CS 332 - Algorithms Dynamic programming Longest Common Subsequence

7/17/2011

1

Dynamic programming 

It is used, when the solution can be recursively described in terms of solutions to subproblems (optimal substructure) Algorithm finds solutions to subproblems and stores them in memory for later use More efficient than ³brute-force methods´, which solve the same subproblems over and over again
2  

7/17/2011

Longest Common Subsequence (LCS)
Application: comparison of two DNA strings Ex: X= {A B C B D A B }, Y= {B D C A B A} Longest Common Subsequence: X= AB C BDAB BDCAB A Y= Brute force algorithm would compare each subsequence of X with the symbols in Y
7/17/2011 3

|Y| = n. we must compare each with Y (n comparisons) So the running time of the brute-force algorithm is O(n 2m) Notice that the LCS problem has optimal substructure: solutions of subproblems are parts of the final solution. Subproblems: ³find LCS of pairs of prefixes of X and Y´ 4 7/17/2011 .LCS Algorithm     if |X| = m. then there are 2m subsequences of x.

j  1]. Yj to be the prefixes of X and Y of length i and j respectively Define c[i. j ]) other ise ° 7/17/2011 5 . Define Xi. j ] ! ¯ max(c[i.LCS Algorithm     First we¶ll find the length of LCS.n] i x[i ] ! y[ j ]. c ®[i  1. c[i  1.j] to be the length of LCS of Xi and Yj Then the length of LCS of X and Y will be c[m. j  1]  1 c[i. Later we¶ll modify the algorithm to find LCS itself.

c[i  1.e. j] = c[i. ®[i  1. j ]) other ise °  We start with i = j = 0 (empty substrings of x and y) Since X0 and Y0 are empty strings.LCS recursive solution c i x[i ] ! y[ j ]. so for every i and j: c[0. j  1].0] = 0) LCS of empty string and any other string is empty. c[0. their LCS is always empty (i. j  1]  1 c[i.0] = 0 6   7/17/2011 . j ] ! ¯ max(c[i.

c[i. c[i 1. j ] ¯ max(c[i.LCS recursive solution c[i 1. j 1]. j ]) otherwise °  When we calculate c[i. we consider two cases: First case: x[i]=y[j]: one more symbol in strings X and Y matches.j]. j 1] 1 if x[i ] y[ j ]. so the length of LCS Xi and Yj equals to the length of LCS of  smaller strings Xi-1 and Yi-1 . plus 1 7/17/2011 7 .

and the length of LCS(Xi . c[i  1. j  1]. maximum of LCS(Xi.Yj) Why not just take the length of LCS(Xi-1.e. Yj) is the same as before (i. j  1]  1 c[i. Yj-1) and LCS(Xi-1.LCS recursive solution c if x[i ] ! y[ j ]. j ] ! ¯ max(c[i. Yj-1) ? 8 7/17/2011 . ®[i  1. our solution is not improved. j ]) other ise °   Second case: x[i] != y[j] As symbols don¶t match.

0] = 0 // special case: Y0 4.j-1] ) 7/17/2011 10. for j = 1 to n c[0. n = length(Y) // get the # of symbols in Y 3.j] = max( c[i-1. c[i. for i = 1 to m // for all Xi 6. if ( Xi == Yj ) 8. for j = 1 to n // for all Yj 7. for i = 1 to m c[i.j-1] + 1 9. return c 9 . c[i. Y) 1.j] = 0 // special case: X0 5.j] = c[i-1.LCS Length Algorithm LCS-Length(X.j]. else c[i. m = length(X) // get the # of symbols in X 2.

LCS Example We¶ll see how LCS algorithm works on the following example:  X = ABCB  Y = BDCAB What is the Longest Common Subsequence of X and Y? LCS(X. Y) = BCB X=AB C B 7/17/2011 Y= BDCAB 10 .

4] 7/17/2011 11 .LCS Example (0) j i 0 1 2 3 4 Xi A B C B 0 Yj 1 B 2 D 3 C 4 A ABCB BDCAB 5 B X = ABCB. n = |Y| = 5 Allocate array c[5. m = |X| = 4 Y = BDCAB.

