SUBJECT CODE : 210252 As per Revised Syllabus of SAVITRIBAI PHULE PUNE UNIVERSITY Choice Based Credit System (CBCS) S.E. (Computer) Semester - IV DaTA STRUCTURES AND ALGORITHMS Anuradha A. Puntambekar M.E. (Computer) Formerly Assistant Professor in PE.S. Modem College of Engineering, Pune Minal P. Nerkar ME. (Computer Science) Assistant Professor in AISSMS Institute of Information Technology, Pune ==> TECHNICAL PUBLICATIONS An Up-Thrust for Knowledge ri 5 iDATA STRUCTURES AND ALGORITHMS Subject Code : 210252 S.E. (Computer Engineering) Semester - IV © Copyright with A.A. Puntambekar All publishing rights (printed and ebook version) reserved with Technical Publications. No part ofthis book should be reproduced in any form, Electronic, Mechanical, Photocopy or any information storage and retrieval system without prior permission in writing, from Technical Publications, Pune. Published by : - TECHNICAL | °°" °°*2222% tice No.1, 412, Shoniwartetn, PUBLICATIONS | "2 ~ 417090, MS. INDIA, Ph.: +91-020.24495496/97 PRUs oe [email protected] Website: vava:technicalpublications.org Printer = Yogi Printers & Bindes SNe. 10/14, Ghule Indus Estate, Nanded Vilage Road, Tal, - Havel, Dat. - Pune - 411041 ISBN 978-93-90450-38-0 dirs93001450350 9789390450250 [1] GyPREFACE The importance of Data Structures and Algorithms is well known in various engineering fields. Overwhelming response to our books on various subjects inspired us to write this book. The book is structured to cover the key aspects of the subject Data Structures and Algorithms. The book uses plain, lucid language to explain fundamentals of this subject. The book provides logical method of explaining various complicated concepts and stepwise methods to explain the important topics. Each chapter is well supported with necessary illustrations, practical examples and solved problems. All chapters in this book are arranged in a proper sequence that permits each topic to build upon earlier studies. All care has been taken to make students comfortable in understanding the basic concepts of this subject. Representative questions have been added at the end of each section to help the students in picking important points from that section, The book not only covers the entire scope of the subject but explains the philosophy of the subject. This makes the understanding of this subject more clear and makes it more interesting. The book will be very useful not only to the students but also to the subject teachers. The students have to omit nothing and possibly have to cover nothing more We wish to express our profound thanks to all those who helped in making this book a reality. Much needed moral support and encouragement is provided on numerous occasions by our whole family. We wish to thank the Publisher and the entire team of Technical Publications who have taken immense pain to get this book in time with quality printing. Any suggestion for the improvement of the book will be acknowledged and well appreciated Authors A.A. Dantambekar M.D. Neskar Dedicated to God iSYLLABUS Data Structures and Algorithms - 210252 Credit Scheme _| Examination Scheme and Marks Mid_Semester (TH) : 30 Marks End_Semester (TH) : 70 Marks 03 Unit-1 Hashing Hash Table : Concepts - hash table, hash function, basic operations, bucket, collision, probe. synonym, ‘overflow, open hashing, closed hashing, perfect hash function, load density, full table, load factor, rehashing. Issues in hashing. hash functions - properties of good hash function, division, multiplication, extraction, mid - square, folding and universal. collision resolution strategies - open addressing and chaining, hash table overtiow - open addressing and chaining, extendible hashing, closed addressing and separate chaining. Skip List : Representation, searching and operations - insertion, removal, (Chapter - 1) Unit-1 Trees Tree: Basic terminology, General tree and it's representation, representation using sequential anc{ linked organization, Binary tree properties, converting tree to binary tree, binary tree traversals (recursive and non-recursive) - inorder, preorder. post order, depth first and breacith first, Oprations on binary tree. ‘Huffman Tree (Concept and use), Binary Search Tree (BST), BST operations, Threaded binary search tree - concepts, threading, insertion and deletion of nodes in inorder threaded binary search tree, in order traversal of in-order threaded binary search tree. (Chapter - 2) Unit-Il_ Graphs Basic Concepts, Storage representation, Adjacency matrix, Adjacency list, Adjacency multi fist, Inverse adjacency list. Traversals - depth first and breadth first, Minimum spanning tree. Greedy algorithms for ‘computing minimum spanning tree - Prims and Kruskal Algorithms, Dikjtra’s single source shortest path, All pairs shortest paths - Flyod - Warshall Algorithm, Topological ordering. (Chapter - 3) Unit-IV_ Search Trees Symbol Table - Representation of Symbol Tables-Static treee table and Dyanamic tree table. Weight balanced tree - Optimal Binary Search Tree (OBST), OBST as an exmple of Dynamic Programming. Helght Balanced Tree- AVL tree. Red-Black Tree, A tree, K-dimensional tree, Splay Tree. (Chapter -4) Unit-V Indexing and Multiway Trees Indexing and multiway trees - Indexing, Indexing techniques - Primary, secondary, dense, sparse. Multiway search trees. BeTree - Insertion, deletion. B+Tree - Insertion. deletion. use of B+ tree in Indexing, Trie tree, (Chapter - 5) roUnit-VI__ File Organization Files - Concept, need, primitive operations, Sequential file organization - Concept and primitive ‘operations, Direct Access File - Concepts and primitive operations, Indexed sequential file organization - Concept, Types of indices, Structure of index sequential file, Linked organization - Multi list tiles, Coral rings, inverted files ancl celular partitions. (Chapter - 6) oTABLE OF CONTENTS Chapter- 1 Hashing (1 - 1) to (1 - 64) 1.1 Concept. . 11-2 1.1.1 Basic Concepts in Hashing. . . .1-3 ALD lta PUmPCRa Ti ress cass snnessccemenacensncdnssssnmnensininiccnsssesnsensoustsnsnasnsoumsssemsissneconssnsasd = 1.2.1 Division Method. ... 2.00.0. 0 0c cece cece cect eee e eee tree eee e eee LS 1.2.2 Multiplicative Hash Function .... been e ere eeeeee on oaseessass kee. 1.2.3 Extraction . seen . . oanee 1-6 1.2.4 Mid Square... .. weed -7 1.2.5 Folding 1-7 1.2.6 Universal Hashing . seane . - . sees 1-8 1.3 Properties of Good Hash Function. 1-9 1.4 Collision Resolution Strategies .. 11-9 1.4.1 Open and Closed Hashing..........0000eecseeeveeeeeeeteeereeeeeeee 1-10 VA2 Chaining ....0. 60 ccc eects a essesevevers seee1-10 1.4.3 Open Addressing ..... 0... .ceeecseeeseeeeeeeeeeeeseeeseeeseeeseees D223, 1.5 Extensible Hashing 1-50 1.6 Applications of Hashing 1-53 17 Skip List 1.7.1 Operations on Skip List... 6.66.66 ec eee eevee eee eeeeeee eet eeeeeee 1-55 1.7.2 Searching of aNode...........6.0e0008 beeen veeeees cee 155 1.7.3 Insertion of ANOdE ......e eve eeeeee ee eveeeeeeeereeeeeessseeseees L257 1.7.4 Removal of aNOde 0.2... 66 cece cece ee eee e eee e ete settee ete eeeeee 1-58 1.7.5 Features of Skip Lists. 1-59 1.7.6 Comparison between Hashing and Skip Lists 1-59 1.8 Multiple Choice Questions .........+..++ 1-60 woChapter-2 Trees (2 - 1) to (2-94) 2.1 Basic Terminology...... send 2.2 Properties of Binary Tree... 12-6 2.3 Representation of Binary Tree. 2.4 General Tree and its Representation...... 2.5 Converting Tree into Binary Tree... 2.6 Binary Tree Traversals .. 2.6.1 Non Recursive Traversals 2.7 Depth and Level Wise Traversals... 2.7.1 Depth First Search (DFS) 2.7.2 Breadth First Search (BFS)...........+ 2.8 Operations on Binary Tree........ 2.9 Huffman's Tree...... 2.9.1 Algorithm. .. 2.10 Binary Search Tree (BST)... 2.11 BST Operations... 2.12 BST as ADT oss 2.13 Threaded Binary Tree .. 2.13.1 Concept ... 2.13.2 Insertion and Deletion.........0...c0eeceeeeeneeeeeeseeeeseseeenes 2°76 2.13.3 Inorder Traversal 2-79 2.13.4 Advantages and Disadvantages ©... 666. 0.seeeeee es eeeeeee ee eeee ees 2°89) 2.14 Multiple Choice Questions..... Lbs p08 Chapter-3 Graphs (B- 1) to (3 - 124) 3.1 Basic Concept... 3-2 win3.1.1 Comparison between Graph and Tree 3-2 BAZ Types of Graph... eect e eee 3-3 3.1.3 Properties Of Graph... ..666ceeesceteeeetetsettetetettreterereeeeee BS 3.2 Storage Representation 3.2.1 Adjacency Matrix... 3.2.2 Adjacency List...... 3.2.3 Adjacency Multilist 3.2.4 Inverse Adjacency List. . 3.3 Graph Operations. 3.4 Storage Structure... 3.5 Traversals.... 3.5.1 BFS Traversal of Graph . 3.5.2 DFS Traversal of Graph 0.0.0... eee ceeee cece eeeeeeseeeesev estes 3.6 Introduction to Greedy Strategy. 3.6.1 Applications of Greedy Method ........0.00csceceeveceeeeeeeseeeeeees 3 +56 3.7 Minimum Spanning Tree 3.7.1 Difference between Prim's and Kruskal's Algorithm .........6...00e0005 3-67 3.8 Dijkstra's Shortest Path.. 3.9 All Pair Shortest Path (Warshall and Floyd's Algorithm)... 3.9.1 Warshall's Algorithm ...... 6600.0. cc cece cceee eee eeceeeee ee teeeeeee 3.9.2 Floyd's Algorithm .........0ceceeeceeeseseteeeeeeseeseeeeiereeees 3.10 Topological Ordering 3.11 Case Study ... 3.11.1 Webgraph and Google Map .. 3.12 Multiple Choice Questions Chapter-4 Search Trees (4-1) to (4 - 104) 4.1 Symbol Table - Representation... wh ro)4.2 Static Tree Table .. 4.3 Dynamic Tree Table .. 4.4 Introduction to Dynamic Programming. 4.4.1 Problems that can be Solved using Dynamic Programming. .... 4-6 4.4.2 Principle of Optimality. ........00c00ceeecccseeeeeesesseeeeeeeseaeee 4.5 Weight-Balance Tree 4.6 Optimal Binary Search Tree (OBST) .. QL AMBOrthM 6... eee cece cece cece ence ee tees sense et eeeeneneneneaee 4,7 Height Balance Tree (AVL)... 4.7.1 Height of AVL Tree. 4.7.2 Representation of AVL Tree «2... . 1. seeee weet ee eee eee ee ee d= 26 4.7.3 Algorithms and Analysis of AVL Trees ....... 00 ceceeeeeeeeeeeeeee ees 4-28 ATA Insertion 0... .ee cece ccc eeee eee teeeees es eeeeseeteneeeesee senses A= 28 ATS Deletion 66s eeeeeeeeveeeeeee ee eeeeeeeeeeeeees ste eeee eee = 36 ATG SCATCHING eee eee e eee e ee eeee eee eeeeeee eens A237 4.7.7 AVLTree Implementation ........006ceeeeseveeeeeeeeeeeeeeeee een ee eA 38 4.7.8 Comparison between BST, OBST and AVL Tree. 4-47 4.7.9 Comparison of AVL Tree with Binary Search Tree ...........0esseeee ees 4-63 4.8 Red Black Tree. 4.8.1 Properties of Red Black Tree. ...... sce eeeceeeeeeeeeeeeee eee e eee A= 64 4.8.2 Representation . . 4-65 4.8.3 Insertion in Red Black Tree... 60... 0 cece cece eesee essen eens ence nese 4 66 4.8.4 Deletion Operation ......... 000. 00ceee cece cess esse stew sees nesses A= 72 4.9 AATree .. 4.9.1 Insertion Operation............ oy toesterievens wee 4-81 4.9.2 Deletion Operation ..... 0... .eeeceee cree eee eee etter eee e eee eee 2 BS 4.10 k-dimensional Tree... 4.11 Splay Tree... 4.11.1 Splay Operations... eee eee e eee cece ee eee 4.12 Multiple Choice Questions .... wwChapter-5 — Indexing and Multiway Trees (5 - 1) to (5 - 70) 5.1 Indexing .. wo 22 5.2 Indexing Techniques... wn 3 5.2.1 Primary Indexing Techniques woven 5-3 5.2.2 Secondary Indexing Techniques.......66006000eeecc vette SM 5.2.3 Dense and Sparse Indexing........-..0s0c0ceeeeeeceeeeeeeeeee 5-5 5.3 Types of Search Tree. we 6 5.4 Multiway Search Tree... 5 6 5.5 B Tree... wo 7 5.5.1 Insertion .. ae 7 . : . 7 5-7 S5.2DEItON oe ceeeeceeeeeeeeeesee esse tees ee eeeeeeeeeeteneee ee SAIS, 5.5.3 Searching ... . . rere ere t ers wane tenet eeeene woees 518 5.5.4 Height of B-Trees. . . . 5-20 5.5.5 Variants of B-Trees . 5.5.6 Implementation of B-Trees............... 