Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd Logo
Upload

Welcome to Scribd!

  • 91 views

    Fundamental Algorithms, Assignment 4 Solutions

    Uploaded by

    Aashish D
    The document contains solutions to algorithm analysis problems involving recursion and the Master Theorem. 1) It calculates values for a recursion and uses the Master Theorem to determine its asymptotic runtime is Θ(n^2 log n). 2) It uses the Master Theorem to determine the asymptotic runtimes of several recursions are Θ(n log n), Θ(n^5), and Θ(n^2 log n). 3) It analyzes the runtime of the Toom-3 algorithm for multiplying large numbers and determines its runtime is Θ(nlog3 5) which is faster than Karatsuba for large n. 4) It expresses several sums in asymptotic notation by

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Fundamental Algorithms, Assignment 4 Solutions

    Uploaded by

    Aashish D
    91 views3 pages
    The document contains solutions to algorithm analysis problems involving recursion and the Master Theorem. 1) It calculates values for a recursion and uses the Master Theorem to determine its asymptotic runtime is Θ(n^2 log n). 2) It uses the Master Theorem to determine the asymptotic runtimes of several recursions are Θ(n log n), Θ(n^5), and Θ(n^2 log n). 3) It analyzes the runtime of the Toom-3 algorithm for multiplying large numbers and determines its runtime is Θ(nlog3 5) which is faster than Karatsuba for large n. 4) It expresses several sums in asymptotic notation by

    Original Title

    sol4.pdf

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    The document contains solutions to algorithm analysis problems involving recursion and the Master Theorem. 1) It calculates values for a recursion and uses the Master Theorem to determine its asymptotic runtime is Θ(n^2 log n). 2) It uses the Master Theorem to determine the asymptotic runtimes of several recursions are Θ(n log n), Θ(n^5), and Θ(n^2 log n). 3) It analyzes the runtime of the Toom-3 algorithm for multiplying large numbers and determines its runtime is Θ(nlog3 5) which is faster than Karatsuba for large n. 4) It expresses several sums in asymptotic notation by

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    91 views3 pages

    Fundamental Algorithms, Assignment 4 Solutions

    Uploaded by

    Aashish D
    The document contains solutions to algorithm analysis problems involving recursion and the Master Theorem. 1) It calculates values for a recursion and uses the Master Theorem to determine its asymptotic runtime is Θ(n^2 log n). 2) It uses the Master Theorem to determine the asymptotic runtimes of several recursions are Θ(n log n), Θ(n^5), and Θ(n^2 log n). 3) It analyzes the runtime of the Toom-3 algorithm for multiplying large numbers and determines its runtime is Θ(nlog3 5) which is faster than Karatsuba for large n. 4) It expresses several sums in asymptotic notation by

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 3

    Fundamental Algorithms, Assignment 4

    Solutions
    1. Consider the recursion T (n) = 9T (n/3) + n2 with initial value T (1) =
    1. Calculate the precise values of T (3), T (9), T (27), T (81), T (243).
    Make a good (and correct) guess as to the general formula for T (3i )
    and write this as T (n). (Dont worry about when n is not a power
    of three.) Now use the Master Theorem to give, in Thetaland, the
    asymptotics of T (n). Check that the two answers are consistent.
    Solution:T (3) = 9(1) + 33 = 18 = 2(9), T (9) = 9(18) + 92 = 243 =
    3(81), T (27) = 9(243) + 729 = 2916 = 4(729), T (81) = 32805 =
    5(6561), T (243) = 354294 = 6(59049). In general, T (3i ) = (i + 1)32i .
    With n = 3i we have 32i = n2 and i = log3 n so the formula is
    T (n) = n2 (1 + log3 n). In Thetaland, T (n) = (n2 lg n). With the
    Master Theorem, as log3 9 = 2 we are in the special case which gives
    indeed T (n) = (n2 lg n).
    Another approach is via the auxilliary function S(n) discussed in class.
    Here S(n) = T (n)/n2 . Dividing the original recursion by n2 gives
    T (n) T (n/3)
    2
    = +1
    n (n/3)2
    so that

