Chapter 6

Dynamic Programming

Slides by Kevin Wayne.

Copyright © 2005 Pearson-Addison Wesley.
All rights reserved.

1
 Algorithmic Paradigms

Greed. Build up a solution incrementally, myopically optimizing some

local criterion.

Divide-and-conquer. Break up a problem into two sub-problems, solve

each sub-problem independently, and combine solution to sub-problems
to form solution to original problem.

Dynamic programming. Break up a problem into a series of overlapping

sub-problems, and build up solutions to larger and larger sub-problems.

Bellman. Pioneered the systematic study of dynamic programming in

the 1950s.

2
 Dynamic Programming Applications

Areas.
Bioinformatics.
Control theory.
Information theory.
Operations research.
Computer science: theory, graphics, AI, systems, ….

Some famous dynamic programming algorithms.

Viterbi for hidden Markov models.
Unix diff for comparing two files.
Smith-Waterman for sequence alignment.
Bellman-Ford for shortest path routing in networks.
Cocke-Kasami-Younger for parsing context free grammars.

3
6.1 Weighted Interval Scheduling
 Weighted Interval Scheduling

Weighted interval scheduling problem.

Job j starts at sj, finishes at fj, and has weight or value vj .
Two jobs compatible if they don't overlap.
Goal: find maximum weight subset of mutually compatible jobs.

h
Time
0 1 2 3 4 5 6 7 8 9 10 11
5
 Unweighted Interval Scheduling Review

Recall. Greedy algorithm works if all weights are 1.

Consider jobs in ascending order of finish time.
Add job to subset if it is compatible with previously chosen jobs.

Observation. Greedy algorithm can fail spectacularly if arbitrary

weights are allowed.

weight = 999 b

weight = 1 a
Time
0 1 2 3 4 5 6 7 8 9 10 11
6
 Weighted Interval Scheduling

Notation. Label jobs by finishing time: f1  f2  . . .  fn .

Def. p(j) = largest index i < j such that job i is compatible with j.

Ex: p(8) = 5, p(7) = 3, p(2) = 0.

8
Time
0 1 2 3 4 5 6 7 8 9 10 11

7
 Dynamic Programming: Binary Choice

Notation. OPT(j) = value of optimal solution to the problem consisting

of job requests 1, 2, ..., j.

Case 1: OPT selects job j.

– can't use incompatible jobs { p(j) + 1, p(j) + 2, ..., j - 1 }
– must include optimal solution to problem consisting of remaining
compatible jobs 1, 2, ..., p(j)
optimal substructure

Case 2: OPT does not select job j.

– must include optimal solution to problem consisting of remaining
compatible jobs 1, 2, ..., j-1

 0 if j = 0
OPT( j) = 
max  v j + OPT( p( j)), OPT( j −1)  otherwise

 8
Weighted Interval Scheduling: Brute Force

9
 Weighted Interval Scheduling: Brute Force

Observation. Recursive algorithm fails spectacularly because of

redundant sub-problems  exponential algorithms.

Ex. Number of recursive calls for family of "layered" instances grows

like Fibonacci sequence.

1 4 3

2
3 2 2 1
3
4
2 1 1 0 1 0
5
1 0
p(1) = 0, p(j) = j-2

10
 Weighted Interval Scheduling: Memoization

Memoization. Store results of each sub-problem in a cache; lookup as

needed.

11
 Weighted Interval Scheduling: Running Time

Claim. Memoized version of algorithm takes O(n log n) time.

Sort by finish time: O(n log n).
Computing p() : O(n logn) via binary search.

M-Compute-Opt(j): each invocation takes O(1) time and either

– (i) returns an existing value M[j]
– (ii) fills in one new entry M[j] and makes two recursive calls

Progress measure  = # nonempty entries of M[].

– initially  = 0, throughout   n.
– (ii) increases  by 1  at most 2n recursive calls.

Overall running time of M-Compute-Opt(n) is O(n).

