1

Linear Algebra
Assignment

Systems of Linear Equations

 2

Systems of Linear Equations

Linear Equations, Solutions, 2 × 2 Systems
1. Determine whether each of the following equations is linear:
(a) 5x + 7y – 8yz = 16, (b) x + πy + ez = log 5, (c) 3x + ky – 8z = 16
(a) No, because the product yz of two unknowns is of second degree.
(b) Yes, because π, e, and log 5 are constants.
(c) As it stands, there are four unknowns: x, y, z, k. Because of the term ky it is not a linear equation.
However, assuming k is a constant, the equation is linear in the unknowns x, y, z.
2. Determine whether the following vectors are solutions of x1 + 2x2 – 4x3 + 3x4 = 15:
(a) u = (3,2,1,4) and (b) ν = (1, 2, 4, 5).
(a) Substitute to obtain 3 + 2(2) – 4(1) + 3(4) = 15, or 15 = 15; yes, it is a solution.
(b) Substitute to obtain 1 + 2(2) – 4(4) + 3(5) = 15, or 4 = 15; no, it is not a solution.
3. Solve (a) ex = π, (b) 3x – 4 – x = 2x + 3, (c) 7 + 2x – 4 = 3x + 3 – x
(a) Because e ≠ 0, multiply by 1 /e to obtain x = π/e.
(b) Rewrite in standard form, obtaining 0x = 7. The equation has no solution.
(c) Rewrite in standard form, obtaining 0x = 0. Every scalar k is a solution.
4. Prove Theorem 3.4: Consider the equation ax = b.
(a) If a ≠ 0, then x = b/a is a unique solution of ax = b.
(b) If a = 0 but b ≠ 0, then ax = b has no solution.
(c) If a = 0 and b = 0, then every scalar k is a solution of ax = b.
Suppose a ≠ 0. Then the scalar b/a exists. Substituting b/a in ax = b yields a(b/a) = b, or b = b; hence, b/a
is a solution. On the other hand, suppose x0 is a solution to ax = b, so that ax0 = b. Multiplying both sides
by 1/a yields x0 = b/a. Hence, b/a is the unique solution of ax = b. Thus, (i) is proved.
On the other hand, suppose a = 0. Then, for any scalar k, we have ak = 0k = 0. If b ≠ 0, then ak ≠ b.
Accordingly, k is not a solution of ax = b, and so (ii) is proved. If b = 0, then ak = b. That is, any scalar k is
a solution of ax = b, and so (iii) is proved.
5. Solve each of the following systems:
2x − 5 y =
11 2x − 3y = 8 2x − 3y = 8
(a) (b) (c)
3x + 4 y =
5 −6 x + 9 y = 6 −4 x + 6 y = −16
(a) Eliminate x from the equations by forming the new equation L = –3L1 + 2L2. This yields the equation
23y = –23, and so y = –1
Substitute y = – 1 in one of the original equations, say L1, to get
2x – 5(–1) = 11 or 2x + 5 = 11 or 2x = 6 or x = 3
Thus, x = 3, y = – 1 or the pair u = (3, –1) is the unique solution of the system.
 3

(b) Eliminate x from the equations by forming the new equation L = 3L1, + L2. This yields the equation
0x + 0y = 30
This is a degenerate equation with a nonzero constant; hence, this equation and the system have no
solution. (Geometrically, the lines corresponding to the equations are parallel.)
(c) Eliminate x from the equations by forming the new equation L = 2L1 + L2. This yields the equation
0x + 0y = 0
This is a degenerate equation where the constant term is also zero. Thus, the system has an infinite
number of solutions, which correspond to the solution of either equation. (Geometrically, the lines
corresponding to the equations coincide.)
To find the general solution, set y = a and substitute in L1 to obtain
3
2x – 3a = 8 or 2x = 3a + 8 or x = a+4
2
Thus, the general solution is
3 3 
x
= a + 4 , y = a or
= u  a + 4, a 
2 2 
where a is any scalar.
6. Consider the system
x + ay = 4
ax + 9y = b
(a) For which values of a does the system have a unique solution?
(b) Find those pairs of values (a, b) for which the system has more than one solution.
(a) Eliminate x from the equations by forming the new equation L = –aL1 + L2. This yields the equation
(9 – a2)y = b – 4a (1)
The system has a unique solution if and only if the coefficient of y in (1) is not zero—that is, if 9 – a2
≠ 0 or if a ≠ ±3.
(b) The system has more than one solution if both sides of (1) are zero. The left-hand side is zero when a
= ±3. When a = 3, the right-hand side is zero when b – 12 = 0 or b = 12. When a = –3, the right- hand
side is zero when b + 12 – 0 or b = –12. Thus, (3,12) and (–3, –12) are the pairs for which the system
has more than one solution.