    S(n) = S(n/3) + 1 with initial value S(1) = T (1)/12 = 1

    so that

    S(n) = 1 + log3 n and so T (n) = n2 (1 + log3 n)

    2. Use the Master Theorem to give, in Thetaland, the asymptotics of


    these recursions:

    (a) T (n) = 6T (n/2) + n n
    ln 6
    Solution:As log2 6 = ln 2 = 2.58 > 3/2 we have Low Over-
    head and T (n) = (n log 2 6 ).
    (b) T (n) = 4T (n/2) + n5
    Solution:log2 4 = 2 < 5 so we have High Overhead and T (n) =
    (n5 ).
    (c) T (n) = 4T (n/2) + 7n2 + 2n + 1
    Solution:log2 4 = 2 and the Overhead is (n2 ) so T (n) =
    (n2 lg n.
    3. Toom-3 is an algorithm similar to the Karatsuba algorithm discussed in
    class. (Dont worry how Toom-3 really works, we just want an analysis
    given the information below.) It multiplies two n digit numbers by
    making five recursive calls to multiplication of two n/3 digit numbers
    plus thirty additions and subtractions. Each of the additions and
    subtractions take time O(n). Give the recursion for the time T (n) for
    Toom-3 and use the Master Theorem to find the asymptotics of T (n).
    Compare with the time (nlog2 3 ) of Karatsuba. Which is faster when
    n is large?
    Solution:T (n) = 5T (n/3) + O(n) as the thirty is absorbed into the
    big oh n term. From the master theorem T (n) = (nlog3 5 ). As

    ln 5
    log3 5 = = 1.46 < 1.58 = log2 3
    ln 3
    it is better that the (nlog2 3 ) of Karatsuba. (In practice unless n is
    really large Karatsuba does better because Toom-3 has large constant
    factors.)

    4. Write the following sums in the form (g(n)) with g(n) one of the
    standard functions. In each case give reasonable (they neednt be
    optimal) positive c1 , c2 so that the sum is between c1 g(n) and c2 g(n)
    for n large.

    (a) n2 + (n + 1)2 + . . . + (2n)2


    Solution:(n3). There are n terms all between n2 and 4n2
    so the sum is between n3 (1 + o(1)) and 4n3 (1 + o(1)).
    (b) lg2 (1) + lg2 (2) + . . . + lg2 (n)
    Solution:(n lg2 n). There are n terms all at most lg2 (n) so
    an upper bound is n lg2 (n). Lopping off the bottom half of the
    terms we still have n/2 terms and each is at least lg2 (n/2) =
    (lg(n) 1)2 lg2 n so the lower bound is (1 + o(1))( n2 ) lg2 n.
    (c) 13 + . . . + n3 .
    Solution:T (n) = (n4 ). Upper bound n4 as n terms, each at
    most n. Lopping off bottom half yields n/2 terms, each at least
    (n/2)3 giving a lower bound (n/2)(n/2)3 = n4 /16.

    5. Give an algorithm for subtracting two n-digit decimal numbers. The


    numbers will be inputted as A[0 N ] and B[0 N ] and the output
    should be C[0 N ]. How long does your algorithm take, expressing
    your answer in one of the standard (g(n)) forms.
    Solution:Here is one way, the term BORROW being the truth value of
    whether you have borrowed.
    BORROW=false;
    FOR I=0 TO N;
    IF BORROW=false THEN X=A[I]-B[I];
    IF BORROW=true THEN X=A[I]-1-B[I];
    IF X 0 THEN
    C[I]=X;
    BORROW=false;
    ELSE
    C[I]=X+10;
    BORROW=true;
    ENDFOR
    IF BORROW=true THEN ERROR;
    END
    This takes only a single pass and so is a linear time, that is (N )
    algorithm.

    You might also like