5.6 B+ Tree 5.7 Trie Tree... 5.8 Heap - Basic Concept 5.8.1 Heaps using Priority Queues. 5.9 Multiple Choice Questions Chapter-6 — File Organization (6 - 1) to (6 - 60) 6.1 File Definition and Concept 6-2 6.2 File Handling in C++... son = 4 6.3 File Organization. 6-10 6.4 Sequential Organization ....... ..6-10 ro)6.5 Direct Access File....... 6.6 Index Sequential File Organization... 6.6.1 Types of Indices ........ -6-29 6.7 Linked Organization... 6.8 External Sort .. 6.8.1 Consequential Processing and Merging Two Lists .. 6.8.2 Multiway Merge... 0.00 cseeeceeeeeeeeeeeeeeeeeeteetetereeeeeee 6-49 6.8.3 K way Merge Algorithm.............0008 6-53 6.9 Multiple Choice Questions... 6-57 Laboratory Experiments (L- 1) to (L- 86) Solved Model Question Papers (M- 1) to (M- 4) oii)UNIT - I Hashing Syllabus Hash Table : Concepts - hash table, hash function, basic operations, bucket, collision, probe, synonym, overflow, open hashing, closed hashing, perfect hash fiction, load density, full table, load factor rehashing, issues in hashing, hash functions - properties of good hash function, division, multiplication, extraction, mid - square, folding and universal, collision resolution strategies - open addressing and chaining, hash table overflow - open addressing and chaining, extendible hashing, closed addressing and separate chaining. Skip List : Representation, searching and operations - insertion, removal, Contents 1.1 Concept 1.2 Hash Funetions «+ May-10, 12, Dec.-12, 1.3 Properties of Good Hash Function... . May-05, 11, 13, Dec.-07, 08, 13, 1.4 Collision Resolution Strategies. ..... . May-07, 08, 09, 14, 17, 18, 19 Dec.-06, 07, 09, 10, 11, 13, 17, 19, 1.5. Extensible Hashing 1.6 Applications of Hashing 1.7 Skip List 1.8 Multiple Choice Questions a-nData Structures and Algorithms 1-2 Hashing Concept Hashing is an effective way to Position Key pocord reduce the number of comparisons. — FeeroogL > 0 AB67000 Actually hashing deals with the idea 1 of proving the direct address of the 2[__ Ba2T002 record where the record is likely to 3 store. To understand the idea clearly let us take an example - — Suppose the manufacturing company has an inventory file that consists of less than 1000 parts. Each 395 part is having unique 7 digit 396 | 4618396 number. The number is called ‘key’ 397 | 4957397 and the particular keyed record 398 consists of that part name. If there een toe ES are less than 1000 parts then a 1000 Table | 494 element array can be used to store 402 the complete file, Such an array will 403, be indexed from 0 to 999. Since the 404 key number is 7 digit it is converted 405 to 3 digits by taking only last three 406 digits of a key. This is shown in the a Fig. 1.1.1, Observe in Fig. 1.1.1 that the first —— key 496700 and it is stored at 0” position. The second key is 8421002. The last three digits indicate the position 2" in the array. Let us search the element 4957397. Naturally tee it will be obtained at position 397. i This method of searching is called 993 | 0047998 hashing. The function that converts 994 the key (7 digit) into array position is 995| 9846995, called hash function. 996 0 Here hash function is ooo aoe (key) = key % 1000 999[ 0001999 Where key % 1000 will be the hash function and the key obtained by hash function is called hash key. Fig. 1.1.1 Hashing TECHNICAL PUBLICA’ Tions® - An up thrust for knowledgeData Structures and Algorithms 1-3 Hashing Basic Concepts in Has! 9 1) Hash Table : Hash table is a data structure used for storing and retrieving data quickly. Every entry in the hash table is made using Hash function. 2) Hash Function : * Hash function is a function used to place data in hash table. * Similarly hash function is used to retrieve data from hash table. * Thus the use of hash function is to implement hash table. For example : Consider hash function as key mod 5. The hash table of size 5. Step 1: Insert 33 of 2 Step 3: Insert 25 [a3 mod 5]- 3 1 Hash function ~~ 25mod S20 Step 2: Ga Insert 54 3 | Gay] — Bucket 54mod4=4—~4 [54 Hash Table 3) Bucket : The hash function H(key) is used to map several dictionary entries in the hash table. Each position of the hash table is called bucket. 4) Collision : Collision is situation in which hash function returns the same address for more than one record. For example we want to insert 55 the But at 0" location 25 is already placed and now 55 is demanding the same location. Hence we say “collision occurs” Fig, 1.1.2 5) Probe : Each calculation of an address and test for success is known as a probe. 6) Synonym : The set of keys that has to the same location are called synonyms. For example - In above given hash table computation 25 and 55 are synonyms. 7) Overflow : When hash table becomes full and new record needs to be inserted then it is called overflow. TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-4 Hashing For example - 25 31 42 63, 49 Fig. 1.1.3 Overflow situation 8) Perfect hash function : The perfect hash function is a function that maps distinct key elements into the hash table with no collisions. Advantages of perfect hash function 1) A perfect hash function with limited set of elements can be used for efficient lookup operation. 2) There is no need to apply collision resolution technique. 9) Load factor and load density : Consider the hash table as given below - Slot Slot2 o[C& ‘Ai_} }+— Bucket 1 a{_e¢ C2 s[_ 0 Dt 4 Hash Table Fig. 4.1.4 Let, * be the total number of elements in the table. * Tis the total number of possible elements. Element density : The element density of hash table is the ratio - Load density of load factor : The load density or load factor of hash table is n “* &) TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-5 Hashing EEA Hash Funct There are various types of hash functions or hash methods which are used to place the elements in hash table. Cn Folding Division | [Muttipication] [Extraction and method | |_method method universal method Fig. 1.2.4 Types of hash functions Division Method The hash function depends upon the remainder of division. Typically the divisor is table length. For example :- If the record 54, 72, 89, 37 is to be placed in the hash table and if the table size is 10 then Hash Table Hash Function 6 4=54%10 1 2 2 2=72%10 3 4 54 9 = 89% 10 5 6 7=37%10 7 a7 8 9 89 Multiplicative Hash Function The multiplicative hash function works in following steps 1) Multiply the key 'k’ by a constant A where A is in the range 0 < A < 1. Then extract the fractional part of kA. TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-6 Hashing 2) Multiply this fractional part by m and take the floor. The above steps can be formulated as tk) = [m {kA} | Fractional part Donald Knuth suggested to use A = 0.61803398987 Example : Let key k = 107, assume m = 50. A = 0.61803398987 hik) = | nm [07 = 0,61803308087] | | 66.12 | 0.12 | Fractional part h(k) = 50» 0.12 =6 bik) =6 That means 107 will be placed at index 6 in hash table. Advantage : The choice of m is not critical Extraction In this method some digits are extracted from the key to form the address location in hash table. For example : Suppose first, third and fourth digit from left is selected for hash key. @s@O@24 478 at 478 location in the hash table of size 1000 the key can be stored. TECHNICAL PUBLICA’ Tions® - An up thrust for knowledgeData Structures and Algorithms 1-7 This method works Hashing in following steps 1) Square the key 2) Extract middle part of the result. This will indicate the location of the key element in the hash table. Note that if the key element is a string then it has to be preprocessed to produce a number. Let key = 3111 ony For the hash table of size of 1000 H@I11) = 783 Folding There are two folding techniques i) Fold shift ii) Fold boundary i) Fold shift : In this method the key is divided into separate parts whose size matches with the size of required address. Then left and right parts are shifted and added with the middle part. ii) Fold boundary : In this method the key is divided into separate parts. The leftmost and rightmost parts are folded on fixed boundary and added with the middle part. For example 345678123 Key Digit reversed Index = 146 (@s42_ = Index = 542 (a) Fold shitt (b) Fold boundary Fig. 1.2.2 Folding techniques TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-8 Hashing Universal Hashing Universal hashing is a hashing technique in which randomized algorithm is used to select the hash function at random, from family of hash functions with the help of certain mathematical properties. The universal hashing guarantees lower number of collisions. Let U be the set of universe keys H_be the finite collection of Hash functions mapping U into {0, 1,...., m-1}, Then H is called universal if, for x, y ¢U (xy) [8] [{heH: h)=hy)} | = St ™ That means, probability of a collision for two different keys x and y given a hash function randomly chosen from H is —. m Theorem : If h is chosen from a universal class of hash functions and is used to hash n keys into a table of size m, where n TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-9 Hashing Then hg (K) = ‘s a; Kj |modsize (0 h(xyz) = (35 24 + 100 « 25 + 12 16) % 11 (xyz) = 0 That means key = xyz is stored at index 0 in Hash table. wt Ori 1. What is a hashing function ? Explain any 4 types of hashing functions. SU See ed 2. What is hash function ? Explain the following hash functions ’) Mid square ii) Modulo division iti) Folding method iv) Digit analysis Sao See EE] Properties of Good Hash Function RR RA Rules for choosing good hash function 1. The hash function should be simple to compute. 2. Number of collisions should be less while placing the record in the hash table, Ideally no collision should occur. Such a function is called perfect hash function. 3. Hash function should produce such keys which will get distributed uniformly over an array 4, The hash function should depend on every bit of the key. Thus the hash function that simply extracts the portion of a key is not suitable University Questions 1. What is hashing function ? What are different ways to design a hash function ? Explain any one of the method in detail 2. What is hashing ? What are the characteristics of good hashing function ? Explain any two types of hash functions, ERD TAS stack aes eae ees 3. What is the use of hash tables ? Enlist the characteristics of good hash fienction. SSDs eee Collision Resolution Strate: SPPU : May-07, 08, 09, 14, 17, 18, 19, Dec.-06, 07, 09, 10, 11, 13, 17, 19, Marks 12 Definition : If collisions occur then it should be handled by applying some techniques, such techniques are called collision handling techniques. TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-10 Hashing Collision handling techniques ‘Open hashing (chaining) Closed hashi (open addressing) Chaining Linear probing] [Quadratic] [~ Double probing | | hashing Fig. 1.4.1 Collision resolution techniques Open and Closed Hashing * The open hashing is also called as separate chaining and closed hashing is called open addressing. * The difference between open hashing and closed hashing is that in open hashing the collisions are stored outside the table and in closed hashing the collisions are stored in the same table at some other slot. Chaining Index Data Chain 4. Chaining without replacement o a [4 In collision handling method chaining is a u ue Z concept which introduces an additional field 2 2 with data ie. chain. A separate chain table is a a eu maintained for colliding data. When collision 4 4 a occurs we store the second colliding data by 5 6 7 linear probing method. The address of this, 6 6 a colliding data can be stored with the first 7 7 er colliding element in the chain table, without 8 -1 replacement. 