12
 Weighted Interval Scheduling: Finding a Solution

Q. Dynamic programming algorithms computes optimal value. What if

we want the solution itself?
A. Do some post-processing.

# of recursive calls  n  O(n).

13
 Weighted Interval Scheduling: Bottom-Up

Bottom-up dynamic programming. Unwind recursion.

14
6.3 Segmented Least Squares
 Segmented Least Squares

Least squares.
Foundational problem in statistic and numerical analysis.
Given n points in the plane: (x1, y1), (x2, y2) , . . . , (xn, yn).
Find a line y = ax + b that minimizes the sum of the squared error:

n
y
SSE =  ( yi − axi − b)2
i=1


x

Solution. Calculus  min error is achieved when

n i xi yi − (i xi ) (i yi ) i yi − a i xi
a= 2
, b=
2
n i xi − (i xi ) n

16
 Segmented Least Squares

Segmented least squares.

Points lie roughly on a sequence of several line segments.
Given n points in the plane (x1, y1), (x2, y2) , . . . , (xn, yn) with
x1 < x2 < ... < xn, find a sequence of lines that minimizes f(x).

Q. What's a reasonable choice for f(x) to balance accuracy and

parsimony?
goodness of fit

number of lines

x
17
 Segmented Least Squares

Segmented least squares.

Points lie roughly on a sequence of several line segments.
Given n points in the plane (x1, y1), (x2, y2) , . . . , (xn, yn) with
x1 < x2 < ... < xn, find a sequence of lines that minimizes:
– the sum of the sums of the squared errors E in each segment
– the number of lines L
Tradeoff function: E + c L, for some constant c > 0.

x
18
 Dynamic Programming: Multiway Choice

Notation.
OPT(j) = minimum cost for points p1, p2 , . . . , pj.
e(i, j) = minimum sum of squares for points pi, pi+1 , . . . , pj.

To compute OPT(j):
Last segment uses points pi, pi+1 , . . . , pj for some i.
Cost = e(i, j) + c + OPT(i-1).

 0 if j = 0
OPT( j) =  min e(i, j) + c + OPT(i −1) otherwise
1  i  j  

19
 Segmented Least Squares: Algorithm

Running time. O(n3).

Bottleneck = computing e(i, j) for O(n2) pairs, O(n) per pair using
previous formula.

20
6.4 Knapsack Problem
 Knapsack Problem

Knapsack problem.
Given n objects and a "knapsack."
Item i weighs wi > 0 and has value vi > 0.
Knapsack has capacity of W.
Goal: fill knapsack so as to maximize total value.

Ex: { 3, 4 } has value 40. Item Value Weight

1 1 1
2 6 2
W = 11
3 18 5
4 22 6
5 28 7

Greedy: repeatedly add item with maximum ratio vi / wi.

Ex: { 5, 2, 1 } achieves only value = 35  greedy not optimal.

22
 Dynamic Programming: Adding a New Variable

Def. OPT(i, w) = max profit subset of items 1, …, i with weight limit w.

Case 1: OPT does not select item i.

– OPT selects best of { 1, 2, …, i-1 } using weight limit w

Case 2: OPT selects item i.

– new weight limit = w – wi
– OPT selects best of { 1, 2, …, i–1 } using this new weight limit

 0 if i = 0

OPT(i, w) = OPT(i −1, w) if wi  w
max OPT(i −1, w), v + OPT(i −1, w − w ) otherwise
  i i 

23
Knapsack Problem: Bottom-Up

24
 Knapsack Algorithm

W+1

0 1 2 3 4 5 6 7 8 9 10 11

 0 0 0 0 0 0 0 0 0 0 0 0
{1} 0 1 1 1 1 1 1 1 1 1 1 1

n+1 { 1, 2 } 0 1 6 7 7 7 7 7 7 7 7 7
{ 1, 2, 3 } 0 1 6 7 7 18 19 24 25 25 25 25
{ 1, 2, 3, 4 } 0 1 6 7 7 18 22 24 28 29 29 40
{ 1, 2, 3, 4, 5 } 0 1 6 7 7 18 22 28 29 34 34 40

Item Value Weight

1 1 1
OPT: { 4, 3 } 2 6 2
value = 22 + 18 = 40
W = 11 3 18 5
4 22 6
5 28 7
25
 Knapsack Problem: Running Time

Running time. (n W).