Systems in Triangular and Echelon Form

7. Determine the pivot and free variables in each of the following systems:
2x1 – 3x2 — 6x3 – 5x4 + 2x5 = 7 2x – 6y + 7z = 1 x + 2y – 3z = 2
x3 + 3x4 – 7x5 = 6 4y + 3z = 8 2x + 3y + z = 4
x4 – 2x5 = 1 2z = 4 3x + 4y + 5z = 8
(a) (b) (c)
 4

(a) In echelon form, the leading unknowns are the pivot variables, and the others are the free variables.
Here x1, x3, x4 are the pivot variables, and x2 and x5 are the free variables.
(b) The leading unknowns are x, y, z, so they are the pivot variables. There are no free variables (as in any
triangular system).
(c) The notion of pivot and free variables applies only to a system in echelon form.
8. Solve the triangular system in Problem 3.7(b).
Because it is a triangular system, solve by back-substitution.
(i) The last equation gives z = 2.
1
(ii) Substitute z = 2 in the second equation to get 4y + 6 = 8 or y =
2
1
(iii) Substitute z = 2 and y = in the first equation to get
2
1
1 or 2x + 11 = 1 or x = –5
2 x − 6   + 7(2) =
2
1 1
Thus, x = –5, y = , z = 2 or u = (–5, 2) is the unique solution to the system.
2 2
9. Solve the echelon system in Problem 3.7(a).
Assign parameters to the free variables, say x2 = a and x5 = b, and solve for the pivot variables by back-
substitution.
(i) Substitute x5 = b in the last equation to get x4 – 2b = 1 or x4 = 2b + 1.
(ii) Substitute x5 = b and x4 = 2b + 1 in the second equation to get
x3 + 3(2b + 1) – 7b = 6 or x3 – b + 3 = 6 or x3 = b + 3
(iii) Substitute x5 = b, x4 = 2b + 1, x3 = b + 3, x2 = a in the first equation to get
3
2x1 – 3a – 6(b + 3) – 5(2b + 1) + 2b = 7 or 2x1 – 3a – 14b – 23 = 7 or x1 = a + 7b + 15
2
Thus,
3
x1 = a + 7b + 15, x2 = a, x3 = b + 3, x4 = 2b + 1, x5 = b
2
3 
or u=  a + 7b + 15, a, b + 3, 2b + 1, b 
2 
is the parametric form of the general solution.
Alternatively, solving for the pivot variable x1, x3, x4 in terms of the free variables x2 and x5 yields the
following free-variable form of the general solution:
3
x1 = x2 + 7x5 + 15, x3 = x5 + 3, x4 = 2x5 + 1
2
 5

10. Prove Theorem 3.6. Consider the system (3.4) of linear equations in echelon form with r equations and n
unknowns.
(i) If r = n, then the system has a unique solution.
(ii) If r < n, then we can arbitrarily assign values to the n – r free variable and solve uniquely for the r
pivot variables, obtaining a solution of the system.
(i) Suppose r = n. Then we have a square system AX = B where the matrix A of coefficients is (upper)
triangular with nonzero diagonal elements. Thus, A is invertible. By Theorem 3.10, the system has a
unique solution.
(ii) Assigning values to the n – r free variables yields a triangular system in the pivot variables, which, by
(i), has a unique solution.