9 al For example consider elements, 131, 3, 4, 21, 61, 6, 71, 8,9 Fig. 1.4.2 Chaining without replacement From the example, you can see that the chain is maintained the number who demands for location 1. First number 131 comes we will place at index 1, Next comes 21 but collision occurs so by linear probing we will place 21 at index 2, and chain is maintained by writing 2 in chain table at index 1 similarly next comes 61 by linear TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms t-it Hashing probing we can place 61 at index 5 and chain will be maintained at index 2. Thus any element which gives hash key as 1 will be stored by linear probing at empty location but a chain is maintained so that traversing the hash table will be efficient. The drawback of this method is in finding, the next empty location. We are least bothered about the fact that when the element which actually belonging to that empty location cannot obtain its location. This means logic of hash function gets disturbed. Let us now see a program which implements chaining without replacement. [HHS En ES nin nnn nin niniiiiiiniiiaiaiaieiaiaiee Program to create hash table and to handle the collision using chianing without replacement.In this program hash function is (number %10) #include #include #include #define MAX 10 class WO_chain { private: int a[MAX||2]; public: WO_chain(); int create(int); void chain(int,int),display(); hk , The constructor defined / WO_chain::WO_chain() { int i; for(i=0i >num; key=h.create(num);//retums hash key h.chain(key,num);//collision handled by chaining without replacement cout<<"\n Do U Wish To Continue?(y/n)", ans=getche(); TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-15 Hashing 2. Chi ing with replacement ‘As previous method has a drawback of loosing the meaning of the hash function, to overcome this drawback the method known as changing with replacement is introduced. Let us discuss the example to understand the method. Suppose we have to store following elements : 131, 21, 31, 4, 5 TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-16 Hashing eiaiaw 9 Now next element is 2. As hash function will indicate hash key as 2 but already at index 2. We have stored element 21. But we also know that 21 is not of that position at which currently it is placed. Hence we will replace 21 by 2 and accordingly chain table will be updated. See the table : Index Data Chain a =il ea 1 131 6 2 2 -1 3 31 =il 4 4 ea 5 5 ea 6 21 3 7 -1 -1 8 =il =il 9 =il Ea ‘The value -1 in the hash table and chain table indicate the empty location. The advantage of this method is that the meaning of hash function is preserved. But each time some logic is needed to test the element, whether it is at its proper position. [DOOR DUD RRA Program to create hash table and to handle the collision using chaining with replacement.In this program hash function is (number %10) diatesersnsvasanocensaussoounnnuceaneensauaseneenanaueseuannnseenseeynsns/ #include #include #include #define MAX 10 class W_chain t private: int a[MAX][2]; TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms = An up thrust for knowledgeData Structures and Algorithms = An up thrust for knowledgeData Structures and Algorithms = An up thrust for knowledgeData Structures and Algorithms = An up thrust for knowledgeData Structures and Algorithms = An up thrust for knowledgeData Structures and Algorithms = An up thrust for knowledgeData Structures and Algorithms 1-23 Hashing Open Addressing Open addressing is a collision handling technique in which the entire hash table is searched in systematic way for empty cell to insert new item if col . Various techniques used in open addressing are 1. Linear probing 2. Quadratic probing 3. Double hashing 4. Linear probing When collision occurs ie. when two records demand for the same location in the hash table, then the collision can be solved by placing second record linearly down. wherever the empty location is found. For example Fig. 1.4.3 Linear probing TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-24 Hashing In the hash table given in Fig. 14.3 the hash function used is number % 10. If the first number which is to be placed is 131 then 131 % 10 = 1 ie. remainder is 1 so hash key = 1. That means we are supposed to place the record at index 1. Next number is 21 which gives hash key = 1 as 21 % 10 = 1. But already 131 is placed at index 1. That means collision is occurred. We will now apply linear probing. In this method, we will search the place for number 21 from location of 131. In this case we can place 21 at index 2. Then 31 at index 3. Similarly 61 can be stored at 6 because number 4 and 5 are stored before 61. Because of this technique, the searching becomes efficient, as we have to search only limited list to obtain the desired number. HI HRS SRS non nnn In nnnnnaaaiiaininaiainiaaiaiaiaiiit Program to create hash table and to handle the collision using linear probing.In this Program hash function is (number%10) #include #include #include #define MAX 10 class Hash { private: int a[MAX]; public; Hash(); int create(int); void linear_prob(int,int),display(); hk ; The constructor defined y Hash::Hash() { int i; for(i=0;i>num; key=h.create(num);//returns hash key hilinear_prob(key,num);//collision handled by linear probing cout<<"\n Do U Wish To Continue?(y/n)"; TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms Problem with linear probing One major problem with linear probing is primary clustering. Primary clustering is a process in which a block of data is formed in the hash table when collision is resolved. + An up thrust for knowledgeData Structures and Algorithms 1-28 Hashing For example : 0 39 Cluster 19% 10= 9 1 29 18% 10= 8 2 39% 10=9 3 29% 10=9 4 8%10=8 5 6 7 8 18 9 19 This clustering problem can be solved by quadratic probing. 2. Quadratic probing Quadratic probing operates by taking the original hash value and adding successive values of an arbitrary quadratic polynomial to the starting value. This method uses following formula - Hy(key) = (Hash(key)+i2)%m where m can be a table size or any prime number. For example : If we have to insert following elements in the hash table with table size 10: 37, 90, 55, 22, 11, 17, 49, 87. We will fill the hash table step by step 0 | 90 37%10 = 7 1 in 90%10 2 |» 55%10 = 5 3 22%10 4 11%10 = 5 | 55 6 TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-29 37. Now if we want to place 17 a collision will occur as 17%10 = 7 and bucket 7 has already an element 37. Hence we will apply quadratic probing to insert this record in the hash table. Hi(key) = (Hash(key)+i?)%m we will choose value i = 0, 1, 2, ... whichever is applicable, Consider i 0 then (17407) %10 (17+#17) %10 7 8,when i=1 The bucket 8 is empty hence we will place the element at index 8 Then comes 49 which will be placed at index 9. 49%10 = 9 Now to place 87 we will use quadratic probing, (87 + 0)%10 = 7 (87 + 1)%10 = 8 ... but already occupied (87 + 27)%10 = 1 ... already occupied (87 + 37)%10 = 6... this slot is free». We place 87 at 6"" index. It is observed that if we want to place all the necessary elements in the hash table the size of divisor (m) should be twice as large as total number of elements. 3. Double hashing Double hashing is technique in which a second hash function is applied to the key when a collision occurs. By applying the second hash function we will get the number of positions from the point of collision to insert. There are two important rules to be followed for the second function : « It must never evaluate to zero. * Must make sure that all cells can be probed. The formula to be used for double hashing is Hy(key) = key mod tablesize H,(key) = M - ( key mod M ) Hashing Fig. 1.4.4 TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-30 Hashing where M is a prime number smaller than the size of the table. Consider the following elements to be placed in the hash table of size 10 37, 90, 45, 22, 17, 49, 55 Initially insert the elements using the formula for Hy(key). Insert 37, 90, 45, 22. 0| 90 H,(37) = 37%10 = 7 1 H,(90) = 90%10 = 0 al 22 H4(45) = 45%10 = 5 3 Hy(22) = 22% 10 = 2 4 Hy(49) = 49% 10 = 9 5|_45 6 7|_37 8 9|_49 Now if 17 is to be inserted then H,(Q7) = 17%10=7 [0 Hy(key) = M ~ (key%M) ‘1 2] 22 Here M is a prime number smaller than the size of 7 the table. Prime number smaller than table size 10 is 7. 1 Hence M = 7 5) 45 H,(17) = 7-(17%7)=7-3=4 6 ‘That means we have to insert the element 17 at 4 7a places from 37. In short we have to take 4 jumps. a Therefore the 17 will be placed at index 1 s+ Now to insert number 55. Fig. 1.4.6 H,(55) = 55%10 = 5 ... collision HA(55) = 7 - (65%7) =7-6=1 That means we have to take one jump from index 5 to place 55. Finally the hash table will be - of 90 1, 47 22 2| 3| 45, TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-31 Hashing 49. Comparison of quadratic probing and double hashing The double hashing requires another hash function whose probing efficiency is same as some another hash function required when handling random collision. The double hashing is more complex to implement than quadratic probing. The quadratic probing is fast technique than double hashing. 4, Rehashing Rehashing is a technique in which the table is resized, i.e,, the size of table is doubled by creating a new table. It is preferable if the total size of table is a prime number. There are situations in which 0 Sr ohact a. Transferring 1 the rehashing is required the contents 2 * When table is completely full. * With quadratic probing when the table is filled half. * When insertions fail due 9 to overflow. Old table In such situations, we have to transfer entries from old table to the new table by recomputing their _ positions New table Fig. 1.4.7 Rehashing using suitable hash functions. Consider we have to insert the elements 37, 90, 55, 22, 17, 49 and 87. The table size is 10 and will use hash function, H(key) = key mod tablesize 37%10 = 7 90%10 = 0 8 55%10 = 5 TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-32 Hashing 22%10 = 2 17% 10 = 7 Collision solved by goxio 9 linear probing 55, ea 37. 9 7 49 Now this table is almost full and if we try to insert more elements collisions will occur and eventually further insertions will fail. Hence we will rehash by doubling the table size. The old table size is 10 then we should double this size for new table, that becomes 20. But 20 is not a prime number, we will prefer to make the table size as 23. And new hash function will be H(key) = key mod 23 37%23 = 14 90%23 = 21 55%23 = 9 22%23 = 22 17%23 = 17 49%23 = 3 87%23 = 18 Now the hash table is sufficiently large to accommodate new insertions. 10) u a 13 14 15) 16 v| 18 9) 20 a 22 49. 37. 7 87, 90. 2 TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-33 Hashing Advantages 1. This technique provides the programmer a flexibility to enlarge the table size if required. 2. Only the space gets doubled with simple hash function which avoids occurrence of collisions. Give the input (4371, 1323, 6173, 4199, 4344, 9679, 1989} and hash function A(X) = X(mod 10), show the results for the following : i) Open addressing hash table using linear probing ii) Open addressing hash table using quadratic probing iti)Open addressing hash table with second hash function h2(X) - (X mod 7). Corer Solution : i) Open addressing hash table using linear probing : We assume mod function as mod 10. 