Chapter 1 Systems of Linear Equa ons

row echelon form helps us iden fy all types of solu ons. We’ll explore the topic of un-
derstanding what the reduced row echelon form of a matrix tells us in the following
sec ons; in this sec on we focus on finding it.

.
Defini on 3 Reduced Row Echelon Form

A matrix is in reduced row echelon form if its entries sa sfy

the following condi ons.

1. The first nonzero entry in each row is a 1 (called a lead-

ing 1).

.
2. Each leading 1 comes in a column to the right of the
leading 1s in rows above it.

3. All rows of all 0s come at the bo om of the matrix.

4. If a column contains a leading 1, then all other entries
in that column are 0.
A matrix that sa sfies the first three condi ons is said to be
in row echelon form.

. Example 3 .Which of the following matrices is in reduced row echelon form?

[ ] [ ]
1 0 1 0 1
a) b)
0 1 0 1 2
[ ] [ ]
0 0 1 1 0
c) d)
0 0 0 0 1
   
1 0 0 1 1 2 0 0
e)  0 0 0 0 f)  0 0 3 0
0 0 1 3 0 0 0 4
   
0 1 2 3 0 4 1 1 0
g)  0 0 0 0 1 5 h)  0 1 0
0 0 0 0 0 0 0 0 1

S The matrices in a), b), c), d) and g) are all in reduced row echelon
form. Check to see that each sa sfies the necessary condi ons. If your ins ncts were
wrong on some of these, correct your thinking accordingly.
The matrix in e) is not in reduced row echelon form since the row of all zeros is
not at the bo om. The matrix in f) is not in reduced row echelon form since the first

14
 1.3 Elementary Row Opera ons and Gaussian Elimina on

nonzero entries in rows 2 and 3 are not 1. Finally, the matrix in h) is not in reduced
row echelon form since the first entry in column 2 is not zero; the second 1 in column
2 is a leading one, hence all other entries in that column should be 0.
We end this example with a preview of what we’ll learn in the future. Consider
the matrix in b). If this matrix came from the augmented matrix of a system of linear
equa ons, then we can readily recognize that the solu on of the system is x1 = 1 and
x2 = 2. Again, in previous examples, when we found the solu on to a linear system,
we were unwi ngly pu ng our matrices into reduced row echelon form. .

We began this sec on discussing how we can manipulate the entries in a matrix
with elementary row opera ons. This led to two ques ons, “Where do we go?” and
“How do we get there quickly?” We’ve just answered the first ques on: most of the
me we are “going to” reduced row echelon form. We now address the second ques-
on.
There is no one “right” way of using these opera ons to transform a matrix into
reduced row echelon form. However, there is a general technique that works very well
in that it is very efficient (so we don’t waste me on unnecessary steps). This technique
is called Gaussian elimina on. It is named in honor of the great mathema cian Karl
Friedrich Gauss.
While this technique isn’t very difficult to use, it is one of those things that is eas-
ier understood by watching it being used than explained as a series of steps. With this
in mind, we will go through one more example highligh ng important steps and then
we’ll explain the procedure in detail.

. Example 4 .Put the augmented matrix of the following system of linear equa-
ons into reduced row echelon form.

−3x1 − 3x2 + 9x3 = 12

2x1 + 2x2 − 4x3 = −2
−2x2 − 4x3 = −8

S We start by conver ng the linear system into an augmented ma-

trix.
 
−3 −3 9 12
 2 2 −4 −2 
0 −2 −4 −8

Our next step is to change the entry in the box to a 1. To do this, let’s mul ply row
1 by − 13 .

 
1 1 −3 −4
− 13 R1 → R1  2 2 −4 −2 
0 −2 −4 −8

15
Chapter 1 Systems of Linear Equa ons

We have now created a leading 1; that is, the first entry in the first row is a 1.
Our next step is to put zeros under this 1. To do this, we’ll use the elementary row
opera on given below.
 
