AMS210 Homework 8

4.6:3,8,16 4.9:2,3,12,13 DUE: Wednesday, Nov. 1, Beginning of Class

Solutions to Problems:


4.6:3)  rank A = number of pivot columns.
>From the echelon form B, we see that columns 1,3, and 4 are pivot columns. Thus
	rank A = 3 .
Using the Rank Theorem, or that dim(Nul A) = the # of nonpivot columns = 
# of free variables for Ax = 0; we get
	dim(Nul A) = 2.
A basis for Col A is the pivot columns of A, which    { [ 2 ] , [ 6 ] , [ 2 ] }
from the echelon form B are identified as the		[-2 ]   [-3 ]   [-3 ]
1st,3rd, and 4th columns of A.   			[ 4 ]   [ 9 ]   [ 5 ]
							[-2 ]   [ 3 ]   [-4 ]

For Bx = 0 we get 
  x_2, x_5 free (note columns 2 and 5 are nonpivot columns).
To get a basis for Nul A, row reduce B to REDUCED echelon form:

[ 2  -3   6   2   5 ]	 [ 1  -3/2 0   0 -9/2]    [ 1  -3/2 0   0 -9/2]
[ 0   0   3  -1   1 ]	 [ 0   0   3   0   4 ]    [ 0   0   1   0  4/3]
[ 0   0   0   1   3 ]  ~ [ 0   0   0   1   3 ]  ~ [ 0   0   0   1   3 ] 
[ 0   0   0   0   0 ]    [ 0   0   0   0   0 ]    [ 0   0   0   0   0 ]

Thus [x_1] = [(3/2)x_2 + (9/2)x_5 ] = x_2 [3/2] + x_5 [ 9/2]
     [x_2]   [        x_2         ]       [ 1 ]	      [  0 ]
     [x_3]   [    (-4/3)x_5	  ]       [ 0 ]       [-4/3]
     [x_4]   [       -3x_5	  ]       [ 0 ]	      [ -3 ]
     [x_5]   [	      x_5         ] 	  [ 0 ]	      [  1 ]

Thus a basis for Nul A is { (3/2,1,0,0,0) , (9/2,0,-4/3,-3,1) }.


4.6:8) A (a 5x6 matrix) has 4 pivot columns, so
   rankA = dim(Col A) = 4.
   By the Rank Thm.  dim(Nul A) = n - rankA = 6 - 4 = 2.
   No, Col A does not equal R^4.  Note the pivot columns of A consist of 4 
linearly independent column vectors in R^5 which comprise a basis for Col A.
Thus Col A is a 4 dimensional subspace of R^5.  However, these pivot column
vectors each have 5 entries (recall A has 5 rows) so are  not in R^4.


4.6:16) A is given to be 6x4 (6 rows, 4 columns).
By the Rank Theorem,
	dim(Nul A) = 4 - rankA = 4 - (# pivot columns of A).
  A could have as many as 4 pivot columns; thus dim(Nul A) could be 0 (smallest
possible).


4.9.2) a)
	       From
	Healthy      Ill

 P =  [   .95       .45  ]   Healthy
      [   .05       .55  ]   Ill	To

	b) The initial probability state vector (on Monday) is 

   x_0  =  [ proportion healthy ]  =  [ .80  ]
	   [ proportion   ill   ]     [ .20  ]

The state vector for Tues. is 

   x_1  =  P x_0  =  [ .95   .45 ]  [ .80 ]  =  [ .85 ]   Thus 15% are ill
		     [ .05   .55 ]  [ .20 ]     [ .15 ]      on Tues.
The state vector for Wed. is 

   x_2  =  P x_1  =  [ .95   .45 ]  [ .85 ]  =  [ .875 ]   Thus 12.5% are ill
		     [ .05   .55 ]  [ .15 ]     [ .125 ]      on Wed.

c) To find the probability that a student is healthy in two days given that he
 is healthy today, find the state vector x_2 given the initial state vector

   x_0  =  [ probability healthy ]  =  [ 1  ]
	   [ probability   ill   ]     [ 0  ]

