Dynamic

Dynamic Programming
Programming

Comp 122, Fall 2004

 Longest Common Subsequence
 Problem: Given 2 sequences, X = x1,...,xm and
Y = y1,...,yn, find a common subsequence whose
length is maximum.

springtime ncaa tournament basketball

printing north carolina krzyzewski

Subsequence need not be consecutive, but must be in order.

Comp 122, Spring Lin / Devi
nprog - 2
2004
 Other sequence questions
 Edit distance: Given 2 sequences, X = x1,...,xm
and Y = y1,...,yn, what is the minimum number of
deletions, insertions, and changes that you must do
to change one to another?
 Protein sequence alignment: Given a score
matrix on amino acid pairs, s(a,b) for
a,b{}A,
and 2 amino acid sequences, X = x1,...,xmAm
and Y = y1,...,ynAn, find the alignment with
lowest score…
Comp 122, Spring Lin / Devi
nprog - 3
2004
 More problems
Optimal BST: Given sequence K = k1 < k2 <··· < kn
of n sorted keys, with a search probability pi for
each key ki, build a binary search tree (BST) with
minimum expected search cost.
Matrix chain multiplication: Given a sequence of
matrices A1 A2 … An, with Ai of dimension mini,
insert parenthesis to minimize the total number of
scalar multiplications.
Minimum convex decomposition of a polygon,
Hydrogen placement in protein structures, …
Comp 122, Spring Lin / Devi
nprog - 4
2004
 Dynamic Programming
 Dynamic Programming is an algorithm design technique for
optimization problems: often minimizing or maximizing.
 Like divide and conquer, DP solves problems by combining
solutions to subproblems.
 Unlike divide and conquer, subproblems are not independent.
» Subproblems may share subsubproblems,
» However, solution to one subproblem may not affect the solutions to other
subproblems of the same problem. (More on this later.)
 DP reduces computation by
» Solving subproblems in a bottom-up fashion.
» Storing solution to a subproblem the first time it is solved.
» Looking up the solution when subproblem is encountered again.
 Key: determine structure of optimal solutions

Comp 122, Spring Lin / Devi

nprog - 5
2004
 Steps in Dynamic Programming
1. Characterize structure of an optimal solution.
2. Define value of optimal solution recursively.
3. Compute optimal solution values either top-
down with caching or bottom-up in a table.
4. Construct an optimal solution from computed
values.
We’ll study these with the help of examples.

Comp 122, Spring Lin / Devi

nprog - 6
2004
 Longest Common Subsequence
 Problem: Given 2 sequences, X = x1,...,xm and
Y = y1,...,yn, find a common subsequence whose
length is maximum.

springtime ncaa tournament basketball

printing north carolina snoeyink

Subsequence need not be consecutive, but must be in order.

Comp 122, Spring Lin / Devi
nprog - 7
2004
 Naïve Algorithm
 For every subsequence of X, check whether it’s a
subsequence of Y .
 Time: Θ(n2m).
» 2m subsequences of X to check.
» Each subsequence takes Θ(n) time to check:
scan Y for first letter, for second, and so on.

Comp 122, Spring Lin / Devi

nprog - 8
2004
 Optimal Substructure
Theorem
Theorem
LetZZ==zz1,,......,,zzkbe
Let
1 k beany
anyLCS
LCSof
ofXXand
andYY..
1.1.IfIfxxm ==yyn,,then z = x = y and Z isisan anLCS
LCSof
ofXXm-1 and Y ..
m n then zkk = xmm = ynn and Zk-1
k-1 m-1 and Yn-1
n-1
2.2.IfIfxxm yyn,,then eitherzzk xxm and
theneither andZZisisan
anLCS
LCSof
ofXXm-1 and
andYY..
m n k m m-1
3.3. or zzkkyynnand
or andZZisisan
anLCS
LCSof
ofXXand
andYYn-1
n-1
..
Notation:
prefix Xi = x1,...,xi is the first i letters of X.
This says what any longest common subsequence must look like;
do you believe it?

Comp 122, Spring Lin / Devi

nprog - 9
2004
 Optimal Substructure
Theorem
Theorem
LetZZ==zz1,,......,,zzkbe
Let
1 k beany
anyLCS
LCSof
ofXXand
andYY..
1.1.IfIfxxm ==yyn,,then z = x = y and Z isisan anLCS
LCSof
ofXXm-1 and Y ..
m n then zkk = xmm = ynn and Zk-1
k-1 m-1 and Yn-1
n-1
2.2.IfIfxxm yyn,,then eitherzzk xxm and
theneither andZZisisan
anLCS
LCSof
ofXXm-1 and
andYY..
m n k m m-1
3.3. or zzkkyynnand
or andZZisisan
anLCS
LCSof
ofXXand
andYYn-1
n-1
..
Proof: (case 1: xm = yn)
Any sequence Z’ that does not end in xm = yn can be made longer by adding xm = yn to the
end. Therefore,
(1) longest common subsequence (LCS) Z must end in xm = yn.
(2) Zk-1 is a common subsequence of Xm-1 and Yn-1, and
(3) there is no longer CS of Xm-1 and Yn-1, or Z would not be an LCS.

