Algorithms and Theory of Computation. Lecture 2: Big-O Notation Graph Algorithms

Transcription

1 Algorithms and Theory of Computation Lecture 2: Big-O Notation Graph Algorithms Xiaohui Bei MAS 714 August 14, 2018 Nanyang Technological University MAS 714 August 14, / 20

2 O, Ω, and Θ Nanyang Technological University MAS 714 August 14, / 20

3 O, Ω, and Θ Let T, f are two monotone increasing functions that maps from N to R +. Asymptotic Upper Bounds We say T(n) = O(f(n)) if there exists constants c > 0 and n 0 0, such that for all n n 0, we have T(n) c f(n). Nanyang Technological University MAS 714 August 14, / 20

4 O, Ω, and Θ Let T, f are two monotone increasing functions that maps from N to R +. Asymptotic Upper Bounds We say T(n) = O(f(n)) if there exists constants c > 0 and n 0 0, such that for all n n 0, we have T(n) c f(n). Examples T(n) = 1000n Nanyang Technological University MAS 714 August 14, / 20

5 O, Ω, and Θ Let T, f are two monotone increasing functions that maps from N to R +. Asymptotic Upper Bounds We say T(n) = O(f(n)) if there exists constants c > 0 and n 0 0, such that for all n n 0, we have T(n) c f(n). Examples T(n) = 1000n = T(n) = O(n) Nanyang Technological University MAS 714 August 14, / 20

6 O, Ω, and Θ Let T, f are two monotone increasing functions that maps from N to R +. Asymptotic Upper Bounds We say T(n) = O(f(n)) if there exists constants c > 0 and n 0 0, such that for all n n 0, we have T(n) c f(n). Examples T(n) = 1000n = T(n) = O(n) set c = 1001 and n0 = 100 Nanyang Technological University MAS 714 August 14, / 20

7 O, Ω, and Θ Let T, f are two monotone increasing functions that maps from N to R +. Asymptotic Upper Bounds We say T(n) = O(f(n)) if there exists constants c > 0 and n 0 0, such that for all n n 0, we have T(n) c f(n). Examples T(n) = 1000n = T(n) = O(n) set c = 1001 and n0 = 100 T(n) = pn 2 + qn + r for constants p, q, r > 0 Nanyang Technological University MAS 714 August 14, / 20

8 O, Ω, and Θ Let T, f are two monotone increasing functions that maps from N to R +. Asymptotic Upper Bounds We say T(n) = O(f(n)) if there exists constants c > 0 and n 0 0, such that for all n n 0, we have T(n) c f(n). Examples T(n) = 1000n = T(n) = O(n) set c = 1001 and n0 = 100 T(n) = pn 2 + qn + r for constants p, q, r > 0 = T(n) = O(n 2 ) Nanyang Technological University MAS 714 August 14, / 20

9 O, Ω, and Θ Let T, f are two monotone increasing functions that maps from N to R +. Asymptotic Upper Bounds We say T(n) = O(f(n)) if there exists constants c > 0 and n 0 0, such that for all n n 0, we have T(n) c f(n). Examples T(n) = 1000n = T(n) = O(n) set c = 1001 and n0 = 100 T(n) = pn 2 + qn + r for constants p, q, r > 0 = T(n) = O(n 2 ) set c = p + q + r and n0 = 1 also correct to say T(n) = O(n 3 ) Nanyang Technological University MAS 714 August 14, / 20

10 Some Remarks Equals sign. O(f(n)) is a set of functions, but computer scientists often write T(n) = O(f(n)) instead of T(n) O(f(n)). Consider f(n) = 5n 3 and g(n) = 3n 2, we write f(n) = O(n 3 ) = g(n), but it does not mean f(n) = g(n). Domain. The domain of f(n) is typically the natural numbers {0, 1, 2,...}. Sometimes we restrict to a subset of the natural numbers. Other times we extend to the reals. Nonnegative functions. When using big-o notation, we assume that the functions involved are (asymptotically) nonnegative. Nanyang Technological University MAS 714 August 14, / 20

