MATH2033 (2023–2024)

Introduction to Scientific Computation

Linear Systems: Direct Methods II

PROBLEM SHEET SOLUTIONS

Solution to Problem 1
a We start with the augmented matrix

1 −5
 
1 7
 10 0 20 6  .
5 0 −1 4
10
We perform R2 − 1
R1 → R2 and R3 − 51 R1 → R3 to obtain

1 −5
 
1 7
 0 50 10 −64  .
0 25 −6 −31
25
We perform R3 − R
50 2
→ R3 to obtain

1 −5
 
1 7
 0 50 10 −64  .
0 0 −26 97

b We start with the augmented matrix

1 −5
 
1 7
 10 0 20 6  .
5 0 −1 4
Since max{|1|, |10|, |5|} = 10, we perform R1 ↔ R2 to obtain
 
10 0 20 6
 1 −5 1 7 .
5 0 −1 4
1 5
We perform R2 − R
10 1
→ R2 and R3 − 10 R1 → R3 to obtain
 
10 0 20 6
 0 −5 −1 6.4  .
0 0 −11 1

Page 1 of 9
 Since max{| − 5|, |0|} = 5, we do not interchange rows. We perform
0
R3 − −5 R2 → R3 to obtain
 
10 0 20 6
 0 −5 −1 6.4  .
0 0 −11 1

c We start with the augmented matrix

1 −5
 
1 7
 10 0 20 6  .
5 0 −1 4

The scale factors are s1 = max{|1|, | − 5|, |1|} = 5, s2 =

max{|10|,
n |0|, |20|}
o = 20 and s3 = max{|5|, |0|, | − 1|} = 5. Since
|1| |10| |5| |5|
max 5
, 20 , 5 = 5
, we perform R1 ↔ R3 to obtain

0 −1 4
 
5
 10 0 20 6 
1 −5 1 7
10
and note that now s1 = 5, s2 = 20 and s3 = 5. We perform R2 − 5
R1 →
R2 and R3 − 15 R1 → R3 to obtain

0 −1
 
5 4
 0 0 22 −2  .
0 −5 1.2 6.2
n o
|0| |−5| |−5|
Since max 20
, 5 = 5
, we perform R2 ↔ R3 to obtain

0 −1
 
5 4
 0 −5 1.2 6.2 
0 0 22 −2
0
and note that now s2 = 5 and s3 = 20. We perform R3 − R
−5 2
→ R3 to
obtain
0 −1
 
5 4
 0 −5 1.2 6.2  .
0 0 22 −2

d We start with the augmented matrix

1 −5
 
1 7
 10 0 20 6  .
5 0 −1 4

Page 2 of 9
Since max{|1|, | − 5|, |1|, |10|, |0|, |20|, |5|, |0|, | − 1|} = 20, we perform
R1 ↔ R2 to obtain  
10 0 20 6
 1 −5 1 7 
5 0 −1 4
and then perform C1 ↔ C3 to obtain
 
20 0 10 6
 1 −5 1 7 
−1 0 5 4

and note that the solution to the linear system corresponding to this aug-
mented matrix is  
x3
 x2  .
x1
1 −1
We perform R2 − R
20 1
→ R2 and R3 − R
20 1
→ R3 to obtain
 
20 0 10 6
 0 −5 0.5 6.7  .
0 0 5.5 4.3

Since max {| − 5|, |0.5|, |0|, |5.5|} = 5.5, we perform R2 ↔ R3 to obtain

 
20 0 10 6
 0 0 5.5 4.3 
0 −5 0.5 6.7

and then perform C2 ↔ C3 to obtain

 
20 10 0 6
 0 5.5 0 4.3 
0 0.5 −5 6.7

and note that the solution to the linear system corresponding to this aug-
mented matrix is  
x3
 x1  .
x2
0.5
We perform R3 − R
5.5 2
→ R3 to obtain
 
20 10 0 6
 0 5.5 0 4.3 .
0 0 −5 6.30909 . . .