1 1 −3 −4
−2R1 + R2 → R2 0 0 2 6 
0 −2 −4 −8

Once this is accomplished, we shi our focus from the leading one down one row,
and to the right one column, to the posi on that is boxed. We again want to put a 1
in this posi on. We can use any elementary row opera ons, but we need to restrict
ourselves to using only the second row and any rows below it. Probably the simplest
thing we can do is interchange rows 2 and 3, and then scale the new second row so
that there is a 1 in the desired posi on.
 
1 1 −3 −4
R2 ↔ R3  0 −2 −4 −8 
0 0 2 6
 
1 1 −3 −4
− 12 R2 → R2 0 1 2 4 
0 0 2 6

We have now created another leading 1, this me in the second row. Our next
desire is to put zeros underneath it, but this has already been accomplished by our
previous steps. Therefore we again shi our a en on to the right one column and
down one row, to the next posi on put in the box. We want that to be a 1. A simple
scaling will accomplish this.
 
1 1 −3 −4
1
R → R3
2 3
0 1 2 4 
0 0 1 3

This ends what we will refer to as the forward steps. Our next task is to use the
elementary row opera ons and go back and put zeros above our leading 1s. This is
referred to as the backward steps. These steps are given below.
 
1 1 0 5
3R3 + R1 → R1  0 1 0 −2 
−2R3 + R2 → R2
0 0 1 3
 
1 0 0 7
−R2 + R1 → R1 0 1 0 −2 
0 0 1 3

It is now easy to read off the solu on as x1 = 7, x2 = −2 and x3 = 3. .

16
 1.3 Elementary Row Opera ons and Gaussian Elimina on

We now formally explain the procedure used to find the solu on above. As you
read through the procedure, follow along with the example above so that the expla-
na on makes more sense.

Forward Steps

1. Working from le to right, consider the first column that isn’t all zeros that hasn’t
already been worked on. Then working from top to bo om, consider the first
row that hasn’t been worked on.

2. If the entry in the row and column that we are considering is zero, interchange
rows with a row below the current row so that that entry is nonzero. If all entries
below are zero, we are done with this column; start again at step 1.

3. Mul ply the current row by a scalar to make its first entry a 1 (a leading 1).

4. Repeatedly use Elementary Row Opera on 1 to put zeros underneath the lead-
ing one.

5. Go back to step 1 and work on the new rows and columns un l either all rows
or columns have been worked on.

If the above steps have been followed properly, then the following should be true
about the current state of the matrix:

1. The first nonzero entry in each row is a 1 (a leading 1).

2. Each leading 1 is in a column to the right of the leading 1s above it.

3. All rows of all zeros come at the bo om of the matrix.

Note that this means we have just put a matrix into row echelon form. The next
steps finish the conversion into reduced row echelon form. These next steps are re-
ferred to as the backward steps. These are much easier to state.

Backward Steps

1. Star ng from the right and working le , use Elementary Row Opera on 1 re-
peatedly to put zeros above each leading 1.

The basic method of Gaussian elimina on is this: create leading ones and then use
elementary row opera ons to put zeros above and below these leading ones. We can
do this in any order we please, but by following the “Forward Steps” and “Backward
Steps,” we make use of the presence of zeros to make the overall computa ons easier.
This method is very efficient, so it gets its own name (which we’ve already been using).

17
Chapter 1 Systems of Linear Equa ons

.
Defini on 4 Gaussian Elimina on

Gaussian elimina on is the technique for finding the re-

duced row echelon form of a matrix using the above pro-
cedure. It can be abbreviated to:

1. Create a leading 1. .
2. Use this leading 1 to put zeros underneath it.

3. Repeat the above steps un l all possible rows have

leading 1s.

4. Put zeros above these leading 1s.

Let’s prac ce some more.

. Example 5 .Use Gaussian elimina on to put the matrix A into reduced row ech-
elon form, where  
−2 −4 −2 −10 0
A= 2 4 1 9 −2  .
3 6 1 13 −4

S We start by wan ng to make the entry in the first column and first
row a 1 (a leading 1). To do this we’ll scale the first row by a factor of − 12 .
 
1 2 1 5 0
− 21 R1 → R1  2 4 1 9 −2 
3 6 1 13 −4
Next we need to put zeros in the column below this newly formed leading 1.
 