11 O, Ω, and Θ Let T, f are two monotone increasing functions that maps from N to R +. Asymptotic Lower Bounds We say T(n) = Ω(f(n)) if there exists constants ɛ > 0 and n 0 0, such that for all n n 0, we have T(n) ɛ f(n). Nanyang Technological University MAS 714 August 14, / 20

12 O, Ω, and Θ Let T, f are two monotone increasing functions that maps from N to R +. Asymptotic Lower Bounds We say T(n) = Ω(f(n)) if there exists constants ɛ > 0 and n 0 0, such that for all n n 0, we have T(n) ɛ f(n). Examples T(n) = pn 2 + qn + r for constants p, q, r > 0 Nanyang Technological University MAS 714 August 14, / 20

13 O, Ω, and Θ Let T, f are two monotone increasing functions that maps from N to R +. Asymptotic Lower Bounds We say T(n) = Ω(f(n)) if there exists constants ɛ > 0 and n 0 0, such that for all n n 0, we have T(n) ɛ f(n). Examples T(n) = pn 2 + qn + r for constants p, q, r > 0 = T(n) = Ω(n 2 ) Nanyang Technological University MAS 714 August 14, / 20

14 O, Ω, and Θ Let T, f are two monotone increasing functions that maps from N to R +. Asymptotic Lower Bounds We say T(n) = Ω(f(n)) if there exists constants ɛ > 0 and n 0 0, such that for all n n 0, we have T(n) ɛ f(n). Examples T(n) = pn 2 + qn + r for constants p, q, r > 0 = T(n) = Ω(n 2 ) set ɛ = p and n0 = 1 also correct to say T(n) = Ω(n) Nanyang Technological University MAS 714 August 14, / 20

15 O, Ω, and Θ Let T, f are two monotone increasing functions that maps from N to R +. Asymptotic Lower Bounds We say T(n) = Ω(f(n)) if there exists constants ɛ > 0 and n 0 0, such that for all n n 0, we have T(n) ɛ f(n). Examples T(n) = pn 2 + qn + r for constants p, q, r > 0 = T(n) = Ω(n 2 ) set ɛ = p and n0 = 1 also correct to say T(n) = Ω(n) Meaningful Statement. Any compare-based sorting algorithm requires Ω(n log n) compares in the worst case. Nanyang Technological University MAS 714 August 14, / 20

16 O, Ω, and Θ Let T, f are two monotone increasing functions that maps from N to R +. Asymptotic Lower Bounds We say T(n) = Ω(f(n)) if there exists constants ɛ > 0 and n 0 0, such that for all n n 0, we have T(n) ɛ f(n). Examples T(n) = pn 2 + qn + r for constants p, q, r > 0 = T(n) = Ω(n 2 ) set ɛ = p and n0 = 1 also correct to say T(n) = Ω(n) Meaningful Statement. Any compare-based sorting algorithm requires Ω(n log n) compares in the worst case. Meaningless Statement. Any compare-based sorting algorithm requires O(n log n) compares in the worst case. Nanyang Technological University MAS 714 August 14, / 20

17 O, Ω, and Θ Let T, f are two monotone increasing functions that maps from N to R +. Asymptotic Tight Bounds We say T(n) = Θ(f(n)) if there exists constants c 1, c 2 > 0 and n 0 0, such that for all n n 0, we have c 1 f(n) T(n) c 2 f(n). Nanyang Technological University MAS 714 August 14, / 20

18 O, Ω, and Θ Let T, f are two monotone increasing functions that maps from N to R +. Asymptotic Tight Bounds We say T(n) = Θ(f(n)) if there exists constants c 1, c 2 > 0 and n 0 0, such that for all n n 0, we have c 1 f(n) T(n) c 2 f(n). Alternative definition T(n) = Θ(f(n)) if T(n) is both O(f(n)) and also Ω(f(n)). Nanyang Technological University MAS 714 August 14, / 20

