hidden geometric correlations in real multiplex networks

53
Correlations in real MULTIPLEXES HIDDEN GEOMETRIC Kaj Kolja KLEINEBERG Universitat de Barcelona @KoljaKleineberg [email protected]

Upload: kolja-kleineberg

Post on 14-Apr-2017

754 views

Category:

Science


0 download

TRANSCRIPT

Correlationsin real

MULTIPLEXES

HIDDEN GEOMETRIC

Kaj Kolja KLEINEBERGUniversitat de Barcelona

@[email protected]

You have no message

routingthe problem of

map relies on underlying

metric space

real systems interact:

multiplexes

between maps of layers?Relation

Hidden metric spaces underlying real complex networksprovide a fundamental explanation of their observed topologies

Nature Physics 5, 74–80 (2008)

Hidden metric spaces underlying real complex networksprovide a fundamental explanation of their observed topologies

How can we infer the coordinates of nodes embeddedin hidden metric spaces?

Hyperbolic geometry emerges from Newtonian modeland similarity× popularity optimization in growing network

S1

H2 growing

1

2

3

p(κ) ∝ κ−γ

ri = R− 2 ln κi

κmin t = 1, 2, 3 . . .

r = 1

1+[d(θ,θ′)µκκ′

]1/T

p(xij) =1

1+exij−R

2T

mins [s×∆θst]

PRL 100, 078701

PRE 82, 036106 Nature 489, 537–540

Hyperbolic geometry emerges from Newtonian modeland similarity× popularity optimization in growing network

S1 H2

growing

1

2

3

p(κ) ∝ κ−γ ri = R− 2 ln κi

κmin

t = 1, 2, 3 . . .

r = 1

1+[d(θ,θ′)µκκ′

]1/T p(xij) =1

1+exij−R

2T

mins [s×∆θst]

PRL 100, 078701 PRE 82, 036106

Nature 489, 537–540

Hyperbolic geometry emerges from Newtonian modeland similarity× popularity optimization in growing network

S1 H2 growing

1

2

3

p(κ) ∝ κ−γ ri = R− 2 ln κi

κmin t = 1, 2, 3 . . .

r = 1

1+[d(θ,θ′)µκκ′

]1/T p(xij) =1

1+exij−R

2T

mins [s×∆θst]

PRL 100, 078701 PRE 82, 036106 Nature 489, 537–540

Constituent network layers of real multiplex systemsare embedded into separate hyperbolic spaces

Invert model→ embed real networks: PRE 92, 022807 (2015)

Internet Air and train Drosophila

C. Elegans Human brain arXiv

Rattus Physicians SacchPomb

Constituent network layers of real multiplex systemsare embedded into separate hyperbolic spaces

Invert model→ embed real networks: PRE 92, 022807 (2015)

Internet Air and train Drosophila

C. Elegans Human brain arXiv

Rattus Physicians SacchPomb

Constituent network layers of real multiplex systemsare embedded into separate hyperbolic spaces

Internet IPv4 network Internet IPv6 network

Constituent network layers of real multiplex systemsare embedded into separate hyperbolic spaces

Internet IPv4 network Internet IPv6 network

Are coordinates of same nodes in different layerscorrelated?

Radial and angular coordinates are correlatedbetween different layers in many real multiplexes

0

π

θ1

0

π

θ2

50100150200

Multiplexes are not random superpositions of layersbut instead are dictated by geometric correlations.

Radial and angular coordinates are correlatedbetween different layers in many real multiplexes

0

π

θ1

0

π

θ2

50100150200

Multiplexes are not random superpositions of layersbut instead are dictated by geometric correlations.

Distance between pairs of nodes in one layer isan indicator of the connection probability in another layer

Hyperbolic distance in IPv4

Co

nn

ecti

on

pro

b. i

n IP

v6

P(2|1)

0 5 10 15 20 25 30 35 40

10-4

10-3

10-2

10-1

100

Geometric correlations enable precise trans-layerlink prediction.

Distance between pairs of nodes in one layer isan indicator of the connection probability in another layer

Hyperbolic distance in IPv4

Co

nn

ecti

on

pro

b. i

n IP

v6

P(2|1)

0 5 10 15 20 25 30 35 40

10-4

10-3

10-2

10-1

100

Pran(2|1)

Geometric correlations enable precise trans-layerlink prediction.