1 2 1 5 0
−2R1 + R2 → R2  0 0 −1 −1 −2 
−3R1 + R3 → R3
0 0 −2 −2 −4
Our a en on now shi s to the right one column and down one row to the posi on
indicated by the box. We want to put a 1 in that posi on. Our only op ons are to either
scale the current row or to interchange rows with a row below it. However, in this case
neither of these op ons will accomplish our goal. Therefore, we shi our a en on to
the right one more column.
We want to put a 1 where there is a –1. A simple scaling will accomplish this; once
done, we will put a 0 underneath this leading one.
 
1 2 1 5 0
−R2 → R2 0 0 1 1 2 
0 0 −2 −2 −4

18
 1.3 Elementary Row Opera ons and Gaussian Elimina on

 
1 2 1 5 0
2R2 + R3 → R3 0 0 1 1 2
0 0 0 0 0
Our a en on now shi s over one more column and down one row to the posi on
indicated by the box; we wish to make this a 1. Of course, there is no way to do this,
so we are done with the forward steps.
Our next goal is to put a 0 above each of the leading 1s (in this case there is only
one leading 1 to deal with).
 
1 2 0 4 −2
−R2 + R1 → R1 0 0 1 1 2 
0 0 0 0 0
This final matrix is in reduced row echelon form. .

. Example 6 Put the matrix

 
1 2 1 3
2 1 1 1
3 3 2 1
into reduced row echelon form.

S Here we will show all steps without explaining each one.

 
1 2 1 3
−2R1 + R2 → R2  0 −3 −1 −5 
−3R1 + R3 → R3
0 −3 −1 −8
 
1 2 1 3
− 13 R2 → R2  0 1 1/3 5/3 
0 −3 −1 −8
 
1 2 1 3
3R2 + R3 → R3  0 1 1/3 5/3 
0 0 0 −3
 
1 2 1 3
− 13 R3 → R3  0 1 1/3 5/3 
0 0 0 1
 
1 2 1 0
−3R3 + R1 → R1
 0 1 1/3 0 
− 53 R3 + R2 → R2
0 0 0 1
 
1 0 1/3 0
−2R2 + R1 → R1  0 1 1/3 0 
0 0 0 1
.

19
Chapter 1 Systems of Linear Equa ons

The last matrix in the above example is in reduced row echelon form. If one thinks
of the original matrix as represen ng the augmented matrix of a system of linear equa-
ons, this final result is interes ng. What does it mean to have a leading one in the
last column? We’ll figure this out in the next sec on.

. Example 7 .Put the matrix A into reduced row echelon form, where
 
2 1 −1 4
A =  1 −1 2 12  .
2 2 −1 9

S We’ll again show the steps without explana on, although we will
stop at the end of the forward steps and make a comment.
 
1 1/2 −1/2 2
1
R → R1
2 1
 1 −1 2 12 
2 2 −1 9
 
1 1/2 −1/2 2
−R1 + R2 → R2  0 −3/2 5/2 10 
−2R1 + R3 → R3
0 1 0 5
 
1 1/2 −1/2 2
− 23 R2 → R2 0 1 −5/3 −20/3 
0 1 0 5
 
1 1/2 −1/2 2
−R2 + R3 → R3 0 1 −5/3 −20/3 
0 0 5/3 35/3
 
1 1/2 −1/2 2
3
R → R3
5 3
0 1 −5/3 −20/3 
0 0 1 7
Let’s take a break here and think about the state of our linear system at this mo-
ment. Conver ng back to linear equa ons, we now know

x1 + 1/2x2 − 1/2x3 = 2
x2 − 5/3x3 = −20/3 .
x3 = 7
Since we know that x3 = 7, the second equa on turns into

x2 − (5/3)(7) = −20/3,

telling us that x2 = 5.
Finally, knowing values for x2 and x3 lets us subs tute in the first equa on and find

x1 + (1/2)(5) − (1/2)(7) = 2,

20
 1.3 Elementary Row Opera ons and Gaussian Elimina on

so x1 = 3.

