Solving by Iteration

Embed Size (px)

Citation preview

  • 8/13/2019 Solving by Iteration

    1/9

    Section 2: Solving Recurrence

    Relations by Iteration As we noted at the end of the last lecture, when

    analyzing recurrence relations, we want to rewrite

    the general term as a function of the index andindependent of predecessor terms.

    This will allow us to compute any arbitrary term

    in the sequence without having to compute all the

    previous terms.

    In this section, we will look at one method for

    solving recurrence relations.

    8.2.1

  • 8/13/2019 Solving by Iteration

    2/9

    The Method of Iteration

    Let {ai} be the sequence defined by:ak= ak1+ 2 with a0= 1.

    Plugging values of kinto the relation, we get:

    a1= a0+ 2 = 1 + 2

    a2= a1+ 2 = 1 + 2 + 2 = 1 + 2(2)

    a3= a2+ 2 = 1 + 2 + 2 + 2 = 1 + 3(2)

    a4= a3+ 2 = 1 + 2 + 2 + 2 + 2 = 1 + 4(2)

    Continuing in this fashion reinforces the apparentpattern that an= 1 + n(2) = 1 + 2n.

    This brute forcetechnique is theMethod of

    Iteration.

    8.2.2

  • 8/13/2019 Solving by Iteration

    3/9

    The Tower of Hanoi

    Recall the Tower of Hanoi relation:mk= 2mk1+ 1 with m1= 1.

    Plugging values of kinto the relation, we get:

    m2= 2m1+ 1 = 2 + 1

    m3= 2m2+ 1 = 2(2 + 1) + 1 = 22+ 2 + 1

    m4= 2m3+ 1 = 2(22+ 2 + 1) + 1

    = 23+ 22+ 2 + 1

    m5= 2m4+ 1 = 2(23

    + 22

    + 2 + 1)= 24+ 23+ 22+ 2 + 1.

    Thus, weguess: mn= 2n1+ 2n2+...+ 22+ 2 + 1.

    This is a Geometric Series, so mn= 2

    n

    1.

    8.2.3

  • 8/13/2019 Solving by Iteration

    4/9

    Another Example

    Let {ai} be the sequence given by:ak= ak1+ k with a0= 0.

    Solve this recurrence relation and find a100.

    Now, a1= a0+ 1 = 1 + 0a2= a1+ 2 = 2 + 1 + 0

    a3= a2+ 3 = 3 + 2 + 1 + 0

    a4= a3+ 4 = 4 + 3 + 2 + 1 + 0 Thus an= n+ (n1) + (n2) +...+ 3 + 2 + 1 + 0

    so an= n(n+1)/2.

    Plugging in n= 100: a100

    = 100(101)/2 = 5050.

    8.2.4

  • 8/13/2019 Solving by Iteration

    5/9

    A Geometric Sequence

    Let aand bbe non-zero constants, and consider:

    sk= ask-1 withs0= b.

    Thus: s1= as0= ab

    s2= as1= a(ab) = a2b

    s3= as2= a(a2b) = a3b

    s4= as3= a(a3b) = a4b.

    From this, we can make the conjecture that:sn= a

    nb.

    Note: if b= 1, thensn= an.

    8.2.5

  • 8/13/2019 Solving by Iteration

    6/9

    A Perturbed Geometric Sequence

    Let abe non-zero constants, and consider:

    sk= ask-1 + 1 withs0= 1.

    Thus: s1= as0 + 1 = a+ 1

    s2= as1 + 1 = a(a+ 1) + 1 = a2 + a+ 1

    s3= as2 + 1 = a(a2 + a+ 1) + 1

    = a3 +a2 + a+ 1

    s4= as3 + 1 = a(a3 +a2 + a+ 1)= a4+ a3 +a2 + a+ 1.

    It appears thatsn= an + an1 +...+a2 + a+ 1, so

    sn= (a

    n+1

    1) / (a

    1) .

    8.2.6

  • 8/13/2019 Solving by Iteration

    7/9

    Some Observations

    In solving these recurrence relations, we point out

    the following observations:

    1. Each recurrence relation looks only 1 step back; that is

    each relation has been of the formsn= F(sn1);2. We have relied on luckto solve the relation, in that we

    have needed to observe a pattern of behavior and

    formulated the solution based on the pattern;

    3. The initial condition has played a role in making thispattern evident;

    4. Generating a formula from the generalization of the

    pattern looks back to our study of induction.

    8.2.7

  • 8/13/2019 Solving by Iteration

    8/9

    Verifying Solutions

    Once we guess the form of the solution for a

    recurrence relation, we need to verify it is, in fact,

    the solution.

    We use Mathematical Induction to do this. For example, in the Tower of Hanoi game, we

    conjecture that the solution is mn= 2n1.

    Basis Step: m1= 1 (by playing the game), and211 = 2 1 = 1, therefore m1= 2

    11.

    Inductive Step: Recall the recurrence relation is

    mk= 2mk1+ 1. Assume mk= 2k1.

    8.2.8

  • 8/13/2019 Solving by Iteration

    9/9

    Verifying Solutions (contd.)

    Inductive Step: Show mk+1= 2k+11.

    Now, mk+1= 2mk+ 1

    = 2(2k1) + 1

    = 22k2 + 1= 2k+11.

    Therefore mk= 2k1 is the solution to

    mk= 2mk1+ 1, when m1= 1. QED A similar verification process will work for all

    the other formulas we discovered in this section.

    8.2.9