CMPUT 204 Practice Test with Solutions
Question 1.
Consider array A = [3, 8, 5, 9, 2, 7, 6, 4].
Part (a)
To sort the array A by the call MergeSort(A, 1, 8), how many recursive calls to MergeSort are
made? Show the first four such calls.
Answer. If MergeSort(A, 1, 8) itself is counted, then totally 15 calls to MergeSort are made.
The first four such calls are as follows.
MergeSort(A, 1, 8);
MergeSort(A, 1, 4);
MergeSort(A, 1, 2);
MergeSort(A, 1, 1);
Part (b)
Show the resulting array by the call Build-Max-Heap(A), with the original array A above.
Answer. Build-Max-Heap(A) returns [9, 8, 7, 4, 2, 5, 6, 3].
Part (c)
(i) Show the resulting array by the call Partition(A, 1, 8), with the original array A.
(ii) What loop invariant does the procedure Partition maintain? In your work for question (i),
use the iteration at j = 5 (and therefore A[j] = 2) to explain how the maintenance works in this
case.
Answer. (i) Partition(A, 1, 8) returns [3, 2, 4, 9, 8, 7, 6, 5]
(ii) Loop invariant: Note that the call to partition is Partition(A, p, r) with pivot = A[r].
LI: At the start of each iteration j,
A[p..i] holds original keys ≤ pivot,
A[i+ 1..(j 1)] holds original keys > pivot, and
i < j.
When j = 5, i = 1 and the working array is [3, 8, 5, 9, 2, 7, 6, 4]. The assumption before the
iteration is: A[1] ≤ pivot and A[k] > pivot where k = 2, 3, 4. Now since A[j] ≤ pivot, according
to the Partition code, i = 2, and we exchange A[i] and A[j], which results in the working array
[3, 2, 5, 9, 8, 7, 6, 4]. The maintenance works because A[k] ≤ pivot for k = 1, 2, and A[k] > pivot for
k = 3, 4, 5 before the next iteration (more precisely, at the start of the next iteration).
Question 2.
For each pair of functions below, indicate their relation in terms of orders of growth using the
notations O(·), o(·), Θ(·), (·), or ω(·). Use the most precise notation, e.g., 2n ∈ o(12n2) is the
correct answer, even though 2n ∈ O(12n2) is not wrong (but it is a wrong answer for this question).
No proof or justification is needed. If the two functions cannot be related in this way, just say so.
Below, answers are accompanied with detailed justifications for the convenience of your study.
(A short justification: When n approaches infinity, the first term is approaching zero and the
second is approaching infinity.)
- n4(log n)3 n3.99(log n)4
Answer. n4(log n)3 ∈ ω(n3.99(log n)4).
Question 3.
Find asymptotic tight bounds (i.e. Θ) for the following recurrences. Assume that in each case,
T (n) is some positive constant for sufficiently small n. Your answer should consist of two parts (i) a
Θ-bound, and (ii) a brief justification. If you apply the master theorem, indicate which case; if you
use iterated substitution, give derivations leading to your conclusion; if you use a recurrence tree,
draw such a tree and briefly explain it. In any case, you do not need to prove that your closed-form
is correct.
for some ∈ (0, logb a]. The Master Theorem (case 1) is applicable to this recurrence relation,
and it solves it to T (n) ∈ Θ(nlog2 3).
b) T (n) = 4T (n3 ) + n
2 + n
Answer.
(i) T (n) ∈ Θ(n2).
(ii) Let a = 4, b = 3, and f(n) = n2 + n. We have T (n) = aT (nb ) + f(n), f(n) ∈ ?(nlogb a+)
for some ∈ (0, 2? logb a], and af(nb ) ≤ δf(n) for some δ ∈ ( ab2 , 1) and all sufficiently large n
(n ≥ (abδ)/(δab2 )). The Master Theorem (case 3) is applicable to this recurrence relation,
and it solves it to T (n) ∈ Θ(n2).
c) T (n) = T (n? 3) + 2 log n
Answer.
(i) T (n) ∈ Θ(n log n).
(ii) Without loss of generality, we suppose 3 | n. Using iterated substitution, we have
So we have T (n) ∈ O(n log n) and T (n) ∈ ?(n log n). We therefore have T (n) ∈ Θ(n log n).
Page 3
Question 4.
We are given two sets of integers S (with size n) and T (with size m). We want to find out if
S and T are disjoint, i.e. if S ∩ T = ?.
Part (a) [6 marks]
Suppose that m ≤ n. Describe a simple algorithm that takes O(n log n) time (briefly justify the
running time). You do not have to give a pseudocode in your description.
Answer.
- Sorting S[1..n] in ascending order, which takes O(n log n) time.
- For each T [i] in T [1..m], using binary search in S to determine whether T [i] ∈ S or not. If
T [i] ∈ S, then terminate and return S ∩T 6= ?. Otherwise, continue binary search in S for the next
element T [i+ 1]. - If we have T [i] 6∈ S for every T [i] in T [1 . . .m], then return S ∩ T =.
The step 2 will repeat at most m times, and each binary search in S takes O(log n) time. So
the total running time is at most n log n+m log n ≤ 2n log n ∈ O(n log n).
Part (b) [8 marks]
Suppose that m is substantially smaller than n, say m ∈ O(log n). Give an algorithm that takes
only O(n log log n) time (briefly justify the running time). Your algorithm should be given in a
pseudocode, in which any algorithm from the textbook or lecture notes can be called directly
without showing the code.
Answer. The pseudocode of the algorithm is as follows,
Checking whether S ∩ T = ?
**Precondition: S[1 . . . n], T [1 . . .m] are two arrays of integers.
**Postcondition: Returns whether S ∩ T = ? or not.
QuickSort(T, 1,m) takes O(m logm) operations. For each S[i] in S[1 . . . n], we use binary
search in T to determine whether S[i] ∈ T or not. This will repeat at most n times, and each
binary search in T takes logm operations. Totally, the time complexity is m logm + n logm ≤
2n logm ∈ O(n log logn).