19 O, Ω, and Θ Let T, f are two monotone increasing functions that maps from N to R +. Asymptotic Tight Bounds We say T(n) = Θ(f(n)) if there exists constants c 1, c 2 > 0 and n 0 0, such that for all n n 0, we have c 1 f(n) T(n) c 2 f(n). Alternative definition T(n) = Θ(f(n)) if T(n) is both O(f(n)) and also Ω(f(n)). Examples T(n) = pn 2 + qn + r for constants p, q, r > 0 Nanyang Technological University MAS 714 August 14, / 20

20 O, Ω, and Θ Let T, f are two monotone increasing functions that maps from N to R +. Asymptotic Tight Bounds We say T(n) = Θ(f(n)) if there exists constants c 1, c 2 > 0 and n 0 0, such that for all n n 0, we have c 1 f(n) T(n) c 2 f(n). Alternative definition T(n) = Θ(f(n)) if T(n) is both O(f(n)) and also Ω(f(n)). Examples T(n) = pn 2 + qn + r for constants p, q, r > 0 = T(n) = Θ(n 2 ) Nanyang Technological University MAS 714 August 14, / 20

21 O, Ω, and Θ Let T, f are two monotone increasing functions that maps from N to R +. Asymptotic Tight Bounds We say T(n) = Θ(f(n)) if there exists constants c 1, c 2 > 0 and n 0 0, such that for all n n 0, we have c 1 f(n) T(n) c 2 f(n). Alternative definition T(n) = Θ(f(n)) if T(n) is both O(f(n)) and also Ω(f(n)). Examples T(n) = pn 2 + qn + r for constants p, q, r > 0 = T(n) = Θ(n 2 ) set c1 = p, c 2 = p + q + r and n 0 = 1 T(n) is neither Θ(n) nor Θ(n 3 ) Nanyang Technological University MAS 714 August 14, / 20

22 Properties of Asymptotic Growth Rates Nanyang Technological University MAS 714 August 14, / 20

23 Properties of Asymptotic Growth Rates 1 If f = O(g) and g = O(h), then f = O(h). Nanyang Technological University MAS 714 August 14, / 20

24 Properties of Asymptotic Growth Rates 1 If f = O(g) and g = O(h), then f = O(h). 2 If f = Ω(g) and g = Ω(h), then f = Ω(h). Nanyang Technological University MAS 714 August 14, / 20

25 Properties of Asymptotic Growth Rates 1 If f = O(g) and g = O(h), then f = O(h). 2 If f = Ω(g) and g = Ω(h), then f = Ω(h). 3 If f = Θ(g) and g = Θ(h), then f = Θ(h). Nanyang Technological University MAS 714 August 14, / 20

26 Properties of Asymptotic Growth Rates 1 If f = O(g) and g = O(h), then f = O(h). 2 If f = Ω(g) and g = Ω(h), then f = Ω(h). 3 If f = Θ(g) and g = Θ(h), then f = Θ(h). 4 If f = O(h) and g = O(h), then f + g = O(h). Nanyang Technological University MAS 714 August 14, / 20

27 Properties of Asymptotic Growth Rates 1 If f = O(g) and g = O(h), then f = O(h). 2 If f = Ω(g) and g = Ω(h), then f = Ω(h). 3 If f = Θ(g) and g = Θ(h), then f = Θ(h). 4 If f = O(h) and g = O(h), then f + g = O(h). 5 If g = O(f), then f + g = Θ(f). Nanyang Technological University MAS 714 August 14, / 20

28 Properties of Asymptotic Growth Rates 1 If f = O(g) and g = O(h), then f = O(h). 2 If f = Ω(g) and g = Ω(h), then f = Ω(h). 3 If f = Θ(g) and g = Θ(h), then f = Θ(h). 4 If f = O(h) and g = O(h), then f + g = O(h). 5 If g = O(f), then f + g = Θ(f). f(n) 6 If lim = c > 0, then f = Θ(g). n g(n) Nanyang Technological University MAS 714 August 14, / 20

