Lecture 1

Introduction and Peak Finding 6.006 Fall 2011

Lecture 1: Introduction and Peak Finding

Lecture Overview
Administrivia

Course Overview

Peak nding problem 1D and 2D versions

Course Overview
This course covers:

Ecient procedures for solving problems on large inputs (Ex: U.S. Highway Map,
Human Genome)

Scalability

Classic data structures and elementary algorithms (CLRS text)

Real implementations in Python

Fun problem sets!

The course is divided into 8 modules each of which has a motivating problem and problem
set(s) (except for the last module). Tentative module topics and motivating problems are
as described below:

1. Algorithmic Thinking: Peak Finding

2. Sorting & Trees: Event Simulation

3. Hashing: Genome Comparison

4. Numerics: RSA Encryption

5. Graphs: Rubiks Cube

6. Shortest Paths: Caltech MIT

7. Dynamic Programming: Image Compression

8. Advanced Topics

1
Lecture 1 Introduction and Peak Finding 6.006 Fall 2011

Peak Finder
One-dimensional Version
Position 2 is a peak if and only if b a and b c. Position 9 is a peak if i h.

1 2 3 4 5 6 7 8 9
a b c d e f g h i

Figure 1: a-i are numbers

Problem: Find a peak if it exists (Does it always exist?)

Straightforward Algorithm

6 7 4 3 2 1 4 5

Start from left

might be peak

(n) complexity worst case

Figure 2: Look at n/2 elements on average, could look at n elements in the worst case

What if we start in the middle? For the conguration below, we would look at n/2 elements.
Would we have to ever look at more than n/2 elements if we start in the middle, and choose
a direction based on which neighboring element is larger that the middle element?

n/2

2
Lecture 1 Introduction and Peak Finding 6.006 Fall 2011

Can we do better?

1 2 . . . n/2 -1 n/2 n/2 +1 . . . n-1 n

look at n/2 position

Figure 3: Divide & Conquer

If a[n/2] < a[n/2 1] then only look at left half 1 . . . n/2 1 to look for peak

Else if a[n/2] < a[n/2 + 1] then only look at right half n/2 + 1 . . . n to look for peak

Else n/2 position is a peak: WHY?

a[n/2] a[n/2 1]
a[n/2] a[n/2 + 1]

What is the complexity?

T (n) = T (n/2) + (1) = (1) + . . . + (1) (log2 (n) times) = (log2 (n))
' -v "
to compare a[n/2] to neighbors

In order to sum up the (i)s as we do here, we need to nd a constant that works for all.
If n = 1000000, (n) algo needs 13 sec in python. If algo is (log n) we only need 0.001 sec.
Argue that the algorithm is correct.

Two-dimensional Version

b a d
n rows
e

m columns

Figure 4: Greedy Ascent Algorithm: (nm) complexity, (n2 ) algorithm if m = n

a is a 2D-peak i a b, a d, a c, a e

3
Lecture 1 Introduction and Peak Finding 6.006 Fall 2011

14 13 12
15 9 11 17

16 17 19 20

Figure 5: Circled value is peak.

Attempt # 1: Extend 1D Divide and Conquer to 2D

j = m/2

Pick middle column j = m/2.

Find a 1D-peak at i, j.

Use (i, j) as a start point on row i to nd 1D-peak on row i.

Attempt #1 fails

Problem: 2D-peak may not exist on row i

10

14 13 12
15 9 11

16 17 19 20

End up with 14 which is not a 2D-peak.

4
Lecture 1 Introduction and Peak Finding 6.006 Fall 2011

Attempt # 2

Pick middle column j = m/2

Find global maximum on column j at (i, j)

Compare (i, j 1), (i, j), (i, j + 1)

Pick left columns of (i, j 1) > (i, j)

Similarly for right

(i, j) is a 2D-peak if neither condition holds WHY?

Solve the new problem with half the number of columns.

When you have a single column, nd global maximum and youre done.

Example of Attempt #2

go with

10 8 10 10 10 10 10

14 13 12 11 12 11 11

15 9 11 21 11 21 21 find 21

16 17 19 20 19 20 20

pick this column pick this column

Complexity of Attempt #2

If T (n, m) denotes work required to solve problem with n rows and m columns

T (n, m) = T (n, m/2) + (n) (to nd global maximum on a column (n rows))

Question: What if we replaced global maximum with 1D-peak in Attempt #2? Would that
work?

5
MIT OpenCourseWare
http://ocw.mit.edu

6.006 Introduction to Algorithms

For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.