4371 mod 10 = 1 1323 mod 10 = 3 6173 mod 10 = 3 collision occurs Hence by linear probing we will place 6173 at next empty location. That is, at location 4. 4199 mod 10 = 9 4344 mod 10 = 4 but location 4 is not empty. Hence we will place 4344 at next empty location ie. 5, 9679 mod 10 = 9 collision occurs so place at next location at 0. The hash table is of size 10. Hence we find the next empty location by rolling the table in forward direction. 1989 mod 10 = 9 collision occurs, so we find the next empty location at index 2. The hash table will then be Index Keys 0 9679 1 4371 2 1989 3 1323 4 6173 5 4344 6 TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-34 Hashing 9 4199 ii) Open addressing hash table using quadratic probing In quadratic probing we consider the original hash key and then add an arbitrary polynomial. This sum is then considered for hash function. The hash function will be H(Key) = (Key +i?) %m where m can be a table size If we assume m = 10, then the numbers can be inserted as follows - Step 1: Step 2: Thdex | Keys Thdex | Keys oO 0 4371 4374 1323 0] @|~Jo]a]s]eo] no] 1823 mod 10 4371 mod 11 Step 3: 6173 mod 10 = 3 As collision occurs. ‘we will apply quadratic probing. H(key) = (H{key) +i) % m_ Consider |= 0 H (6173) = (+0) % 10 =$ collision Hence consider H 6173) As index 4 is an empty slot, we will place 6173 at index 4. TECHNICAL PUBLICA’ Tions® - An up thrust for knowledgeData Structures and Algorithms 1-35 Hashing Step 4: index] Keys, 4371 4199 mod 10 = 9 1323 As this slot is empty 5173} we will place 4199 at index 9 Step 5: 4344 mod 10= 4 But index 4 shows an ‘occupied slot. Hence collision occurs. Therefore we will place 4344 using quadratic probing, Hkey) = (H(key) +7) % m 2, H (key) = (4 40.) % 10 = 4 colision H (key) = (4 #1") % 10 ‘The index 5 is an empty slot. Hence we will place 4344 at index 5. Step 6 : 9679 mod 10 =9 collision occurs. Hence we will place the element using quadratic probing, Hikey) = (Htkey) +) 5% m Hikey) = 8, iwill be 0 or 1 or 2 m=10 2 H (key) = (940) %7_ v 1=0 9 collision H (key) = (9 + 1°) % 10=0 "-. Place 9679 at index 0 TECHNICAL PUBLICA’ Tions® - An up thrust for knowledgeData Structures and Algorithms 1-36 Hashing Step 7: 1989 mod 10 = 9 collision occurs. Hence we will place the element using quadratic probing, 2 = H{key) = (H{key) +i) %m. Hikey) = 9, hence H (key) = (9 +0°) % 10, =9 collision H (key) = (9 +1") % 10 =0 collision (9+2)%10 v i=2 =3 collison H (key) = (9 +3°) % 10 = i= 3 =8 Insert 1989 at index 8 H (key) iii) Open addressing hash table with second hash function Step 1: Index Key 1 4371 4371 mod 10 = 1 3 1323 1323 mod 10 = 3 TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-37 Hashing Step 2: Key 4371 1323 6173 Step 3: Key 4371 1323 6173 4199, 6173 mod 10 = 3 collision occurs. Hence we will apply second hash. function, 209 = 7 - (X mod 7) 12(6173) = 7 ~ (6173 mod 7) =7-6 =1 ‘That means we have to take 1 jump from the index 3 (the place at which collision occurs). Hence we will place 6173 at index 4. 4199 mod 10 = 9. As this slot is empty, we will place 4199 at index 9 TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-38 Hashing Step 4: Index Key | 4344 mod 10 = 4 collision occurs. Hence 0 h(a3ia) = 7 - (4344 mod 7) 7-4 1 4971 3 2 i.e. Take 3 jumps from index 4 ie, place 4444 at index 7. 3 1323 | Similarly element 4 6173 | 9679 will be placed at index 5 using second i 2679 | hash function. 6 7 ou | hace There is no place for 8 1989, 9 4199 Explain linear probing with and without replacement using the following data: 12, 01, 04, 03, 07, 08, 10, 02, 05, 14, 06, 28 Assume buckets from 0 to 9 and each bucket has one slot. Calculate average cost/aumber of comparison for both. Se Marks 10 Solution : We assume hash function = key mod 10. As buckets are from 0 to 9 and each bucket has one slot, the keys can be inserted in the hash table using our hash function as follows - Index Key 0 10 12 % 10 =2 1 1 01% 10 = 2 2 04% 10 = 4 3 03 03 % 10 = 3 4 04 07 % 10 =7 5 02 08 % 10 = 8 6 05 10% 10 7 7 02 % 10 = 2 collision oceurs 8 08 |. By linear probing place 02 at index 05 % 10 = 5 collision occurs 9 14 ©. Place 05 at index 6 14 % 10 = 4 collision occurs * Place 14 at index 9 TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-39 Hashing Now as hash table is full we can not insert 06 and 28 keys in the hash table. Now if we want to search any key then following analysis is made Key 12 OL 04 «03 07 (8 10 02 05 14 06 2B ‘Number of 1 1 1 1 1 1 1 4 2 6 10 10 comparisons made for each key Thus total number of comparisons made = 14+141414+14+14+14+44+24+6+10+10 = 39 buckets get examined. That means average number of buckets that get examined per key Totalnumber ofcomparisons _ 39 Number of keys 2 = 3.25 We must find loading density which denoted by n Number ofkeys _ 12 a = 12 The average number of key comparisons Totalnumber of comparisons Total number ofbuckets The average cost can be computed by calculating successful search (denoted by Sn) and unsuccessful search (denoted by Un). 1 1 sn = ("aay 1 1 - (twa) 1 1 « {*-03) Sn = -2 1 1 tn Wie 2 Hosts] TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-40 Hashing Total cost = n Total cost What do you understand by collision in hashing ? Represent the following keys in memory using linear probing with or without replacement. Use modulo (10) as your hashing function : (24, 13, 16, 15, 19, 20, 22, 14, 17, 26, 84, 96) SPPU : De a Sr Solution : Collision in hashing - Refer section 1.4. i) Linear probing with replacement We will consider the hash function as modulo (10) for the hash table size 12, that is from 0 to 11. In linear probing with replacement, we first find the probable position of the key clement using hash function. If the location which we obtain from hash function is empty then place the corresponding key element at that location. If the location is not empty and the key element which is present at that location belongs to that location only then, move down in search of empty slot. Place the record at the empty slot. If the location contains a record which does not belong to that location then replace that record by the current key element. Place the replaced record at some empty slot, which can be obtained by moving linearly down. 0 20 1 2 2 3 4 2 5 16 6 16 7 4 Place 14 here because 14 % 10= 4. 8 The index 4 contain the element 24 9 19 This is a proper record atits place. Hence to place 14 we have to move down 10 in search of an empty slot. " TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-41 Hashing We get the empty location at index 7. Hence 14 is placed at index 7. J<— 17 % 10 =7. This position was occupied by 14, Bul 14 is not the proper record al this place, Hence replace 14 by 17. Then piace 14 al next empty slot 0 20 1 2 22, 3 18 4 24 5 15 6 16 |+— 26 % 10 = 6. But 16 is a proper record at this place. 7 7 8 14 9 19 10] 26 | Hence at next empty location 26 is placed, TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-42 Hashing 0 20 1 2 22 3 13 4 24 |e 84% 10 = 4. But 24 is occupying its own location 5 15 6 16 7 7 8 14 9 19 10] 26 " 84. [+— Then by moving linearly down we can place 84 at the empty location found 96 }+— As 96 % 10 = 6. But index 6 holds a record 16 which is correct for that location. By moving down linearly 22 we get no empty slot. Hence we roll back and get the empty slot at index 1 3 Hence 96 will be replaced at index 1. TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-43 Hashing ii) Linear probing without replacement 10 " 0 20 22 13 24 15 Collision occurs at [7 inex 4. Hence probing 14 at | the next empty slot ‘Atindex 7, the 14 is already placed. Hence at next empty slot 17 is placed, PL 25 % 10 = 6. The collision occurs. Hence 26 is placed at next ematy sit }+— 96% 10= 6, But at location 6, the element 16 is placed. Hence we go linearly down in search of an emply slot. Bul since table gets full, we may not get an empty slot. Therefore ‘oll back to search an empty slot Atindex 1, we can then place 98. }e— 84% 10= 4. But index 4 contains key element 24, Hence by linear probing, at empty slot 11 ‘the element 84 is placed GEER) Assume the size of hash table as 8. The hash function to be used to calculate the hash value of the data X is X%8. Insert the following values in hash table : 10, 12, 20, 18, 15. Use linear probing without replacement for handling collision. Solution : Table size 8 Hash function is x % 8 Dresser We will handle collision of the elements using linear probing, without replacement. TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-44 Hashing 10% 8 = 2 ° 12%8 = 4 ' 2 10 20% 8 = 4- collision occurs 3 Fal place at 10 C 5. 4 12 18 % 8 = 2- collision occurs 5 20) 6 . place at 10 € 3. 7 15 15%8 = 7 Fig. 1.4.8 Hash table EEEEEREE) Construct hash table of size 10 using linear probing with replacement strategy for collision resolution. The hash function is h(x) = x % 10. Calculate total numbers of comparisons required for searching. Consider slot per bucket is 1, 25, 3, 21, 13, 1, 2, 7,12, 4,8 a Solution : 25 % 10 = of 3% 10=3 1 21 % 10 2, | 13 % 10 = 3 Collision i Hence by linear probing : eed we will insert 13 sl) at index 3. 7 a 9 1% 10 = 1 collission | <. Insert 1 at index 2 1 es 2% 10 = 2, But index 2 2 is not empty it ' 4 13 is occupied by 1. s bs Replace 1 by 2, place 6 Ia 1 at next empty location al? ie. 6 P| 7% 10 =7 ol) Ta TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-45 Hashing 12 % 10 = 2 Again collision occurs. Hence place 12 at next empty location ie. 8 4% 10 = 4. But index 4 is occupied by 13. We will replace it by 4 and probe 13 at next empty location i.e. 9. 8 % 10 = 8. But index 8 is already occuppied by element 12. Hence replace 12 by 8. Then probe 8 at next empty location 21 13 25 12 21 25, 12 13 12 21 25 13, TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-46 Hashing For the given set of values : 11, 33, 20, 88, 79, 98, 44, 68, 66, 22. Create a hash table with size 10 and resolve collision using chaining with replacement and without replacement. Use the modulus Hash function. (key % size). Solution : Step 1: Without replacement 6 7 Se Tas Key Chain 20 a at zy ail 11% 10 =1 33 a 35% 10 = 3 =f 20% 10 = 0 a 88% 10 = 8 a 79% 10 = 9 a Step 2: Now insert 98. The hash key for occupied. So we search for next empty position. The position at index 2 is empty. Hence insert 98 at index 2, adjust chain table of 88 key. Then insert 44. 7” 98 is 98% 10 = 8. But position 8 is already Step 3: Insert 68 at position index 5. Adjust chain table. Insert 66 at index 6. Now insert 22 at index 7. The hash table will be as follows : TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-47 Hashing 2 8B R|R|B £ Bl x Chaining with replacement : In this technique we place the actual clement to its belonging position. If the position of element is occupied by an element which does not belong to that corresponding location then we make replacement. Step 1: Insert 11, 33, 20, 88, 79. Key Chain 0 20 a 1 sr a 2 3 33, - 4 5 6 7 8 88 9 7” Step 2: Now to insert 98, the collision occurs. 98 % 10 = 8. At index 8 we have stored element 88 which deserves that place, hence we cannot replace 88 by 98. Hence by linear probing. we can insert 98 at location 2. Then insert 44, and 68, 66. Adjust the chain table accordingly. TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-48 Hashing Key Chain 0 20 - 1 n = 2 98 3 3B 4 44 5 68 6 66 -1 7 8 88 2 9 2B - Step 3: Now to insert 22 we get 22 % 10 = 2. At the index 2 element 98 is stored which does not belong to location 2. Hence replace it by 22 and adjust chain table appropriately, Key Chain 0 20 1 n 2 2 3 B =i 4 44 -1 5 68 -1 6 66 -1 7 98 5 8 88 7 9 ce) -1 Construct hash table of size 10 using linear probing without replacement trategy for collision resolution. The hash function is h(x) = x % 10. Consider slot per bucket is 1. 