Scribd
Upload
54 views3 pages

HW 1

This document contains an introduction to algorithms homework assignment with 4 problems. Problem 1 asks to determine if a power algorithm runs in linear or non-linear time. Problem 2 involves calculating probabilities of drawing balls from an urn. Problem 3 analyzes the running time of the Euclid's algorithm. Problem 4 proves properties of a multi-level insertion sort algorithm.

Uploaded by

Eden
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
54 views3 pages

HW 1

This document contains an introduction to algorithms homework assignment with 4 problems. Problem 1 asks to determine if a power algorithm runs in linear or non-linear time. Problem 2 involves calculating probabilities of drawing balls from an urn. Problem 3 analyzes the running time of the Euclid's algorithm. Problem 4 proves properties of a multi-level insertion sort algorithm.

Uploaded by

Eden
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

Introduction to Algorithms Homework 1

Instructors: Xue Chen due: 3.16

Problem 1 (20 points).


Answer: non-linear
The input length l = log k, cause we want be compute the power, so we assume n, q is fixed.
Assuming that the product and modulo can be computed in Θ(1) , so the 6th line takes Θ(1),
and the for loop iterate k times, so the total time T = Θ(k) = Θ(2log k ) = Θ(2l ) non-linear

Algorithm 1 POWER in linear time


function POWER(n, k, q)
if k = 0
return 1
if k = 1
return n
if k mod 2 = 0
return POWER(n ∗ n mod q, k/2, q)
else
return POWER(n ∗ n mod q, k/2, q)∗n mod q
end function

Correctness: base case: POWER work right for k = 0, 1.


Say 2p ≤ k < 2p+1 , and assume that POWER work right for k/2p+1 (= 0), k/2p (= 1), ..., k/2i ,
i = 1, .., p. Now we want want to prove it is right for k ′ = k/2i−1
′
if k ′ is even, POWER(n ∗ n mod q, k ′ /2, q) ≡ nk
′
if k ′ is odd, POWER(n ∗ n mod q, k/2, q)∗n mod q ≡ nk
By conclusion we can prove for k ′ = k/20 = k, so it is correct.
Runtime Analysis: Let T (k) be the time of POWER(n, k, q). We have T (k) = T (k/2) + O(1).
log k
X
T (k) = O(1) = O(log k)
i=1

The input length is log k, so it’s linear


Problem 2 (20 points).
N!
Answer: (a). (N −n)!N n (b). p1
(a).The first ball have N options, and N −1 for the second ball, so the number of all the possible
cases are N (N − 1)...(N − n + 1), so the probability is

N (N − 1)...(N − n + 1) N!
P = n
=
N (N − n)!N n
(b).Let random varible X be the times of coin-throwing
∞
X 1
EX = i(1 − p)i−1 p =
i=1
p

Problem 3 (20 points).


Answer: O(log a + log b)
Without losing the generality of analysis, let a ≥ b ,and let T (a, b) be the time of EUCLID(a, b)
Lemma: a mod b < a/2
Proof. if b > a/2, a mod b = a − b < a/2; else if b ≤ a/2, a mod b < b ≤ a/2
We have T (a, b) = T (b, r1 ) + O(1), where r1 = a mod b . let r0 = b, we have

T (ri , ri+1 ) = T (ri+1 , ri+2 ) + O(1)

here ri+2 < ri /2, so r1 < a/2, r3 < a/22 ... r2 < b/2, r4 < b/22 ... until ri = 0
So at most we have ⌈log a⌉ + ⌈log b⌉ times O(1), so the running time is O(log a + log b)
Problem 4 (40 points).
1. We prove for h2 = 3, h1 = 2
After calling INSERTIONSORT(h2 ),we have S[1] ≤ S[4] ≤ S[7]..., S[2] ≤ S[5] ≤ S[8]..., S[3] ≤
S[6] ≤ S[9]... And INSERTIONSORT(h1 ) sort independently on S1 = {S[1], S[3], ...} and
S2 = {S[2], S[4], ...} to form a new array A[1], A[2], A[3].... What we want to prove is that
A[i] ≤ A[i + 3], 1 ≤ i ≤ n − 3
We have |S1 | = (n + 1)/2, |S2 | = n/2. When i is odd, because S[1] ≤ S[4], S[3] ≤ S[6]..., there
are (n − 2)/2 elements in S1 smaller than a specific element in S2 , after sort independently,
A[i + 3] is bigger than (i + 1)/2 elements in S2 , so it is bigger than (i + 1)/2 elements in S1 , so
A[i] ≤ A[i + 3] ,that is to say, A[1] ≤ A[4], A[3] ≤ A[6]. Similarly, we can prove for i is even.
2. Lemma:if ht+1 , ht is co-prime, ∀n ≥ ht ht+1 , ∃x, y ∈ N, s.t. xht+1 + yht = n
Proof. When gcd(ht+1 , ht )=1, we have a′ , b′ ∈ Z s.t. a′ ht+1 + b′ ht = 1, let x′ = a′ n, y ′ = b′ n,
we have x′ ht+1 + y ′ ht = n = xht+1 + yht , so x = x′ + kht , y = y ′ − kht+1 , k ∈ Z so n can be
written in the unique form
n = xht+1 + yht , 0 ≤ x < ht
When n > ht ht+1 , cause 0 ≤ x < ht , we must have y ≥ 0
Now we want to prove that each element will be moved by at most ht+1 ht /ht−1 steps when
ht+1 , ht is co-prime.
First we show that ∀j, l < j − ht+1 ht ,we have A[l] ≤ A[j]. This is because j − l > ht ht+1 , so
∃x, y ∈ N s.t. xht+1 + yht = j − l, so

A[l] ≤ A[l + xht+1 ] ≤ A[l + xht+1 + yht ] = A[j]


And the increment is ht−1 , so each element will be moved by at most ht+1 ht /ht−1 steps, and each
step takes O(1), there are O(n) elements to be moved, so the time of INSERTIONSORT(ht−1 )
is O( nhht+1
t−1
ht
)

You might also like