AMS210 Homework 3

1.7:3,12,19,32; 1.8:8,13,14,16,37; 1.9:11,12 DUE: Wednesday, FEb. 14, Beginning of Class

Solutions to Problems:


1.7:3)  Solve the associated system Ax = b.  Row reduce the augmented matrix 

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

	[  1   0   0   4 ]   Thus x = [ 4 ]  is the unique x satisfying Tx = b
    ~	[  0   1   0   3 ]	      [ 3 ]
	[  0   0   1   4 ]	      [ 4 ]
 	

1.7:12)  b is in the range of the linear transformation if and only if
Ax = b is consistent (has a solution).  Row reduce the augmented matrix [A b]
to an echelon form to determine consistency.

	[  1   2  -7   5   9 ]             [  1   2  -7   5   9 ]
	[  0   1  -4   0   5 ]      ~      [  0   1  -4   0   5 ]    ~
	[  1   0   1   6   0 ]		   [  0  -2   8   1  -9 ]
	[  2  -1   6   8  -9 ]             [  0  -5  20  -2  -27]

	[  1   2  -7   5   9 ]             [  1   2  -7   5   9 ]
	[  0   1  -4   0   5 ]      ~      [  0   1  -4   0   5 ]    
	[  1   0   0   1   1 ]		   [  0   0   0   1   1 ]
	[  0   0   0  -2  -2 ]             [  0   0   0   0   0 ]

   The system is consistent so b is in the range of the transformation x -> Ax.
   (In fact, we see that there is a free variable x_3 so infinitely many
 solutions, but computation of the solutions is not necessary for the 
conclusion that b is in the range of the linear transformation).


1.7:19)  Utilize the linearity of T.

	T(2u) = 2 T(u) = 2 [ 2 ] = [ 4 ]
			   [ 0 ]   [ 0 ]

	T(3v) = 3 T(v) = 3 [ 1 ] = [ 3 ]
			   [-4 ]   [-12]
  
   Using above result,

	T(2u + 3v) = T(2u) + T(3v) = [ 4 ] + [ 3 ] = [ 7 ]
				     [ 0 ]   [-12]   [-12]


1.7:32) T(x_1,x_2) = ( 4x_1 - 2x_2, 3|x_2| )
   To show that T is not linear, you need to find any counterexample, 
that is, either

	1) Find vectors u,v in R^2 such that T(u+v) does not equal T(u)+T(v)

    or  2) Find a real number c and a vector u such that T(cu) not equal cT(u).

We will show 2):
	Let c = -1, and u be any vector (x_1,x_2) with x_2 > 0, e.g.
u = (0,1).  Then
	T(-u) = T{(0,-1)} = (2,3)
However,
	-T(u) = -T{(0,1)} = -(-2,3) = (2,-3)
As these are not equal, T is not linear.


1.8:8)  To determine the standard matrix representation A of transformation T,
we need to determine what T does to
each of the unit vectors e_1 = [ 1 ],  e_2 = [ 0 ]
			       [ 0 ]	     [ 1 ]
 
   T(e_1) = e_1 = [ 1 ],   T(e_2) = e_2 - 3e_1 = [-3 ]
		  [ 0 ] 			 [ 1 ]

The standard matrix has column 1 equal to T(e_1) and column 2 equal to T(e_2)

	A  =  [ 1  -3 ]
	      [ 0   1 ]


1.8:13) In this and the next problem, you should use plane geometry to
visualize what T does to the vectors e_1 and e_2.
T is a composition of linear transformations: T = VU , where
   U rotates points counterclockwise pi/4 radians:

	U(e_1) = [ 1/sqrt{2} ]  =  (1/sqrt{2}) (e_1  +  e_2)
		 [ 1/sqrt{2} ]

	U(e_2) = [-1/sqrt{2} ]  =  -(1/sqrt{2}) e_1  +  (1/sqrt{2}) e_2
		 [ 1/sqrt{2} ]

   V reflects a point about x_2 axis, that is V(x_1,x_2) = (-x_1, x_2).Hence
	
	V(e_1) = -e_1  ,   V(e_2) = e_2
   
   The rotation U followed by the reflection V is the composition T = VU.
   Thus T(e_1) = V{U(e_1)} = (1/sqrt{2}){V(e_1) + V(e_2)} = (1/sqrt{2}) [-1 ]
									[ 1 ]
    and T(e_2) = V{U(e_2)} = (1/sqrt{2}){-V(e_1) + V(e_2)} = (1/sqrt{2}) [ 1 ]
									 [ 1 ]
Therefore the standard matrix is

	A  = [ T(e_1)  T(e_2) ] =  (1/sqrt{2}) [ -1   1 ]
			                       [  1   1 ]

Equivalently, A can be found by the multiplication of, respectively, the
standard matrix of V and the standard matrix of U:

	[-1   0 ]  1/sqrt{2} [ 1  -1 ]  =  1/sqrt{2} [ -1   1 ]
	[ 0   1 ]	     [ 1   1 ]	             [  1   1 ]


1.8:14) T is a composition of linear transformations; T = VU, where
U reflects points through the origin, that is U(x_1, x_2) = (-x_1, -x_2)
(identical to reflection about x_1 axis AND x_2 axis), so
	
	U(e_1) = -e_1  ,  U(e_2) = -e_2

V rotates vectors pi/2 clockwise, so

	V(e_1) = [ 0 ]  =  -e_2
		 [-1 ]
	V(e_2) = [ 1 ]  =   e_1
		 [ 0 ] 

Therefore T(e_1) = V{U(e_1)} = V(-e_1) = -V(e_1) = -(-e_2) = e_2 
	  T(e_2) = V{U(e_2)} = V(-e_2) = -V(e_2) =  -e_1 

Thus the standard matrix of T has column 1 equal to  e_2 and column 2 = -e_1: 

	A  =  [ 0  -1 ]
	      [ 1   0 ]


1.8:16) [ 5x_1 - 6x_2 ]	    =	x_1 [ 5 ]  +  x_2 [-6 ] = [ 5  -6 ]  [ x_1 ]
	[     x_2     ]   	    [ 0 ] 	  [ 1 ]	  [ 0   1 ]  [ x_2 ]
	[ 2x_1 - 2x_2 ]		    [ 2 ]         [-2 ]   [ 2  -2 ]


1.8:37) Determine if the columns of the standard matrix are linearly
independent (implies that T is 1-1).
	Row reduce to an echelon form.  Columns are lin. independent iff 
every column is a pivot column; equivalently for this square matrix check if
pivot position in every row.

	[ -5  10  -5   4 ]         [ -5  10  -5   4 ]	    [-5  10  -5   4 ]
	[  8   3  -4   7 ]    ~    [  0  65 -60  67 ]   ~   [ 0  65 -60  67 ]
	[  4  -9   5  -3 ]         [  0 -10  10   2 ]	    [ 0 -10  10   2 ]
	[ -3  -2   5   4 ]	   [  0  40 -40  -8 ]	    [ 0   0   0   0 ]

No further reduction necessary.
No pivot position in row 4 so columns not linearly independent so T is not 1-1.


1.9:11)
	a) Let x_0 be the initial vector:
	
   x_0  =  [ Population   In CA   in 1990 ]  =  [  29,716,000 ]
	   [ Population Out of CA in 1990 ]	[ 218,994,000 ]

After 1 year:

Proportion In CA who stay In  = (29,716,000 - 509,500)/29,716,000 = .98285
Proportion In CA who move Out  =  509,500/29,716,000 =  1 - .98285 = .01715
Proportion Out of CA who move In  =  564,100/218,994,000 = .00258
Proportion Out of CA who stay Out =(218,994,000 - 564,100)/218,994,000 = .99742

Therefore the migration matrix M is

	   In      Out

 M =  [ .98285   .002576 ]   In
      [ .01715    .99742 ]   Out

	b) Utilize the relation x_(k+1) = M x_k ,  k = 0,1,2,...
Year 2000 is 10 years hence from 1990, so we find x_10:

   x_10  =  [ Population   In CA   in 2000 ]  = M^10 x_0 =  [  30,215,000 ]
	    [ Population Out of CA in 2000 ]	            [ 218,495,000 ]
							(to the nearest 1,000)
Use a scientific calculator/computer to calculate M^10 above, where M^10 is the
matrix M multiplied by itself 10 times:

	M^10  =  [ .84295  .02359 ]
		 [ .15705  .97641 ]

	c)The proportion of total population In CA in 2000 is, from (b),
	
	30,215,000/(218,495,000 + 30,215,000)  =  .1215, i.e. 12.15%.


1.9:12)  Let x_0 be the initial status vector of the cars. We are given

	x_0 = [ # of cars at  airport on Mon. ]  =  [ 304 ]
	      [ # of cars at eastside on Mon. ]     [  48 ]
	      [ # of cars at westside on Mon. ]     [  98 ]

The migration matrix M is given in text.

	M  =  [ .97   .05   .10  ]
	      [ .00   .90   .05  ]
	      [ .03   .05   .85  ]
	      
Wednesday is 2 days hence from Mon, 1 day hence from Tues, so we wish to find 

	x_2  =  M (M x_0)  = M (x_1) =  M [ 307 ]  =  [ 309.69 ]
					  [  48 ]     [  47.95 ]
					  [  95 ]     [  92.36 ]
					  ( Tues )     ( Wed )

Round entries to the nearest integer (# cars in whole numbers) to get  [ 310 ]
							               [  48 ]
							               [  92 ]
the state vector for Wednesday.