29 Some Common Functions 1 Polynomials. Let f be a polynomial of degree d, in which the coefficient a d is positive. Then f = O(n d ). Asymptotic rate of growth is determined by their high-order term. Nanyang Technological University MAS 714 August 14, / 20

30 Some Common Functions 1 Polynomials. Let f be a polynomial of degree d, in which the coefficient a d is positive. Then f = O(n d ). Asymptotic rate of growth is determined by their high-order term. Polynomial-Time Algorithm A polynomial-time algorithm is one with running time O(n d ) for some constant d. Nanyang Technological University MAS 714 August 14, / 20

31 Some Common Functions 1 Polynomials. Let f be a polynomial of degree d, in which the coefficient a d is positive. Then f = O(n d ). Asymptotic rate of growth is determined by their high-order term. Polynomial-Time Algorithm A polynomial-time algorithm is one with running time O(n d ) for some constant d. 2 Logarithms. For every a, b > 1 and every x > 0, we have log a n = Θ(log b n) = O(n x ). Nanyang Technological University MAS 714 August 14, / 20

32 Some Common Functions 1 Polynomials. Let f be a polynomial of degree d, in which the coefficient a d is positive. Then f = O(n d ). Asymptotic rate of growth is determined by their high-order term. Polynomial-Time Algorithm A polynomial-time algorithm is one with running time O(n d ) for some constant d. 2 Logarithms. For every a, b > 1 and every x > 0, we have log a n = Θ(log b n) = O(n x ). No need to specify base (assuming it s a constant). Logarithms are always better than polynomials. Nanyang Technological University MAS 714 August 14, / 20

33 Some Common Functions 1 Polynomials. Let f be a polynomial of degree d, in which the coefficient a d is positive. Then f = O(n d ). Asymptotic rate of growth is determined by their high-order term. Polynomial-Time Algorithm A polynomial-time algorithm is one with running time O(n d ) for some constant d. 2 Logarithms. For every a, b > 1 and every x > 0, we have log a n = Θ(log b n) = O(n x ). No need to specify base (assuming it s a constant). Logarithms are always better than polynomials. 3 Exponentials. For every r > 1 and every d > 0, we have n d = O(r n ). Nanyang Technological University MAS 714 August 14, / 20

34 Some Common Functions 1 Polynomials. Let f be a polynomial of degree d, in which the coefficient a d is positive. Then f = O(n d ). Asymptotic rate of growth is determined by their high-order term. Polynomial-Time Algorithm A polynomial-time algorithm is one with running time O(n d ) for some constant d. 2 Logarithms. For every a, b > 1 and every x > 0, we have log a n = Θ(log b n) = O(n x ). No need to specify base (assuming it s a constant). Logarithms are always better than polynomials. 3 Exponentials. For every r > 1 and every d > 0, we have n d = O(r n ). Polynomials are always better than exponentials. Nanyang Technological University MAS 714 August 14, / 20

35 More Notations Asymptotic Smaller T(n) We say T(n) = o(f(n)) if lim n f(n) = 0. Asymptotic Larger f(n) We say T(n) = ω(f(n)) if lim n T(n) = 0. Nanyang Technological University MAS 714 August 14, / 20

36 O, Ω, and Θ with multiple variables Asymptotic Upper Bounds We say T(m, n) = O(f(m, n)) if there exists constants c > 0, m 0 0 and n 0 0, such that for all m m 0 and n n 0, we have T(m, n) c f(m, n). Similar definitions for Ω and Θ. Nanyang Technological University MAS 714 August 14, / 20

37 O, Ω, and Θ with multiple variables Asymptotic Upper Bounds We say T(m, n) = O(f(m, n)) if there exists constants c > 0, m 0 0 and n 0 0, such that for all m m 0 and n n 0, we have T(m, n) c f(m, n). Similar definitions for Ω and Θ. Examples T(m, n) = 32mn mn + 32n 3 Nanyang Technological University MAS 714 August 14, / 20

