 6, (2.4.5)
(4x1 + 6x2 − 7x3 ) + 2(x1 + 2x2 + 4x3 ) = 8 + 2(2). (2.4.6)

We can verify that, if (2.4.4)–(2.4.6) are satisfied, then so are (2.4.1)–(2.4.3), and
vice versa. It follows that the system of equations (2.4.4)–(2.4.6) has the same solution
set as the original system of equations (2.4.1)–(2.4.3).

More generally, similar reasoning can be used to show that the following three
operations can be performed on any m × n system of linear equations without altering
the solution set:

1. Permute equations.
2. Multiply an equation by a nonzero constant.
3. Add a multiple of one equation to another equation.

Since these operations involve changes only in the system coefficients and constants
(and not changes in the variables), they can be represented by the following operations
on the rows of the augmented matrix of the system:

1. Permute rows.
2. Multiply a row by a nonzero constant.
3. Add a multiple of one row to another row.

i i

i i
 i i “main”
2007/2/16
page 141

i i

2.4 Elementary Row Operations and Row-Echelon Matrices 141

These three operations, called elementary row operations, will be a basic computational
tool throughout the text, even in cases when the matrix under consideration is not derived
from a system of linear equations. The following notation will be used to describe
elementary row operations performed on a matrix A.

1. Pij : Permute the ith and j th rows in A.

2. Mi (k): Multiply every element of the ith row of A by a nonzero scalar k.
3. Aij (k): Add to the elements of the j th row of A the scalar k times the corresponding
elements of the ith row of A.

Furthermore, the notation A ∼ B will mean that matrix B has been obtained from
matrix A by a sequence of elementary row operations. To reference a particular elemen-
tary row operation used in, say, the nth step of the sequence of elementary row operations,
n
we will write ∼ B.

Example 2.4.2 The one-step operations performed on the system in Example 2.4.1 can be described as
follows using elementary row operations on the augmented matrix of the system:
   
1 2 4 2 2 −5 3 6
 2 −5 3 6  ∼ 1 
1 2 4 2 1. P12 . Permute (2.4.1) and (2.4.2).
4 6 −7 8 4 6 −7 8
   
1 2 4 2 1 2 4 2
 2 −5 3 6  ∼  10 −25 15 30  1. M2 (5). Multiply (2.4.2) by 5.
1

4 6 −7 8 4 6 −7 8
   
1 2 4 2 1 2 4 2
 2 −5 3 6  ∼  2 −5 3 6 
1
1. A13 (2). Add 2 times (2.4.1) to (2.4.3).

4 6 −7 8 6 10 1 12

It is important to realize that each elementary row operation is reversible; we can “un-
do” a given elementary row operation by another elementary row operation to bring the
modified linear system back into its original form. Specifically, in terms of the notation
introduced above, the reverse operations are determined as follows (ERO refers here to
“elementary row operation”):

ERO Applied to A Reverse ERO Applied to B

A∼B B∼A
Pij Pj i : Permute row j and i in B.
Mi (k) Mi (1/k): Multiply the ith row of B by 1/k.
Aij (k) Aij (−k): Add to the elements of the j th row
of B the scalar −k times the corresponding
elements of the ith row of B

We introduce a special term for matrices that are related via elementary row operations.

DEFINITION 2.4.3
Let A be an m × n matrix. Any matrix obtained from A by a finite sequence of
elementary row operations is said to be row-equivalent to A.

i i

i i
 i i “main”
2007/2/16
page 142

i i

142 CHAPTER 2 Matrices and Systems of Linear Equations

Thus, all of the matrices in the previous example are row-equivalent. Since elemen-
tary row operations do not alter the solution set of a linear system, we have the next
theorem.

Theorem 2.4.4 Systems of linear equations with row-equivalent augmented matrices have the same
solution sets.

