1 maximal independent set. 2 independent set (is): in a graph g=(v,e), |v|=n, |e|=m, any set of...
TRANSCRIPT
1
Maximal Independent Set
2
Independent Set (IS):
In a graph G=(V,E), |V|=n, |E|=m, any set of nodes that are not adjacent
3
Maximal Independent Set (MIS):
An independent set that is nosubset of any other independent set
4
Maximum Independent Set:
A MIS of maximum size
A graph G… …a MIS of G… …a MIS of max size
Remark: finding a maximum IS is an NP-hard problem, while a MIS can be found in polynomial time!
5
Applications in DS: network topology
•In a network graph consisting of nodes representing processors, a MIS defines a set of processors which can operate in parallel without interference
•For instance, in wireless ad hoc networks, to avoid interferences, a conflict graph is built, and a MIS on that defines a clustering of the nodes enabling efficient routing
6
Applications in DS: network monitoring
• A MIS is always a Dominating Set (DS) of the graph (the converse in not true, unless the DS is independent), namely every node in G is at distance at most 1 from at least one node in the MIS (otherwise the MIS could be enlarged, against the assumption of maximality!)
In a network graph G consisting of nodes representing processors, a MIS defines a set of processors which can monitor the correct functioning of all the nodes in G (in such an application, one should find a minimum MIS)
7
A Sequential Greedy algorithm
Suppose that will hold the final MISI
Initially I
G
8
Pick a node and add it to I1v
1v
Phase 1:
1GG
9
Remove and neighbors )( 1vN1v
1G
10
Remove and neighbors )( 1vN1v
2G
11
Pick a node and add it to I2v
2v
Phase 2:
2G
12
2v
Remove and neighbors )( 2vN2v
2G
13
Remove and neighbors )( 2vN2v
3G
14
Repeat until all nodes are removed
Phases 3,4,5,…:
3G
15
Repeat until all nodes are removed
No remaining nodes
Phases 3,4,5,…,x:
1xG
16
At the end, set will be a MIS of I G
G
17
Worst case graph (for number of phases):
n nodes, n-1 phases
Running time of the algorithm: Θ(m)
Number of phases of the algorithm: O(n)
18
Homework
Can you see a distributed version of the algorithm just given?
19
A General Algorithm For Computing MIS
Same as the sequential greedy algorithm, but at each phase, instead of a single node, we now select any independent set (this selection should be seen as a black box at this stage)
The underlying idea is that this approach will be useful for a distributed algorithm
20
Suppose that will hold the final MISI
Initially I
Example:
G
21
Find any independent set 1I
Phase 1:
And insert to :1I I 1III
1GG
22
1I )( 1INremove and neighbors
1G
23
remove and neighbors 1I )( 1IN
1G
24
remove and neighbors 1I )( 1IN
2G
25
Phase 2:
Find any independent set 2I
And insert to :2I I 2III
On new graph
2G
26
remove and neighbors 2I )( 2IN
2G
27
remove and neighbors 2I )( 2IN
3G
28
Phase 3:
Find any independent set 3I
And insert to :3I I 3III
On new graph
3G
29
remove and neighbors 3I )( 3IN
3G
30
remove and neighbors 3I )( 3IN
No nodes are left
4G
31
Final MIS I
G
32
1. The algorithm is correct, since independence and maximality follow by construction
2. The number of phases depends on the choice of the independent set in each phase: The larger the subgraph removed at the end of a phase, the smaller the residual graph, and then the faster the algorithm. Then, how do we choose such a set, so that independence is guaranteed and the convergence is fast?
Analysis
33
Example:If is MIS, 1 phase is needed
1I
Example:If each contains one node, phases may be needed
kI)(n
(sequential greedy algorithm)
34
A Randomized Sync. Distributed Algorithm
•Follows the general MIS algorithm paradigm, by choosing randomly at each phase the independent set, in such a way that it is expected to remove many nodes from the current residual graph•Works with synchronous, anonymous, uniform models
Remark: It is randomized in a Las Vegas sense, i.e., it uses randomization only to reduce the expected running time, but always terminates with a correct result
35
Let be the maximum node degree in the whole graph
d
1 2 d
Suppose that d is known to all the nodes (this may require a pre-processing)
36
Elected nodes are candidates forindependent set
Each node elects itself with probability
At each phase :k
kI
dp
1
1 2 d
kGz
z
37
However, it is possible that neighbor nodes are elected simultaneously (nodes can check it out by testing their neighborhood)
Problematic nodeskG
38
All the problematic nodes step back to the unelected status, and proceed to the next phase. The remaining elected nodes formindependent set kI
kGkI
kIkI
kI
39
Success for a node in phase : disappears at end of phase(enters or )
Analysis:
kGz
kI
1 2 y
No neighbor elects itself
z
z
k)( kIN
k
A good scenariothat guaranteessuccess
elects itself
40
Basics of ProbabilityLet A and B denote two events in a probability space;
let
1. A (i.e., not A) be the event that A does not occur;2. AՈB be the event that both A and B occur;3. AUB be the event that A or (non-exclusive) B occurs.
Then, we have that:1. P(A)=1-P(A);2. P(AՈB)=P(A)P(B) (if A and B are independent)3. P(AUB)=P(A)+P(B)-P(AՈB) (if A and B are
mutually exclusive, then P(AՈB)=0, and P(AUB)=P(A)+P(B)).
41
Fundamental inequalities
tn
t ent
nt
e
11
21n
nt ||
10 p
1k
k
kp
p
11
42
Probability for a node z of success in a phase: P(success z) = P((z enters Ik) OR (z enters N(Ik))
i.e., is at least the probability that it enters Ik, i.e., it elects itself AND no neighbor elects itself, i.e. (notice that events are independent):
1 2 y
p
p1
p1p1
z
No neighbor should elect itself
elects itself
P(success z) ≥ p(1-p)y (recall that p=1/d)
43
Probability of success for a node in a phase:
2ed1
d1
1ed1
d1
1d1
p)p(1p1pd
dy
At least
For 2d
First (left) ineq. with t =-1
44
Therefore, node disappears at the end of phase with probability at least
1 2 y
z
z
ed21k
Node z does not disappear at the end of phase k with probability at most
2ed1
1
45
after phases
Definition: Bad event for node :
ned ln4
node did not disappear
This happens with probability P(ANDk=1,..,4ed lnn (z does not disappear at the end of phase
k))
i.e., at most:
2ln2
ln22ln411
21
12
11
needed n
nedned
z
z
(first (right) ineq. with t =-1 and n =2ed)
Independent events
46
after phases
Bad event for G:
ned ln4
at least one node did not disappear
This happens with probability (notice that events are not mutually exclusive): P(ORxG(bad event for x)) ≤
nnnx
Gx
11) for event P(bad 2
47
within phases
Good event for G:
ned ln4
all nodes disappear
This happens with probability:
n1
-1G] for event bad of ty[probabili1
(i.e., with high probability)
48
Total number of phases:
)log(ln4 ndOned
# rounds for each phase: 31. In round 1, each node tries to elect itself and
notifies neighbors;2. In round 2, each node receives notifications
from neighbors, decide whether it is in Ik, and notifies neighbors;
3. In round 3, each node receiving notifications from elected neighbors, realizes to be in N(Ik),
notifies its neighbors about that, and stops. total # of rounds: )log( ndO
(w.h.p.)
Time complexity
49
Homework
Can you provide a good bound on the number of messages?