katarzyna sznajd-weron katedra fizyki teoretycznejkatarzynaweron/students/intro... · 6 two basic...

Post on 24-Jan-2020

0 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

TRANSCRIPT

Complex networks

Katarzyna Sznajd-WeronKatedra Fizyki Teoretycznej

Internet Movie DataBase(IMDB)

A Few Good Man

Robert Wagner

Austin Powers: The spy who shagged me

Wild Things

Let’s make it legal

Barry Norton

What Price Glory

Monsieur Verdoux

3

Try to play at https://oracleofbacon.org/

4

Network science

• A.L. Barabasi, Network Science, Cambridge University Press (2016)

• Online at http://networksciencebook.com/

5

Interdisciplinary research

Arkadiusz JędrzejewskiBachelor: PhysicsMaster: Applied MathPhD: Physics (statistical physics)

Several figures from his thesis will be used during this lecture.

6

Two basic network parameters

• Number of nodes (𝑵 – the size of the network): the number of components in the system.

• To distinguish the nodes, we label them with 𝑖 = 1, 2, … ,𝑁. • Number of links (𝑳): the total number of interactions

between the nodes. • Links are rarely labeled, as they can be identified through the

pair of nodes they connect: 1,2 , 2,3 , 2,4 , 4,5 , (5,7)

node (vertex)

link (edge)

7

Network science and graph theory

Network science Graph theory

network graph

node vertex

link edge

• Two terminologies are used interchangeably!• A subtle distinction between them:

• the terms {network, node, link} often refers to real systems

• the terms {graph, vertex, edge} refers to mathematical representation of networks

8

The same graph for different networks

© 2017 Marcin Weron

The social network

The technological network

The corresponding graph

9

Adjacency Matrix

A complete description of a network

10

• The simplest way: a complete list of the links

– {(1, 2), (1, 3), (2, 3), (2, 4)}

• Hint: for contact processes (opinion dynamics)

– For each node a vector of neighbours

Nn[1] = [2,3]

Nn[2] = [1,3]

Nn[3] = [1,2]

Nn[4] = [2]

Another description of a network

11

Examples of networks

Network Nodes Links Dir / Undir N L

Internet Routers Internet connections Undirected 192,244 609,066

WWW Webpages Links Directed 325,729 1,497,134

Mobile-Phone Calls

Subscribers Calls Directed 36,595 91,826

Email Email add Emails Directed 57,194 103,731

Science Collaboration

Scientists Co-authorships Undirected 23,133 93,437

Actor Network Actors Co-acting Undirected 702,388 29,397,908

Citation Network Papers Citations Directed 449,673 4,689,479

12

J. J. Potterat et al. (2002): Risk network structure in the early epidemic phase of HIV transmission in Colorado Springs

A social network of sexual contacts

13

Dating in a high school

nodes: studentsBlue: boysPink: girlslinks: they were on a date

14

Other examples - puzzle

15

• Basic properties of a node

– Degree

– Clustering Coefficient

• Basic properties of a network

– Degree distribition

– Average degree

– Average Path Length

– Average Clustering Coefficient

Basic properties

16

• Degree of 𝑖 − 𝑡ℎ node (𝑘𝑖): number of links attached (in/out in directed) to the node

• Example: 𝑘1 = 1, 𝑘2 = 3, 𝑘3 = 1, 𝑘4 = 2, 𝑘5 =2, 𝑘6 = 0, 𝑘7 = 1

• The total number of links:

𝐿 =1

2

𝑖=1

𝑁

𝑘𝑖

Basic properties of a node: Degree

17

Directed networks

• Incoming degree (𝑘𝑖𝑖𝑛): the number of links that

point to node 𝑖

• Outgoing degree (𝑘𝑖𝑜𝑢𝑡): the number of links that

point from node 𝑖 to other nodes

• A node’s total degree

𝑘𝑖 = 𝑘𝑖𝑖𝑛 + 𝑘𝑖

𝑜𝑢𝑡

• The total number of links in a directed network:

𝐿 =

𝑖=1

𝑁

𝑘𝑖𝑖𝑛 =

𝑖=1

𝑁

𝑘𝑖𝑜𝑢𝑡

18

Basic properties of a network: Degreedistribution and average degree

• The probability that a randomly selected node in the network has degree 𝑘

• Average degree:

𝑘 =1

𝑁

𝑖=1

𝑁

𝑘𝑖 =2𝐿

𝑁

19

Basic properties: average path length

• A path: a route that runs along the links of the network

• A path’s length: the number of links the path contains

• The shortest path between nodes 𝑖 and 𝑗: the path with the fewest number of links that connect nodes 𝑖 and 𝑗

• Diameter: the longest shortest path in a graph

• Average path length

20

Basic properties: clustering coefficient

the number of links between the 𝑘𝑖 neighbors of node 𝑖

𝐶 =1

𝑁

𝑖=1

𝑁

𝐶𝑖

21

„Small world” by Stanley Milgram

Stanley Milgram, "The Small World Problem", Psychology Today, vol. 1, no. 1,

May 1967, pp61‐67 (8542 cyt.)