LCS Example (1) j i 0 1 2 3 4 Xi A B C B 0 Yj 0 0 0 0 0 1 B 0 2 D 0 3 C 0 4 A 0 ABCB BDCAB 5 B 0 for i = 1 to m for j = 1 to n 7/17/2011 c[i.0] = 0 c[0.j] = 0 12 .

j] = c[i-1.j].LCS Example (2) j i 0 1 2 3 4 Xi A B C B 0 Yj 0 0 0 0 0 1 B 0 0 2 D 0 3 C 0 4 A 0 ABCB BDCAB 5 B 0 if ( Xi == Yj ) c[i.j-1] + 1 else c[i.j] = max( c[i-1. c[i.j-1] ) 7/17/2011 13 .

c[i.j] = c[i-1.j-1] + 1 else c[i.j-1] ) 7/17/2011 14 .j] = max( c[i-1.j].LCS Example (3) j i 0 1 2 3 4 Xi A B C B 0 Yj 0 0 0 0 0 1 B 0 0 2 D 0 0 3 C 0 0 4 A 0 ABCB BDCAB 5 B 0 if ( Xi == Yj ) c[i.

j] = max( c[i-1.j-1] ) 7/17/2011 15 .LCS Example (4) j i 0 1 2 3 4 Xi A B C B 0 Yj 0 0 0 0 0 1 B 0 0 2 D 0 0 3 C 0 0 4 A 0 1 ABCB BDCAB 5 B 0 if ( Xi == Yj ) c[i.j] = c[i-1. c[i.j-1] + 1 else c[i.j].

j-1] ) 7/17/2011 16 .LCS Example (5) j i 0 1 2 3 4 Xi A B C B 0 Yj 0 0 0 0 0 1 B 0 0 2 D 0 0 3 C 0 0 4 A 0 1 ABCB BDCAB 5 B 0 1 if ( Xi == Yj ) c[i.j] = max( c[i-1.j] = c[i-1. c[i.j].j-1] + 1 else c[i.

j] = c[i-1.j] = max( c[i-1.j].j-1] ) 7/17/2011 17 .j-1] + 1 else c[i. c[i.LCS Example (6) j i 0 1 2 3 4 Xi A B C B 0 Yj 0 0 0 0 0 1 B 0 0 1 2 D 0 0 3 C 0 0 4 A 0 1 ABCB BDCAB 5 B 0 1 if ( Xi == Yj ) c[i.

j-1] ) 7/17/2011 18 .j-1] + 1 else c[i.j] = max( c[i-1.LCS Example (7) j i 0 1 2 3 4 Xi A B C B 0 Yj 0 0 0 0 0 1 B 0 0 1 2 D 0 0 1 3 C 0 0 1 4 A 0 1 1 ABCB BDCAB 5 B 0 1 if ( Xi == Yj ) c[i. c[i.j].j] = c[i-1.

j] = max( c[i-1.j-1] + 1 else c[i.j-1] ) 7/17/2011 19 .j]. c[i.LCS Example (8) j i 0 1 2 3 4 Xi A B C B 0 Yj 0 0 0 0 0 1 B 0 0 1 2 D 0 0 1 3 C 0 0 1 4 A 0 1 1 ABCB BDCAB 5 B 0 1 2 if ( Xi == Yj ) c[i.j] = c[i-1.

LCS Example (10) j i 0 1 2 3 4 Xi A B C B 0 Yj 0 0 0 0 0 1 B 0 0 1 1 2 D 0 0 1 1 3 C 0 0 1 4 A 0 1 1 ABCB BDCAB 5 B 0 1 2 if ( Xi == Yj ) c[i.j-1] + 1 else c[i.j] = c[i-1. c[i.j] = max( c[i-1.j].j-1] ) 7/17/2011 20 .

c[i.j-1] ) 7/17/2011 21 .LCS Example (11) j i 0 1 2 3 4 Xi A B C B 0 Yj 0 0 0 0 0 1 B 0 0 1 1 2 D 0 0 1 1 3 C 0 0 1 2 4 A 0 1 1 ABCB BDCAB 5 B 0 1 2 if ( Xi == Yj ) c[i.j].j] = max( c[i-1.j-1] + 1 else c[i.j] = c[i-1.

j] = max( c[i-1. c[i.j] = c[i-1.j-1] ) 7/17/2011 22 .j].LCS Example (12) j i 0 1 2 3 4 Xi A B C B 0 Yj 0 0 0 0 0 1 B 0 0 1 1 2 D 0 0 1 1 3 C 0 0 1 2 4 A 0 1 1 2 ABCB BDCAB 5 B 0 1 2 2 if ( Xi == Yj ) c[i.j-1] + 1 else c[i.