38 O, Ω, and Θ with multiple variables Asymptotic Upper Bounds We say T(m, n) = O(f(m, n)) if there exists constants c > 0, m 0 0 and n 0 0, such that for all m m 0 and n n 0, we have T(m, n) c f(m, n). Similar definitions for Ω and Θ. Examples T(m, n) = 32mn mn + 32n 3 T(m, n) is both O(mn 2 + n 3 ) and O(mn 3 ). T(m, n) is neither O(n 3 ) nor O(mn 2 ). Nanyang Technological University MAS 714 August 14, / 20

39 Data Structures We discussed algorithms in peusdocode, but how the data will be represented in an actual implementation of the algorithm? Nanyang Technological University MAS 714 August 14, / 20

40 Data Structures We discussed algorithms in peusdocode, but how the data will be represented in an actual implementation of the algorithm? Data Structures: a particular way of organizing data such that it can be used efficiently in an algorithm. Appropriately designed data structures can help to give more efficient algorithms. Data structure in Number Addition : arrays. Nanyang Technological University MAS 714 August 14, / 20

41 Data Structures We discussed algorithms in peusdocode, but how the data will be represented in an actual implementation of the algorithm? Data Structures: a particular way of organizing data such that it can be used efficiently in an algorithm. Appropriately designed data structures can help to give more efficient algorithms. Data structure in Number Addition : arrays. Program = Algorithm + Data Structure Nanyang Technological University MAS 714 August 14, / 20

42 Common Data Structures Elementary Data Structures Arrays Linked Lists Stacks and Queues Priority Queues Hash Tables Binary Search Trees many others Nanyang Technological University MAS 714 August 14, / 20

43 Graph Algorithms Nanyang Technological University MAS 714 August 14, / 20

44 Graphs One of the most important combinatoric object in Computer Science, Optimization, Combinatorics. Figure: Seven Bridges of Königsberg Nanyang Technological University MAS 714 August 14, / 20

45 Graphs Graph A graph G consists of a set V of vertices and a set E of edges Directed Graph: Each edge e E is an ordered pair (u, v) for some u, v V. Undirected Graph: Each edge e E is an unordered pair {u, v} for some u, v V. Nanyang Technological University MAS 714 August 14, / 20

46 Graphs Graph 1 A graph G consists of a set V of vertices and a set E of edges Directed Graph: Each edge e E is an ordered pair (u, v) for some u, v V Undirected Graph: Each edge e E is an unordered pair {u, v} for some u, v V. 6 Example In the above undirected graph G = (V, E): V = {1, 2, 3, 4, 5, 6} E = {{1, 2}, {1, 3}, {2, 3}, {2, 4}, {3, 4}, {3, 5}, {4, 5}, {4, 6}} Nanyang Technological University MAS 714 August 14, / 20

47 Applications Transportation Networks The map of routes served by an airline carrier/railway company. vertices are airports/train stations; an edges from u to v if there is a flight/railway track between them. Nanyang Technological University MAS 714 August 14, / 20

48 Applications Transportation Networks The map of routes served by an airline carrier/railway company. vertices are airports/train stations; an edges from u to v if there is a flight/railway track between them. Information Networks The World Wide Web can be viewed as a directed graph. vertices are Web pages; an edge from u to v if u has a hyperlink to v. Nanyang Technological University MAS 714 August 14, / 20

49 Applications Transportation Networks The map of routes served by an airline carrier/railway company. vertices are airports/train stations; an edges from u to v if there is a flight/railway track between them. Information Networks The World Wide Web can be viewed as a directed graph. vertices are Web pages; an edge from u to v if u has a hyperlink to v. Social Networks A collection of people who interact with each other forms an undirected graph. vertices are people an edge joining u and v if they are friends. Nanyang Technological University MAS 714 August 14, / 20