Row-Echelon Matrices
Our methods for solving a system of linear equations will consist of using elementary
row operations to reduce the augmented matrix of the given system to a simple form.
But how simple a form should we aim for? In order to answer this question, consider the
system

x1 + x2 − x3 = 4, (2.4.7)
x2 − 3x3 = 5, (2.4.8)
x3 = 2. (2.4.9)

This system can be solved most easily as follows. From Equation (2.4.9), x3 = 2.
Substituting this value into Equation (2.4.8) and solving for x2 yields x2 = 5 + 6 = 11.
Finally, substituting for x3 and x2 into Equation (2.4.7) and solving for x1 , we obtain
x1 = −5. Thus, the solution to the given system of equations is (−5, 11, 2), a single
vector in R3 . This technique is called back substitution and could be used because the
given system has a simple form. The augmented matrix of the system is
 
1 1 −1 4
0 1 −3 5
0 0 1 2

We see that the submatrix consisting of the first three columns (which corresponds
to the matrix of coefficients) is an upper triangular matrix with the leftmost nonzero
entry in each row equal to 1. The back-substitution method will work on any system of
linear equations with an augmented matrix of this form. Unfortunately, not all systems
of equations have augmented matrices that can be reduced to such a form. However,
there is a simple type of matrix to which any matrix can be reduced by elementary row
operations, and which also represents a system of equations that can be solved (if it has
a solution) by back substitution. This is called a row-echelon matrix and is defined as
follows:

DEFINITION 2.4.5
An m × n matrix is called a row-echelon matrix if it satisfies the following three
conditions:

1. If there are any rows consisting entirely of zeros, they are grouped together at
the bottom of the matrix.
2. The first nonzero element in any nonzero row4 is a 1 (called a leading 1).
3. The leading 1 of any row below the first row is to the right of the leading 1 of
the row above it.

4
A nonzero row (nonzero column) is any row (column) that does not consist entirely of zeros.

i i

i i
 i i “main”
2007/2/16
page 143

i i

2.4 Elementary Row Operations and Row-Echelon Matrices 143

Example 2.4.6 Examples of row-echelon matrices are

 
    1 −1 6 5 9
1 −2 3 7 0 0 1 0 0 1 2 5
0 1 5 0  , 0 0 0 , and  
0 0 0 1 0 ,
0 0 0 1 0 0 0
0 0 0 0 0

whereas
 
  1 0 0
1 0 −1 0
0 1 2  0 0
and 0 1 −1
0 1 −1
0 0 1

are not row-echelon matrices.

The basic result that will allow us to determine the solution set to any system of
linear equations is stated in the next theorem.

Theorem 2.4.7 Any matrix is row-equivalent to a row-echelon matrix.

According to this theorem, by applying an appropriate sequence of elementary row
operations to any m × n matrix, we can always reduce it to a row-echelon matrix. When
a matrix A has been reduced to a row-echelon matrix in this way, we say that it has been
reduced to row-echelon form and refer to the resulting matrix as a row-echelon form
of A. The proof of Theorem 2.4.7 consists of giving an algorithm that will reduce an
arbitrary m × n matrix to a row-echelon matrix after a finite sequence of elementary
row operations. Before presenting such an algorithm, we first illustrate the result with
an example.
 
2 1 −1 3
 1 −1 2 1
Example 2.4.8 Use elementary row operations to reduce  
−4 6 −7 1 to row-echelon form.
2 0 1 3
Solution: We show each step in detail.
Step 1: Put a leading 1 in the (1, 1) position.
This is most easily accomplished by permuting rows 1 and 2.
   
2 1 −1 3 1 −1 2 1
 1 −1 2 1 1  2 1 −1 3
   
−4 6 −7 1 ∼ −4 6 −7 1
2 0 1 3 2 0 1 3

Step 2: Use the leading 1 to put zeros beneath it in column 1.

This is accomplished by adding appropriate multiples of row 1 to the remaining
rows.
  
1 −1 2 1 
Add −2 times row 1 to row 2.
2 
 0 3 −5 1 
∼ Step 2 row operations: Add 4 times row 1 to row 3.
0 2 1 5 

0 2 −3 1 Add −2 times row 1 to row 4.

Step 3: Put a leading 1 in the (2, 2) position.

We could accomplish this by multiplying row 2 by 1/3. However, this would intro-
duce fractions into the matrix and thereby complicate the remaining computations. In

i i

i i
 i i “main”
2007/2/16
page 144

i i

144 CHAPTER 2 Matrices and Systems of Linear Equations

hand calculations, fewer algebraic errors result if we avoid the use of fractions. In this
case, we can obtain a leading 1 without the use of fractions by adding −1 times row 3
to row 2.
 
