DSA UNIT-TT SEARCH TREES SYMBOL TABLE: + Symbol able is an imporkant data structure ct i 0 cated & maint 6 compilers in order to store intormation about she occurrence ained id entities (variables, functions , Objects , classes) - * + compiler stores Ane identifvers of an phe prog: in the symbo) table. sthe identifiers are stored in the form of name, value, address 4 type- e Thus, compiler can Keep track of QU the identifiers with all the Necessary information. Global syrnbol Table =» C4” be accessed by — au the procedures & value _\var_fint scope symbol table Pro.one |proc| int Pro-two | proc] int |, pro.one symbol table / \ pro-two symbol table one-4 | var | int jot | var | int| one-2 | var | int 4uyo-2| var | int one-5 | var} Int ywo-5| var | ink / \_ [one war | int one | var | int +wo-3 | var | int [one -4| var | int [one=1 | vor | int} ywo-4 | var| int inner scope £ tnner scope 2 inner scope 3 + Types of symbol fable? rP i) static symbol table > stores fixed amount of inygmatie (OBS ts i 3 stores dynamic inpormation. Y 3 Dynamic Syenbo! table iy i eo (AvL) coding) Advantages OH, using Symbol. {ees ° ie hi operations = ~ Fast 100K up of (dentifrers duting compilation. ear e the Tuntime allocation for identifiers is managed Insertion wsing symbol tables - Handles issues like scope of identifiers & implicit atdarations. Deletion Séarching* WEIGHT- BALANCE TREE: (wB8T) = WET is a BST whose balance is based on the sizes a the subtrets in each node. = OPTIMAL BINARY SEARCH TREE: (08ST) “Tne element having more probability of appearance should be placed nearer to the root pst & the element whtth appears for \eaat number Mes it Lesser probability should be laud awa om ane rock * 4 i q e + ye ts & OST that minienizes the expected search cost - (no: m comparisons required to search yor a given Key) « + Time complexity for searching a key = > OBST 1S constructed yom rlid) — jesource LT) > the value oy Je destination. yoat — weight C(i,j) = ne {cUik-s) + e(Kd)y + w(i,j) Cost c(ii) = 0 vi) =0 w(ii) = Ui) wij) = eG) + 4G) + (i, 5-4) Your | ‘2 Wor= 12 ]wy3= 4 ee neh 2 \or= 14 [Oye 2 y 7 (a, 02,03 ,44) = (Ao, i6, int while) . ! a y= 1) etd) = G30) 3 ere fe i q (0:4) = (25,04) Noss 2 [tye 2 ) (it) 4 \Coy= 32 oy? @) Gb (3) Con) rootX DYNAMIC PROGRAMMING + (oBsT) : pane prog 5 method in which the sol? +0 a problem is obtained by maxing sequence of aecislons - i * Dynamic programming works on the prindple % optimality . Solve problems - “Tt Is typically applied 40 optimization problems. “The principle optimality states that "tn an optimal sequence choices , each stbsequence are also be optimal. % + QBAT Th OBST, we always celece the value % K whith © obtained from minimum art (while @nstuetton) , thus fe follows principle Oy optimality - + OBST is. an CKAMPIE A, dynamic programming: , < problems thar Can be Solved bY dynarmme prog! = wmputing nh Fibonace’ number - = wmpulng binomial wey: ~ warshall’s & Floyd's algorithm . 7 OBST: X WEVGAT BALANCE TREE! > AVL TREE* A binar +The FE dynamically watanced Aree to be proposed - Retrieval of any node = [ottegn)] “TT fs O non-empty BT with TL & Tr the (ett & right subtrees, qnen Tis height valanced if & only if: i) TL & Te are oF same heigt: 4) pifference % In height of Te 4 TR chouldh'é be mor than 6A ont. it; (Potance factor = (1,0, Balance - Heignt of —_ Height oF Factor (eft subtree right subhree BE) * neght of \taf node =0- Y tree structure that is balanced with respect to height of subtrees.AVL TREE OPERATIONS : ANL TREE OPERATIONS 1) Insertion 2) deletion 3) Searching ROTATIONS OF AVL TREE FOR REBALANCING: (LL, RR, LR RL) i) eft Rotation (tt) : = every Mode Moves One position to (@tt trom the current Position. ~ Single rotation . = Right unbaraned tree- 3) veet- Right Rotation (12) : Every node moves one position to (tft, then one position to the vight from current positron. - Pouble rotation. INSERTION ALGORITHM: ) tnsert a new node as new leaf just 3) Trace tne path from n encountered» cheek TE herghts a) 5 6) otherwise restructure by new node towards root. yor a lefe(a) 4 right(n) differ by atmos 1. es, move towards parent (n). doing Uther single or double rotation. 