Download - 1 dorogovtsev
![Page 1: 1 dorogovtsev](https://reader033.vdocuments.net/reader033/viewer/2022060116/557d7828d8b42a58788b45d2/html5/thumbnails/1.jpg)
Basic epidemiologyof complex networks
Sergey Dorogovtsev
University of Aveiro and Ioffe Institute
Avalanche collapse in interdependent networks,
Baxter, S.D., Goltsev, and Mendes,
Phys. Rev. Lett. 109, 248701 (2012);
arXiv:1207.0448
![Page 2: 1 dorogovtsev](https://reader033.vdocuments.net/reader033/viewer/2022060116/557d7828d8b42a58788b45d2/html5/thumbnails/2.jpg)
Solid state physics vs science of networks:
Researcher: “I have a dream” —
(object – new material) =⇒ device or technol.process
⇓ ⇑understanding → · · · · · · · · →
⇑ ⇓(object – new network) =⇒ algorithm
Google PageRank
(arXiv:1207.0448) 2 / 28
![Page 3: 1 dorogovtsev](https://reader033.vdocuments.net/reader033/viewer/2022060116/557d7828d8b42a58788b45d2/html5/thumbnails/3.jpg)
The big data problem
(arXiv:1207.0448) 3 / 28
![Page 4: 1 dorogovtsev](https://reader033.vdocuments.net/reader033/viewer/2022060116/557d7828d8b42a58788b45d2/html5/thumbnails/4.jpg)
Lords of the world: intellect & information
The big data problem.
Privacy is actually reduced to the big data problem.
90% of the data in the world has been generated in thepast two years.
Google invested $1.6bn on building data centres just inthe three months from April to June.
Technology firms to spend $140bn on building new datacentres this year.
(arXiv:1207.0448) 4 / 28
![Page 5: 1 dorogovtsev](https://reader033.vdocuments.net/reader033/viewer/2022060116/557d7828d8b42a58788b45d2/html5/thumbnails/5.jpg)
Progress: 1997
Google’s servers in 1997(arXiv:1207.0448) 5 / 28
![Page 6: 1 dorogovtsev](https://reader033.vdocuments.net/reader033/viewer/2022060116/557d7828d8b42a58788b45d2/html5/thumbnails/6.jpg)
Progress: 2001
Google’s servers in 2001(arXiv:1207.0448) 6 / 28
![Page 7: 1 dorogovtsev](https://reader033.vdocuments.net/reader033/viewer/2022060116/557d7828d8b42a58788b45d2/html5/thumbnails/7.jpg)
Progress: now
Google’s server room in council Bluffs, Iowa.(arXiv:1207.0448) 7 / 28
![Page 8: 1 dorogovtsev](https://reader033.vdocuments.net/reader033/viewer/2022060116/557d7828d8b42a58788b45d2/html5/thumbnails/8.jpg)
A central cooling plant in Google’s data centre in Georgia.
(arXiv:1207.0448) 8 / 28
![Page 9: 1 dorogovtsev](https://reader033.vdocuments.net/reader033/viewer/2022060116/557d7828d8b42a58788b45d2/html5/thumbnails/9.jpg)
Estimating their share in the world’seconomy
(arXiv:1207.0448) 9 / 28
![Page 10: 1 dorogovtsev](https://reader033.vdocuments.net/reader033/viewer/2022060116/557d7828d8b42a58788b45d2/html5/thumbnails/10.jpg)
Lords of the world: intellect & information
The big data problem.
Privacy is actually reduced to the big data problem.
90% of the data in the world has been generated in thepast two years.
Google invested $1.6bn on building data centres just inthe three months from April to June.
Technology firms to spend $140bn on building new datacentres this year.
(arXiv:1207.0448) 10 / 28
![Page 11: 1 dorogovtsev](https://reader033.vdocuments.net/reader033/viewer/2022060116/557d7828d8b42a58788b45d2/html5/thumbnails/11.jpg)
Information:
Automated processing of big data.
Software is more expensive than hardware.
Expensive writing the code.
2012 is the first year when both Apple and Google spentmore on patent wars than on innovations.
(arXiv:1207.0448) 11 / 28
![Page 12: 1 dorogovtsev](https://reader033.vdocuments.net/reader033/viewer/2022060116/557d7828d8b42a58788b45d2/html5/thumbnails/12.jpg)
Mapping the global social network
Vertices - 94 entity types.
Edges - 164 relationship types.
- a huge coloured graph (edges and vertices are ofdifferent kinds).