1 −1 2 1
3 0 1 −6 −4
∼ 0 2 1 5 
 Step 3 row operation: Add −1 times row 3 to row 2.
0 2 −3 1

Step 4: Use the leading 1 in the (2, 2) position to put zeros beneath it in column 2.
We now add appropriate multiples of row 2 to the rows beneath it. For row-echelon
form, we need not be concerned about the row above it, however.
 
1 −1 2 1
4  0 1 −6 −4  Add −2 times row 2 to row 3.
∼0 0 13 13
 Step 4 row operations:
Add −2 times row 2 to row 4.
0 0 9 9

Step 5: Put a leading 1 in the (3, 3) position.

This can be accomplished by multiplying row 3 by 1/13.
 
1 −1 2 1
5 0 1 −6 −4
∼ 0 0 1 1 


0 0 9 9

Step 6: Use the leading 1 in the (3, 3) position to put zeros beneath it in column 3.
The appropriate row operation is to add −9 times row 3 to row 4.
 
1 −1 2 1
6 0 1 −6 −4
∼ 0 0 1 1 


0 0 0 0

This is a row-echelon matrix, hence the given matrix has been reduced to row-echelon
form. The specific operations used at each step are given next, using the notation intro-
duced previously in this section. In future examples, we will simply indicate briefly the
elementary row operation used at each step. The following shows this description for the
present example.

1. P12 2. A12 (−2), A13 (4), A14 (−2) 3. A32 (−1)

4. A23 (−2), A24 (−2) 5. M3 (1/13) 6. A34 (−9)

Remarks

1. Notice that in steps 2 and 4 of the preceding example we have performed multi-
ple elementary row operations of the type Aij (k) in a single step. With this one
exception, the reader is strongly advised not to combine multiple elementary row
operations into a single step, particularly when they are of different types. This is
a common source of calculation errors.

i i

i i
 i i “main”
2007/2/16
page 145

i i

2.4 Elementary Row Operations and Row-Echelon Matrices 145

2. The reader may have noticed that the particular steps taken in the preceding exam-
ple are not uniquely determined. For instance, we could have achieved a leading 1
in the (1, 1) position in step 1 by multiplying the first row by 1/2, rather than per-
muting the first two rows. Therefore, we may have multiple strategies for reducing
a matrix to row-echelon form, and indeed, many possible row-echelon forms for
a given matrix A. In this particular case, we chose not to multiply the first row by
1/2 in order to avoid introducing fractions into the calculations.

The reader is urged to study the foregoing example very carefully, since it illustrates
the general procedure for reducing an m×n matrix to row-echelon form using elementary
row operations. This procedure will be used repeatedly throughout the text. The idea
behind reduction to row-echelon form is to start at the upper left-hand corner of the
matrix and proceed downward and to the right in the matrix. The following algorithm
formalizes the steps that reduce any m × n matrix to row-echelon form using a finite
number of elementary row operations and thereby provides a proof of Theorem 2.4.7.
An illustration of this algorithm is given in Figure 2.4.1.

Algorithm for Reducing an m × n Matrix A to Row-Echelon Form

1. Start with an m × n matrix A. If A = 0, go to step 7.

2. Determine the leftmost nonzero column (this is called a pivot column,
and the topmost position in this column is called a pivot position).
3. Use elementary row operations to put a 1 in the pivot position.
4. Use elementary row operations to put zeros below the pivot position.
5. If there are no more nonzero rows below the pivot position go to step 7,
otherwise go to step 6.
6. Apply steps 2 through 5 to the submatrix consisting of the rows that lie
below the pivot position.
7. The matrix is a row-echelon matrix.

* Elementary row * Elementary row *

* operations applied * operations applied *
0 . . . 0 * * * * to pivot column 0 . . . 0 1 * * * to rows beneath 0 . . . 0 1 * * *
* in submatrix * pivot position 0 *
* ~ * ~ . *
ze

ze

ze
* Pivot position * . *
ro

ro

ro
s

s
* * 0 *
Nonzero elements New submatrix
Submatrix Pivot column

Figure 2.4.1: Illustration of an algorithm for reducing an m × n matrix to row-echelon form.

Remark In order to obtain a row-echelon matrix, we put a 1 in each pivot position.

However, many algorithms for solving systems of linear equations numerically are based
around the preceding algorithm, except that in step 3 we place a nonzero number (not
necessarily a 1) in the pivot position. Of course, the matrix resulting from an application
of this algorithm differs from a row-echelon matrix, since it will have arbitrary nonzero
elements in the pivot positions.
 