The state vector one day hence is 

   x_1  =  P x_0  =  [ .95   .45 ]  [ 1  ]  =  [ .95 ]   
		     [ .05   .55 ]  [ 0  ]     [ .05 ] 
  
The state vector two days hence is 

   x_2  =  P x_1  =  [ .95   .45 ]  [ .95  ]  =  [ .925 ]   The first entry
		     [ .05   .55 ]  [ .05  ]     [ .075 ]   .925 is answer


4.9:3) a)
		From
	    #1   #2   #3
   P  =  [ .50  .25  .25 ]  #1
	 [ .25  .50  .25 ]  #2  To
	 [ .25  .25  .50 ]  #3

The probability that the animal chooses food #2 on second trial given that it 
chose food #1 on initial trial is the 2nd entry of x_2 given the initial state
x_0 = (1,0,0).

  x_2  =  P x_1  =  P (P x_0)  =  P^2 x_0  =  P^2 [ 1 ]
						  [ 0 ]
						  [ 0 ]

 =  1st column of the matrix P^2  =  [ .5(.5) + .25(.25) + .25(.25) ] = [.3750]
				     [.25(.5) +  .5(.25) + .25(.25) ]   [.3125]
				     [.25(.5) + .25(.25) +  .5(.25) ]   [.3125]

Thus the desired probability (2nd entry) is .3125.
Alternatively,

  x_2  =  P x_1  =  P (P x_0)  =   P [ .50 ]  =  [.3750]
				     [ .25 ]	 [.3125]
				     [ .25 ]	 [.3125]

4.9:12) a) Note P is a regular stochastic matrix (here all entries of P are
postive) so a unique steady state probability vector x exists (Theorem 18).
  Solve
 	   Px  =  x
    ==> (P-I)x =  0 
    ==> [ -.05   .45 ] [ x_1 ]  =  [ 0 ]
	[  .05  -.45 ] [ x_2 ]	   [ 0 ]

Thus   .05 x_1  =  .45 x_2  ==>  x_1 = 9x_2,  x_2 free, so x = x_2 [ 9 ].
								   [ 1 ]
 
Now the fact that [ x_1 ] is a probability vector implies x_1 + x_2 = 1
		  [ x_2 ]			     ==>  x_2 = 1 - x_1.
Thus x_1 = 9 x_2 = 9(1 - x_1) = 9 - 9 x_1  ==>  10 x_1 = 9  ==> [ x_1 ] = [.9] 
								[ x_2 ] = [.1]
This last step is equivalent to picking x_2 any non-zero # (say 1) and 
dividing the resultant vector ( [ 9 ] ) by the sum of the entries (= 10).
				[ 1 ]
 
b)  After "many" days, the state probabilities approach those of the steady
state vector, regardless of the initial state of the person (so no, it doesn't
matter whether the person is ill today).  Thus the probability that the person
is ill many days in future is the 2nd entry of steady state vector, i.e.  .1.


4.9:13) Find the steady state vector from (P-I)x = 0.

[ -.5  .25  .25  0 ]         [ -1  1/2  1/2  0 ]        [ 1  -1/2  -1/2   0 ]
[ .25  -.5  .25  0 ]    ~    [  1  -2    1   0 ]   ~    [ 0  -3/2   3/2   0 ]
[ .25  .25  -.5  0 ]         [  1   1   -2   0 ]	[ 0   3/2  -3/2   0 ]

     [ 1   0  -1   0 ]   Thus x_1  =  x_3
  ~  [ 0   1  -1   0 ]	      x_2  =  x_3
     [ 0   0   0   0 ]        x_3  free

Since the probabilities sum to 1,   x_1 + x_2 + x_3 = 1  ==>
				    x_3 =  1/3  =  x_2  =  x_1.
Thus after many days, the animal has no food preference (all foods equally
preferred), i.e. the steady state probability vector is [ 1/3 ].
							[ 1/3 ]
							[ 1/3 ]