Distance between pairs of nodes in one layer isan indicator of the connection probability in another layer

Hyperbolic distance in IPv4

Co

nn

ecti

on

pro

b. i

n IP

v6

P(2|1)

0 5 10 15 20 25 30 35 40

10-4

10-3

10-2

10-1

100

Pran(2|1)

Geometric correlations enable precise trans-layerlink prediction.

navigation?How do geometric correlations

affect

Greedy routing in single network using hyperbolic spaceallows efficient navigation relying only on local knowledge

Greedy routing in single network using hyperbolic spaceallows efficient navigation relying only on local knowledge

Greedy routing in single network using hyperbolic spaceallows efficient navigation relying only on local knowledge

Greedy routing in single network using hyperbolic spaceallows efficient navigation relying only on local knowledge

Greedy routing in single network using hyperbolic spaceallows efficient navigation relying only on local knowledge

Greedy routing in single network using hyperbolic spaceallows efficient navigation relying only on local knowledge

Mutual greedy routing allows efficient navigationusing several network layers and metric spaces

Mutual greedy routing allows efficient navigationusing several network layers and metric spaces

Mutual greedy routing allows efficient navigationusing several network layers and metric spaces

Mutual greedy routing allows efficient navigationusing several network layers and metric spaces

Mutual greedy routing allows efficient navigationusing several network layers and metric spaces

Mutual greedy routing allows efficient navigationusing several network layers and metric spaces

Mutual greedy routing allows efficient navigationusing several network layers and metric spaces

Mutual greedy routing allows efficient navigationusing several network layers and metric spaces

How do geometriccorrelations affectperformance?

Do more layersimprove

performance?

How closeto the optimum isthe Internet?

Develop a model with tunable geometriccorrelations preserving constituent layer topologies.

Mutual greedy routing allows efficient navigationusing several network layers and metric spaces

How do geometriccorrelations affectperformance?

Do more layersimprove

performance?

How closeto the optimum isthe Internet?

Develop a model with tunable geometriccorrelations preserving constituent layer topologies.

Mutual greedy routing allows efficient navigationusing several network layers and metric spaces

How do geometriccorrelations affectperformance?

Do more layersimprove

performance?

How closeto the optimum isthe Internet?

Develop a model with tunable geometriccorrelations preserving constituent layer topologies.

Mutual greedy routing allows efficient navigationusing several network layers and metric spaces

How do geometriccorrelations affectperformance?

Do more layersimprove

performance?

How closeto the optimum isthe Internet?

Develop a model with tunable geometriccorrelations preserving constituent layer topologies.

Geometric multiplex model allows to tunecorrelations independently from layer topologies

Constituent layertopologies by

hyperbolic model

Radial correlationscontrolled byν ∈ [0, 1]

Angular corr.controlled byg ∈ [0, 1]

Geometric multiplex model allows to tunecorrelations independently from layer topologies

Constituent layertopologies by

hyperbolic model

Radial correlationscontrolled byν ∈ [0, 1]

Angular corr.controlled byg ∈ [0, 1]

Geometric multiplex model allows to tunecorrelations independently from layer topologies

Constituent layertopologies by

hyperbolic model

Radial correlationscontrolled byν ∈ [0, 1]

Angular corr.controlled byg ∈ [0, 1]

Geometric correlations determine the improvement ofmutual greedy routing by increasing the number of layers

0.0 0.2 0.4 0.6 0.8 1.0

0.2

0.4

0.6

0.8

1.0Hyperbolic routing

0.980

0.985

0.990

0.995

P

1 2 3 40

2

4

6

Layers

Mit

igat

ion

fact

or

opt. correlateduncorrelated

Angular correlations

Rad

ial c

orr

elat

ion

s

Reduction of failure rate

Routing is perfected by using many networkssimultaneously if correlations are present.

Geometric correlations determine the improvement ofmutual greedy routing by increasing the number of layers

0.0 0.2 0.4 0.6 0.8 1.0

0.2

0.4

0.6

0.8

1.0Hyperbolic routing

0.980

0.985

0.990

0.995

P

1 2 3 40

2

4

6

Layers

Mit

igat

ion

fact

or

opt. correlateduncorrelated

Angular correlations

Rad

ial c

orr

elat

ion

s

Reduction of failure rate

Routing is perfected by using many networkssimultaneously if correlations are present.

Geometric correlations determine the improvement ofmutual greedy routing by increasing the number of layers

0.0 0.2 0.4 0.6 0.8 1.0

0.2

0.4

0.6

0.8

1.0Hyperbolic routing

0.980

0.985

0.990

0.995

P

1 2 3 40

2

4

6

Layers

Mit

igat

ion

fact

or

opt. correlateduncorrelated

Angular correlations

Rad

ial c

orr

elat

ion

s

Reduction of failure rate

Routing is perfected by using many networkssimultaneously if correlations are present.

Internet?Do correlations help to navigate

the real

Internet multiplex model allows to study the performanceof mutual greedy routing for arbitrary correlations

Real Model

• Node only in IPv4 • Node in IPv4 and IPv6

Existing correlations in the Internet multiplex increaseperformance of mutual greedy routing significantly

0.0 0.2 0.4 0.6 0.8 1.00.0

0.2

0.4

0.6

0.8

1.0

g

νAngular

0.0 0.2 0.4 0.6 0.8 1.00.0

0.2

0.4

0.6

0.8

1.0

Hyperbolic

0.63 0.67 0.71 0.75

Success rate

0.85 0.91

Success rate

uncorr. emp. opt. uncorr. emp. opt.