31, 3, 4, 21, 61, 6, 71, 8, 9, 25 Solution: The hash table after inserting elements using linear probing without replacement. TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-49 Hashing o| 25 1} 31 2) 2 3} 3 4} 4 5} 61 6} 6 7) a” s| 8 9} 9 For the given set of values 35, 36 25, 47, 2501, 129, 65, 29, 16, 14, 99. Create a hash table with size 15 and resolve collision using open addressing techniques. Za 9, Mar Solution : Table size = 15 Assume hash function is element - value % 15. We use open addressing using linear probing. 10 u R B u uu 16 a7 29 25% 15 = 10 47 % 15 =2 2501 % 1 65 % 15 = 5 — Collision occurs :. Place 65 at index 7 by linear probing 29 % 15 16% 15 =1 14% 15 = 14 < Collision occurs Store 14 at index 0 99 % 15 = 9 — Collision occurs Store 99 at index 12 TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-50 (inom eens 1. What is the probing in hash table ? What is linear probing ? How does it differs from quairatic probing ? Explain with suitable example. 2. What is collision ? What are different collision resolution techniques ? Explain any two methods in detail. SPPU : Dec.-10, 11, M Hashing 14, Marks 8 Ha Extensible Hashing * Extensible hashing is a technique which handles a large amount of data. The data to be placed in the hash table is by extracting certain number of bits. + Extensible hashing grow and shrink similar to B-trees. * In extensible hashing referring the size of directory the elements are to be placed in buckets. The levels are indicated in parenthesis. For example : Directory (1) (1) Levels oo Data to o10 be placed in bucket Fig. 1.5.1 * The bucket can hold the data of its global depth. If data in bucket is more than global depth then, split the bucket and double the directory. Insertion Operation : Consider an example for understanding the insertion operation Consider we have to insert 1, 4, 5,7, 8, 10. Assume each page can hold 2 data entries (2 is the depth). Step 1 et 4 [] 1 = 001 o 4= 100 001 We will examine last bit of data and aaa insert the data in bucket, Fig. 1.5.2 TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-81 Hashing Insert 5. The bucket is full. Hence double the directory. 1=001 4= 100 5=101 Based on last bit the data is inserted. Step 2 : Insert 7. =i 4 0 a) (1) 100 001 101 Fig. 1.5.3 Insert 7 : But as depth is full we can not insert 7 here. Then double the directory and split the bucket. After insertion of 7. Now consider last two bits. 00 [ on [2 ] 1" [o Je Ja 100 001 ci 101 Fig. 1.5.4 Step 3 : Insert 8 ie. 1000. 00 1 10 4 “ @ 100 001 m1 1000 401 Fig. 1.5.5 TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-52 Hashing Step 4: Insert 10. ic. 1010 00 o1 10 11 a) @) 2) (2) 100 oo1 1010 "1 1000 401 Fig. 1.5.6 ‘Thus the data is inserted using extensible hashing. Deletion operation If we want to delete 10 then, simply make the bucket of 10 empty. 00 | o1 10 ] W [o Ja i 100 001 411 4000 401 Fig. 1.5.7 Delete 7. ) (ae ~~ Note that the level was increased 100 when we insert 7. Now on deletion oo G27. the tvel shoul get decorates Fig. 1.5.8 TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-53 Hashing Delete 8. Remove entry from directory 00. 00 01 10 " (1) (1) 001 100 101 Fig. 1.5.9 EES Applications of Hashing 1. In compilers to keep track of declared variables. 2. For online spelling checking the hashing functions are used. 3. Hashing helps in Game playing programs to store the moves made. 4, For browser program while caching the web pages, hashing is used. Skip List © Skip list is a variant list for the linked list * Skip lists are made up of a series of nodes connected one after the other. Each node contains a key and value pair as well as one or more references, or pointers, to nodes further along in the list * The number of references each node contains is determined randomly. The number of references a node contains is called its node level. * There are two special nodes in the skip list one is head node which is the starting node of the list and tail node is the last node of the list. * The skip list is an efficient implementation of dictionary using sorted chain. This is because in skip list each node consists of forward references of more than one node at a time. Following Fig. 1.7.1 shows the skip list. Po tote 1 2 3 6 NIL head tail node node Fig. 41.7.1 Skip list TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-54 Hashing Consider a sorted chain as given below - head tail Leelee eL el Ae el ed Fig. 1.7.2 Now to search any node from above given sorted chain we have to search the sorted chain from head node by visiting each node. But this searching time can be reduced if we add one level in every alternate node. This extra level contains the forward pointer of some node. That means in the sorted chain some nodes can hold pointers to more than one node. For example, : -ao-eo- aoe ut] A bol F-BL t-te el eel elf Fig. 1.7.3 If we want to search node 50 from above chain there we will require comparatively less time. This search again can be made efficient if we add few more pointers of forward references. For example, head tail Fig. 1.7.4 Skip List In above sorted chain every second having one level of chain is added. For instance node 20 has two next chains one pointing to 30 and another pointing to node 40. This makes searching more efficient. In skip list the hierarchy of chains is maintained. The level 0 is a sorted chain of all pairs. The level i chain consists of a subset of pairs in every level i-1 chains. TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-55 Hashing Node structure of skip list ‘The head node contains maximum number of level chains of the skip list whereas the tail node contains simply NULL value and no pointer. Each node in the skip list consists of pair of key and value given by element and a next pointer which is basically an array of pointers. template struct skipNode { typedef pair pair_type; pair_type element; skipNode **next; skipNode(const pair_type &New_pair,int MAX):element(New_pair) a next=new skipNode*[MAXI; } The individual node will look like this - [7 Element next Fig. 1.7.5 Operations on Skip List Various operations that can be performed on skip lists are 1, Searching a node from the skip list. 2. Insertion of a node from skip list. 3. Deletion of a node from skip list. Searching of a Node When we search for the 'key’ node then follow these 3 rules - 1) If key current — key then the node is found. 2) If key < next > key, just go down one level. 3) If key > next + key, go right along the link. TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-56 Hashing For example : Find 50 from the given skip list. tail header We will start search from level 3. 1) Key = 50, next + key = 60, just go down one level, 2) Now we are at level 2. Here next + key = 40 go right along the link. But this link indicates next > key = 60. Hence go down one more level. 3) Now we are at level 1. Here next > key = 60. Hence go down one more level 4) Now we are at level 0. The next —> key = key. Thus the node 50 is said to be present in the skip list. The algorithm for the same is as given below - Algorithm search(K& Key val) { skipNode *Forward_Node = header; for (int i = levels; i >= 0; i~) { while (Forward_Node->next[iJ->element.key < Key_val) Forward_Node = Forward_Node->next[i]; lastli] = Forward_Node; t retum Forward_Node->next[0}; The searching of a node takes O (log n) time. TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-57 Hashing Insertion of a Node * While inserting a new node in the skip list, it is necessary to find its appropriate location in the skip lists. Note that after inserting a new node in the skip list, the sorted order need to be maintained. * The level of the new node is determined randomly, For example : Consider following skip list in which we want to insert 55. [40] 50] Before insertion of 55 {s0] 40 [50] Ater insertion of 55 The algorithm for the same is as given below - void insert(pair& New_pair) {// Insert New_pair in the skip list if (New_pair.key >= tailKey) { } // if pair with the key value of New node //i already present skipNode* temp = search(New_Pair.key); if (temp->element.key == New_Pair-key) {// update temp->element.value temp->element.value = New_pair.value; return; i cout<<"Key is Too Large"; // Tf the Key_val is not already present //then determine level for new node randomly int New_Level = randomlevel(); // evel of retrieved for new node // Adjust New_Level to be <= levels + 1 if (New_Level > levels) i TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-58 Hashing New_Level = + +levels; last[New_Level] = header; t // get and insert new node just after node temp /Jallocating memory for Newnode skipNode *newNode = new skipNode(New_Pair, New Level + 1); for (int i = 0; 1 <= New Level; i++) { // insert into level i chain newNode->next[i] = last[i]->nextli]; last[i]->next|i] = newNode; t /{ number of pairs in the dictionary will be incremented by 7/1, As one node is inserted in the dictionary len+ The deletion of a node works in two steps - 1) Search the node to be deleted from the skip list 2) On obtaining the desired node, remove the node from the list and adjust the pointers. header tail NULL 7L-BL-BL +E elH Before: The node 40 is to be deleted header tail cas | -l-S-eo+ After: The node 40 is deleted Fig. 1.7.6 Deletion operation void delet(K& Key_val) { // Delete the pair, if any, whose key equals Key val if (Key_val >= tailKey) // too large TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-59 Hashing retum; // see if matching pair present skipNode* temp = search(Key_val); //temp node is to be deleted if (temp->element.key != Key_val) // node is not present retum; // delete node from skip list for (int i <= levels;it +) { ifflastli]->nextli] = lastlil->next[i] } // update levels while (levels > 0 && header->next{levels] levels— 5 temp) temp->next!i]; ail) delete temp;//ftee memory of the node to be deleted len-—//the pair is removed from the dictionary f¥ Features of Skip Lists 1) It is a randomized data structure. 2) This is a kind of linked list which works with levels. 3) It is ordered linked list which is called as sorted chain. 4) The bottommost list of the skip list contains all the nodes. 