This process of subs tu ng known values back into other equa ons is called back
subs tu on. This process is essen ally what happens when we perform the backward
steps of Gaussian elimina on. We make note of this below as we finish out finding the
reduced row echelon form of our matrix.
 
5
R + R2 → R2
3 3
1 1/2 −1/2 2
(knowing x3 = 7 allows us 0 1 0 5
to find x2 = 5) 0 0 1 7
1
R + R1 → R1  
2 3 1 0 0 3
− 12 R2 + R1 → R1 0 1 0 5
(knowing x2 = 5 and x3 =7
0 0 1 7
allows us to find x1 = 3)

We did our opera ons slightly “out of order” in that we didn’t put the zeros above
our leading 1 in the third column in the same step, highligh ng how back subs tu on
works. .

In all of our prac ce, we’ve only encountered systems of linear equa ons with ex-
actly one solu on. Is this always going to be the case? Could we ever have systems
with more than one solu on? If so, how many solu ons could there be? Could we
have systems without a solu on? These are some of the ques ons we’ll address in
the next sec on.

Exercises 1.3
 
In Exercises 1 – 4, state whether or not the 1 0 0
given matrices are in reduced row echelon (c)  0 0 1
form. If it is not, state why. 0 0 0
 
[ ] [ ] 1 0 0 −5
1. (a)
1 0
(c)
1 1 (d)  0 1 0 7 
0 1 1 1 0 0 1 3
[ ] [ ]
(b)
0 1
(d)
1 0 1  
1 0 0 1 2 2 0 0 2
[ ] [ ] 4. (a) 0 2 0 2
1 0 0 0 0 0 0 0 2 2
2. (a) (c)
0 0 1 1 0 0  
[ ] [ ] 0 1 0 0
1 0 1 0 0 0 0
(b) (d) (b) 0 1 0
0 1 1 0 0 0
0 0 0 0
   
1 1 1 0 0 1 −5
3. (a)  0 1 1 (c) 0 0 0 0 
0 0 1 0 0 0 0
   
1 0 0 1 1 0 0 1 1
(b)  0 1 0 (d) 0 0 1 0 1 1
0 0 0 0 0 0 1 0 0

21
Chapter 1 Systems of Linear Equa ons

 
In Exercises 5 – 22, use Gaussian Elimina on 1 2 1
to put the given matrix into reduced row ech- 15.  1 3 1
elon form. −1 −3 0
[ ]
1 2  
5. 1 2 3
−3 −5
[ ] 16.  0 4 5
2 −2 1 6 9
6.
3 −2
[ ]  
4 12 1 1 1 2
7. 17.  2 −1 −1 1
−2 −6
[ ] −1 1 1 0
−5 7
8.  
10 14
[ ] 2 −1 1 5
−1 1 4 18.  3 1 6 −1 
9.
−2 1 1 3 0 5 0
[ ]
7 2 3  
10.
3 1 2 1 1 −1 7
[ ] 19.  2 1 0 10 
3 −3 6 3 2 −1 17
11.
−1 1 −2
[ ]  
4 5 −6 4 1 8 15
12.
−12 −15 18 20.  1 1 2 7 
  3 1 5 11
−2 −4 −8
13.  −2 −3 −5  [ ]
2 2 1 3 1 4
2 3 6 21.
1 1 1 3 1 4
 
2 1 1 [ ]
14.  1 1 1  1 −1 3 1 −2 9
22.
2 1 2 2 −2 6 1 −2 13

1.4 Existence and Uniqueness of Solu ons

.
AS YOU READ ...
.

1. T/F: It is possible for a linear system to have exactly 5 solu ons.

2. T/F: A variable that corresponds to a leading 1 is “free.”

3. How can one tell what kind of solu on a linear system of equa ons has?

4. Give an example (different from those given in the text) of a 2 equa on, 2 un-
known linear system that is not consistent.

5. T/F: A par cular solu on for a linear system with infinite solu ons can be found
by arbitrarily picking values for the free variables.

22
 1.4 Existence and Uniqueness of Solu ons

So far, whenever we have solved a system of linear equa ons, we have always
found exactly one solu on. This is not always the case; we will find in this sec on that
some systems do not have a solu on, and others have more than one.
We start with a very simple example. Consider the following linear system:

x − y = 0.

