Step 3 is just like steps
1 and
2, except we now have three more
assumptions to add to the starting relations.
We set
P33 and its transpose equal to 2,
and we set the inverse of
P33 and its transpose equal to 1/2.
Then we set
P23 and its transpose equal to zero.
All of this is done in the file
problem2.step3.m.
In this step the NCProcess command creates a spreadsheet
file called
``Banswer3.dvi''.
As usual, we only need to concentrate on the intersting categories.
The first two categories involve
P12 and its transpose.
The two equations in the
P12 category tell us that
P12 times
A11 transpose to any power, times
C1 transpose is zero.
The expressions with unknown variables
and knowns
The expressions with unknown variables
and knowns
The first two equations in the
P12 transpose category give a similar
result for
P12 transpose.
If we make certain assumptions about controlability, then we can
prove that
P12 and its transpose are equal to zero.
The next two categories tell us about the last two elements of the P
matrix,
P11 and
P22.
The expressions with unknown variables
and knowns
The first equation in the above category indicates that perhaps
P11 equals
.
The other equations in this category give more evidence that this is
the case. Remember, that we can run for more iterations and generate
equations like this of as high of a degree as we please.
Again, we would need to make some sort of assumption about
controlability before we could prove that
P11 equals
,
but the NCProcess does generate what we would need to prove
it rigorously.
The expressions with unknown variables
and knowns