This preview shows pages 1–2. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: 6.896 Sublinear Time Algorithms September 21, 2001 Lecture 3: Estimating the Number of Connected Components Lecturer: Ronitt Rubinfeld Scribe: Kevin Matulef 1 Testing for Connectedness Reviewing from last class, we had the following tester for connectedness: 1. Choose m = O ( 1 d ) nodes. 2. For every chosen node s , perform a BFS until (a) ≥ 8 d new nodes are found. (b) s is found to be in a component of size ≤ 8 d . 3. If (b) occurs, reject, otherwise accept. Note that this algorithm requires O ( 1 2 d ) total queries. For a BFS to discover up to 8 d new nodes, it requires up to O ( 8 d · d ) = O ( 1 ) queries. Since we do O ( 1 d ) BFSs, the total number of queries is O ( 1 2 d ). To analyze the validity of the algorithm, we had the following lemma: Lemma 1 G isfar from P ( d ) n ⇒ G has ≥ dn 4 components From this, we derive the following corollary: Corollary 2 G isfar from P ( d ) n ⇒ G has ≥ dn 8 “small” components of size ≤ 8 d . Proof Let = number of components of size ≤ 8 d and = number of components of size > 8 d . From the lemma, we know that + ≥ dn 4 . We can bound the number of large components with the equation 8 d ≤ n . This simply comes from the fact that there can’t be more nodes in large components than the total number of nodes. We can rearrange this to say ≤ dn 8 . Combining our lower bound on the total number of components with our upper bound on the number of large components, we get the lower bound ≥ dn 4 − dn 8 = dn 8 on the number of small components....
View
Full
Document
This note was uploaded on 04/02/2010 for the course CS 6.896 taught by Professor Ronittrubinfeld during the Spring '04 term at MIT.
 Spring '04
 RonittRubinfeld
 Algorithms

Click to edit the document details