Linear Algebra

Grinshpan

Reconstructing a system from its solution

 
1
Example 1. Find a system of linear equations whose solution set is given by span .
2

Answer. Let x1 and x2 be the unknowns.

  Since
  the solution set consists of all possible
1 0 1
multiples of , the zero vector = 0· is a solution. So the system must be
2 0 2
homogeneous. How many equations are needed? Clearly, we can get away with one equation:
 
  x1
2 −1 =0
x2
or
2x1 − x2 = 0.
The solution set of this equation is exactly as prescribed. In fact, any
 number
 of equations is
possible: choose any n×2 matrix such that its rows are multiples of 2 −1 and at least one
of the rows is nonzero. Then the resulting system in two unknowns has one basic variable
and one free variable and its augmented matrix is row equivalent to
2 −1 0
 
0 0 0 
. . . .
 .. .. .. 
0 0 0
For instance,
−2x1 + x2 = 0


x1 − 0.5x2 = 0
is just as good of an answer.
   
−1 1
Example 2. Find a linear system whose solution set is given by + span .
1 2
 
0
Answer. This time the system is nonhomogeneous: the zero vector does not belong to
0
     
−1 1 t−1
the solution set of vectors +t = , because t cannot equal 1 and −1/2 at
1 2 2t + 1
the same time. Observe that the solution set has “particular + homogeneous” form and that
we already know how to deal with the homogeneous part. It remains to find b such that
 
  x1
2 −1 =b
x2
has the solution set as prescribed. Clearly,
 
  −1
b = 2 −1 = −3.
1
Thus 2x1 − x2 = −3 is a possible answer.
    
−1  1 
Example 3. Find a linear system whose solution set is given by 1 + span 1 .
 
3 1
 

Answer. The solution set does not contain the zero vector, so the system must be nonho-
mogeneous. How many equations are needed? Since there are three variables and only one
of them is free, one equation is insufficient to create two basic variables. We would need two
equations. Consider the homogeneous part first  and write the reduced augmented matrix:
  1
1 0 ∗ 0
. How to choose ∗’s? Of course, for 1 to be a solution, ∗ = −1.
0 1 ∗ 0
1
 
  1
1 0 −1 0
The solution set of the augmented matrix consists of all multiples of 1.
0 1 −1 0
1
   
b1 1 0 −1 b1
Now, for the nonhomogeneous part, we need to determine b = such that
b 0 1 −1 b2
  2 
−1   −1  
1 0 −1   −4
is satisfied by the particular solution 1 : b =
  1 = .
0 1 −1 −2
3 3
Putting things together, we obtain an answer:
x1 − x3 = −4


x2 − x3 = −2.
    
−1  1 2 
Example 4. Find a linear system whose solution set is given by 1 +span
  1 , −2 .
 
3 1 1
 

Answer. Let’s use another method. A typical solution vector is of the form
     
−1 1 2
x = 1 + s 1 + t −2 ,
    
3 1 1
for some scalars s and t. Thus 
 x1 = −1 + s + 2t

x2 = 1 + s − 2t

 x = 3 + s + t.
3

Now eliminate s and t. Observe that (x1 + 1) − (x2 − 1) = 4t and (x1 + 1) − (x3 − 3) = t. So
x1 − x2 + 2 = 4(x1 − x3 + 4)
or
3x1 + x2 − 4x3 = −14.
The preceding equation has the solution set as prescribed. Indeed, both x1 = x2 = x3 = 1
and x1 = 2, x2 = −2, x3 = 1 satisfy the homogeneous equation 3x1 + x2 − 4x3 = 0 and
x1 = −1, x2 = 1, x3 = 3 is a particular solution of 3x1 + x2 − 4x3 = −14.