Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· -...

44
Leveraging Linial’s Locality limit Christoph Lenzen and Roger Wattenhofer Presented by Yam Chemel

Transcript of Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· -...

Page 1: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

Leveraging Linial’s Locality limit

Christoph Lenzen and Roger Wattenhofer

Presented by Yam Chemel

Page 2: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

Linial (1992)

3-coloring and MIS

on a ring (𝑅𝑛)

takes at least logβˆ—(𝑛) rounds

Reminder

Page 3: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

Prove logβˆ—(𝑛) Lower bound for MaxIS algorithms in rings

Our Goal:

How?

MaxIS π‘œ(logβˆ— 𝑛 )

3-coloring π‘œ logβˆ— 𝑛

Contradicts Linial!

Page 4: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

MaxIS vs. MIS - Example

Independent Set MIS MaxIS

(general graphs)

MIS ⟸ MaxIS

Page 5: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

MaxIS in rings

1 2

3 4

1 2

3

4

5

Q: What if each even node returns 1, and each odd node returns 0?

1 2

3 4

𝑅4 MaxIS

1 2

3

4

5 𝑅5

MaxIS

Page 6: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

1 3

2 4

MaxIS

Identifiers are not necessarily in order!

1 3

2 4

Page 7: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

MaxIS vs. MIS – in rings

MIS ⟸ MaxIS

Algorithm 𝐴 MaxIS

π‘œ(logβˆ— 𝑛 )

Algorithm 𝐴 MIS

π‘œ logβˆ— 𝑛

Contradicts Linial!

Page 8: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

f-approximation algorithm for MaxIS

𝑀 -MaxIS solution 𝐼 – algorithm A solution

𝐼 = 𝑀

If 𝐴 is 𝑓(𝑛)-approximation: 𝐼 ≀ 𝑀

and also:

𝐼 βˆ— 𝑓(𝑛) β‰₯ |𝑀|

𝐴 𝑅𝑛 = 𝑅𝑛

A is 2-approximation for MaxIS

𝐼 =𝑛

4

𝑛

4βˆ— 2 β‰₯ 𝑀 =

𝑛

2

Page 9: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

Leveraging Linial’s Locality limit

MaxIS approximation π‘œ(π‘™π‘œπ‘”βˆ— 𝑛 )

3-coloring π‘œ π‘™π‘œπ‘”βˆ— 𝑛

Page 10: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

Leveraging Linial’s Locality limit

MaxIS approximation 𝜎 𝑛

3-coloring 𝜎(𝑛)

Page 11: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

Leveraging Linial’s Locality limit

𝜎(𝑛)-alternating 𝜎(𝑛)

MaxIS approximation 𝜎 𝑛

3-coloring 𝜎(𝑛)

Page 12: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

𝜎(𝑛)-alternating algorithm - Definition:

Suppose A is an algorithm operating on 𝑅𝑛 which assigns each node 𝑣𝑖 ∈ 𝑉𝑛 a value 𝑐(𝑣𝑖) ∈ {0, 1}. We call A 𝜎(n)-alternating, if the length of any monochromatic sequence

𝑐 𝑣𝑖 = 𝑐 𝑣𝑖+1 = β‹― = 𝑐 𝑣𝑖+π‘˜ is smaller than 𝝈(n).

< 𝜎(𝑛)

Page 13: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

𝐴 𝑅𝑛 = 𝑅𝑛

𝜎(𝑛)-alternating algorithm - examples

n is even

Longest monochromatic sequence = 1

n is odd

Longest monochromatic sequence = 2

A is 3-alternating (also 4-alternating, 5-alternating…)

(Not necessarily MaxIS algorithm)

Page 14: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

𝐴 𝑅𝑛 = 𝑅𝑛

A is (n+1)-alternating

(Not necessarily MaxIS algorithm)

𝜎(𝑛)-alternating algorithm - examples

Page 15: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

Lemma - Modified MaxIS approximation Suppose an f-approximation algorithm A for the MaxIS problem on the ring 𝑅𝑛 running in at most g(n) β‰₯ 1 rounds is given, where we have 𝑓(𝑛)𝑔(𝑛) ∈ π‘œ(logβˆ—(𝑛)). Then an π‘œ(logβˆ—(𝑛))-alternating algorithm Aβ€² requiring π‘œ logβˆ— 𝑛 communication rounds exists.