3 2 −5 2
Example 2.4.9 Reduce 1 1 −2 1 to row-echelon form.
1 0 −3 4

i i

i i
 i i “main”
2007/2/16
page 146

i i

146 CHAPTER 2 Matrices and Systems of Linear Equations

Solution: Applying the row-reduction algorithm leads to the following sequence of

elementary row operations. The specific row operations used at each step are given at
the end of the process.
Pivot position
     
Pivot position - 3h2 −5 2 1 1 −1 1 1 1 −1 1
1 1 −1 1 ∼1  m−2 −1
3 2 −5 2 ∼ 0 −1
2

1 0 −3 4 1 0 −3 4 0 −1 −2 3
6 6
Pivot column
Pivot column Pivot column
   @R  
@ 
1 1 −1 1 1 1 −1 1 1 1 −1 1
∼ 0 1 2 1 ∼ 0 1 2 1 ∼ 0 1 2 1
3 4 5

0 −1 −2 3 0 0 0 4j 0 0 0 1
6
Pivot position

This is a row-echelon matrix and hence we are done. The row operations used are
summarized here:

1. P12 2. A12 (−3), A13 (−1) 3. M2 (−1) 4. A23 (1) 5. M3 (1/4)

The Rank of a Matrix

We now derive some further results on row-echelon matrices that will be required in the
next section to develop the theory for solving systems of linear equations.
We first observe that a row-echelon form for a matrix A is not unique. Given one
row-echelon form for A, we can always obtain a different one by taking the first row-
echelon form for A and adding some multiple of a given row to any rows above it. The
result is still in row-echelon form.
However, even though the row-echelon form of A is not unique, we do have the
following theorem (in Chapter 4 we will see how the proof of this theorem arises naturally
from the more sophisticated ideas from linear algebra yet to be introduced).

Theorem 2.4.10 Let A be an m × n matrix. All row-echelon matrices that are row-equivalent to A have
the same number of nonzero rows.
Theorem 2.4.10 associates a number with any m × n matrix A—namely, the number
of nonzero rows in any row-echelon form of A. As we will see in the next section, this
number is fundamental in determining the solution properties of linear systems, and
indeed it plays a central role in linear algebra in general. For this reason, we give it a
special name.

DEFINITION 2.4.11
The number of nonzero rows in any row-echelon form of a matrix A is called the
rank of A and is denoted rank(A).

 
3 1 4
Example 2.4.12 Determine rank(A) if A = 4 3 5.
2 −1 3

i i

i i
 i i “main”
2007/2/16
page 147

i i

2.4 Elementary Row Operations and Row-Echelon Matrices 147

Solution: In order to determine rank(A), we must first reduce A to row-echelon form.

         
3 1 4 1 2 1 1 2 1 1 2 1 1 2 1
 4 3 5 ∼ 1 
4 3 5 ∼ 0 −5 1 ∼ 0 1 − 15  ∼ 0 1 − 15  .
2 3 4

2 −1 3 2 −1 3 0 −5 1 0 −5 1 0 0 0

Since there are two nonzero rows in this row-echelon form of A, it follows from Defini-
tion 2.4.11 that rank(A) = 2.

1. A31 (−1) 2. A12 (−4), A13 (−2) 3. M2 (−1/5) 4. A23 (5)

In the preceding example, the original matrix A had three nonzero rows, whereas any
row-echelon form of A has only two nonzero rows. We can interpret this geometrically as
follows. The three row vectors of A can be considered as vectors in R3 with components

a1 = (3, 1, 4), a2 = (4, 3, 5), a3 = (2, −1, 3).

In performing elementary row operations on A, we are taking combinations of these

vectors in the following way:

c1 a1 + c2 a2 + c3 a3 ,

and thus the rows of a row-echelon form of A are all of this form. We have been combining
the vectors linearly. The fact that we obtained a row of zeros in the row-echelon form
means that there are values of the constants c1 , c2 , c3 such that

c1 a1 + c2 a2 + c3 a3 = 0,

where 0 denotes the zero vector (0, 0, 0). Equivalently, one of the vectors can be written
in terms of the other two vectors, and therefore the three vectors lie in a plane. Reducing
the matrix to row-echelon form has uncovered this relationship among the three vectors.
We shall have much more to say about this in Chapter 4.