There are obviously infinite solu ons to this system; as long as x = y, we have a so-
lu on. We can picture all of these solu ons by thinking of the graph of the equa on
y = x on the tradi onal x, y coordinate plane.
Let’s con nue this visual aspect of considering solu ons to linear systems. Con-
sider the system

x+y=2
x − y = 0.

Each of these equa ons can be viewed as lines in the coordinate plane, and since their
slopes are different, we know they will intersect somewhere (see Figure 1.1 (a)). In
this example, they intersect at the point (1, 1) – that is, when x = 1 and y = 1, both
equa ons are sa sfied and we have a solu on to our linear system. Since this is the
only place the two lines intersect, this is the only solu on.
Now consider the linear system

x+y=1
2x + 2y = 2.

It is clear that while we have two equa ons, they are essen ally the same equa on;
the second is just a mul ple of the first. Therefore, when we graph the two equa ons,
we are graphing the same line twice (see Figure 1.1 (b); the thicker line is used to
represent drawing the line twice). In this case, we have an infinite solu on set, just as
if we only had the one equa on x + y = 1. We o en write the solu on as x = 1 − y to
demonstrate that y can be any real number, and x is determined once we pick a value
for y.

(a) (b) (c)

. linear equa ons with two unknowns.
Figure 1.1: The three possibili es for two

23
Chapter 1 Systems of Linear Equa ons

Finally, consider the linear system

x+y=1
x + y = 2.

We should immediately spot a problem with this system; if the sum of x and y is 1,
how can it also be 2? There is no solu on to such a problem; this linear system has no
solu on. We can visualize this situa on in Figure 1.1 (c); the two lines are parallel and
never intersect.
If we were to consider a linear system with three equa ons and two unknowns, we
could visualize the solu on by graphing the corresponding three lines. We can picture
that perhaps all three lines would meet at one point, giving exactly 1 solu on; per-
haps all three equa ons describe the same line, giving an infinite number of solu ons;
perhaps we have different lines, but they do not all meet at the same point, giving
no solu on. We further visualize similar situa ons with, say, 20 equa ons with two
variables.
While it becomes harder to visualize when we add variables, no ma er how many
equa ons and variables we have, solu ons to linear equa ons always come in one of
three forms: exactly one solu on, infinite solu ons, or no solu on. This is a fact that
we will not prove here, but it deserves to be stated.

.
Theorem 1 Solu on Forms of Linear Systems
.
Every linear system of equa ons has exactly one solu on,
infinite solu ons, or no solu on.

This leads us to a defini on. Here we don’t differen ate between having one so-
lu on and infinite solu ons, but rather just whether or not a solu on exists.

.
Defini on 5 Consistent and Inconsistent Linear Systems

.
A system of linear equa ons is consistent if it has a solu on
(perhaps more than one). A linear system is inconsistent if
it does not have a solu on.

How can we tell what kind of solu on (if one exists) a given system of linear equa-
ons has? The answer to this ques on lies with properly understanding the reduced
row echelon form of a matrix. To discover what the solu on is to a linear system, we
first put the matrix into reduced row echelon form and then interpret that form prop-
erly.
Before we start with a simple example, let us make a note about finding the re-

24
 1.4 Existence and Uniqueness of Solu ons

duced row echelon form of a matrix.

Technology Note: In the previous sec on, we learned how to find the reduced row
echelon form of a matrix using Gaussian elimina on – by hand. We need to know how
to do this; understanding the process has benefits. However, actually execu ng the
process by hand for every problem is not usually beneficial. In fact, with large systems,
compu ng the reduced row echelon form by hand is effec vely impossible. Our main
concern is what “the rref” is, not what exact steps were used to arrive there. Therefore,
the reader is encouraged to employ some form of technology to find the reduced row
echelon form. Computer programs such as Mathema ca, MATLAB, Maple, and Derive
can be used; many handheld calculators (such as Texas Instruments calculators) will
perform these calcula ons very quickly.
As a general rule, when we are learning a new technique, it is best to not use
technology to aid us. This helps us learn not only the technique but some of its “inner
workings.” We can then use technology once we have mastered the technique and are
now learning how to use it to solve problems.
From here on out, in our examples, when we need the reduced row echelon form
of a matrix, we will not show the steps involved. Rather, we will give the ini al matrix,
then immediately give the reduced row echelon form of the matrix. We trust that the
reader can verify the accuracy of this form by both performing the necessary steps by
hand or u lizing some technology to do it for them.