Comp 122, Spring Lin / Devi

nprog - 10
2004
 Optimal Substructure
Theorem
Theorem
LetZZ==zz1,,......,,zzkbe
Let
1 k beany
anyLCS
LCSof
ofXXand
andYY..
1.1.IfIfxxm ==yyn,,then z = x = y and Z isisan anLCS
LCSof
ofXXm-1 and Y ..
m n then zkk = xmm = ynn and Zk-1
k-1 m-1 and Yn-1
n-1
2.2.IfIfxxm yyn,,then eitherzzk xxm and
theneither andZZisisan
anLCS
LCSof
ofXXm-1 and
andYY..
m n k m m-1
3.3. or zzkkyynnand
or andZZisisan
anLCS
LCSof
ofXXand
andYYn-1
n-1
..
Proof: (case 2: xm  yn, and zk  xm)
Since Z does not end in xm,
(1) Z is a common subsequence of Xm-1 and Y, and
(2) there is no longer CS of Xm-1 and Y, or Z would not be an LCS.

Comp 122, Spring Lin / Devi

nprog - 11
2004
 Recursive Solution
 Define c[i, j] = length of LCS of Xi and Yj .
 We want c[m,n].

00 ifif ii 00or or jj 00,,


cc[[ii,, jj]]cc[[ii11,, jj11]]11 ifif ii,, jj 00and
and xxi i  yyj j,,
max(c[i  1, j ], c[i, j  1]) ifif ii,, jj 00and
max(c[i  1, j ], c[i, j  1]) and xxi i  yyj j..

This gives a recursive algorithm and solves the problem.

But does it solve it well?

Comp 122, Spring Lin / Devi

nprog - 12
2004
 Recursive Solution
00 ififempty
emptyor or empty
empty,,

cc[[,,]]cc[[prefix
prefix,,prefix
prefix]]11 end())end(
ifif end( end()),,
max(c[ prefix ,  ], c[ , prefix ])
max(c[ prefix ,  ], c[ , prefix ]) end())end(
ifif end( end())..

c[springtime, printing]

c[springtim, printing] c[springtime, printin]

[springti, printing] [springtim, printin] [springtim, printin] [springtime, printi]

[springt, printing] [springti, printin] [springtim, printi] [springtime, print]

Comp 122, Spring Lin / Devi

nprog - 13
2004
 Recursive Solution
00 ififempty
emptyor or empty
empty,,

cc[[,,]]cc[[prefix
prefix,,prefix
prefix]]11 end())end(
ifif end( end()),,
max(c[ prefix ,  ], c[ , prefix ])
max(c[ prefix ,  ], c[ , prefix ]) end())end(
ifif end( end())..
p r i n t i n g
•Keep track of c[] in a
table of nm entries: s

•top/down p
r
•bottom/up i
n
g
t
i
m
Comp 122, Spring Lin / Devi
nprog - 14 e
2004
 Computing the length of an LCS
LCS-LENGTH
LCS-LENGTH(X, (X,Y) Y)
1.1. mm←←length[X]
length[X]
2.2. nn←←length[Y]
length[Y]
3.3. for
forii←
←11to tomm
4.4. do
doc[i,
c[i,0]
0]← ←00
5.5. for
forjj←
←00to tonn
6.6. do
doc[0,
c[0,jj]]← ←00 b[i, j ] points to table entry
7.7. for
forii←
←11to tomm whose subproblem we used
8.8. do
dofor
forjj← ←11to tonn
9.9. do in solving LCS of Xi
doififxxi i==yyj j
10.
10. then
thenc[i, c[i,jj]]← ←c[i c[i1,1,j
j1]1]++11 and Yj.
11.
11. b[i,
b[i,jj]]← ←““ ””
12.
12. else
elseififc[i c[i1,1,jj]]≥≥c[i,
c[i,j j1]
1] c[m,n] contains the length
13.
13. then
thenc[i, c[i,jj]]←←c[i c[i1,1,jj]] of an LCS of X and Y.
14.
14. b[i,
b[i,jj]]← ←“↑” “↑”
15.
15. else
elsec[i,
c[i,jj]]← ←c[i,c[i,j j1]
1]
16.
16. b[i,
b[i,jj]]← ←“←” “←”
Time: O(mn)
17.
17.return
returnccandandbb
Comp 122, Spring Lin / Devi
nprog - 15
2004
 Constructing an LCS
PRINT-LCS
PRINT-LCS(b, (b,X,
X,i,i,j)j)
1.1. ififii==00ororjj==00
2.2. then thenreturn
return
3.3. ififb[i,
b[i,jj]]==““ ””
4.4. then thenPRINT-LCS(b,
PRINT-LCS(b,X, X,ii1,
1,jj1)
1)
5.5. print
printxxi i
6.6. elseif
elseifb[i,
b[i,jj]]==“↑”
“↑”
7.7. then
thenPRINT-LCS(b,
PRINT-LCS(b,X, X,ii1,
1,j)j)
8.8. else
elsePRINT-LCS(b,
PRINT-LCS(b,X, X,i,i,jj1)
1)

•Initial call is PRINT-LCS (b, X,m, n).