Gaussian Elimination
11. Solve each of the following systems:
x + 2y – 4z = –4 x + 2y – 3z = –1 x + 2y – 3z = 1
2x + 5y – 9z = –10 –3x + y – 2z = –7 2x + 5y – 8z = 4
3x – 2y + 3z = 11 5x + 3y – 4 = 2 3x + 8y – 13z = 7
(a) (b) (c)
Reduce each system to triangular or echelon form using Gaussian elimination:
(a) Apply "Replace L2 by –2L1 + L2’’ and "Replace L3 by –3L1 + L3" to eliminate x from the second and
third equations, and then apply "Replace L3 by 8L2 + L3" to eliminate y from the third equation. These
operations yield
x + 2y – 4z = –4 x + 2y – 4z = –4
y–z=–2 and then y – z = –2
–8y + 15z = 23 7z = 7
The system is in triangular form. Solve by back-substitution to obtain the unique solution u = (2, –1,
1).
(b) Eliminate x from the second and third equations by the operations "Replace L2 by 3L1 + L2" and
"Replace L3 by –5L1 + L3." This gives the equivalent system
x + 2y – 3z = –1
7y – 11z = –10
–7y + 11z = 7
The operation "Replace L3 by L2 + L3" yields the following degenerate equation with a nonzero
constant:
0x + 0y + 0z = –3
This equation and hence the system have no solution.
(c) Eliminate x from the second and third equations by the operations "Replace L2 by –2L1 + L2" and
"Replace L3 by –3L1 + L3." This yields the new system
 6

x + 2 y − 3z =1
x + 2 y − 3z =
1
y − 2z = 2 or
y − 2z = 2
2 y − 4z = 4
(The third equation is deleted, because it is a multiple of the second equation.) The system is in echelon
form with pivot variables x and y and free variable z.
To find the parametric form of the general solution, set z = a and solve for x and y by back- substitution.
Substitute z = a in the second equation to get y = 2 + 2a. Then substitute z = a and y = 2 + 2a in the
first equation to get
x + 2(2 + 2a) – 3a = l or x + 4 + a = l or x = –3 – a
Thus, the general solution is
x = –3 – a, y = 2 + 2a, z = a or u = (–3 – a, 2 + 2a, a)
where a is a parameter.
12. Solve each of the following systems:
x1 – 3x2 + 2x3 – x4 + 2x5 = 2 x1 + 2x2 – 3x3 + 4x4 = 2
3x1 – 9x2 + 7x3 – x4 + 3x5 = 7 2x1 + 5x2 – 2x3 + x4 = 1
2x1 – 6x2 + 7x3 + 4x4 – 5x5 = 7 5x1 + 12x2 – 7x3 + 6x4 = 3
(a) (b)
Reduce each system to echelon form using Gaussian elimination:
(a) Apply "Replace L2 by –3L1 + L2" and "Replace L3 by –2L1 + L3" to eliminate x from the second and
third equations. This yields
x1 − 3 x2 + 2 x3 − x4 + 2 x5 =
2
x1 − 3 x2 + 2 x3 − x4 + 2 x5 =
2
x3 + 2 x4 − 3 x5 =
1 or
x3 + 2 x4 − 3 x5 =
1
3 x3 + 6 x4 − 9 x5 =
3
(We delete L3, because it is a multiple of L2.) The system is in echelon form with pivot variables x1
and x3 and free variables x2,x4,x5.
To find the parametric form of the general solution, set x2 = a, x4 = b, x5 = c, where a, b, c are
parameters. Back-substitution yields x3 = 1 – 2b + 3c and xl = 3a + 5b – 8c. The general solution is
x1 = 3a + 5b – 8c, x2 = a, x3 = 1 – 2b + 3c, x4 = b, x5 = c
or, equivalently, u = (3a + 5b – 8c, a, 1 – 2b + 3c, b, c).
(b) Eliminate x1 from the second and third equations by the operations "Replace L2 by –2L1 + L2" and
"Replace L3 by –5L1 + L3." This yields the system
x1 + 2x2 – 3x3 + 4x4 = 2
x2 + 4x3 – 7x4 = –3
2x2 + 8x3 – 14x4 = –7
The operation "Replace L3 by –2L2 + L3" yields the degenerate equation 0 = –1. Thus, the system has
no solution (even though the system has more unknowns than equations).
 7

13. Solve using the condensed format:

2y + 3z = 3
x+y+z=4
4x + 8y – 3z = 35
The condensed format follows:
Number Equation Operation
(2) (1) 2y +3z =3 Zj ^—* Z>2

(1) (2) x + y + z= 4 Zj ^ ^ l2

(3) 4x + 8y – 3z = 35
(3') 4y – 7z = 19 Replace L3 by –4L1 + L3
(3") –13z = 13 Replace L3 by–2L2 + L3

