Amortized Rigidness in Dynamic Cartesian Trees Iwona Białynicka-Birula and Roberto Grossi...

Preview:

Citation preview

Amortized Rigidness in Dynamic Cartesian TreesIwona Białynicka-Birula and Roberto GrossiUniversità di Pisa

STACS 2006

23 February 2006 23rd International Symposium on Theoretical Aspects of Computer Science 2

Cartesian trees Vuillemin 1980 Nodes store points ⟨x, y⟩

y value can be viewed as priority Recursive definition

Root stores point with greatest y value x value partitions remaining points(left and right subtrees)

23 February 2006 23rd International Symposium on Theoretical Aspects of Computer Science 3

Cartesian tree example ⟨2, 22⟩ ⟨21, 20⟩⟨18, 19⟩⟨6, 17⟩ ⟨20, 16⟩⟨7, 15⟩⟨8, 13⟩⟨5, 12⟩ ⟨9, 11⟩ ⟨17, 10⟩⟨3, 9⟩ ⟨16, 8⟩⟨10, 7⟩ ⟨22, 6⟩⟨15, 4⟩⟨12, 3⟩⟨1, 2⟩ ⟨11, 1⟩

⟨19, 21⟩⟨4, 18⟩

⟨14, 5⟩

23 February 2006 23rd International Symposium on Theoretical Aspects of Computer Science 4

Applications Priority queue Randomized searching (treaps) Range and dominance searching RMQ (Range Maximum Query) LCA (Least Common Ancestor) Integer sorting Memory management Suffix trees ...

23 February 2006 23rd International Symposium on Theoretical Aspects of Computer Science 5

From RMQ to LCA

2, 22, 9, 18, 12, 17, 15, 13, 11, 7, 1, 3, 5, 4, 8, 10, 19, 21, 16, 20, 6

22 201917 16151312 11 109 87 6432 1

2118

59, 18, 12, 17, 15, 13, 11, 7, 1

17151312 119 71

18

23 February 2006 23rd International Symposium on Theoretical Aspects of Computer Science 6

From LCP array to suffix tree$I$IPPI$ISSIPPI$ISSISSIPPI$MISSISSIPPI$PI$PPI$SIPPI$SISSIPPI$SSIPPI$SSISSIPPI$

0 1 1 4 0 0 1 0 2 1 3$ I M... P S

$ P... SSI

P... S...

I$ PI$ I SI

P... S...

P... S...

1

23

4

56

78 91011

12

121185211097463

01140010213

23 February 2006 23rd International Symposium on Theoretical Aspects of Computer Science 7

History Static setting

O(n) construction time, provided elements already sorted Randomized

Random priority values – treaps O(log n) expected height O(log n) expected update time

Other distributions yield O(√n) or even O(n) height (Devroye 1994) Dynamic and deterministic

???

23 February 2006 23rd International Symposium on Theoretical Aspects of Computer Science 8

Our result Dynamic Cartesian tree

Supports insertion Supports weak deletion Maintains actual tree structure between each operation O(log n) amortized time per operation

23 February 2006 23rd International Symposium on Theoretical Aspects of Computer Science 9

Solution outline Combinatorial analysis

How many tree elements change due to n insertions? Notion of entropy is exploited

Auxiliary structure for accessing tree Needed to quickly access tree elements which need to change Based on the interval tree

23 February 2006 23rd International Symposium on Theoretical Aspects of Computer Science 10

Insertion ⟨2, 22⟩ ⟨21, 20⟩⟨18, 19⟩⟨6, 17⟩ ⟨20, 16⟩⟨7, 15⟩⟨8, 13⟩⟨5, 12⟩ ⟨9, 11⟩ ⟨17, 10⟩⟨3, 9⟩ ⟨16, 8⟩⟨10, 7⟩ ⟨22, 6⟩⟨15, 4⟩⟨12, 3⟩⟨1, 2⟩ ⟨11, 1⟩