j] = c[i-1.j-1] + 1 else c[i.LCS Example (13) j i 0 1 2 3 4 Xi A B C B 0 Yj 0 0 0 0 0 1 B 0 0 1 1 1 2 D 0 0 1 1 3 C 0 0 1 2 4 A 0 1 1 2 ABCB BDCAB 5 B 0 1 2 2 if ( Xi == Yj ) c[i.j-1] ) 7/17/2011 23 .j] = max( c[i-1. c[i.j].

j].j-1] ) 7/17/2011 24 .j] = c[i-1.j-1] + 1 else c[i.j] = max( c[i-1. c[i.LCS Example (14) j i 0 1 2 3 4 Xi A B C B 0 Yj 0 0 0 0 0 1 B 0 0 1 1 1 2 D 0 0 1 1 1 3 C 0 0 1 2 2 4 A 0 1 1 2 2 ABCB BDCAB 5 B 0 1 2 2 if ( Xi == Yj ) c[i.

j-1] ) 7/17/2011 25 .j].j] = c[i-1.LCS Example (15) j i 0 1 2 3 4 Xi A B C B 0 Yj 0 0 0 0 0 1 B 0 0 1 1 1 2 D 0 0 1 1 1 3 C 0 0 1 2 2 4 A 0 1 1 2 2 ABCB BDCAB 5 B 0 1 2 2 3 if ( Xi == Yj ) c[i.j] = max( c[i-1.j-1] + 1 else c[i. c[i.

j] is calculated in constant time.n] So what is the running time? O(m*n) since each c[i.LCS Algorithm Running Time   LCS algorithm calculates the values of each entry of the array c[m. and there are m*n elements in the array 7/17/2011 26 .

but not LCS itself. we have just found the length of LCS.j] and c[i.j] depends on c[i-1. here c[i.How to find actual LCS So far.  We want to modify this algorithm to make it output Longest Common Subsequence of X and Y Each c[i.j] we can say how it was acquired:  2 7/17/2011 2 3 2 For example.j] = c[i-1. j-1] For each c[i.j-1] or c[i-1.j-1] +1 = 2+1=3 27 .

j  1]. j ] ! ¯ max(c[i.How to find actual LCS . j  1]  1 c[i.continued  Remember that c i x[i ] ! y[ j ].e. remember x[i] (because x[i] is a part of LCS) When i=0 or j=0 (i. ®[i  1. we reached the beginning). j ]) other ise °    So we can start from c[m.j] = c[i-1. output remembered letters in reverse order 28 7/17/2011 . c[i  1. j-1]+1.n] and go backwards Whenever c[i.

Finding LCS j i 0 1 2 3 4 Xi A B C B 0 Yj 0 0 0 0 0 1 B 0 0 1 1 1 2 D 0 0 1 1 1 3 C 0 0 1 2 2 4 A 0 1 1 2 2 5 B 0 1 2 2 3 7/17/2011 29 .

Finding LCS (2) j i 0 1 2 3 4 Xi A B C B 0 Yj 0 0 0 0 0 1 B 0 0 1 1 1 2 D 0 0 1 1 1 3 C 0 0 1 2 2 4 A 0 1 1 2 2 5 B 0 1 2 2 3 LCS (reversed order): B C B LCS (straight order): B C B 7/17/2011 string turned out to be a palindrome) (this 30 .

Each item has some weight and some value. Total weight that we can carry is no more than some fixed number W.Knapsack problem Given some items. So we must consider weights of items as well as their value. Item # 1 2 3 Weight Value 1 8 3 6 5 5 7/17/2011 31 . pack the knapsack to get the maximum total value.

you either take an item or not. Solved with a greedy algorithm. Solved with dynamic programming (2) Items are divisible: you can take any fraction of an item.Knapsack problem There are two versions of the problem: (1) ³0-1 knapsack problem´ and (2) ³Fractional knapsack problem´ (1) Items are indivisible. 7/17/2011 32 .

Sign up to vote on this title
UsefulNot useful

Master Your Semester with Scribd & The New York Times

Special offer for students: Only $4.99/month.

Master Your Semester with a Special Offer from Scribd & The New York Times

Cancel anytime.