22

Results

Stanley Milgram, "The Small World Problem", Psychology Today, vol. 1, no. 1,

May 1967, pp61‐67 (8542 cyt.)

23Sociometry, Vol. 32, No. 4 (Dec., 1969), pp. 425-443

24

Results

25

Small world of facebook

The New York Times, 21.11.2011: 4.74The world is shrinking!

26

It began in 1998? Not really ...

27

• 𝐺(𝑁, 𝐿): 𝑁 labeled nodes are connected with 𝐿 randomly placed links[Erdős & Rényi, 1959-1968]

• 𝑮(𝑵, 𝒑): Each pair of 𝑁 labeled nodes is connected with probability 𝑝 [Gilbert, 1959]

Erdős–Rényi random graphs, 1960

28

1) Start with N isolated nodes

2) Select a node pair and generate 𝑟 ~𝑈 0,1

3) If 𝑝 < 𝑟 then connect the selected pair

4) Otherwise leave them disconnected

5) Repeat step 2) for each of the 𝑁 𝑁−1

2node

pairs

Random graph algorithm 𝑮(𝑵, 𝒑)

29

1) Start with N isolated nodes

2) Select a node pair and generate 𝑟 ~𝑈 0,1

3) If 𝑝 < 𝑟 then connect the selected pair

4) Otherwise leave them disconnected

5) Repeat step 2) for each of the 𝑁 𝑁−1

2node

pairs

Random graph algorithm 𝑮(𝑵, 𝒑)

30

Erdős–Rényi random graphs, 1960

(c) 2016, Arkadiusz Jędrzejewski

𝑘 = 𝑝 𝑁 − 1

𝑝𝑘 =𝑁 − 1𝑘

𝑝𝑘 1 − 𝑝 𝑁−1−𝑘

Most real networks are sparse, i.e. 𝑘 ≪ 𝑁:

𝑝𝑘 = exp(− 𝑘 )𝑘 𝑘

𝑘!Poisson Distribution

31

• Small diameter

• High clustering coefficient

Properties of a real social network

Source: Wolfram|Alpha Personal Analytics for Facebook

Ego graph: Kasia is not present in the graph

32

Watts-Strogatz model: small-world network

Watts, D. J.; Strogatz, S. H. (1998)"Collective dynamics of 'small-world' networks„ Nature 393: 440–442.

(c) 2016, Arkadiusz Jędrzejewski

33

34

1/2

1/2

2/4

1/4

1/4

1/6

3/6

1/6

1/6

1/8

3/8

2/8

1/8

1/8

...

Barabasi-Albert network

A.-L.Barabási, R. Albert, Science 286, 509 (1999)

35

Barabási, Albert-László; Albert, Réka (1999) "Emergence of scaling in random networks" Science 286, 509–512.

Scale-free networks

hub

36

Network robustness

• The Impact of Node Removal1. Attacks 2. Errors

• How to measure robustness?

• Percolation theory

37

The Impact of Node Removal

How you define an error and how you define an attack?

38

Think about two types of networks

carbon nanotubessource: https://arstechnica.com

Internet

• The Impact of Node Removal1. Attacks 2. Errors

39

The Impact of Node Removal

Diameter of the network

40

Network fragmentation under failures and attacks

the average size of the isolated clusters

the relative size of the largest clusters S

the fraction of removed nodes f

41

• Spreading phenomena on networks

• Temporal networks

• Multiplex

Petter Holme, Jari Saramäki

Physics Reports

Vol 519, Pages 97–125 (2012)

There is more …

42

Multiplex (Anna Chmiel)CS-AARHUS

The multiplex social network consists of five kinds of online and offline relationships

(Facebook, Leisure, Work, Co-authorship, Lunch) between the employees of Computer Science department at Aarhus

5 layers Multiplex

Nodes: 61

Edges: 620

Matteo Magnani, Barbora Micenkova, Luca Rossi - Combinatorial Analysis of Multiple Networks. arXiv:1303.4986 (2013)

43

More on http://networksciencebook.com/

top related