50 Applications Transportation Networks The map of routes served by an airline carrier/railway company. vertices are airports/train stations; an edges from u to v if there is a flight/railway track between them. Social Networks A collection of people who interact with each other forms an undirected graph. vertices are people an edge joining u and v if they are friends. Information Networks The World Wide Web can be viewed as a directed graph. vertices are Web pages; an edge from u to v if u has a hyperlink to v. Dependency Networks A directed graph that captures the interdependencies among a collection of objects, e.g., in an university: vertices are courses offered an edge from u to v if u is a prerequisite for v. Nanyang Technological University MAS 714 August 14, / 20

51 Notation and Convention usually use n = V and m = E u and v are the end points of an edge e = {u, v} multi-graphs may have loops which are edges (u, u) multi-edges which are different edges between same pair of vertices in most of this class we only consider simple graphs Nanyang Technological University MAS 714 August 14, / 20

52 Graph Representation Nanyang Technological University MAS 714 August 14, / 20

53 Graph Representation Adjacency Matrix Represent a graph G = (V, E) by an n n adjacency matrix A, where A[u, v] = 1 if edge (u, v) E, and A[u, v] = 0 otherwise. The matrix is symmetric for undirected graphs. Nanyang Technological University MAS 714 August 14, / 20

54 Graph Representation Adjacency Matrix Represent a graph G = (V, E) by an n n adjacency matrix A, where A[u, v] = 1 if edge (u, v) E, and A[u, v] = 0 otherwise. The matrix is symmetric for undirected graphs. Advantages: can check if an edge (u, v) E in O(1) time can do linear algebra on the matrix Nanyang Technological University MAS 714 August 14, / 20

55 Graph Representation Adjacency Matrix Represent a graph G = (V, E) by an n n adjacency matrix A, where A[u, v] = 1 if edge (u, v) E, and A[u, v] = 0 otherwise. The matrix is symmetric for undirected graphs. Advantages: can check if an edge (u, v) E in O(1) time can do linear algebra on the matrix Disadvantages: require Ω(n 2 ) space even when m = o(n 2 ) cannot examine all edges incident to a given node efficiently Nanyang Technological University MAS 714 August 14, / 20

56 Graph Representation Adjacency List Represent a graph G = (V, E) by adjacency lists, which is an array Adj of length n, where each entry Adj[v] is a list containing all vertices adjacent to vertex v Nanyang Technological University MAS 714 August 14, / 20

57 Graph Representation Adjacency List Represent a graph G = (V, E) by adjacency lists, which is an array Adj of length n, where each entry Adj[v] is a list containing all vertices adjacent to vertex v Advantages: take O(n + m) space, which is close to optimal can browse all edges incident to a given vertex efficiently Nanyang Technological University MAS 714 August 14, / 20

58 Graph Representation Adjacency List Represent a graph G = (V, E) by adjacency lists, which is an array Adj of length n, where each entry Adj[v] is a list containing all vertices adjacent to vertex v Advantages: take O(n + m) space, which is close to optimal can browse all edges incident to a given vertex efficiently Disadvantages: hard to find a specific edge Standard representation for graphs. Nanyang Technological University MAS 714 August 14, / 20

59 A Concrete Representation Linked List Linked lists are a data structure to represent sequence whose length is arbitrary and changeable. each entry in a linked list consists of a cell for data and a pointer that points to the next entry there is a pointer to the first element the last entry points to NIL Nanyang Technological University MAS 714 August 14, / 20

60 A Concrete Representation Linked List Linked lists are a data structure to represent sequence whose length is arbitrary and changeable. each entry in a linked list consists of a cell for data and a pointer that points to the next entry there is a pointer to the first element the last entry points to NIL Advantages: easy to add an element into a linked list, and to sequentially read the list. Disadvantages: no random access. Nanyang Technological University MAS 714 August 14, / 20

61 Adjacency List Example Example of a Linked List Nanyang Technological University MAS 714 August 14, / 20

62 Adjacency List Example Example of a Linked List Example of an Adjacency List Nanyang Technological University MAS 714 August 14, / 20