A

MaxIS, 𝑓(𝑛)-approximation,

𝑔(𝑛) rounds, 𝑓 𝑛 𝑔 𝑛 ∈ π‘œ π‘™π‘œπ‘”βˆ— 𝑛

A’ π‘œ π‘™π‘œπ‘”βˆ— 𝑛 -alternating,

π‘œ π‘™π‘œπ‘”βˆ— 𝑛 rounds

Page 16: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

Modified MaxIS approximation - Proof

General idea

𝑆𝑛

𝑅𝑛

𝑅𝑛′ 𝑛

𝑆𝑛′

Page 17: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

Observation: For each node 𝑣𝑖, 𝑐(𝑣𝑖) is a function of:

𝑣𝑖

𝑔(𝑛) 𝑔(𝑛)

𝑔 𝑛 = #π‘Ÿπ‘œπ‘’π‘›π‘‘π‘ 

β€’ The IDs of its 𝑔(𝑛) neighbors on each side

β€’ 𝑛 β€’ Its ID

𝑔(𝑛) 𝑔(𝑛)

Modified MaxIS approximation - Proof

Page 18: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

β€’ 𝑆𝑒𝑑 𝜎 𝑛 ← 10𝑓(𝑛)𝑔(𝑛)

𝑆𝑛 =

𝑣1, … , 𝑣𝑔 𝑛 , 𝑣𝑔 𝑛 +1, …, 𝑣𝑔 𝑛 +𝜎 𝑛 , 𝑣𝑔 𝑛 +1+𝜎 𝑛 , … , 𝑣2𝑔 𝑛 +𝜎 𝑣

| βˆ€π‘– ∈ 𝑣𝑔 𝑛 +1, …, 𝑣𝑔 𝑛 +𝜎 𝑛 𝑐 𝑣𝑖 = 0

0 0

?

?

0

? ?

? ?

𝑔(𝑛) 𝑔(𝑛) 𝜎(𝑛)

Modified MaxIS approximation - Proof

β€’ Define exactly the set of sequences preventing that A is 𝜎(𝑛)-alternating

Page 19: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

𝑔(𝑛) 𝜎(𝑛)

? ? ? 0 0 0 ? ? ?

Id=2 Id=11 Id=8 Id=9 Id=14 Id=21 Id=7 Id=6 Id=3

𝑔(𝑛)

𝑠 = 2, 11, 8, 9, 14, 21, 7, 6, 3 β†’ 𝑆𝑛

Take 2𝑔 𝑛 + 𝜎(𝑛) consecutive nodes in 𝑅𝑛

Assign them identifiers

Run A on the sequence / on 𝑅𝑛 containing the sequence.

If the 𝜎(𝑛) center nodes compute 0, add s to 𝑆𝑛

Id=2 Id=11 Id=8 Id=9 Id=14 Id=21 Id=7 Id=6 Id=3

Building 𝑆𝑛

2𝑔 𝑛 + 𝜎(𝑛)

Page 20: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

Construct a sequence of identifiers for 𝑅𝑛:

𝑠2

𝑠1 𝑛 βˆ’ 𝑗

𝑅𝑛 π‘ π‘˜

Modified MaxIS approximation - Proof

1. Choose an arbitrary sequence s from 𝑆𝑛 and assign the identifiers to 𝑣1, … , 𝑣 𝑠 .

2. Assuming we already assigned labels to the nodes 𝑣1, … , 𝑣𝑗:

While there exists a sequence 𝑠 ∈ 𝑆𝑛 that can be appended to 𝑣1, … , 𝑣𝑗 without duplicating an identifier, we do so.