Page 3 of 9
Solution to Problem 2
a We start with the augmented matrix

1 −1 1
 
1
 1 1 4 2 .
2 −1 2 3

We perform R2 − 11 R1 → R2 and R3 − 21 R1 → R3 to obtain

1 −1 1
 
1
 0 0 5 1 .
0 −3 4 1
We perform R2 ↔ R3 to obtain

1 −1 1
 
1
 0 −3 4 1 .
0 0 5 1
0
We perform R3 − R
−3 2
→ R3 to obtain

1 −1 1
 
1
 0 −3 4 1 .
0 0 5 1

b We start with the augmented matrix

1 −1 1
 
1
 1 1 4 2 .
2 −1 2 3

Since max{|1|, |1|, |2|} = 2, we perform R1 ↔ R3 to obtain

2 −1
 
2 3
 1 1 4 2 .
1 1 −1 1

We perform R2 − 12 R1 → R2 and R3 − 12 R1 → R3 to obtain

2 −1
 
2 3
 0 1.5 3 0.5  .
0 1.5 −2 −0.5

Since max{|1.5|, |1.5|} = 1.5, we do not interchange rows. We perform

R3 − 11 R2 → R3 to obtain

2 −1
 
2 3
 0 1.5 5 3.5  .
0 0 −5 −1

Page 4 of 9
c We start with the augmented matrix

1 −1 1
 
1
 1 1 4 2 .
2 −1 2 3

The scale factors are s1 = max{|1|, |1|, | − 1|} = 1, s2 =

n |1|, |4|}o = 4 and s3 = max{|2|, | − 1|, |2|} = 2. Since
max{|1|,
|1| |1| |2| |1|
max 1
, 4, 2 = 1
, we do not interchange rows. We perform R2 −
1 2
R
1 1
→ R2 and R3 − R → R3 to obtain
1 1

1 −1 1
 
1
 0 0 5 1 .
0 −3 4 1
n o
|0| |−3| |−3|
Since max 4
, 2 = 2
, we perform R2 ↔ R3 to obtain

1 −1 1
 
1
 0 −3 4 1 
0 0 5 1
0
and note that now s2 = 2 and s3 = 4. We perform R3 − R
−3 2
→ R3 to
obtain
1 −1 1
 
1
 0 −3 4 1 .
0 0 5 1

d We start with the augmented matrix

1 −1 1
 
1
 1 1 4 2 .
2 −1 2 3

Since max{|1|, |1|, | − 1|, |1|, |1|, |4|, |2|, | − 1|, |2|} = 4, we perform R1 ↔
R2 to obtain  
1 1 4 2
 1 1 −1 1 
2 −1 2 3
and then perform C1 ↔ C3 to obtain
 
4 1 1 2
 −1 1 1 1 
2 −1 2 3

Page 5 of 9
 and note that the solution to the linear system corresponding to this aug-
mented matrix is  
x3
 x2  .
x1
−1
We perform R2 − 4
R1 → R2 and R3 − 24 R1 → R3 to obtain
 
4 1 1 2
 0 1.25 1.25 1.5  .
0 −1.5 1.5 2

Since max {|1.25|, |1.25|, | − 1.5|, |1.5|} = 1.5, we perform R2 ↔ R3 to

obtain  
4 1 1 2
 0 −1.5 1.5 2  .
0 1.25 1.25 1.5
1.25
We perform R3 − R
−1.5 2
→ R3 to obtain
 
4 1 1 2
 0 −1.5 1.5 2 .
0 0 2.5 3.16666 . . .

Solution to Problem 3
a We start with the augmented matrix
 
0.03 58.9 59.2
.
5.31 −6.10 47.0

Using 3-digit arithmetic with rounding yields

5.31 0.03 = 177,

−6.10 177⊗58.9 = −6.10 f l(10425.3) = −6.10 10400 = f l(−10431.4) = −10400

and

47.0 177⊗59.2 = 47.0 f l(10478.4) = 47.0 10500 = f l(−10453) = −10500.