Remark If A is an m×n matrix, then rank(A) ≤ m and rank(A) ≤ n. This is because

the number of nonzero rows in a row-echelon form of A is equal to the number of pivots
in a row-echelon form of A, which cannot exceed the number of rows or columns of A,
since there can be at most one pivot per row and per column.

Reduced Row-Echelon Matrices

In the future we will need to consider the special row-echelon matrices that arise when
zeros are placed above, as well as beneath, each leading 1. Any such matrix is called a
reduced row-echelon matrix and is defined precisely as follows.

DEFINITION 2.4.13
An m × n matrix is called a reduced row-echelon matrix if it satisfies the following
conditions:

1. It is a row-echelon matrix.
2. Any column that contains a leading 1 has zeros everywhere else.

i i

i i
 i i “main”
2007/2/16
page 148

i i

148 CHAPTER 2 Matrices and Systems of Linear Equations

Example 2.4.14 The following are examples of reduced row-echelon matrices:

     
1 3 0 0   1 0 5 3 1 0 0
0 0 1 0 , 1 −1 7 0 0 1 2 1 , 0 1 0  .
, and

0 0 0 1
0 0 0 1 0 0 0 0 0 0 1

Although an m × n matrix A does not have a unique row-echelon form, in reducing

A to a reduced row-echelon matrix we are making a particular choice of row-echelon
matrix, since we arrange that all elements above each leading 1 are zeros. In view of this,
the following theorem should not be too surprising:

Theorem 2.4.15 An m × n matrix is row-equivalent to a unique reduced row-echelon matrix.

The unique reduced row-echelon matrix to which a matrix A is row-equivalent will
be called the reduced row-echelon form of A. As illustrated in the next example, the
row-reduction algorithm is easily extended to determine the reduced row-echelon form
of A—we just put zeros above and beneath each leading 1.

 
3 −1 22
Example 2.4.16 Determine the reduced row-echelon form of A = −1 5 2.
2 4 24
Solution: We apply the row-reduction algorithm, but put 0s above and below the
leading 1s. In so doing, it is immaterial whether we first reduce A to row-echelon form
and then arrange 0s above the leading 1s, or arrange 0s both above and below the leading
1s as we proceed from left to right.
         
3 −1 22 1 9 26 1 9 26 1 9 26 1 0 8
A = −1 5 2 ∼ −1 5 2 ∼ 0 14 28 ∼ 0 2 ∼ 0 1 2
1 2 3 4
1
2 4 24 2 4 24 0 −14 −28 0 −14 −28 0 0 0

which is the reduced row-echelon form of A.

1. A21 (2) 2. A12 (1), A13 (−2) 3. M2 (1/14) 4. A21 (−9), A23 (14)

Exercises for 2.4

Key Terms • Be able to find the rank of a matrix.
Elementary row operations, Row-equivalent matrices, Back
substitution, Row-echelon matrix, Row-echelon form, Lead- True-False Review
ing 1, Pivot, Rank of a matrix, Reduced row-echelon matrix. For Questions 1–9, decide if the given statement is true or
false, and give a brief justification for your answer. If true,
Skills you can quote a relevant definition or theorem from the text.
If false, provide an example, illustration, or brief explanation
• Be able to perform elementary row operations on a of why the statement is false.
matrix.
• Be able to determine a row-echelon form or reduced 1. A matrix A can have many row-echelon forms but only
row-echelon form for a matrix. one reduced row-echelon form.

i i

i i
 i i “main”
2007/2/16
page 149

i i

2.4 Elementary Row Operations and Row-Echelon Matrices 149

 
2. Any upper triangular n × n matrix is in row-echelon 0 1 0 0
form. 8.  0 0 1 0 .
0 0 0 0
3. Any n × n matrix in row-echelon form is upper trian-
gular. For Problems 9–18, use elementary row operations to reduce
the given matrix to row-echelon form, and hence determine
4. If a matrix A has more rows than a matrix B, then
the rank of each matrix.
rank(A) ≥ rank(B).
 