3. If no further sequence fits, we add 𝑛 βˆ’ 𝑗 arbitrary identifiers not yet present in 𝑣1, … , 𝑣𝑗 to

complete the labeling (𝑣1, … , 𝑣𝑛) of 𝑅𝑛.

Page 21: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

Assume for contradiction, that for arbitrarily large n it is possible to

label 𝑅𝑛 as described, with at least 𝑛 βˆ’π‘›

5𝑓 𝑛 identifiers

stemming from sequences out of 𝑆𝑛.

# π‘›π‘œπ‘‘π‘’π‘  π‘‘β„Žπ‘Žπ‘‘ π‘π‘œπ‘šπ‘π‘’π‘‘π‘’ 1 𝑖𝑛 𝑅𝑛 ≀ 𝑛 βˆ’πœŽ 𝑛 βˆ— 𝑛

𝜎 𝑛 + 2𝑔 π‘›βˆ’

𝑛

5𝑓 𝑛

β‰₯ π‘›π‘œπ‘‘π‘’π‘  π‘‘β„Žπ‘Žπ‘‘ π‘π‘œπ‘šπ‘π‘’π‘‘π‘’ 0 𝑖𝑛 𝑠 ∈ 𝑆𝑛

π‘›π‘œπ‘‘π‘’π‘  𝑖𝑛 𝑠 ∈ π‘†π‘›βˆ— 𝑛 βˆ’ #(π‘›π‘œπ‘‘π‘’π‘  π‘›π‘œπ‘‘ 𝑖𝑛 𝑠 ∈ 𝑆𝑛)

Modified MaxIS approximation - Proof

# π‘›π‘œπ‘‘π‘’π‘  π‘‘β„Žπ‘Žπ‘‘ π‘π‘œπ‘šπ‘π‘’π‘‘π‘’ 0 𝑖𝑛 𝑅𝑛 β‰₯

=𝜎 𝑛

𝜎 𝑛 + 2𝑔(𝑛)βˆ— 𝑛 βˆ’

𝑛

5𝑓 𝑛

Page 22: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

# π‘›π‘œπ‘‘π‘’π‘  π‘‘β„Žπ‘Žπ‘‘ π‘π‘œπ‘šπ‘π‘’π‘‘π‘’ 1 𝑖𝑛 𝑅𝑛 ≀ 𝑛 βˆ’πœŽ 𝑛 βˆ— 𝑛

𝜎 𝑛 + 2𝑔 π‘›βˆ’

𝑛

5𝑓 𝑛

= 𝑛 1 βˆ’πœŽ 𝑛

𝜎 𝑛 + 2𝑔 𝑛+

1

5𝑓 𝑛

𝜎 𝑛 = 10𝑓 𝑛 𝑔(𝑛)

Modified MaxIS approximation - Proof

= 𝑛 1 βˆ’10𝑓 𝑛 𝑔 𝑛

10𝑓 𝑛 𝑔 𝑛 + 2𝑔 𝑛+

1

5𝑓 𝑛

= 𝑛 1 βˆ’5𝑓 𝑛

5𝑓 𝑛 + 1+

1

5𝑓 𝑛

= 𝑛 1 βˆ’ 1 βˆ’1

5𝑓 𝑛 + 1+

1

5𝑓 𝑛

= 𝑛1

5𝑓 𝑛 + 1+

1

5𝑓 𝑛

≀ 𝑛1

5𝑓 𝑛 + 0+

1

5𝑓 𝑛

=2𝑛

5𝑓 𝑛

Page 23: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

𝐼 = # π‘›π‘œπ‘‘π‘’π‘  π‘‘β„Žπ‘Žπ‘‘ π‘π‘œπ‘šπ‘π‘’π‘‘π‘’ 1 𝑖𝑛 𝑅𝑛 ≀2𝑛

5𝑓 𝑛

However, according to f-approximation definition: 𝑓 𝑛 𝐼 β‰₯ 𝑀

(M -an arbitrary MaxIS of G)

𝑅6

Contradicts the assumption! (β€œFor arbitrarily large n it is possible to label 𝑅𝑛 as described, with at least

𝑛 βˆ’π‘›

5𝑓 𝑛 identifiers stemming from sequences out of 𝑆𝑛.”)