5) The expected time complexity for insertion, deletion and search operation is O(logn) Comparison between Hashing and Skip Lists Hashing, This method is used to carry out dictionary operations using randomized processes. Skip List Skip lists are used to implement dictionary operations using randomized processes It is based on hash function. If the sorted data is given then hashing is not an. effective method to implement dictionary. It does not require hash function. ‘The sorted data improves the performance of skip list. The space requirement in hashing is for hash table and a forward pointer is required per node. Hashing is an efficient method than skip lists. Skip lists are more versatile than hash table. The forward pointers are required for every level of skip The skip lists are not that much efficient. Worst case space requirement is larger for skip list than hashing TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-60 Hashing 1 Explain about skip list with an example. Gi sive applications of Cera EEX Multiple choice Questions Qt a] key b cc) index (dj Q.2 One of the most commonly used metho a_ addition b | division (dj a3 [a] more then one different keys[b] [fe] different hash functions [| Q4 Assuming that the hash function for a time ne: a) O(1) b| c | Ollogn) {d] as probing sequences where m is the size a| Linear probing {[b] | Chaining {dj Q.6 Consider a hash table of size seven, closed hashing ? Note that "_’ denotes [b] di in building hash function is, Tn hashing a record is located using : function none of these subtraction multiplication Collision occurs when the same hash value is obtained from, the equal keys resizing hash tables. table works well and the size of the hash table is reasonably large compared to the number of items in the table, the expected (average) led to find an item in a hash table containing n items is, Om) O(nlogn) Which of the following hashing technique has a potential to generate ©-(m2) different if hash table ? Double hashing All of the above with starting index zero and a hash function (3x + 4) mod 7. Assuming the hash table is initially empty, which of the following the contents of the table when the sequence 1, 3, 8, 10 is inserted into the table using an empty location in the table. 18,10,___3 110,83 TECHNICAL PUBLICATIOI S® - An up thrust for knowledgeData Structures and Algorithms 1-61 Hashing a7 as ag Q.10 att A hash table with 10 buckets with one Symbols $1 to S7 are initially entered using a hashing function with linear probing. Maximum number of comparisons needed in searching an item that is not present is slot per bucket is depicted in following diagram. 0 s7 SI sa $2 g wlalsjalalejelrle o Z| wa deletion is easier bi €) worst case comple d) none of these One major problem of linear probing is. table size DI €) too many computations. — [4 vacant position [a] chaining (B] ©) double hashing, d a] second hash function is applied. [b| chains are used. table size is changed The advantage of chained hash table over open addressing scheme is, ty of search operation is less In following collision resolution technique the key causing collision is placed at first Rehashing is a technique in which, space used is less primary clustering none of these quadratic probing linear probing none of these TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-62 Hashing Q.12 Q13 Q14 Qs Q16 Qi7 Double hashing is a technique in which, second hash function is applied. [al (b] chains are used. table size is changed d) none of these Requirement of additional data structure is the drawback of. linear probing quadratic probing double hashing chaining elele Adding the objects to the hash table for the list of elements which are already sorted give following result Placing of the elements in the table becomes time efficient b| There is no need to resize the table | Placing of the elements in the table becomes time efficient d) No effect A good hashing function must have. minimize collisions b] easy and quick to compute [| distribute the keys evenly over the hash table all of the above. In which of the hashing function the arithmetic or logical function is applied on different field value to calculate hash address ? a] Division method [b] Folding [e] Open hashing [d] Chaining Which of the following is a collision resolution method 2 [a] Open addressing [b] Division method c| Folding [d) All of the above TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-63 Hashing Q.48 The advantage of chained hash table over open addressing is, . space required is less [b| removal or deletion operation is simple [e] worst case complexity of search operation is less [d] none of these Q.19 Following hashing technique is preferred when huge amount of data is available __. [a] double hashing ‘b] rehashing ¢ | quadratic probing [d) extensible hashing Q.20 In browser programs for caching the web pages __is used. a] linked list B] arrays or buffers hashing [d] none of these Q.24 Which of the following is hashing technique makes use of bit prefix ? (a) Chaining [b] Linear probing fe] Quadratic probing [d] Extensible hashing Q.22 What do you understand by collision resolution by open addressing ? a] When collision happens we create a new memory location outside the existing table and use a chain to link to the new memory location When collision occurs we enlarge the hash table. When collision occurs we look for an unoccupied memory location in the existing hash table. [d] We use an extra hash table to collect all the collicled data. Q.23 The time complexity for insertion, deletion, and searching operation of skip list is [a] Om) [b] om?) ¢) 02") [d] Ofogn) Answer Keys for Multiple Choice Questions on b 2 < a3 a Q4 a Qs b Q6 b a7 « as a Qs b TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 1-64 Hashing Q.0 a Qu © Qu2 a Q13 a Qu d Qus d Q.16 b Quy a Qus b 19 a 20 c 21 d Q.22 © 02 a | | Explanations for Multiple Choice Questions : Q.4 All the conditions of good hash functions are specified for the hash table. Hence the positon value returned by the hash function will be used and the desired record can be retrieved. Thus within one hash function computation the record can be found. Hence the time complexity is O(1). Q.6 The hash function is (3x +4) % 7. ol4 The insertion of 1,3, 8 and 10 is as follows 1 | 8 xsl 2 | 10 GBx1+ 4)%7 = 0 3 x=3 4 (3x3+4)%7 = 6 5 x=8 6 [3 (3x84+4)%7 = 0 Collision occurs at 0" index. Place 8 at index 1 x= 10 (x 1044) %7 = 6 Collision occurs at 6" index. Hence we roll back and search for next empty location. The location is found at 2°4 index. Q.7 Consider searching of $8, we will search the locations 8 then 9, then 0, then 1 and finally 2. Q.13 Insert the keys 5, 28, 19, 15, 30, 33, 12, 17, 10 into a hash table with collisions resolved by chaining. The table has 9 slots. The hash function h(k) = k mod 9. Q.14 Consider the hash table of size 11. The hash function h(i) = (2i + 5)%11. The elements to be placed in this table are - 12, 44, 13, 88, 23, 94, 11, 39, 20, 16, 5. The given hash function will return some index positions. Q00 TECHNICAL PUBLICATIONS® - An up thrust for knowledgeUNIT - Il Trees Syllabus Basic terminology, General tree and it’s representation, representation using sequential and linked organization, Binary tree - properties, converting tree 10 binary tree, binary tree traversals (recursive ‘and non-recursive) - inorder, preorder, post order, depth first and breadth first, Oprations on binary tree. Huffman Tree (Concept and use), Binary Search Tree (BST). BST operations, Threaded binary search tree - concepts, threading, insertion and deletion of nodes in inorder threaded binary search tree, in order traversal of in-order threaded binary search tree. Contents 21 Basic Terminology May-10, Dec.-10, 2.2 Properties of Binary Tree cess ssee, May-06, 2.3 Representation of Binary Tree <<, May-10, Dec11, 12, 13, 24 General Tree and its Representation 2.5 Converting Tree into Binary Tree Dec.-11, 13, May-12, 14, 19, - 2.6 Binary Tree Traversals oo May-11, 14, Dec-12, 17, °° 2.7 Depth and Level Wise Traversals. May-07, Dec.-10, 2.8 Operations on Binary Tree 29 Huffman's Tree 2.10 Binary Search Tree (BST) 2.11 BST Operations -....,May-07, 09, 11, 14, Dec.-08, 09, 10,» 2.12 BST as ADT . .. May-06, 09, 10, Dec.-06,--- 2.13 Threaded Binary Tree Dec.-19, 2.14 Multiple Choice Questions @-1)Data Structures and Algorithms 2-2 Trees Co Basic Terminology Definition of a Tree : A tree is a finite set of one or more nodes such that - i) Root node : It is a specially designated node. ii) There are remaining n nodes which can be partitioned into disjoint sets T,, Ts, Ty Ty where Ty, Ty, Ty... Ty are called subtrees of the root. The concept of tree can be represented by following figure - Fig, 2.1.1 Tree Let us get introduced with some of the definitions or terms which are normally used. Fig. 2.1.2 Binary tree From Fig. 2.1.2, 4, Root Root is a unique node in the tree to which further subtrees are attached. For above given tree, node 10 is a root node. 2, Parent node The node having further sub-branches is called parent node. In Fig. 2.1.3, 20 is parent node of 40, 50 and 60. TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 2-3 Trees Geo © @ Fig. 2.1.3 Binary tree representing parent node 3. Child nodes The child nodes in above given tree are marked as shown below, @) ©» ® © Fig. 2.1.4 Binary tree repres ting child node 4. Leaves These are the terminal nodes of the tree. For example Fig. 24.5 ary tree with left nodes 5. Degree of the node The total number of sub-trees attached to that node is called the degree of a node. For example Node with @) ©) © © Node with degree 0 Fig. 2.1.6 Binary tree TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 2-4 Trees 6. Degree of tree ‘The maximum degree in the tree is degree of tree. This is the node with maximum degree i.e. 3.Hence degree of tree Fig, 2.1.7 Binary tree 7. Level of the tree The root node is always considered at level zero. The adjacent nodes to root are supposed to be at level 1 and so on. Go) = Note ateveto Nodes at level 1 (20,30) = — Nodes at level 2 (40,50,60,70) |< — Nodes at level 3 (80,90) Fig. 2.1.8 Binary tree in which levels are shown 8. Height of the tree The maximum level is the height of the tree. In Fig. 2.1.8 the height of tree is 3. Sometimes height of the tree is also called depth of tree. 9. Predecessor While displaying the tree, if some particular node occurs previous to some other node then that node is called predecessor of the other node. For example : While displaying the tree in Fig. 2.1.8 if we read node 20 first and then if we read node 40, then 20 is a predecessor of 40. 40. Successor Successor is a node which occurs next to some node. TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 2-5 Trees For example : While displaying tree in Fig. 2.1.8 if we read node 60 after reading node 20 then 60 is called successor of 20. 44, Internal and external nodes Leaf node means a node having no child node. As leaf nodes are not having further links, we call leaf nodes external nodes and non leaf nodes are called internal nodes. [a hn noses ED exeratrces Fig, 2.1.9 Representing internal and external nodes 42. Sibling The nodes with common parent are called siblings or brothers. For example @) The nodes 40,50 and So are sbings of © eachother Fig. 2.1.10 In this chapter we will deal with special type of trees called binary trees. Let us understand it Uni eS 1. Write the concept of ~ Skewed binary tee 2. What is binary tree ? How is it different than a basic tree ? Explain with figures. Sos Cees TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 2-6 Trees EZA Properties of Binary Tree In the trees, we can have any number of child nodes to a parent node. But if we impose some restriction on number of child nodes ie. if we allow at the most two children nodes then that type of tree is called binary tree. Let us formally define binary trees, Definition of a binary tree : A binary tree is a finite set of nodes which is either empty or consists of a root and t2vo disjoint binary trees called the left sub-tree and right sub-tree. The binary tree can be as shown below. Go) Foot node Node with one child Node with no }=— child or leat node Fig. 2.2.1 Binary tree Fig, 2.2.2 Not a binary tree Left and Right Skewed Trees The tree in which each node is attached as a left child of parent node then it is left skewed tree. The tree in which each node is attached as a right child of parent node then it is called right skewed tree. TECHNICAL PUBLICA’ Tions® - An up thrust for knowledgeData Structures and Algorithms Trees Left skewed tree Right skewed tree Fig, 2.2.3 Skewed trees Properties of Binary Trees Various properties of binary trees are - i) For a binary tree maximum number of nodes at level L are 2". Proof : Basis of Induction : As we know, root node is a unique node present at 0'" level That means, if L.