⟨19, 21⟩⟨4, 18⟩

⟨14, 5⟩

⟨13, 14⟩

23 February 2006 23rd International Symposium on Theoretical Aspects of Computer Science 11

Insertion – worst case⟨1, 16⟩

⟨7, 4⟩⟨8, 2⟩

⟨17, 15⟩⟨16, 13⟩⟨15, 11⟩⟨14, 9⟩⟨13, 7⟩⟨12, 5⟩⟨11, 3⟩⟨10, 1⟩

⟨2, 14⟩⟨3, 12⟩⟨4, 10⟩⟨5, 8⟩⟨6, 6⟩

⟨9, 17⟩

23 February 2006 23rd International Symposium on Theoretical Aspects of Computer Science 12

Analysis – main idea Inserting new elements does not require comparing y coordinates of existing points In turn, deleting points does Conclusion: insertions reduce tree information content ... so information entropy can be used as a potential function in an amortized analysis

23 February 2006 23rd International Symposium on Theoretical Aspects of Computer Science 13

>> >

>>>

>

>>

>>

Insertion revisited

23 February 2006 23rd International Symposium on Theoretical Aspects of Computer Science 14

Insertion reversed (deletion)

?????

23 February 2006 23rd International Symposium on Theoretical Aspects of Computer Science 15

Formally... Tree T induces partial order ≺T on nodes

Defined by the heap condition Partial order ≺T has ℒ(T) linear extensions

Linear extensions are permutations satisfying the order, i.e. P[i] ≺T P[j] ⇒ i < j We define missing entropy: ℋ(T)=log ℒ(T)

Information needed to sort nodes given tree topology

> > >>> >>>>A B CD E IG JF H

A B H J F G I D E CH A J F G I D B E CA J D F H G I B E CD J I A H F G E B CH J D A F G I E B CA D H F G J I B E C. . .

23 February 2006 23rd International Symposium on Theoretical Aspects of Computer Science 16

Missing entropy Can be zero ℋ(T)=0 Or can be up to ℋ(T)=O(n log n) When an insertion affects k edges, ℋ(T) increases by at least Ω(k)

23 February 2006 23rd International Symposium on Theoretical Aspects of Computer Science 17

So what now? Amortized number of edge modifications is O(log n) per insertion into an initially empty tree

Node modifications are always constant But how to access the edges to modify?

Without increasing the complexity

23 February 2006 23rd International Symposium on Theoretical Aspects of Computer Science 18

Implementation overview Companion interval tree stores tree edges Edges in Cartesian tree are either disjoint or nested

So the interval tree has additional properties Operations are tailored to the special case of the Cartesian tree

23 February 2006 23rd International Symposium on Theoretical Aspects of Computer Science 19

Insertion once again

1. Find parent2. Edges affected

4. Shrink k3a. Delete 2

3b. Insert 31. Find parent2. Edges affected3a. Delete 23b. Insert 34. Shrink k

23 February 2006 23rd International Symposium on Theoretical Aspects of Computer Science 20

Action implementations1. Find parent

Uses the interval tree as a search tree2. Edges affected Special kind of stabbing query3. Insert and delete Standard interval tree operations4. Shrink Emulating using inserts and deletes would yield O(k∗log n) Amortized argument based on the fact that shrinking edge travels down

O(log n)O(log n+k)

O(1)∗O(log n)k∗O(1)

23 February 2006 23rd International Symposium on Theoretical Aspects of Computer Science 21

Summary Scheme for maintaining a Cartesian tree under insertion and week deletion

Amortized O(log n) time per update At any moment the actual tree structure is accessible

Solution components Combinatorial analysis of tree behavior Auxiliary data structure

First result on dynamic Cartesian trees in a non-stochastic setting

23 February 2006 23rd International Symposium on Theoretical Aspects of Computer Science 22

Further work

Strong deletion Applications

For example RMQ

23 February 2006 23rd International Symposium on Theoretical Aspects of Computer Science 23

Thank you!

Questions?

Recommended