[ Home ] [ Up ] [ Info ] [ Mail ]

Proof of Theorem 6].


We are given n linearly independent eigenvectors p1, p2, ... , pn of n-square matrix A and the corresponding eigenvalues λ1, λ2, ... , λn. An eigenvector is a vector X that a matrix A carries into a multiple of itself according to the equation AX = λX where λ is the corresponding eigenvalue. Thus the eigenvectors p1, p2, ... , pn and eigenvalues λ1, λ2, ... , λn must satisfy the following set of equations:


            Ap1 = λ1p1

1)        Ap2 = λ2p2

            ............

            Apn = λnpn .


Now system 1) is equivalent to


2)        A[p1 p2 ... pn] = [λ1p1 λ2p2 ... λnpn] .


Why? Because of the following theorem:


______________________________________________


Theorem 1. Suppose that the matrix A carries the vector X1 into the vector Y1, the vector X2 into the vector Y2 , etc. i.e.


                                    AX1 = Y1

                                    AX2 = Y2

                                        ......

                                        ......

                                    AXn = Yn

 


 

Then                           

 

                           A [ X1 X2 .... Xn] = [Y1 Y2 ... Yn]

                           


where [ X1 X2 .... Xn] is a matrix whose columns are X1, X2, .... , Xn and [Y1 Y2 ... Yn] is a matrix whose columns are Y1,Y2, ... ,Yn .


_____________________________________________________


Furthermore,


ole.gif





Why? Because of the following theorem:



______________________________________________


Theorem 2. The effect of post-multiplying a matrix



              ole1.gif    



by a diagonal matrix



                ole2.gif  




is that of multiplying the i-th column of matrix A by the factor ole3.gif i.e. the successive columns of the original matrix are simply multiplied by successive diagonal elements of the diagonal matrix. Explicitly:




      ole4.gif      ole5.gif                     


______________________________________________



From equations 2) and 3) we get


            AP = PD


or, equivalently,


4)        A = PDP-1


where we have substituted P for [p1 p2 ... pn] and D for



             ole6.gif


[ Home ] [ Up ] [ Info ] [ Mail ]