Not polynomial in input size!
"Pseudo-polynomial."
Decision version of Knapsack is NP-complete. [Chapter 8]

Knapsack approximation algorithm. There exists a polynomial algorithm

that produces a feasible solution that has value within 0.01% of
optimum. [Section 11.8]

26
6.5 RNA Secondary Structure
 RNA Secondary Structure

RNA. String B = b1b2bn over alphabet { A, C, G, U }.

Secondary structure. RNA is single-stranded so it tends to loop back

and form base pairs with itself. This structure is essential for
understanding behavior of molecule.

C A
Ex: GUCGAUUGAGCGAAUGUAACAACGUGGCUACGGCGAGA
A A

A U G C

C G U A A G

G
U A U U A
G
A C G C U
G

C G C G A G C

G
A U

G
complementary base pairs: A-U, C-G

28
 RNA Secondary Structure

Secondary structure. A set of pairs S = { (bi, bj) } that satisfy:

[Watson-Crick.] S is a matching and each pair in S is a Watson-
Crick complement: A-U, U-A, C-G, or G-C.
[No sharp turns.] The ends of each pair are separated by at least 4
intervening bases. If (bi, bj)  S, then i < j - 4.
[Non-crossing.] If (bi, bj) and (bk, bl) are two pairs in S, then we
cannot have i < k < j < l.

Free energy. Usual hypothesis is that an RNA molecule will form the
secondary structure with the optimum total free energy.
approximate by number of base pairs

Goal. Given an RNA molecule B = b1b2bn, find a secondary structure S

that maximizes the number of base pairs.

29
 RNA Secondary Structure: Examples

Examples.
G
G G G G
G G
C U C U

C G C G C U

A U A U A G

U A U A U A

base pair

A U G U G G C C A U A U G G G G C A U A G U U G G C C A U
4

ok sharp turn crossing

30
 RNA Secondary Structure: Subproblems

First attempt. OPT(j) = maximum number of base pairs in a secondary

structure of the substring b1b2bj.

match bt and bn

1 t n

Difficulty. Results in two sub-problems.

Finding secondary structure in: b1b2bt-1. OPT(t-1)

Finding secondary structure in: bt+1bt+2bn-1. need more sub-problems

31
 Dynamic Programming Over Intervals

Notation. OPT(i, j) = maximum number of base pairs in a secondary

structure of the substring bibi+1bj.

Case 1. If i  j - 4.
– OPT(i, j) = 0 by no-sharp turns condition.

Case 2. Base bj is not involved in a pair.

– OPT(i, j) = OPT(i, j-1)

Case 3. Base bj pairs with bt for some i  t < j - 4.

– non-crossing constraint decouples resulting sub-problems
– OPT(i, j) = 1 + maxt { OPT(i, t-1) + OPT(t+1, j-1) }

take max over t such that i  t < j-4 and

bt and bj are Watson-Crick complements

32
 Bottom Up Dynamic Programming Over Intervals

Q. What order to solve the sub-problems?

A. Do shortest intervals first.

4 0 0 0
3 0 0
i
2 0
1
6 7 8 9

Running time. O(n3).

33
 Dynamic Programming Summary

Recipe.
Characterize structure of problem.
Recursively define value of optimal solution.
Compute value of optimal solution.
Construct optimal solution from computed information.

Dynamic programming techniques.

Binary choice: weighted interval scheduling.
Multi-way choice: segmented least squares.
Adding a new variable: knapsack.
Dynamic programming over intervals: RNA secondary structure.

Top-down vs. bottom-up: different people have different intuitions.