2) Right Rotation (RR) - - every node moves one position to Fight from ine current position. ~ single vot ation - for lett unbalanced tree- 4) Right -Leet Rotation (RL) « 2 tery nodt moves one positton to Fight , then one position to left “wom antrent postion. - Double rotation. as in srdinary BST- tach node 'n’ DELETION ALGORITHM’ i) search the node to be deleted - Jal YH tsa leat node then simply mare fe NULL. nor then the node must 9 4] swapped with its inorder successor - 3) Now traverse back up the path 4owards root checking the Br o, Lach node & performing the rotations auordingly . SEARCHING ALGORITHM + h tek the aaa be searched be ‘key. 2) ¢ Key root then go to the reght Subfrte. ; make that node as ‘current node H neitha cheated Key ==Yoot 3) Gmpdaie key with ourrent node & follow stp @ -X RED-BLACK TREE: (RB-tte) “Tt is G@ tYPE of BST r balandng a ree using dhe colour attribute ~The rook node ts always black- ae (0) =the ettrnal nodes are black . = Two consecatlve red nodes are not allowed . «x children of red nodes 3 black. & ( ©) possible parent o, red node > black. = umber of black modes should be same @) ©) in a path pom root node to external node- = All (eat nodes have the same black depth: od > INSERTION I RB TREE” i) reryrm insertion oy 4 node ‘%! same Uke QST- 2) Newly added node x! (5 red. is Yoot node then change it to black, ay *! 4) 3} newly inserted node is red 4 its parent & qlso red, then ba is needed. tanung 5) a parent of new node ts ved then check the colour of uncle. a) JE uncle = black or NULL, then do rotation & veeo recolour. red then: recolour both (uncle & parent) & 6 uncle = 7 tis not root node then recolour it 4 o check IF grandparent necheck - 6) in RB-Hee balancing iS done by + Tecolouring & voratton. P= parent = unde ge grand parent x= new cild.AN TREE: AA tree is created gro the tdta of RB tree, but contains fewer fotations 4 less cormplex- ~ Time complexity for operations o AA Aree i ~ Lev node (instead o wlor) is used jor balanuing node =black, level iS one less than parent level node = rtd, level Is samt OA parent level. Properties * Devery node can be either red or black. 2) root node (s Always black - 3) Externa) nodes are black. &) Every, path yrom root to Weak node has same nov oF WACK nodea. 5) x, node Is Ta then child must be black. 6) Leet child may nok be red. > _while conversion [yom RB-bee to AA tree, no- oL levels reducer . whenever red node omes, wnnect it Horizontally to the parent hodt on the parent node's leva. ~4give arrows. INSERTION® i) The nodes are inserted similar to gsT- ) J parent £ its child ane on the same level then porzontal tinks ane possible: 3) There ane rome ramdtons in Horizontal links + Horizonta) links G-D-@ © ® (eet horizontal Two consecutive v links are not right links ae @ allowed Nof allowed 6d ® @ apply. split Operation operation A single right rotation when an = A cing lett retation when insertim insertion Creates horizontal link: untatds two horizontal inky ~may Ureate onseuarve mgt = Causes level a, middle node to be increased by ant. horizontal links -X KDIMENSIONAL TREE (KD tree) ~ used for 0 anizing some no OK points in a space with K-dimensions. ~ Ie is pasically % QsT weh constraints - = points are arranged in a space called K~ dienensiona space: = he points left side tm the space represent the (ctt suble 4 ih the réght repraene Mgnt sulbbree- — kK > dimensions. ey TTT ey tre (36), (15), (it5) , rl?) QD 1), (10,9) \ a wmpart the x-value at this (evel y > wMpart y-Value at this level. Applications of Aad KD tree: i) Nearest neighbour search 2) vatabase queries -* SPLAY TREE? A stl€-balancing BST with no explidt balance condition. ~ Recently accessed elementl can be accessed quickly - - A special operation “splaying" (3 performed on splay tree = splaying ? arranging tne elements in a tree in such @ YAY that fine mare “recently accessed node will be placed as root node. cases of spaying: © m 2iq-2aq cast: @® 2i9-2ig sep * @ 2g steP: = UR SHRP ; - Ww skp -LSHP - nae access ¢ @ ® s re ae franiparent | (P) ©) 3) ©) (i) © @® ® 2 66 ‘ Q | i ® ® ) © © © ® ® ‘ oe Ofte ® (4) © © | aaroot Pa tight O} 00 x. P3 right a © 4° right % P | | | | | aigetefe | zag = rgnt |O 0,2,4,6, 813,01 (splay aree) © © > gg » Pp © @, (@) @) (6)