Exact String Matching: KMP & Boyer-Moore Algorithms: R.C.T. Lee and Chin Lung Lu

Exact String Matching: KMP & Boyer-Moore Algorithms
R.C.T. Lee and Chin Lung Lu CS 5313 Algorithms for Molecular Biology
By R.C.T. Lee and C.L. Lu
Exact String Matching p.1
Exact string matching

Given two strings T and P , where |T | = m and |P | = n, nd all the occurrences of P in T ? (Simplied version) Given two strings T and P , if P occurs in T ? Brute-force method: O(mn) KMP algorithm (Knuth, Morris and Pratt, 1977): O(m + n) Boyer-Moore algorithm (Boyer and Moore, 1977): O(m + n)
Brute-force method
m T P
O(m n + 1) times Time complexity of brute-force method: O(mn)

By R.C.T. Lee and C.L. Lu Exact String Matching p.3
Brute-force method
T P a b a c a b a b a b a b a b a b a b a b a b a b a b a b Simple, but not ecient because it cannot avoid rescanning T
How to speedup the brute-force method?

Shift P by more than one place when a mismatch occurs without missing any occurrence of P in T 1. KMP algorithm Proposed by Knuth, Morris and Pratt at 1977 Time complexity: O(m + n) 2. Boyer-Moore algorithm Proposed by Boyer and Moore at 1977 Time complexity: O(m + n)
KMP algorithm
Left-to-right scan Failure function shift rule
KMP algorithm
T P shift a b a c a b a b Left-to-Right Scan a b a b mismatch occurs a b a b a b a b a b a b shift
shift
a b a b
Failure function
If P = p1 p2 pn , then its failure function f is dened as follows, where 1 j n. if such an largest i < j such that p1 pi = pji+1 pj , i 1 exists, f (j) = 0, otherwise. i P 1
i i (j i + 1) j
Examples of failure function

Example 1: P a b a b a b c f (j) 0 0 1 2 3 4 0 Example 2: P a b a c a b a b f (j) 0 0 1 0 1 2 3 2
KMP algorithm
/* 1 2 3 4 5 6 7 8 9 T = T1 T2 Tm and P = p1 p2 pn */ i = 1; j = 1; while i m and j n do if Ti = pj then i = i + 1; j = j + 1; else if j = 1 then i = i + 1; j = 1; else j = f (j 1) + 1;
end if end while if j = n +
1 then a match else no match.
KMP algorithm
T abac f (j) 0 0 1 2 P abab 1 2 3 4 5 6 7 8
abab
(1, 1) (4, 4), j = f (3) + 1 = 2
abab abab abab abab
No this step
(4, 2) (4, 2), j = f (1) + 1 = 1 (4, 1) (4, 1), i = 4 + 1 = 5, j = 1 (5, 1) (8, 4), i = 8 + 1, j = 4 + 1
The correctness of KMP algorithm

Tjk Tj1 T f (j 1) P p1 pk pjk pj1
pj Tj
First mismatch (Tj = pj ) before shift This move is redundant

pj
pj
after shift
Then p1 pk = Tjk Tj1 = pjk pj1 Hence f (j 1) = k > f (j 1), a contradiction.

How to compute failure function?

Let f 1 (j) = f (j) and f k (j) = f (f k1 (j)). Let P = p1 p2 pn . Then, f (1) = 0 and for j 2, k f (j 1) + 1 if there exists k which is the least integer such that f (j) = pf k (j1)+1 = pj , 0 otherwise.

If pf (j1)+1 = pj , then f (j) = f (j 1) + 1. f (j) P f (j 1) pf (j1)+1 f (j 1) f (j) P
f (j) pj
f (j) pj

If pf (j1)+1 = pj and pf (f (j1))+1 = pj , then f (j) = f (f (j 1)) + 1. f (j) f (j) P f (j 1) pf (j1)+1 f (j 1) P f (f (j 1)) + 1
pj
pj

f (j 1) pf (j1)+1 f (j 1) P P Zoom In pj pj
f (f (j 1))
f (j 1)
Boyer-Moore algorithm
Right-to-left scan Bad character shift rule Good sux shift rule
Right-to-left scan
Check whether P occurs in T at some position in the right-to-left scaning manner THE BOYERMOORE ALGORITHM Right-to-Left Scan P BUYERMOORE
Bad character shift rule

What happened if the initial mismatch occurs? THE BOYERMOORE ALGORITHM T Right-to-Left Scan P shift P P
BUYERMOORE movement BUYERMOORE Thisbe skipped can BUYERMOORE


Bad character rule: If x = y, then P is shifted right by max{1, i B(x)} places. B(x): the position of right-most occurrence of x in P (B(x) = 0 if x does not occur in P ) T P before shift P after shift
x x
B(x)
t t y t
y
i
B ) (x

x x
B(x)
T P before shift B(x) < i

i
t
Right-to-Left Scan
y
i
t P after shift y t x P after shift y t

B(x) = 0

If B(x) > i, then bad character rule has no eect.
B ) (x
T P before shift y
x y
i
t x
Right-to-Left Scan
t x
B(x)
t x
P after shift ()

How to compute B(x) for all x in P ? (n = |P |) for i = 1 to n do B(x) = 0; end for for i = n to 1 do if B(P [i]) = 0 then B(P [i]) = i; end for Extended bad character shift rule: For each position i in P and for each x in , nd the position of the closest occurrence of x in P to the left of i.
Good sux rule: Case

If x = y, then nd the right-most copy t of t in P such that t is not a sux of P and z = y T P before shift z t x y t
Right-to-Left Scan
t y t
P after shift
Good sux rule: Case

If t does not exist, then nd the largest prex t of P such that it is equal to a sux of t T P before shift t xt Right-to-Left Scan y t t y t
P after shift
Good sux rule: Case

If t and t do not exist, then T P before shift P after shift xt Right-to-Left Scan y t y t
Good sux rule

T z t xt Right-to-Left Scan y z t t t P before shift y t y t y t
P after shift for case


Exact String Matching: KMP & Boyer-Moore Algorithms: R.C.T. Lee and Chin Lung Lu

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Exact String Matching: KMP & Boyer-Moore Algorithms: R.C.T. Lee and Chin Lung Lu

Uploaded by

Copyright:

Available Formats

Exact String Matching: KMP & Boyer-Moore Algorithms

By R.C.T. Lee and C.L. Lu

Exact String Matching p.1