34
6.6 Sequence Alignment
 String Similarity

How similar are two strings?

ocurrance o c u r r a n c e -
occurrence
o c c u r r e n c e

5 mismatches, 1 gap

o c - u r r a n c e

o c c u r r e n c e

1 mismatch, 1 gap

o c - u r r - a n c e

o c c u r r e - n c e

0 mismatches, 3 gaps

36
 Edit Distance

Applications.
Basis for Unix diff.
Speech recognition.
Computational biology.

Edit distance. [Levenshtein 1966, Needleman-Wunsch 1970]

Gap penalty ; mismatch penalty pq.
Cost = sum of gap and mismatch penalties.

C T G A C C T A C C T - C T G A C C T A C C T

C C T G A C T A C A T C C T G A C - T A C A T

TC + GT + AG+ 2CA 2 + CA

37
 Sequence Alignment

Goal: Given two strings X = x1 x2 . . . xm and Y = y1 y2 . . . yn find

alignment of minimum cost.

Def. An alignment M is a set of ordered pairs x i-yj such that each item
occurs in at most one pair and no crossings.

Def. The pair xi-yj and xi'-yj' cross if i < i', but j > j'.

cost( M ) = x y + i j
 +  
( xi , y j )  M i : x i unmatched j : y j unmatched

mismatch gap

x1 x2 x3 x4 x5 x6

Ex: CTACCG vs. TACATG. C T A C C - G

Sol: M = x2-y1, x3-y2, x4-y3, x5-y4, x6-y6.

- T A C A T G
y1 y2 y3 y4 y5 y6

38
 Sequence Alignment: Problem Structure

Def. OPT(i, j) = min cost of aligning strings x1 x2 . . . xi and y1 y2 . . . yj.

Case 1: OPT matches xi-yj.
– pay mismatch for xi-yj + min cost of aligning two strings
x1 x2 . . . xi-1 and y1 y2 . . . yj-1
Case 2a: OPT leaves xi unmatched.
– pay gap for xi and min cost of aligning x1 x2 . . . xi-1 and y1 y2 . . . yj
Case 2b: OPT leaves yj unmatched.
– pay gap for yj and min cost of aligning x1 x2 . . . xi and y1 y2 . . . yj-1

 j if i = 0
   xi y j + OPT(i −1, j −1)
 
OPT(i, j) =  min   + OPT(i −1, j) otherwise
   + OPT(i, j −1)
 
 i if j = 0

 39
 Sequence Alignment: Algorithm

Analysis. (mn) time and space.

40
6.7 Shortest Paths
 Shortest Paths

Shortest path problem. Given a directed graph G = (V, E), with edge
weights cvw, find shortest path from node s to node t.
allow negative weights

Ex.

2 10 3
9
s
18
6 -16 6
6
30 4 19
11
15 5
-8
6
20 16

7 t
44

42
 Shortest Paths: Failed Attempts

Dijkstra. Can fail if negative edge costs.

u
2 3
s v
1 -6
t

Re-weighting. Adding a constant to every edge weight can fail.

5 5
2 2
s t
6 6
3 3
0
-3

43
 Shortest Paths: Negative Cost Cycles

Negative cost cycle.

-6
-4

Observation. If some path from s to t contains a negative cost cycle,

there does not exist a shortest s-t path; otherwise, there exists
one that is simple.

s t
W

c(W) < 0

44
 Shortest Paths: Dynamic Programming

Def. OPT(i, v) = length of shortest v-t path P using at most i edges.

Case 1: P uses at most i-1 edges.

– OPT(i, v) = OPT(i-1, v)

Case 2: P uses exactly i edges.

– if (v, w) is first edge, then OPT uses (v, w), and then selects best
w-t path using at most i-1 edges

 0 if i = 0

OPT(i, v) =   
 min  OPT(i −1, v) , min  OPT(i −1, w) + cvw   otherwise
 (v, w)  E 

 Remark. By previous observation, if no negative cycles, then

OPT(n-1, v) = length of shortest v-t path.

45
 Shortest Paths: Implementation

Analysis. (mn) time, (n2) space.

Finding the shortest paths. Maintain a "successor" for each table
entry.
46