08.10.2016 Views

Foundations of Data Science

2dLYwbK

2dLYwbK

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

or<br />

The two triangles <strong>of</strong> Part 1 are either<br />

disjoint or share at most one vertex<br />

The two triangles<br />

<strong>of</strong> Part 2 share an<br />

edge<br />

The two triangles in<br />

Part 3 are the same triangle<br />

Figure 4.4: The triangles in Part 1, Part 2, and Part 3 <strong>of</strong> the second moment argument<br />

for the existence <strong>of</strong> triangles in G(n, d n ).<br />

Let’s calculate E(x 2 ) where x is the number <strong>of</strong> triangles. Write x as x = ∑ ijk ∆ ijk,<br />

where ∆ ijk is the indicator variable <strong>of</strong> the triangle with vertices i, j, and k being present.<br />

Expanding the squared term<br />

( ∑ ) 2 ( ∑<br />

)<br />

E(x 2 ) = E ∆ ijk = E ∆ ijk ∆ i ′ j ′ k ′ .<br />

i,j,k<br />

i, j, k<br />

i ′ ,j ′ ,k ′<br />

Split the above sum into three parts. In Part 1, let S 1 be the set <strong>of</strong> i, j, k and i ′ , j ′ , k ′<br />

which share at most one vertex and hence the two triangles share no edge. In this case,<br />

∆ ijk and ∆ i ′ j ′ k ′ are independent and<br />

( ∑ )<br />

E ∆ ijk ∆ i ′ j ′ k ′ = ∑ (<br />

E(∆ ijk )E(∆ i ′ j ′ k ′) ≤ ∑ ) ( ∑ )<br />

E(∆ ijk ) E(∆ i ′ j ′ k ′) = E 2 (x).<br />

S 1 S 1 all<br />

all<br />

ijk<br />

i ′ j ′ k ′<br />

In Part 2, i, j, k and i ′ , j ′ , k ′ share two vertices and hence one edge. See Figure 4.4.<br />

Four vertices and five edges are involved overall. There are at most ( n<br />

4)<br />

∈ O(n 4 ), 4-vertex<br />

subsets and ( 4<br />

2)<br />

ways to partition the four vertices into two triangles with a common edge.<br />

The probability <strong>of</strong> all five edges in the two triangles being present is p 5 , so this part sums<br />

to O(n 4 p 5 ) = O(d 5 /n) and is o(1). There are so few triangles in the graph, the probability<br />

<strong>of</strong> two triangles sharing an edge is extremely unlikely.<br />

In Part 3, i, j, k and i ′ , j ′ , k ′ are the same sets. The contribution <strong>of</strong> this part <strong>of</strong> the<br />

summation to E(x 2 ) is ( )<br />

n<br />

3 p 3 = d3 . Thus, putting all three parts together, we have:<br />

6<br />

which implies<br />

E(x 2 ) ≤ E 2 (x) + d3<br />

6 + o(1),<br />

Var(x) = E(x 2 ) − E 2 (x) ≤ d3<br />

6 + o(1).<br />

78

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!