Here (1), (2), and (3") form a triangular system. (We emphasize that the interchange of L1, and L2 is
accomplished by simply renumbering L1 and L2 as above.)
Using back-substitution with the triangular system yields z = –1 from L3, y = 3 from L2, and x = 2 from
L1. Thus, the unique solution of the system is x = 2, y = 3, z = –1 or the triple u = (2,3, –1).
14. Consider the system
x + 2y + z = 3
ay + 5z = 10
2x + 7y + az = b
(a) Find those values of a for which the system has a unique solution.
(b) Find those pairs of values (a, b) for which the system has more than one solution.
Reduce the system to echelon form. That is, eliminate x from the third equation by the operation "Replace
L3 by —2L1 + L3" and then eliminate y from the third equation by the operation 'Replace L3 by –3L2 + aL3."
This yields
x + 2y + z = 3 x + 2y + z = 3
ay + 5 z =10 and then ay + 5 z =10
3 y + (a − 2) z =b − 6 (a 2 − 2a − 15) z = ab − 6a − 30
Examine the last equation (a2 – 2a – 15) z = ab – 6a – 30.
(a) The system has a unique solution if and only if the coefficient of z is not zero; that is, if
a2 – 2a – 15 = (a – 5)(a + 3) ≠ 0 or a ≠ 5 and a ≠ –3.
(b) The system has more than one solution if both sides are zero. The left-hand side is zero when a = 5 or
a = –3. When a = 5, the right-hand side is zero when 5b – 60 = 0, or b = 12. When a = –3, the right- hand
side is zero when –3b – 12 = 0, or b = –4. Thus, (5,12) and (–3, –4) are the pairs for which the system has
more than one solution.
 8

Echelon Matrices, Row Equivalence, Row Canonical Form

15. Row reduce each of the following matrices to echelon form:
1 2 −3 0   −4 1 −6 
(a) A  2 4 −2 2 
= (b) B  1 2 −5
=
 3 6 −4 3   6 3 −4 

(a) Use a11 = 1 as a pivot to obtain 0's below a11; that is, apply the row operations "Replace R2 by
–2R1 + R2'' and ''Replace R3 by –3R1 + R3.'' Then use a23 = 4 as a pivot to obtain a 0 below a23; that
is, apply the row operation ''Replace R3 by –5R2 + 4R3.'' These operations yield
1 2 −3 0  1 2 −3 0 
A ~ 0 0 4 2  ~ 0 0 4 2 
0 0 5 3  0 0 0 2 

The matrix is now in echelon form.

(b) Hand calculations are usually simpler if the pivot element equals 1. Therefore, first interchange R1 and
R2. Next apply the operations "Replace R2 by 4R1 + R2" and "Replace R3 by –6R1 + R3"; and then apply
the operation "Replace R3 by R2 + R3." These operations yield
 1 2 −5 1 2 −5  1 2 −5 
B ~  −4 1 −6  ~ 0 9 −26  ~ 0 9 −26 
 6 3 −4  0 −9 26  0 0 0 

The matrix is now in echelon form.

16. Describe the pivoting row-reduction algorithm. Also describe the advantages, if any, of using this pivoting
algorithm.
The row-reduction algorithm becomes a pivoting algorithm if the entry in column j of greatest absolute
value is chosen as the pivot a1 j1 and if one uses the row operation

( −aij1 )
/ a1 j1 R1 + Ri → Ri

The main advantage of the pivoting algorithm is that the above row operation involves division by the
(current) pivot a1 j1 , and, on the computer, roundoff errors may be substantially reduced when one divides
by a number as large in absolute value as possible.
 2 −2 2 1 
17. Let A=  −3 6 0 −1 . Reduce A to echelon form using the pivoting algorithm
 1 −7 10 2 

First interchange R1 and R2 so that –3 can be used as the pivot, and then apply the operations "Replace R2
2 1
by R1 + R2 " and "Replace R3 by R1 + R3· " These operations yield
3 3
 9

 
 −3 6 0 −1
 −3 6 0 −1  
1
A ~  2 −2 2 1  ~  0 2 2
 3
 1 −7 10 2   
 0 −5 10 5 
 3
Now interchange R2 and R3 so that –5 can be used as the pivot, and then apply the operation ''Replace
2
R3 by R2 + R3." We obtain
5
 