0.88

Summary: Geometry of multiplex networks can makeinteracting decentralized systems perfectly navigable

0.0 0.2 0.4 0.6 0.8 1.00.0

0.2

0.4

0.6

0.8

1.0

g

ν

0.80

0.82

0.84

0.86

0.88

0.90

Fram

ewor

k

0

π

2 π

θ1

0

π

2 π

θ2

0.20.4

Res

ult

Bas

is

Navigation

Mutual greedy routing(local knowledge)

Geometriccorrelations

Layers embedded inhyperbolic space

Multidimensionalcommunity structure

Precise trans-layerlink prediction

Correlations increaserouting performance

Many correlated layersperfect routing

Distances between nodes are correlated

Hyperbolic distance in IPv4

Co

nn

ecti

on

pro

b. i

n IP

v6

P(2|1)

0 5 10 15 20 25 30 35 40

10-4

10-3

10-2

10-1

100

Pran(2|1)

Nature Physics, doi: 10.1038/NPHYS3782

Node coordinatesare correlated

Multiplexes not randomcombinations of layers

Multiplexorganization

Summary: Geometry of multiplex networks can makeinteracting decentralized systems perfectly navigable

0.0 0.2 0.4 0.6 0.8 1.00.0

0.2

0.4

0.6

0.8

1.0

g

ν

0.80

0.82

0.84

0.86

0.88

0.90

Fram

ewor

k

0

π

2 π

θ1

0

π

2 π

θ2

0.20.4

Res

ult

Bas

is

Navigation

Mutual greedy routing(local knowledge)

Geometriccorrelations

Layers embedded inhyperbolic space

Multidimensionalcommunity structure

Precise trans-layerlink prediction

Correlations increaserouting performance

Many correlated layersperfect routing

Distances between nodes are correlated

Hyperbolic distance in IPv4

Co

nn

ecti

on

pro

b. i

n IP

v6

P(2|1)

0 5 10 15 20 25 30 35 40

10-4

10-3

10-2

10-1

100

Pran(2|1)

Nature Physics, doi: 10.1038/NPHYS3782

Node coordinatesare correlated

Multiplexes not randomcombinations of layers

Multiplexorganization

Summary: Geometry of multiplex networks can makeinteracting decentralized systems perfectly navigable

0.0 0.2 0.4 0.6 0.8 1.00.0

0.2

0.4

0.6

0.8

1.0

g

ν

0.80

0.82

0.84

0.86

0.88

0.90

Fram

ewor

k

0

π

2 π

θ1

0

π

2 π

θ2

0.20.4

Res

ult

Bas

is

Navigation

Mutual greedy routing(local knowledge)

Geometriccorrelations

Layers embedded inhyperbolic space

Multidimensionalcommunity structure

Precise trans-layerlink prediction

Correlations increaserouting performance

Many correlated layersperfect routing

Distances between nodes are correlated

Hyperbolic distance in IPv4

Co

nn

ecti

on

pro

b. i

n IP

v6

P(2|1)

0 5 10 15 20 25 30 35 40

10-4

10-3

10-2

10-1

100

Pran(2|1)

Nature Physics, doi: 10.1038/NPHYS3782

Node coordinatesare correlated

Multiplexes not randomcombinations of layers

Multiplexorganization

Geometric correlations in real multiplexes:Reference and contact information

Thanks to: M. Boguñá, M. A. Serrano, F. Papadopoulos

Reference:

»Hidden geometric correlations in real multiplex networks«Nature Physics, doi: 10.1038/NPHYS3782K.-K. Kleineberg, M. Boguñá, M. A. Serrano, F. Papadopoulos

Kaj Kolja Kleineberg:

[email protected]

• @KoljaKleineberg

• koljakleineberg.wordpress.com

Geometric correlations in real multiplexes:Reference and contact information

Thanks to: M. Boguñá, M. A. Serrano, F. Papadopoulos

Reference:

»Hidden geometric correlations in real multiplex networks«Nature Physics, doi: 10.1038/NPHYS3782K.-K. Kleineberg, M. Boguñá, M. A. Serrano, F. Papadopoulos

Kaj Kolja Kleineberg:

[email protected]

• @KoljaKleineberg← Slides

• koljakleineberg.wordpress.com

IMAGE CREDITS

Compass: Martin FischMessage in bottle: SusanneNilssonOld globe: jayneanddCompass: Creative StallCompass (navigate): CreativeStallInternet router: Thomas Uebe

Train: Naomi Atkinsonfly (drosophila): Daan Kauwenbergworm (celegans): anbileru adalerubain network: parkjisuncoauthor: Matt Wassercommunity: Edward BoatmanLink: Rafaël Massé

Icons: thenounproject

Pictures: flickr