Modified MaxIS approximation - Proof

𝑓(𝑛) 𝐼 ≀2

5𝑛

Choosing every other node in 𝑅𝑛 is a MaxIS solution:

𝑀 =𝑛

2

Combining the equations:

𝑓 𝑛 𝐼 β‰₯𝑛

2

Page 24: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

At least 𝒏

πŸ“π’‡ 𝒏 identifiers remain which cannot form a further

sequence from 𝑆𝑛.

Set 𝑛′ ← max 𝑛0, 5𝑓 𝑛 βˆ™ 𝑛

In 𝑅𝑛′

#π‘Ÿπ‘’π‘šπ‘Žπ‘–π‘›π‘–π‘›π‘” π‘–π‘‘π‘’π‘›π‘‘π‘–π‘“π‘–π‘’π‘Ÿπ‘ 

β‰₯𝑛′

5𝑓 𝑛′

For a large n (𝑛 > 𝑛0) it is possible to label 𝑅𝑛 as described, with

at most 𝑛 βˆ’π‘›

5𝑓 𝑛 identifiers stemming from sequences out of 𝑺𝒏

Modified MaxIS approximation - Proof

𝑠2

𝑠1 𝑛 βˆ’ 𝑗

𝑅𝑛 π‘ π‘˜

β‰₯5𝑓 𝑛 βˆ™ 𝑛

5𝑓 𝑛= 𝑛

Page 25: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

𝑅𝑛 𝑅𝑛′

π‘Ÿπ‘›

Algorithm A’ : π‘Ÿπ‘’π‘› ∢ 𝐴(𝑛′, 𝑛 π‘–π‘‘π‘’π‘›π‘‘π‘–π‘“π‘–π‘’π‘Ÿπ‘  π‘“π‘Ÿπ‘œπ‘š 𝑅𝑛′ )

Modified MaxIS approximation - Proof

Page 26: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

𝑅4 𝑅11

1

2

3

4

3

8

1

4

π‘Ÿ4

Algorithm A’ : 𝑅𝑒𝑛 𝐴(11, 3, 8, 1, 4 )

Example:

Modified MaxIS approximation - Proof

Page 27: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

A’ is 𝝈 𝒏′ -alternating

No 𝝈(𝒏′) consecutive 0’s No 𝝈(𝒏′) consecutive 1’s

𝐴 computes an independent setβ†’ no 2 neighbors in independent set β†’

no 2 consecutive 1’s

No 𝑠 ∈ 𝑆𝑛′ in remaining n nodes of 𝑅𝑛′ β†’ No 𝜎 𝑛′ consecutive nodes compute 0 in

A’ run

Maximum monochromatic sequence : 𝜎 𝑛′ βˆ’ 1 ∈ 𝑂 𝑓 𝑛′ 𝑔 𝑛′ βŠ‚ π‘œ logβˆ— 𝑛′ = π‘œ logβˆ— 𝑛

A’ running time -𝒐(π’π’π’ˆβˆ—(𝒏)) rounds 𝑔 𝑛 ≀ 𝑔 𝑛 𝑓 𝑛 ∈ π‘œ(π‘™π‘œπ‘”βˆ— 𝑛)

𝑔 𝑛′ ∈ π‘œ(π‘™π‘œπ‘”βˆ— 𝑛)

A’ is 𝒐 π’π’π’ˆβˆ— 𝒏 -alternating

Page 28: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

Leveraging Linial’s Locality limit

𝜎(𝑛)-alternating 𝜎(𝑛)

MaxIS approximation 𝜎 𝑛

3-coloring 𝜎(𝑛)

Page 29: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

Leveraging Linial’s Locality limit

𝜎(𝑛)-alternating 𝜎(𝑛)

MaxIS approximation 𝜎 𝑛

3-coloring 𝜎(𝑛)

Page 30: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

2. Proof : Lemma – Given a 𝝈(𝒏)-alternating algorithm A running in 𝑂(𝜎(𝑛)) rounds, a 3-coloring of the ring can be computed in 𝑂(𝜎(𝑛)) rounds.