= 0, Maximum number of nodes are 2!" = 2° = The binary tree is a tree in which each node has maximum two nodes. Hence if level m = 1 then maximum nodes are, 2 = 2' = 2. © @ @ © L=0 then When L = 1, then maximum nodes maximum nodes are1 are two ie. n, and n,. Inductive step : = As, Maximum_nodes (0) = Maximum_nodes (1) = 2 * maximum_nodes (0) Maximum_nodes (k) = 2% maximum_nodes (k+1) = 2+ 2# (24 (2* (..2* maximum-nodes(0))))) =k This proves that maximum number of nodes at level L are 2! TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 2-8 Trees ii) A full binary tree of height h has (2"*11) nodes. Proof : Basis of induction : —— Level o Height = 1 Levelt > Height =2 = Level 2 Fig. 2.2.4 Full binary tree For the binary tree, total number of nodes at level m are 2”. In above figure, total nodes at level 0 are 1, at level 1 are 2 and at level 1 are 4. Thus total number of nodes in the given binary tree are, total_nodes = 2°42"427+ If height h = 0, then total_nodes = 2"! If height h total_nodes = '~ 1 = 3 nodes when h = 1 Induction hypothesis : Assume total_nodes < 2!"""-1 for a full binary tree when h = k. Inductive step : Now a tree T has two subtrees Ty, and Tg. These subtrees are also the full binary trees with the heights h(T,) and h(T,). Hence T T T.- Tr total_nodes (T,) + total_nodes (Tg) + 1 Here 1 is added for root nodes, and T denotes total number of nodes. o T = (2RUL)+1_4)4 (QTR Iq) 41 *s induction hypothesis TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 2-9 Trees = (Qh. +1 ghltR)+1y = 2QmaxthL +L TRL) 4 2-21 max (h(T,) + 1, h(Tp)+1 = h(T) Tp = ght Thus total number of allowed nodes in a full binary tree of height h are 2"*1 -1 is proved. iii) Total number of external nodes in a binary tree are internal nodes + 1. iee=itl Proof: We assume e = External nodes i = Internal nodes. s. We have to prove e =i +1 Basis of induction : If there is only one node i.e. root node @ «228 O+1 e=1 ie. Only one external node is present. Thus, e = i+] is true. Whenever we add the nodes to a binary tree we add two external nodes and then previous external node becomes an internal node. For example Again, i= 1,e=2 © Hence, e=i+1 ie. 2 = 2 is true. Inductive step : (2) (&) ‘As soon as we add two external nodes previous external node becomes an internal node. ie Crow = prev + 1 = (ingey #1) +1 Cnew = inew +1 “+ inew = ipree + 1+ Thus e = i+ 1is true. TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 2-10 Trees 1. Prove that maximaun possible nodes in a binary tree of height és 2h-1 Sea 2. Prove that for binary tree maximum number of nodes at level L are 2L Sa ee EEE] Representation of Binary Tree EL ESOS eee There are two ways of representing the binary tree. 1. Sequential representation. _2. Linked representation. Let us see these representations one by one. 4, Sequential representation of binary trees or array representation : Each node is sequentially arranged from top to bottom and from left to right. Let us understand this matter by numbering each node. The numbering will start from root node and then remaining nodes will give ever increasing numbers in level wise direction. The nodes on the same level will be numbered from left to right The numbering will be as shown in following figure. Fig. 2.3.1 Binary tree Now, observe Fig. 2.3.1 carefully. You will get a point that a binary tree of depth n having 2"-1 number of nodes. The tree is having the depth 4 and total number of nodes are 15. Thus remember that in a binary tree of depth n there will be maximum 2"-1 nodes. And so if we know the maximum depth of the tree then we can represent binary tree using arrays data structure. Because we can then predict the maximum size of an array that can accommodate the tree. Thus array size can be >=n. The root will be at index 0. Its left child will be at index 1, its right child will be at index 2 and so on. Another way of placing the elements in the array is by applying the formula as shown below, TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 2-it Trees When n=0 the root node will placed at 0" location Parent(n) = floor(n—1)/2 Left(n) = n+l) Right(n) = (2n42). Where n>0 O Array Root = Loft child of Ai.e 8 willbe at location, Similarly right child of Ais Cwill be at 2x0 +2=2" location th ig at 8" location \dox 0. n=3 That means parent of lis at 3° location and i.e. D Fig. 2.3.2 Sequential representation of binary tree Advantages of sequential representation The only advantage with this type of representation is that the direct access to any node can be possible and finding the parent, left or right children of any particular node is fast because of the random access. Disadvantages of sequential representation 1. The major disadvantage with this type of representation is wastage of memory. For example : In the skewed tree half of the array is unutilized. You can easily understand this point simply by seeing Fig. 2.3.3 20 ED zn) Fig. 2.3.3 2. In this type of representation the maximum depth of the tree has to be fixed because we have already decided the array size. If we choose the array size quite larger than the depth of the tree, then it will be wastage of the memory. And if we TECHNICAL PUBLICA’ Tions® - An up thrust for knowledgeData Structures and Algorithms 2-12 Trees choose array size lesser than the depth of the tree then we will be unable to represent some part of the tree. 3. The insertions and deletion of any node in the tree will be costlier as other nodes have to be adjusted at appropriate positions so that the meaning of binary tree can be preserved As these drawbacks are there with this sequential type of representation, we will search for more flexible representation, So instead of array we will make use of linked list to represent the tree. 2. Linked representation or node representation of binary trees : In binary tree each node will have left child, right child and data field. Left child Data Right child The left child is nothing but the left link which points to some address of left sub-tree whereas right child is also a right link which points to some address of right sub-tree. And the data field gives the information about the node. Let us see the ‘C’ structure of the node in a binary tree. struct node “left; struct node *right; }bin; The tree with linked representation is as shown below. Left Right child Data child Fig. 2.3.4 Linked representation of binary tree Advantages of linked representation 1. This representation is superior to our array representation as there is no wastage of memory. And so there is no need to have prior knowledge of depth of the tree. Using dynamic memory concept one can create as many memory (nodes) as TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 2-13 Trees required. By chance if some nodes are unutilized one can delete the nodes by making the address free. 2. Insertions and deletions which are the most common operations can be done without moving, the other nodes. Disadvantages of linked representation 1. This representation does not provide direct access to a node and special algorithms are required. 2. This representation needs additional space in each node for storing the left and right sub-trees, GEMMOEER) Represent the following binary tree using array, SSE © ©) © © Fig. 2.3.5 Solution : The root node will be at 0'" location. We will use following formula to find the location of parent node, left child node and right child node. Parent (n) = floor (n-1)/2 Left (n) = (2n+1) Right (n) = (2n+2) Array ofa ® ifs 2, c¢ @® © 3 4 © © s|_> eLe Fig. 2.3.5 (a) University Questions 1. Explain the concept of representation of a binary free using an array. SESE 2. Explain binary tree representation with example. SUED ToS Tae TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 2-14 Trees 3. Explain the following (4) i) What is array representation of given binary tree? (8) ii) What is linked representation of given binary tree ? What are important observations of linked © @ representation ? Sus rks 8 A binary tree EZ] General Tree and its Representation General trees are those trees @ in which there could be any number of nodes that can be (@) attached as a child nodes. The So (15) (18) number of subtrees for each node may be varying, For GY @@ @ O® (5) example - in Fig. 24.1 is a General tree. Fig. 2.4.1 General tree EZ] converting Tree into Binary Tree We can convert the given general tree into equivalent binary tree as follows - i) The root of general tree becomes the root of the binary tree. ii) Find the first child node of the node attach it as a left child to the current node in binary tree. iii) The right sibling can be attached as a right child of that node. Let us convert the given general tree in Fig. 2.5.1 into equivalent binary tree. Fig. 2.5.1 Binary tree TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 2-15 Trees EEEOELE) Convert following generalized tree into a inary tree. AEST Solution : Fig. 2.5.2 (a) BEUCRRES Convert the following tree into Binary tree. Fig. 2.5.3 Solution : The rules are i) The root of general tree becomes root of the binary tree. ii) The first left child becomes left child of the tree. iii) The right sibling is attached as right child. TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 2-16 Trees The binary tree will be - Fig. 2.5.3 (a) Convert the given general tree to its equivalent binary tree Fig. 2.5.4 Solution : Binary tree is O ® TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 2-17 Trees 1. Explain how to convert general trees to binary tree with example. 2. What are the properties for binary trees that distinguish then from general trees ? EG Binary Tree Traversals ESCs Ane) Concept : There are three traversal techniques used for binary tree. These are - Inorder, Preorder and Postorder traversals. Let us discuss these traversals in detail 1. Inorder traversal: In this technique the leftnode, parent and then right node visit is followed. For example 1) Here visit the leftmost node ie. 7, then its parent node 8 then 9 the right child will be visited. Il) The root node (actually a parent node of left sub-branch) will be visited ic. visit 10. Fig. 2.6.1 Binary tr Il) Visit 11 then 12 and finally visit 13. Thus the traversal of right subbranch is performed. Hence, Inorder sequence is 7, 8, 9, 10, 11, 12, 13 2. Preorder traversal : In this method visit parent node the left and then right node. 1) Visit 10 Il) Visit 8, then 7 finally 9 MI) Visit 12 then 11 and finally 13. Hence, Preorder traversal is 10, 8, 7, 9, 12, 11, 13. Fig. 2.6.2 Binary tree 3. Postorder traversal : In this technique visit left node then right node and finally parent node. 1) Visit 7, then 9 and then 8. 11) Visit 11, then 13 and then 12. Ill) Visit node 10. The Postorder sequence is 7, 9, 8, 11, 13, 12, 10. TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 2-18 Trees Algorithms and C Functions 4, Recursive inorder traversal Algorithm : 1. If tree is not empty then a. traverse the left subtree in inorder b. visit the root node c. traverse the right subtree in inorder C Function 2, Recursive preorder traversal Algorithm : 1. If tree is not empty then. a. visit the root node b. traverse the left subtree in preorder c. traverse the right subtree in preorder C Function 3. Recursive postorder traversal Algorithm : 1. If tree is not empty then. TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 2-19 Trees a, traverse the left subtree in postorder b. traverse the right subtree in postorder c. visit the root node C function void postorder(node *temp) { if(temp!