2 1
5. For any matrices A and B of the same dimensions, 9. .
1 −3
rank(A + B) = rank(A) + rank(B).  
2 −4
10. .
6. For any matrices A and B of the appropriate dimen- −4 8
sions,  
2 1 4
rank(AB) = rank(A) · rank(B). 11.  2 −3 4 .
3 −2 6
7. If a matrix has rank zero, then it must be the zero
 
matrix. 0 1 3
8. The matrices A and 2A must have the same rank. 12.  0 1 4 .
0 3 5
9. The matrices A and 2A must have the same reduced  
row-echelon form. 2 −1
13.  3 2 .
Problems 2 5
For Problems 1–8, determine whether the given matrices are  
2 −1 3
in reduced row-echelon form, row-echelon form but not re- 14.  3 1 −2 .
duced row-echelon form, or neither. 2 −2 1
 
1 0 −1 0  
1.  0 0 1 2 . 2 −1 3 4
0 0 0 0 15.  1 −2 1 3 .
  1 −5 0 5
1 0 2 5  
2.  1 0 0 2 . 2 −2 −1 3
0 1 1 0 3 −2 3 1 
16. 
1
.
  −1 1 0 
1 0 0 0 2 −1 2 2
3. .
0 0 0 1
   
0 1 4 7 4 7
4. . 3 5 3 5
1 0 17.  
 2 −2 2 −2 .
 
1 1 5 −2 5 −2
5. .
0 0  
  2 1 3 4 2
1 0 1 2 18.  1 0 2 1 3 .
0 0 1 1
6.  
 0 0 0 1 .
2 3 1 5 7

0 0 0 0 For Problems 19–25, reduce the given matrix to reduced row-

  echelon form and hence determine the rank of each matrix.
0 0 0 0  
7.  0 0 0 0 . 3 2
19. .
0 0 0 0 1 −1

i i

i i
 i i “main”
2007/2/16
page 150

i i

150 CHAPTER 2 Matrices and Systems of Linear Equations

 
3 7 10 the linear algebra package of Maple, the three elementary
20.  2 3 −1 . row operations are
1 2 1
  • swaprow(A, i, j ) : permute rows i and j
3 −3 6 • mulrow(A, i, k) : multiply row i by k
21.  2 −2 4 .
6 −6 12 • addrow(A, i, j ) : add k times row i to row j
   For Problems 26–28, use some form of technology to de-
3 5 −12
 2 3 −7 . termine a row-echelon form of the given matrix.
22.
−2 −1 1 26. The matrix in Problem 14.
 
1 −1 −1 2 27. The matrix in Problem 15.
 3 −2 0 7 
23.  
 2 −1 2 4 . 28. The matrix in Problem 18.
4 −2 3 8  Many forms of technology also have built-in functions
  for directly determining the reduced row-echelon form of
1 −2 1 3
 3 −6 2 7 . a given matrix A. For example, in the linear algebra pack-
24.
age of Maple, the appropriate command is rref(A). In Prob-
4 −8 3 10
lems 29–31, use technology to determine directly the reduced
  row-echelon form of the given matrix.
0 1 2 1
25.  0 3 1 2 .
29. The matrix in Problem 21.
0 2 0 1
30. The matrix in Problem 24.
Many forms of technology have commands for performing
elementary row operations on a matrix A. For example, in 31. The matrix in Problem 25.

2.5 Gaussian Elimination

We now illustrate how elementary row-operations applied to the augmented matrix of a
system of linear equations can be used first to determine whether the system is consistent,
and second, if the system is consistent, to find all of its solutions. In doing so, we will
develop the general theory for linear systems of equations.

Example 2.5.1 Determine the solution set to

3x1 − 2x2 + 2x3 = 9,

x1 − 2x2 + x3 = 5, (2.5.1)
2x1 − x2 − 2x3 = −1.

Solution: We first use elementary row operations to reduce the augmented matrix of
the system to row-echelon form.
     
3 −2 2 9 1 −2 1 5 1 −2 1 5
1 −2 1 5 ∼ 1 
3 −2 2 9 ∼ 0 4 −1 − 6
2

2 −1 −2 −1 2 −1 −2 −1 0 3 −4 −11

     
1 −2 1 5 1 −2 1 5 1 −2 1 5
∼ 0 1 3 5 ∼ 0 1 5 ∼ 0 1 3 5 .
3 4 5
3
0 3 −4 −11 0 0 −13 −26 0 0 12

i i

i i