1. Run A. Let 𝑑(𝑣) ∈ {0, 1} denote the result of this run.

2. Find a pair of neighboring nodes {𝑀1, 𝑀2} with 𝑑 𝑀1 β‰  𝑑 𝑀2 which is closest to v.

0 1 𝒗 𝑀2

1 0 𝒗 𝑀2

If 𝑣 ∈ {𝑀1, 𝑀2}: if 𝑑(𝑣) = 0: set 𝑐 𝑣 ← 𝑏 otherwise: set 𝑐(𝑣) ← π‘Ÿ

0 1 𝒗 𝑀2

1 0 𝒗 𝑀2

1 0 1 1

𝑀2 𝑀1 𝒗 𝑀

1 0 1 1

Else: denote by 𝛿 the distance to the closer node in {𝑀1, 𝑀2}, w.l.o.g. 𝑀1. if 𝛿 ∈ 2β„•: set 𝑐 𝑣 ← 𝑐(𝑀1) else: set 𝑐 𝑣 ← 𝑐(𝑀2)

1 0 1 1

𝑀2 𝑀1 𝒗 𝑀 (𝛿 ∈ 2β„•)

𝛿

Page 31: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

3 8 𝒗 𝑀

8 3 𝒗 𝑀

3. If v has a neighbor w with 𝑐(𝑣) = 𝑐(𝑀) and v > w, set 𝑐(𝑣) ← 𝑔.

3 8 𝒗 𝑀

8 3 𝒗 𝑀

3 8 𝒗 𝑀

1 2 3 8 𝒗 𝑀

1 2 (3)

4. If v has a neighbor w with 𝑐(𝑣) = 𝑐(𝑀) = 𝑔 and v > w, set c(v) to the color none of the neighbors of v has.

3 8 𝒗 𝑀

1 2 (4)

2. Proof : Lemma – Given a 𝝈(𝒏)-alternating algorithm A running in 𝑂(𝜎(𝑛)) rounds, a 3-coloring of the ring can be computed in 𝑂(𝜎(𝑛)) rounds.

Page 32: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

Running time: 𝑢 𝝈 𝒏

Step 1: Running A- 𝑂 𝜎 𝑛

Step 2: Finding a pair of neighbors with different d - 𝜎 𝑛 .

No more than 𝜎(𝑣) consecutive nodes take the same decision d(v)

since A is 𝜎(𝑛)-alternating.

Page 33: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

Valid 3-coloring of 𝑹𝒏

Step 2: Each node 𝑣 chooses different from one of its neighbors,

1 1 0 1 1 1 1 0 1

so at most one of the neighbors of 𝑣 may take the same choice.

Step 3: From each pair of neighbors with the same color one chooses g.

Step 4: If that same color was g, v chooses the color non of its neighbors has.

v v

v

1 1 0

v

1 1 1 1 0 1

v v

Page 34: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

Leveraging Linial’s Locality limit

𝜎(𝑛)-alternating 𝜎(𝑛)

MaxIS approximation 𝜎 𝑛

3-coloring 𝜎(𝑛)

Page 35: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

Therefore, there isn’t a MaxIS approximation algorithms running on a ring

in less than log*(n)

Proof - Summary

Assume by contradiction that there exists a MaxIS approximation algorithm A running in less than π‘™π‘œπ‘”βˆ—(𝑛).

Construct a π‘œ π‘™π‘œπ‘”βˆ— 𝑛 -alternating algorithm running in π‘œ(π‘™π‘œπ‘”βˆ—(𝑛)) using algorithm A.

By lemma 2, a 3-coloring of the ring can be computed in 𝒐(π’π’π’ˆβˆ—(𝒏)) rounds.

This contradicts Linial’s 3-coloring lower bound.

Page 36: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

- Discuss MDS lower bound - Compare MDS and MaxIS difficulty

Our (new) Goal:

No! We’ll show a case where MIS is easier than MDS

MDS on rings O(1)

Is MDS always easier than MIS?