•When b[i, j ] = , we have extended LCS by one character. So
LCS = entries with in them.
•Time: O(m+n)
Comp 122, Spring Lin / Devi
nprog - 16
2004
 Steps in Dynamic Programming
1. Characterize structure of an optimal solution.
2. Define value of optimal solution recursively.
3. Compute optimal solution values either top-
down with caching or bottom-up in a table.
4. Construct an optimal solution from computed
values.
We’ll study these with the help of examples.

Comp 122, Spring Lin / Devi

nprog - 17
2004
 Optimal Binary Search Trees
 Problem
» Given sequence K = k1 < k2 <··· < kn of n sorted keys,
with a search probability pi for each key ki.
» Want to build a binary search tree (BST)
with minimum expected search cost.
» Actual cost = # of items examined.
» For key ki, cost = depthT(ki)+1, where depthT(ki) = depth of ki in
BST T .

Comp 122, Spring Lin / Devi

nprog - 18
2004
 Expected Search Cost
E[search cost in T ]
n
  (depth T (ki )  1)  pi
i 1
n n
  depth T (ki )  pi   pi
i 1 i 1
n Sum of probabilities is 1.
 1   depth T (ki )  pi (15.16)
i 1

Comp 122, Spring Lin / Devi

nprog - 19
2004
 Example
 Consider 5 keys with these search probabilities:
p1 = 0.25, p2 = 0.2, p3 = 0.05, p4 = 0.2, p5 = 0.3.
k2 i depthT(ki) depthT(ki)·pi
1 1 0.25
2 0 0
k1 k4 3 2 0.1
4 1 0.2
5 2 0.6
1.15
k3 k5
Therefore, E[search cost] = 2.15.

Comp 122, Spring Lin / Devi

nprog - 20
2004
 Example
 p1 = 0.25, p2 = 0.2, p3 = 0.05, p4 = 0.2, p5 = 0.3.

k2 i depthT(ki) depthT(ki)·pi
1 1 0.25
2 0 0
k1 k5 3 3 0.15
4 2 0.4
5 1 0.3
1.10
k4
Therefore, E[search cost] = 2.10.

k3 This tree turns out to be optimal for this set of keys.

Comp 122, Spring Lin / Devi
nprog - 21
2004
 Example
 Observations:
» Optimal BST may not have smallest height.
» Optimal BST may not have highest-probability key at
root.
 Build by exhaustive checking?
» Construct each n-node BST.
» For each,
assign keys and compute expected search cost.
» But there are (4n/n3/2) different BSTs with n nodes.

Comp 122, Spring Lin / Devi

nprog - 22
2004
 Optimal Substructure
 Any subtree of a BST contains keys in a contiguous range
ki, ..., kj for some 1 ≤ i ≤ j ≤ n.

T

 If T is an optimal BST and

T contains subtree T with keys ki, ... ,kj ,
then T must be an optimal BST for keys ki, ..., kj.
 Proof: Cut and paste.
Comp 122, Spring Lin / Devi
nprog - 23
2004
 Optimal Substructure
 One of the keys in ki, …,kj, say kr, where i ≤ r ≤ j,
must be the root of an optimal subtree for these keys.
 Left subtree of kr contains ki,...,kr1.
k
 Right subtree of kr contains kr+1, ...,kj.
r

ki kr-1 kr+1 kj
 To find an optimal BST:
» Examine all candidate roots kr , for i ≤ r ≤ j
» Determine all optimal BSTs containing ki,...,kr1 and
containing kr+1,...,kj
Comp 122, Spring Lin / Devi
nprog - 24
2004
 Recursive Solution
 Find optimal BST for ki,...,kj, where i ≥ 1, j ≤ n, j ≥ i1.
When j = i1, the tree is empty.
 Define e[i, j ] = expected search cost of optimal BST for ki,...,kj.

 If j = i1, then e[i, j ] = 0.

 If j ≥ i,
» Select a root kr, for some i ≤ r ≤ j .
» Recursively make an optimal BSTs
• for ki,..,kr1 as the left subtree, and
• for kr+1,..,kj as the right subtree.