=NULL) { postorder(temp->lett); postorder(temp->right); printf(’ %d'' temp->data); } r EEREEELE) Consider the following tree given in the problem. Show a Postorder, Preorder and In order Traversal of tree. SPPU : M DE] @) ( 2) Q2@e@® © Fig. 2.6.4 Solution : Preorder Sequence : 50, 17, 12, 9, 14, 23, 19, 72, 54, 67, 76. Inorder Sequence : 9, 12, 14, 17, 23, 19, 50, 54, 67, 72, 76. Postorder Sequence : 9, 14, 12, 19, 23, 17, 67, 54, 76, 72, 50. From the given traversals construct the binary tree. Pre-order : G, B, Q, A, C, K, F, P, D, E, R, H In-order : Q, B, K, C, F, A, G, P, E, D, H, R Solution : Step 1: The first element in preorder sequence is root or parent node. We will locate this element in inorder sequence. Here the first pre-order element is G. Now, in the inorder sequence the list left GBKGRA PED to G forms left subbranch and the sequence right to G forms the right sub branch. Fig. 26.5 TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 2-20 Trees We will repeat the above procedure for each sub-branch Step 2 Preorder : [B].Q, A, C.K, F, Preorder : [P],D, E.R, H, Inorder: @,[B].K, 6. FA Inorder: [PB]. 0, H, R, EDR Step 3: Preorder: [A],C, KF inorder: K, C, F, @ ® ime Step 4: Preorder: [R].H Inorder: HER] Final binary tree TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 2-21 Trees Non Recursive Traversals We will consider following tree as an example. Initially we assume (10) Root / current root = 10, As current ! = NULL the while statement from the routine CO 6 inorder will get executed. Push 10 move onto left branch (10) =-current = current — left current —- (8) (1) 10 @) (:) Stack Push 8 move onto left branch (1) current = current — left () ©) 8 GD (9) Push 7 move onto left branch current = current > left OQ @® Hy CG © Sack current = NULL TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 2-22 Trees Now as current becomes NULL, we will come out of the while statements and following code fragment will get executed, Code Meaning if (! stempty (top) We will pop the topmost element { from the stack i.e. 7. Then we will poe SPs Meccan print it, And then we will move cout <<"" << current > data; current — right; to right branch or right child of 7, } calling it as new current. Q ® 7 current = NULL. 10 Stack Output OUD] current i Now we will enter the for loop interactively, but this time while statement will not get executed because current = NULL. Hence following code fragment will get executed. Code Meaning, if (1 stempty (top)) We will pop the topmost element { atops, & from the stack ie. 8. Then we will pe See eee gata; Print it. And then we will move on to current = current right; Tight branch, That mean new current = Right child of 8 = Right child of 8 = 9. current = NULL current Stack TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 2-23 Trees Now we will enter the for loop iteratively and while loop will be executed because current = 9. Hence 9 will be pushed onto the stack. And now current = current left. As there is no left child for the node current; current = NULL. Hence following code fragment will be executed. Code Meaning if (1 stempty (top) Now 9 which lies on the top of the { aoe ny stack will be popped off. It will be Pop (&top,s, & current); j cout <<"" << current — data; ed current = current right; current will set to right child of 9 } which is NULL. Q Q@ @ | current = NULL FNOEE] current Stack Output Now we will enter the for loop interactively but this time while statement will not be executed. The if statement will be executed. According to this code fragment, 10 will be popped off, it will be printed as an output and we will move on to the right branch of 10. Hence now, current = 11 As an output we will get 7 8 9 10. Again we will enter the for loop interactively. The while statement will be executed because current = 12. We will push 11 onto the stack. And move onto the left branch of 11. But there is no left child to 11. Hence value of current becomes NULL. Then if statement will get executed. According to it, 11 will be popped off, it will be printed as an output and we will move onto right branch of 11. current = NULL Output Stack TECHNICAL PUBLICA’ Tions® - An up thrust for knowledgeData Structures and Algorithms 2-24 Trees Again we will enter the for loop iteratively. But current = NULL ; hence while statements will not be executed, stack is empty; hence if statements will not be executed. Hence else will be executed, by which control returns to main function. Thus we will get 7 8 9 10 11 as an output of nonrecursive inorder traversal. void TREE_CLASS::inorder(node *root) { node *current,*s{10}; int top=-1; if(root= =NULL) { cout<<"\n Tree is empty\n”; retum; + current=root; for(;;) t while(current!=NULL) { push(current,&top,s); current=current->left; tempty(top)) pop(&top,s,¤t); cout<<" "<data; current=current->right; } else return; } Non-recursive Preorder Traversal The logic for preorder traversal is similar to the inorder traversal. But the only difference is that we will print the value of each visited current node before pushing it onto the stack. Hence : 8 i oo @ @ Output ‘Stack TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 2-25 Trees Visiting each node, calling it as current, printing it as an output, pushing current node onto the stack and moving, on to the left child. These are the sequence of operation which we must follow at every stage. Above is the scenario for stage I, II and III. Now if we move on to left branch of 7 we will get current = NULL. We will pop 7 check if it has any right. As 7 has no right child. We will pop the next element ie. 8. As 8 has a right child ic. 9, we will print 9 as an output and push it on to the stack. The stack will now be 10 8 7 9 9 10 ‘Stack Output Then 9 will be popped off, but 9 has no right child so we will pop 10. The 10 has right child i.e, 11. Hence print 11 as an output and push it onto the stack. 10 8 7 9 1 1 Stack Output Finally 11 will be popped off. The node 11 has no left or right child so nothing will be pushed. As a result we get 10 8 7 9 11 as an output for nonrecursive preorder traversal. void TREE_CLASS::preorder(node *root) i node *current,*s[10]; int top=-1; if(toot==NULL) { cout<<"\n The Tree is empty\n’ return; + current=root; for(::) cout<<" "<data; push(current,&top,s); current=current—>left; } TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 2-26 Trees if(!stempty(top)) { pop(&top,s,¤t); current=current->right; } else retum; + } Nonrecursive Postorder Traversal In the nonrecursive postorder traversal we have to add extra field in the stack structure. This field keeps a check on whether the node is visited once or not. If we visit that node for the first time then we set that check to 1, if we again visits that node then we reset that check. This extra logic we have to put because we traverse to the left node first, then right node and then the parent node void TREE_CLASS::postorder(node *root) t struct stack t node *element;//Here placing the node containing value int check; —_//check 1 means visiting left subtree /{check 0 means visiting right subtree }st[10]; int top=—1; node *current; if(toot==NULL) { cout<<"\n The Tree is empty\n"; return; + current=r00t; for(;;) t while(current!=NULL) { st[top|.element st[top|.check=1;//visiting the left subbranch current=current-> left; } while(st{top].check: { current=st]top].element; top= =; cout<<" "<data; TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 2-27 Trees if(stempty(top)) return; } current=st|top].clement;//pushing the element onto the stack current=current->right; st[top].check=0;//visiting right subtree 1. Write pseudocode for printing the elements of a binary search tree in ascending order non-recursively. Sues eee a 2. Write non-recursive algorithm for traversal of binary tree. ese Depth and Level Wise Traversals EL ae ee The tree can be traversed along with its depth as well as along with its breadth. Let us discuss about it Depth First Search (DFS) Concept : In this traversal technique the tree is traversed according to its depth and the visited vertex in this depthwise traversal are printed. For example : The DFS sequence For displaying the tree in depth first search manner, we have to start from the root node. From that node moving along the edge, and we have to move towards the leaf node and display the data. The depth first traversal is same as preorder traversal. To perform this traversal we need an additional data structure called stack. Algorithm : Step 1: Step 2: Pop the node and display it as output. ‘sit the root node . Push it onto the stack. Step 3: If its right child is not NULL, push it onto the stack TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 2-28 Trees Step 4 : If its left child is not NULL push it onto the stack. Step 5 : Repeat step 2 to Step 4 until stack is not empty. Pseudo Code : Algorithm DFS(root) t temp=root} if(temp!=NULL) { display ‘temp-> data’; DFS(temp->left); DFS(temp->right); } } Consider following tree and obtain its DES sequence. Fig. 2.7.4 Solution : Action Stack Output Step 1: We will visit 10 and push it onto the stack. Pop the node display it as output. 10 10 TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 2-29 Trees Step 2: We will visit right child of popped node push it. Then we will visit left child of 10, push it onto the stack. Pop the node, display it as output. 12 10, 4 Step 3: We will visit right child of node 4, push it onto the stack. Now we will visit left child of node 4, push it onto the stack. Pop the node and display it as output. Bee 10, 4, 3, 8 Step 4: Step 5: Step 6: As 3 has no left or right child, nothing will be pushed. Pop the top node from the stack and display it as output. We will visit right child of 8, push it onto the stack. Then we will visit left child of 6, push it onto the stack. Pop the top node from the stack and display it as output. Visit right child of 6 ie. 7, push it onto the stack. Visit left child of 6 ie. 5, push it onto the stack. Pop the top node from the stack and display it as output. 12 Roo Reve Box 10, 4, 3, 8 10, 4, 3, 8, 6 10, 4, 3, 8, 6, 5 TECHNICAL PUBLICATIONS® - An up thrust for knowledgeData Structures and Algorithms 2-30 Trees Step 7: As 5 has no left or right child nothing will be pushed onto the Pop 9. Display it. stack 9 Pop the top element of stack and 2 EONS display it as output. - Step 8: As 7 has no left or right child nothing will be pushed onto the stack, 12 10, 4, 3, 8, 6, 5, 7,9 Step 9: Again 9 is a leaf node nothing will be pushed. Pop 12, Print it. Step 10: There is no right child of 12. But 12 has left child. So Push 11 onto the stack. Pop 11 and display it. 10, 4, 3, 8, 6, 5,7, 9 12 10, 4,3, 8, 6, 5, 7, 9, 12, 11 Now all the nodes are visited. The stack is empty. Hence we get DFS sequence as. 10, 4, 3, 8, 6, 5, 7, 9, 12, 11. c++ Program Je ee eeeaeere ieee Program to create a binary tree and display it using Depth First Traversal. Both the recursive and non recursive versions of DFS are considered in this program WU rine SARE eMn LAU Sn ERE ARUN SOAS ERAA ALES One ERAAECaN ENR RNAS ETON ON] #include #include #define size 50 /? Declare a node for binary tree */ class tree { private: typedef struct node { int data; struct node *left; struct node “right; TECHNICAL PUBLICATIONS® - An up thrust for knowledge