2 3
In section we showed that A =
is diagonalizable, and A = P DP 1 , where
3 6






3 0
3 1
3/10 1/10
1
D=
,P =
, and P =
.
0 7
1 3
1/10 3/10
   
3
1
2
The columns of P form a basis for R well call B, so B =
,
.
1
3
The mappings x 7 Ax and u 7 Du describe the same linear transformation,
relative to the bases S and B respectively.
 
4
Lets see what this means by considering the vector x =
.
8

  

2 3
4
32
First note that Ax =
=
3 6 8
36
Next, we know that the coordinate vector of x relative to the basis B is
   

2
3/10 1/10 4
1
.
=
[x]B = P x =
2
1/10 3/10 8

  

3 0
2
6
If we compute D[x]B we have D[x]B =
=
.
0 7 2
14





6
32
Notice that
is the coordinate vector of Ax =
relative to the basis B, since
14
36


 

3 1
6
32
Ax = P [Ax]B =
=
.
1 3
14
36
Thus the output of x 7 Ax and u 7 Du are the same, relative to different bases.
Note that D = P 1 AP , and observe that P is the change-of-coordinate matrix that converts
vectors from B to S, and P 1 converts vectors from S to B.
This means we can see how u 7 Du = P 1 AP u works in three steps: P converts the
vector from B to S, then we multiply by A, and lastly P 1 converts the resulting vector
from S back to B, as follows:
 

   
2
3 1 2
4
First, P converts [x]B =
from B to S: P [x]B =
=
= x as above.
2
1 3
2
8


  

2 3
4
32
Next multiply the result by A: Ax =
=
as above.
3 6 8
36
Lastly, P

converts Ax from S to B: P


 

3/10 1/10
32
6
Ax =
=
.
1/10 3/10 36
14