Comp 122, Spring Lin / Devi

nprog - 25
2004
 Recursive Solution
 When the OPT subtree becomes a subtree of a node:
» Depth of every node in OPT subtree goes up by 1.
» Expected search cost increases by
jj
ww((ii,, jj))
ppl l from (15.16)
l l i i

 If kr is the root of an optimal BST for ki,..,kj :

» e[i, j ] = pr + (e[i, r1] + w(i, r1))+(e[r+1, j] + w(r+1, j))
= e[i, r1] + e[r+1, j] + w(i, j). (because w(i, j)=w(i,r1) + p r + w(r + 1, j))

 But, we don’t know kr. Hence,

00 ifif jjii11
ee[[ii, ,jj]]
min {e[i, r  1]  e[r  1, j ]  w(i, j )} ififii jj
r  j {e[i , r  1]  e[ r  1, j ]  w(i , j )}
i min
ir  j
Comp 122, Spring Lin / Devi
nprog - 26
2004
 Computing an Optimal Solution
For each subproblem (i,j), store:
 expected search cost in a table e[1 ..n+1 , 0 ..n]
» Will use only entries e[i, j ], where j ≥ i1.
 root[i, j ] = root of subtree with keys ki,..,kj, for 1 ≤ i ≤ j ≤ n.
 w[1..n+1, 0..n] = sum of probabilities
» w[i, i1] = 0 for 1 ≤ i ≤ n.
» w[i, j ] = w[i, j-1] + pj for 1 ≤ i ≤ j ≤ n.

Comp 122, Spring Lin / Devi

nprog - 27
2004
 Pseudo-code
OPTIMAL-BST(p,
OPTIMAL-BST(p,q,q,n) n)
1.1. for
forii←
←11to tonn++11
2.2. do e[i,ii1]
doe[i, 1]← ←00 Consider all trees with l keys.
3.3. w[i, i  1] ←
w[i, i 1] ← 0 0
4.4. for Fix the first key.
forll←
←11to tonn
5.5. do
dofor
forii← ←11to tonnll++11 Fix the last key
6.6. do
dojj←i ←i++ll11
7.7. e[i,
e[i,jj]←∞
]←∞
8.8. w[i,
w[i,jj]]← ←w[i,w[i,jj1]
1]++ppj j
9.9. for
forrr←i ←ito tojj
10.
10. dodott← ←e[i,e[i,rr1]
1]++e[r
e[r++1,1,jj]]++w[i,
w[i,jj]] Determine the root
11.
11. ififtt<<e[i,
e[i,jj]] of the optimal
12.
12. then
thene[i,
e[i,jj]]←
←tt (sub)tree
13.
13. root[i,
root[i,jj]]←r
←r
14.
14. return
returneeand
androotroot

Time: O(n3)
Comp 122, Spring Lin / Devi
nprog - 28
2004
 Elements of Dynamic Programming
 Optimal substructure
 Overlapping subproblems

Comp 122, Spring Lin / Devi

nprog - 29
2004
 Optimal Substructure
 Show that a solution to a problem consists of making a
choice, which leaves one or more subproblems to solve.
 Suppose that you are given this last choice that leads to an
optimal solution.
 Given this choice, determine which subproblems arise and
how to characterize the resulting space of subproblems.
 Show that the solutions to the subproblems used within
the optimal solution must themselves be optimal. Usually
use cut-and-paste.
 Need to ensure that a wide enough range of choices and
subproblems are considered.
Comp 122, Spring Lin / Devi
nprog - 30
2004
 Optimal Substructure
 Optimal substructure varies across problem domains:
» 1. How many subproblems are used in an optimal solution.
» 2. How many choices in determining which subproblem(s) to use.
 Informally, running time depends on (# of subproblems
overall)  (# of choices).
 How many subproblems and choices do the examples
considered contain?
 Dynamic programming uses optimal substructure bottom
up.
» First find optimal solutions to subproblems.
» Then choose which to use in optimal solution to the problem.

Comp 122, Spring Lin / Devi

nprog - 31
2004
 Optimal Substucture
 Does optimal substructure apply to all optimization
problems? No.
 Applies to determining the shortest path but NOT the
longest simple path of an unweighted directed graph.
 Why?
» Shortest path has independent subproblems.
» Solution to one subproblem does not affect solution to another
subproblem of the same problem.
» Subproblems are not independent in longest simple path.
• Solution to one subproblem affects the solutions to other subproblems.
» Example:

Comp 122, Spring Lin / Devi

nprog - 32
2004
 Overlapping Subproblems
 The space of subproblems must be “small”.
 The total number of distinct subproblems is a polynomial
in the input size.
» A recursive algorithm is exponential because it solves the same
problems repeatedly.
» If divide-and-conquer is applicable, then each problem solved
will be brand new.

Comp 122, Spring Lin / Devi

nprog - 33
2004