 −3 6 0 −1  −3 6 0 −1
 
5   5 
A ~  0 −5 10 ~ 0 −5 10
 3  3
   
0 2 2 1 0 0 6 1
 3
The matrix has been brought to echelon form using partial pivoting.
18. Reduce each of the following matrices to row canonical form:
 2 2 −1 6 4  5 −9 6 
(a)  
A =  4 4 1 10 13  , (b) B = 0 2 3 
8 8 −1 26 23 0 0 7 

(a) First reduce A to echelon form by applying the operations "Replace R2 by –2R1 + R2" and "Replace R3
by –4R1 + R3," and then applying the operation "Replace R3 by –R2 + R3." These operations yield
 2 2 −1 6 4   2 2 −1 6 4 
A ~  0 0 3 −2 5  ~  0 0 3 −2 5 


 0 0 3 2 7   0 0 0 4 2 
Now use back-substitution on the echelon matrix to obtain the row canonical form of A. Specifically,
1
first multiply R3 by to obtain the pivot a34 = 1, and then apply the operations "Replace R2 by 2R3 +
4
R2" and "Replace R1 by –6R3 + R1." These operations yield
   
 2 2 −1 6 4   2 2 −1 0 1
   
A ~  0 0 3 −2 5  ~ 0 0 3 0 6
 1  1
0 0 0 1  0 0 0 1 
 2  2
1
Now multiply R2 by , making the pivot a23 = 1, and then apply "Replace R1 by R2 + R1,'' yielding
3
 10

   
 2 2 −1 0 1  2 2 0 0 3
   
A ~ 0 0 1 0 2  ~ 0 0 1 0 2
 1  1
0 0 0 1  0 0 0 1 
 2  2
1
Finally, multiply R1 by , so the pivot a11 = 1. Thus, we obtain the following row canonical form of
2
A:
 3
1 1 0 0 2 
 
A ~ 0 0 1 0 2 
 1
0 0 0 1 
 2
(b) Because B is in echelon form, use back-substitution to obtain
5 −9 6  5 −9 0  5 −9 0  5 0 0  1 0 0 
B ~ 0 2 3 ~ 0 2 0  ~ 0 1 0  ~ 0 1 0  ~ 0 1 0 
0 0 1  0 0 1  0 0 1  0 0 1  0 0 1 

The last matrix, which is the identity matrix I, is the row canonical form of B. (This is expected,
because B is invertible, and so its row canonical form must be I.)
19. Describe the Gauss–Jordan elimination algorithm, which also row reduces an arbitrary matrix A to its row
canonical form.
The Gauss-Jordan algorithm is similar in some ways to the Gaussian elimination algorithm, except that here
each pivot is used to place 0's both below and above the pivot, not just below the pivot, before working
with the next pivot. Also, one variation of the algorithm first normalizes each row-that is, obtains a unit
pivot-before it is used to produce 0's in the other rows, rather than normalizing the rows at the end of the
algorithm.
1 −2 3 1 2 
20. Let A = 1 1 4 −1 3  . Use Gauss-Jordan to find the row canonical form of A.
 2 5 9 −2 8 
Use a11 = 1 as a pivot to obtain 0's below a11 by applying the operations "Replace R2 by –R1 + R2" and
"Replace R3 by –2R1 + R3." This yields
1 −2 3 1 2 
A ~ 0 3 1 −2 1 
0 9 3 −4 4 

1
Multiply R2 by to make the pivot a22 = 1, and then produce 0's below and above a22 by applying the
3
operations "Replace R3 by –9R2 + R3" and "Replace R1 by 2R2 + R1." These operations yield
 11

 11 1 8
 1 0 −
1 −2 3 1 2 3 3 3
   
1 2 1  1 2 1
A ~ 0 1 − ~ 0 1 −
 3 3 3  3 3 3
0 9  
 3 −4 4  0 0 0 2 1
 
1
Finally, multiply R3 by to make the pivot a34 = 1, and then produce 0's above a34 by applying the
2
2 1
operations "Replace R2 by R3 +R2 " and "Replace R1 by R3 + R1." These operations yield
3 3
 11 1 8  11 17 
1 0 3 − 3 3   1 0
3
0
6
   
1 2 1  1 2
A ~ 0 1 − ~ 0 1 0
 3 3 3  3 3
   
0 0 0 1  1
1 0 0 0 1
 2   2 
which is the row canonical form of A.

