CS 201

Amortized Analysis and Dynamic

Arrays
 Amortized Analysis

• Cost per-operation over a sequence of

operations
– Can take the total work for N operations and
divide by N, i.e. average cost per operation.
– Different than average case
• We don’t average over the possible inputs
• Still consider worst case input

2
 Amortized Analysis Example
A[m] A[m-1] ... A[3] A[2] A[1] A[0] cost
----------------------------------- ----
0 0 0 0 0 0
0 0 0 0 0 1 1
0 0 0 0 1 0 2
0 0 0 0 1 1 1
0 0 0 1 0 0 3
0 0 0 1 0 1 1
0 0 0 1 1 0 2

• Cost is the number of bits that change to go from one

value to the next

3
 Binary Counter Question
A[m] A[m-1] ... A[3] A[2] A[1] A[0] cost
----------------------------------- ----
0 0 0 0 0 0
0 0 0 0 0 1 1
0 0 0 0 1 0 2
0 0 0 0 1 1 1
0 0 0 1 0 0 3
0 0 0 1 0 1 1
0 0 0 1 1 0 2

What are the costs of the next two

increment operations?
A. 1 and 3
B. 1 and 2
C. 1 and 4 0% 0% 0% 0%
D. 2 and 2
3

2
d

d
an

an

an

an
1

2
4
 Binary Counter Analysis
A[m] A[m-1] ... A[3] A[2] A[1] A[0] cost
----------------------------------- ----
0 0 0 0 0 0
0 0 0 0 0 1 1
0 0 0 0 1 0 2
0 0 0 0 1 1 1
0 0 0 1 0 0 3
0 0 0 1 0 1 1
0 0 0 1 1 0 2
0 0 0 1 1 1 1
0 0 1 0 0 0 4

• Over a sequence of N increments, how many times does:

• A[0] change?
 N
• A[1] change
 N/2
• A[2] change
 N/4
5
 Binary Counter Analysis
A[m] A[m-1] ... A[3] A[2] A[1] A[0] cost
----------------------------------- ----
0 0 0 0 0 0
0 0 0 0 0 1 1
0 0 0 0 1 0 2
0 0 0 0 1 1 1
0 0 0 1 0 0 3
0 0 0 1 0 1 1
0 0 0 1 1 0 2
0 0 0 1 1 1 1
0 0 1 0 0 0 4

• Therefore, the total number of changes is:

N + N/2 + N/4 + N/8 … 2 + 1
lg 𝑁 𝑁
= σ𝑖=0 𝑖
2

≤ 𝑁 ∙ σ∞ 1
𝑖=0 2𝑖
σ∞ 1
𝑖=0 2𝑖 = 2
• So the total work is 2N 6
 Binary Counter Banker’s Method
A[m] A[m-1] ... A[3] A[2] A[1] A[0] amortized cost
----------------------------------- ----
0 0 0 0 0 0
0 0 0 0 0 1* 2
0 0 0 0 1* 0 2
0 0 0 0 1* 1* 2
0 0 0 1* 0 0 2
0 0 0 1* 0 1* 2
0 0 0 1* 1* 0 2
0 0 0 1* 1* 1* 2
0 0 1* 0 0 0 2

Different idea, assume the computer runs on tokens:

• Give each operation 2 tokens. Pay for the conversion of
the 0 into a 1 with a token and store the remaining token
there.
• All the other costs are turning 1s into 0s. Pay for those
with the token stored there.

7
 Stack with multi-pop

• Suppose that you have a stack data structure with

the following operations:
– Push(x) adds x to the top of the stack
– Pop() returns (and removes) the item at the top
of the stack
– Multi_pop(k) returns (and removes) the k items at the
top of the stack

What is the amortized cost of the push, pop and multi-

pop operations?

8
 Dynamic Arrays

• Suppose we want to implement arrays

without a fixed size limit.
– insert operation adds a new element to the end of
the array.
– Why, when and how much?
• When do we “re-size” the array?
– When it’s full
• How much space do we add?
– Double
• Why?

9
 Dynamic Arrays

• What is the total cost of a sequence of N=2k

insert operations?
– N for the inserts +
– 2 + 4 + 8 + 16 + … 2k-1 + 2k
– = N + N/2 + N/4 + … 2 < 2N
– So total cost is at most 3N

10
 Dynamic Arrays Bankers Method
• Suppose we give each insert 3 tokens.
– 1st token pays for the insert itself.
– Remaining tokens stored with the item.
– Use the tokens in the full array to pay for the copy.

5 3

Insert 7 then insert 9

5 3 7 ** 9 **

Re-size and copy

5 3 7 9

11
 Tokens Question

Where will the tokens be in the following

array?

5 3 7 9 2 10 4

A. 10 and 4
B. 9,2,10 and 4 0% 0% 0% 0%
C. 2,10, and 4

4
4

d
d

nd

..
D. They got used to play

an
an

o.
a

yD
,
10

10

10

a
2,

2,

pl
9,

to
Donkey Kong

ed
us
t
go
ey
Th
12
 Dynamic Arrays

5 3 7 9 2** 10** 4**

Insert 1

5 3 7 9 2** 10** 4** 1**

Insert 8, re-size first

5 3 7 9 2 10 4 1 8**

The 8 tokens stored at 2,10,4,and 1 “pay” for the copy.

The newly inserted 8 has its two tokens.

13
 Dynamic Arrays Deletion

Suppose we want to add the operation deletion:

The delete operation gets 3 tokens.
• 1 pays for the delete itself
• 2 tokens go to the empty cell

5 3 7 9 2 10

Delete 10

5 3 7 9 2 **

Delete 2

5 3 7 9 ** **

14
 When to downsize?

When the array is ______ full we

should reduce its size.

A. 25%
B. 50% 0% 0% 0%
C. Just keep the money and

%
say we did it.

...
25

50

a nd
ey
on
em
th
ep
ke
st
Ju
15
 Dynamic Arrays

5 3 7 9 ** ** ** **

25 % 25 %

• No guarantee of tokens in the green

• 25% must have 2 tokens each.
• Pays for a copy to an array of ½ the current size.

Result:
5 3 7 9

16
 Dynamic Arrays Summary

• Array always has at least 25% of the positions in use

• 3 Tokens for insert
• 3 Tokens for delete
• O(1) Amortized time per operation.
• O(N) worst case time for any single operation.

17