Our first example explores officially a quick example used in the introduc on of
this sec on.

. Example 8 .Find the solu on to the linear system

x1 + x2 = 1
.
2x1 + 2x2 = 2

S Create the corresponding augmented matrix, and then put the ma-
trix into reduced row echelon form.
[ ] [ ]
1 1 1 −→ 1 1 1
rref
2 2 2 0 0 0
Now convert the reduced matrix back into equa ons. In this case, we only have
one equa on,
x1 + x2 = 1
or, equivalently,

x1 = 1 − x2
x2 is free.

We have just introduced a new term, the word free. It is used to stress that idea
that x2 can take on any value; we are “free” to choose any value for x2 . Once this value

25
Chapter 1 Systems of Linear Equa ons

is chosen, the value of x1 is determined. We have infinite choices for the value of x2 ,
so therefore we have infinite solu ons.
For example, if we set x2 = 0, then x1 = 1; if we set x2 = 5, then x1 = −4. .

Let’s try another example, one that uses more variables.

. Example 9 Find the solu on to the linear system

x2 − x3 = 3
x1 + 2x3 = 2 .
−3x2 + 3x3 = −9

S To find the solu on, put the corresponding matrix into reduced
row echelon form.
   
0 1 −1 3 1 0 2 2
1 0 −→
2 2  rref 0 1 −1 3 
0 −3 3 −9 0 0 0 0
Now convert this reduced matrix back into equa ons. We have

x1 + 2x3 = 2
x2 − x3 = 3

or, equivalently,

x1 = 2 − 2x3
x2 = 3 + x3
x3 is free.

These two equa ons tell us that the values of x1 and x2 depend on what x3 is. As
we saw before, there is no restric on on what x3 must be; it is “free” to take on the
value of any real number. Once x3 is chosen, we have a solu on. Since we have infinite
choices for the value of x3 , we have infinite solu ons.
As examples, x1 = 2, x2 = 3, x3 = 0 is one solu on; x1 = −2, x2 = 5, x3 = 2 is
another solu on. Try plugging these values back into the original equa ons to verify
that these indeed are solu ons. (By the way, since infinite solu ons exist, this system
of equa ons is consistent.) .

In the two previous examples we have used the word “free” to describe certain
variables. What exactly is a free variable? How do we recognize which variables are
free and which are not?
Look back to the reduced matrix in Example 8. No ce that there is only one leading
1 in that matrix, and that leading 1 corresponded to the x1 variable. That told us that
x1 was not a free variable; since x2 did not correspond to a leading 1, it was a free
variable.

26
 1.4 Existence and Uniqueness of Solu ons

Look also at the reduced matrix in Example 9. There were two leading 1s in that
matrix; one corresponded to x1 and the other to x2 . This meant that x1 and x2 were
not free variables; since there was not a leading 1 that corresponded to x3 , it was a
free variable.
We formally define this and a few other terms in this following defini on.

.
Defini on 6 Dependent and Independent Variables

Consider the reduced row echelon form of an augmented

matrix of a linear system of equa ons. Then:
.
a variable that corresponds to a leading 1 is a basic, or
dependent, variable, and

a variable that does not correspond to a leading 1 is a free,

or independent, variable.

One can probably see that “free” and “independent” are rela vely synonymous. It
follows that if a variable is not independent, it must be dependent; the word “basic”
comes from connec ons to other areas of mathema cs that we won’t explore here.
These defini ons help us understand when a consistent system of linear equa ons
will have infinite solu ons. If there are no free variables, then there is exactly one
solu on; if there are any free variables, there are infinite solu ons.

.
Key Idea 2 Consistent Solu on Types

A consistent linear system of equa ons will have exactly

one solu on if and only if there is a leading 1 for each
variable in the system.