How?

Page 37: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

MDS – Minimum Dominating Set

DS MDS

Page 38: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

MDS in rings

Taking every third node gives a minimum dominating set

𝑀 =𝑛

3

Taking every node gives a 3-approximation MDS in 1 round ∈ π‘œ logβˆ— 𝑛

𝑅𝑛

𝑅𝑛

There is no π’π’π’ˆβˆ—(𝒏) bound MDS approximation in rings! MDS approximation in rings takes 𝑢(𝟏) rounds

MDS approximation in rings

Page 39: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

Can we compare MaxIS and MDS difficulty?

MDS

MaxIS

We saw that in 𝑅𝑛 MDS can be computed in 1 round, but MaxIS requires at least π‘™π‘œπ‘”βˆ—(𝑛) round.

Is it always easier to compute MDS than MaxIS?

Page 40: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

MaxIS graph family

𝐺 𝑣 𝑀

Any graph that can be constructed this way

Page 41: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

Lemma (Local computation of a MaxIS on MaxIS Graphs): The set {𝑣 ∈ 𝑉 | |𝑁1

+(𝑣)|π‘šπ‘œπ‘‘ 2 = 1} is a MaxIS for any MaxIS Graph.

Proof – part 1:

- For 𝑣𝑖 ∈ 𝑉𝑖 𝑖 ∈ {3,4}, 𝑣𝑖 has 2|𝑁1+ 𝑣 | neighbors.

- 𝑁1+ 𝑣𝑖 = 2 𝑁1

+ 𝑣 + 1 β‡’ 𝑁1+ 𝑣𝑖 𝑖𝑠 π‘œπ‘‘π‘‘

- For 𝑣𝑖 ∈ 𝑉𝑖 𝑖 ∈ {1,2}, 𝑁1+ 𝑣𝑖 = 4 𝑁1

+ 𝑣 β‡’ 𝑁1+ 𝑣𝑖 𝑖𝑠 𝑒𝑣𝑒𝑛

- {𝑣 ∈ 𝑉 | |𝑁1+(𝑣)|π‘šπ‘œπ‘‘ 2 = 1} = 𝑉3 βˆͺ 𝑉4

Page 42: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

Proof – part 2:

- 𝑉3 βˆͺ 𝑉4 is an Independent set – according to the construction of

MaxIS graph

- (𝑣1, 𝑣3, 𝑣2, 𝑣4) forms a cycle, so for each 4 nodes as such, only 2 can be in the IS.

- MaxIS can’t be larger than 𝑉

2

- 𝑉3 βˆͺ 𝑉4 =𝑉

2

β‡’ 𝑉3 βˆͺ 𝑉4 𝑖𝑠 π‘Ž π‘€π‘Žπ‘₯𝐼𝑆

Lemma (Local computation of a MaxIS on MaxIS Graphs): The set {𝑣 ∈ 𝑉 | |𝑁1

+(𝑣)|π‘šπ‘œπ‘‘ 2 = 1} is a MaxIS for any MaxIS Graph.

Conclusion – MaxIS on a MaxIS graph can be determined locally, without communication (in 𝑂(1) rounds).

Page 43: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

MDS on MaxIS graphs

We can prove that MDS on MaxIS graphs is as efficient as in general graphs, meaning:

Ξ©log 𝑛

log log 𝑛

MaxIS

MDS

Page 44: Christoph Lenzen and Roger Wattenhoferwebcourse.cs.technion.ac.il/236825/Spring2015/ho...Β Β· - MaxIS can’t be larger than 𝑉 2 - 𝑉3βˆͺ𝑉4 = 𝑉 2 ⇒𝑉3βˆͺ𝑉4 𝐼 Lemma

Ring graphs - 𝑅𝑛 MaxIS graphs

MaxIS

MDS 𝑂(1)

Ξ©log 𝑛

log log 𝑛

MDS

MaxIS 𝑂(1)

Ξ© logβˆ— 𝑛

MaxIS and MDS are not comparable in general graphs!