Systems of Linear Equations in Matrix Fonn

21. Find the augmented matrix M and the coefficient matrix A of the following system:
x + 2y – 3z = 4
3y – 4z + 7x = 5
6z + 8x – 9y = 1
First align the unknowns in the system, and then use the aligned system to obtain M and A. We have
x + 2 y − 3z =4 1 2 −3 4  1 2 −3
7x + 3y − 4z = 5 then
= M 7 3 −4 5  and
= A 7 3 −4 
 
8x − 9 y + 6z =1 8 −9 6 1  8 −9 6 

22. Solve each of the following systems using its augmented matrix M:
x + 2y – 3z = 3 x – 2y + 4z = 2 x+ y + 3z = 1
x + 3y + z = 5 2x – 3y + 5z = 3 2x + 3y – z = 3
3x + 8y + 4z = 17 3x – 4y + 6z = 7 5x + 7y + z = 7
(a) (b) (c)
(a) Reduce the augmented matrix M to echelon fonn as follows:
1 2 −1 3  1 2 −1 3  1 2 −1 3 
M = 1 3 1 5  ~ 0 1 2 2  ~ 0 1 2 2 
3 8 4 17  0 2 7 8  0 0 3 4 

Now write down the corresponding triangular system

x + 2y – z = 3
 12

y + 2z = 2
3z = 4
and solve by back-substitution to obtain the unique solution
17 2 4  17 2 4 
− , z =or
x =, y = = u  ,− , 
3 3 3  3 3 3
Alternately, reduce the echelon form of M to row canonical form, obtaining
 13   17 
  1 2 0  1 0 0 3 
1 2 −1 3  3
  
   2  2
M ~ 0 1 2 2 ~ 0 1 0 − ~ 0 1 0 −
 3  3
 4    
0 0 1   4   4 
 3  0 0 1 0 0 1
 3   3 
This also corresponds to the above solution.
(b) First reduce the augmented matrix M to echelon form as follows:
1 −2 4 2  1 −2 4 2  1 −2 4 2 
M =  2 −3 5 3  ~ 0 1 −3 −1 ~ 0 1 −3 −1


 3 −4 6 7  0 2 −6 1  0 0 0 3 
The third row corresponds to the degenerate equation 0x + 0y + 0z = 3, which has no solution. Thus,
"DO NOT CONTINUE." The original system also has no solution. (Note that the echelon form
indicates whether or not the system has a solution.)
(c) Reduce the augmented matrix M to echelon form and then to row canonical form:
 1 1 3 1  1 1 3 1 
    1 0 10 0 
 2 3 −1 3  ~ 0 1 −7 1  ~ 0 1 −7 1 
M=
 
 5 7 1 7  0 2 −14 2 
(The third row of the second matrix is deleted, because it is a multiple of the second row and will result
in a zero row.) Write down the system corresponding to the row canonical form of M and then transfer
the free variables to the other side to obtain the free-variable form of the solution:
x + 10 z =0 x = −10 z
and
y − 7z = 1 y= 1+ 7z
Here z is the only free variable. The parametric solution, using z = a, is as follows:
x = –10a, y = 1 +7a, z = a or u = (–10a, 1 +7a, a)
23. Solve the following system using its augmented matrix M:
x1 + 2x2 – 3x3 – 2x4 + 4x5 = 1
2x1 + 5x2 – 8x3 – x4 + 6x5 = 4
x1 + 4x2 – 7x3 + 5x4 + 2x5 = 8
 13

Reduce the augmented matrix M to echelon form and then to row canonical form:
1 2 −3 −2 4 1  1 2 −3 −2 4 1  1 2 −3 −2 4 1 
M =  2 5 −8 −1 6 4  ~ 0 1 −2 3 −2 2  ~ 0 1 −2 3 −2 2  ~
1 4 −7 5 2 8  0 2 −4 7 −2 7  0 2 0 1 2 3 

1 2 −3 0 8 7  1 0 1 0 24 21
0 1 −2 0 −8 −7  ~ 0 1 −2 0 −8 −7 
   
0 0 0 1 2 3  0 0 0 1 2 3 

Write down the system corresponding to the row canonical form of M and then transfer the free variables
to the other side to obtain the free-variable form of the solution:
x1 + x3 + 24 x5 =
21 x1 = 21 − x3 − 24 x5
x2 − 2 x3 − −7 and x2 =−7 + 2 x3 + 8 x5
8 x5 =
x4 + 2 x5 =
3 x4 = 3 − 2 x5