.
If a consistent linear system of equa ons has a free variable,
it has infinite solu ons.

If a consistent linear system has more variables than leading

1s, then the system will have infinite solu ons.

A consistent linear system with more variables than equa-

ons will always have infinite solu ons.

Note: Key Idea 2 applies only to consistent systems. If a system is inconsistent,

27
Chapter 1 Systems of Linear Equa ons

then no solu on exists and talking about free and basic variables is meaningless.

When a consistent system has only one solu on, each equa on that comes from
the reduced row echelon form of the corresponding augmented matrix will contain
exactly one variable. If the consistent system has infinite solu ons, then there will be
at least one equa on coming from the reduced row echelon form that contains more
than one variable. The “first” variable will be the basic (or dependent) variable; all
others will be free variables.
We have now seen examples of consistent systems with exactly one solu on and
others with infinite solu ons. How will we recognize that a system is inconsistent?
Let’s find out through an example.

. Example 10 Find the solu on to the linear system

x1 + x2 + x3 = 1
x1 + 2x2 + x3 = 2 .
2x1 + 3x2 + 2x3 = 0

S We start by pu ng the corresponding matrix into reduced row

echelon form.
   
1 1 1 1 1 0 1 0
1 2 −→
1 2 rref 0 1 0 0
2 3 2 0 0 0 0 1
Now let us take the reduced matrix and write out the corresponding equa ons.
The first two rows give us the equa ons

x1 + x3 = 0
x2 = 0.

So far, so good. However the last row gives us the equa on

0x1 + 0x2 + 0x3 = 1

or, more concisely, 0 = 1. Obviously, this is not true; we have reached a contradic on.
Therefore, no solu on exists; this system is inconsistent. .

In previous sec ons we have only encountered linear systems with unique solu-
ons (exactly one solu on). Now we have seen three more examples with different
solu on types. The first two examples in this sec on had infinite solu ons, and the
third had no solu on. How can we tell if a system is inconsistent?
A linear system will be inconsistent only when it implies that 0 equals 1. We can
tell if a linear system implies this by pu ng its corresponding augmented matrix into
reduced row echelon form. If we have any row where all entries are 0 except for the
entry in the last column, then the system implies 0=1. More succinctly, if we have a

28
 1.4 Existence and Uniqueness of Solu ons

leading 1 in the last column of an augmented matrix, then the linear system has no
solu on.

.
Key Idea 3 Inconsistent Systems of Linear Equa ons

.
A system of linear equa ons is inconsistent if the reduced
row echelon form of its corresponding augmented matrix
has a leading 1 in the last column.

. Example 11 Confirm that the linear system

x + y = 0
2x + 2y = 4

has no solu on.

S We can verify that this system has no solu on in two ways. First,
let’s just think about it. If x + y = 0, then it stands to reason, by mul plying both sides
of this equa on by 2, that 2x + 2y = 0. However, the second equa on of our system
says that 2x + 2y = 4. Since 0 ̸= 4, we have a contradic on and hence our system has
no solu on. (We cannot possibly pick values for x and y so that 2x + 2y equals both 0
and 4.)
Now let us confirm this using the prescribed technique from above. The reduced
row echelon form of the corresponding augmented matrix is
[ ]
1 1 0
.
0 0 1

We have a leading 1 in the last column, so therefore the system is inconsistent. .

Let’s summarize what we have learned up to this point. Consider the reduced row
echelon form of the augmented matrix of a system of linear equa ons.3 If there is a
leading 1 in the last column, the system has no solu on. Otherwise, if there is a leading
1 for each variable, then there is exactly one solu on; otherwise (i.e., there are free
variables) there are infinite solu ons.
Systems with exactly one solu on or no solu on are the easiest to deal with; sys-
tems with infinite solu ons are a bit harder to deal with. Therefore, we’ll do a li le
more prac ce. First, a defini on: if there are infinite solu ons, what do we call one of
those infinite solu ons?

3 That sure seems like a mouthful in and of itself. However, it boils down to “look at the reduced form of

the usual matrix.”

29