So performing R2 − 5.31 R → R2 using 3-digit arithmetic with rounding

0.03 1
yields  
0.03 58.9 59.2
.
0 −10400 −10500
Performing backward substitution using 3-digit arithmetic with rounding
yields
x2 = (−10500) (−10400) = f l(1.009) = 1.01

Page 6 of 9
 and

x1 = (59.2 58.9 ⊗ x2 ) 0.03

= (59.2 f l(59.48)) 0.03
= (59.2 59.5)) 0.03
= (−0.3) 0.03
= −10.0.

b We start with the augmented matrix

 
0.03 58.9 59.2
.
5.31 −6.10 47.0

Since max{|0.03|, |5.31|} = 5.31, we perform R1 ↔ R2 to obtain

 
5.31 −6.10 47.0
.
0.03 58.9 59.2

Using 3-digit arithmetic with rounding yields

0.03 5.31 = f l(0.005649) = 0.00565,

58.9 0.00565⊗(−6.10) = 58.9 f l(−0.03446) = 58.9 (−0.0345) = f l(58.93) = 58.9

and

59.2 0.00565⊗47.0 = 59.2 f l(0.2655) = 59.2 0.266 = f l(58.93) = 58.9

0.03
So performing R2 − R
5.31 1
→ R2 using 3-digit arithmetic with rounding
yields  
5.31 −6.10 47.0
.
0 58.9 58.9
Performing backward substitution using 3-digit arithmetic with rounding
yields
x2 = 58.9 58.9 = 1.00
and

x1 = (47.0 (−6.10)⊗x2 ) 5.31 = (47.0 (−6.10)) 5.31 = 53.1 5.31 = 10.0.

c We start with the augmented matrix

 
0.03 58.9 59.2
.
5.31 −6.10 47.0

Using 3-digit arithmetic with chopping yields

5.31 0.03 = 177,

Page 7 of 9
 −6.10 177⊗58.9 = −6.10 f l(10425.3) = −6.10 10400 = f l(−10406.1) = −10400
and

47.0 177⊗59.2 = 47.0 f l(10478.4) = 47.0 10400 = f l(−10353) = −10300.

So performing R2 − 5.31 R → R2 using 3-digit arithmetic with chopping

0.03 1
yields  
0.03 58.9 59.2
.
0 −10400 −10300
Performing backward substitution using 3-digit arithmetic with chopping
yields
x2 = (−10300) (−10400) = f l(0.9903) = 0.990
and

x1 = (59.2 58.9 ⊗ x2 ) 0.03

= (59.2 f l(58.31)) 0.03
= (59.2 58.3) 0.03
= 0.9 0.03
= 30.0.

d We start with the augmented matrix

 
0.03 58.9 59.2
.
5.31 −6.10 47.0

Since max{|0.03|, |5.31|} = 5.31, we perform R1 ↔ R2 to obtain

 
5.31 −6.10 47.0
.
0.03 58.9 59.2

Using 3-digit arithmetic with chopping yields

0.03 5.31 = f l(0.005649) = 0.00564,

58.9 0.00564⊗(−6.10) = 58.9 f l(−0.03440) = 58.9 (−0.0344) = f l(58.93) = 58.9

and

59.2 0.00564⊗47.0 = 59.2 f l(0.2650) = 59.2 0.265 = f l(58.93) = 58.9

0.03
So performing R2 − R
5.31 1
→ R2 using 3-digit arithmetic with chopping
yields  
5.31 −6.10 47.0
.
0 58.9 58.9

Page 8 of 9
 Performing backward substitution using 3-digit arithmetic with chopping
yields
x2 = 58.9 58.9 = 1.00
and

x1 = (47.0 (−6.10)⊗x2 ) 5.31 = (47.0 (−6.10)) 5.31 = 53.1 5.31 = 10.0.

References

• Burden, Faires & Burden, Numerical Analysis, 10E

– Section 6.2

Page 9 of 9