(arXiv:1207.0448) 12 / 28
![Page 13: 1 dorogovtsev](https://reader033.vdocuments.net/reader033/viewer/2022060116/557d7828d8b42a58788b45d2/html5/thumbnails/13.jpg)
Interdependent networks:
Networks of networks
(arXiv:1207.0448) 13 / 28
![Page 14: 1 dorogovtsev](https://reader033.vdocuments.net/reader033/viewer/2022060116/557d7828d8b42a58788b45d2/html5/thumbnails/14.jpg)
(arXiv:1207.0448) 14 / 28
![Page 15: 1 dorogovtsev](https://reader033.vdocuments.net/reader033/viewer/2022060116/557d7828d8b42a58788b45d2/html5/thumbnails/15.jpg)
Remaining giant cluster:
(arXiv:1207.0448) 15 / 28
![Page 16: 1 dorogovtsev](https://reader033.vdocuments.net/reader033/viewer/2022060116/557d7828d8b42a58788b45d2/html5/thumbnails/16.jpg)
The SIS model without approximationsOn complex nets, the SIS model 6= contact process.
Deroulers & Monasson (2003), the complete graph of 100 nodes(arXiv:1207.0448) 16 / 28
![Page 17: 1 dorogovtsev](https://reader033.vdocuments.net/reader033/viewer/2022060116/557d7828d8b42a58788b45d2/html5/thumbnails/17.jpg)
The SIS on an individual graph,(but not on a statistical ensemble!)
The evolution equation neglecting correlations betweeninfective and susceptible nodes:
dρi (t)
dt= −ρi (t) + λ[1− ρi (t)]
N∑j=1
Aijρj(t).
The steady state:
ρi =λ
∑j Aijρj
1 + λ∑
j Aijρj.
(arXiv:1207.0448) 17 / 28
![Page 18: 1 dorogovtsev](https://reader033.vdocuments.net/reader033/viewer/2022060116/557d7828d8b42a58788b45d2/html5/thumbnails/18.jpg)
The SIS on an annealed network
The evolution equation neglecting correlations betweeninfective and susceptible nodes:
dρk(t)
dt= −ρk(t)+λ[1−ρk(t)]
∑k
kP(k)ρk(t)/∑
q
qP(q).
The steady state:
ρk =kλ
∑k kP(k)ρk(t)/
∑q qP(q)
1 + kλ∑
k kP(k)ρk(t)/∑
q qP(q).
(arXiv:1207.0448) 18 / 28
![Page 19: 1 dorogovtsev](https://reader033.vdocuments.net/reader033/viewer/2022060116/557d7828d8b42a58788b45d2/html5/thumbnails/19.jpg)
ρi =∑
Λ
c(Λ)fi (Λ).
c(Λ) = λ∑
Λ′
Λ′c(Λ′)N∑
i=1
fi (Λ)fi (Λ′)
1 + λ∑
Λ̃ Λ̃c(Λ̃)fi (Λ̃).
If τ ≡ λΛ1−1�1, then
ρ ≡N∑
i=1
ρi/N ≈ α1τ,
α1 = [N∑
i=1
fi (Λ1)]/[NN∑
i=1
f 3i (Λ1)].
(arXiv:1207.0448) 19 / 28
![Page 20: 1 dorogovtsev](https://reader033.vdocuments.net/reader033/viewer/2022060116/557d7828d8b42a58788b45d2/html5/thumbnails/20.jpg)
Localized and delocalized eigenvectors:
Normalization:∑N
i f 2i (Λ) = 1
The inverse participation ratio:
IPR(Λ) ≡N∑
i=1
f 4i (Λ).
A delocalized state: IPR(Λ)N→∞→ 0.
A localized state: IPR(Λ)N→∞→ const.
It depends on a particular network realization.
(arXiv:1207.0448) 20 / 28
![Page 21: 1 dorogovtsev](https://reader033.vdocuments.net/reader033/viewer/2022060116/557d7828d8b42a58788b45d2/html5/thumbnails/21.jpg)
A delocalized principal eigenvector: fi (Λ) = O(1/√
N),so
α1 = O(1)
A localized principal eigenvector:
α1 = O(1/N)
— the disease is localized on a finite number of verticesin contrast to “localization” within k-cores (a finitefraction of vertices), see Kitsak (2010), Castellano andPastor-Satorras (2012).
(arXiv:1207.0448) 21 / 28
![Page 22: 1 dorogovtsev](https://reader033.vdocuments.net/reader033/viewer/2022060116/557d7828d8b42a58788b45d2/html5/thumbnails/22.jpg)
Unweighted networks:
The first iteration (g(0)=1):
Λ(1)1 =
1
〈q2〉N∑
i ,j
qiAijqj = ΛMF +〈q〉σ2r
〈q2〉,
IPRMF = 〈q4〉/[N〈q2〉2] ∼ O(1/N)
where r is the Pearson coefficient, ΛMF≡〈q2〉/〈q〉.So assortative degree–degree correlations increase Λ1 anddecrease λc .
(arXiv:1207.0448) 22 / 28
![Page 23: 1 dorogovtsev](https://reader033.vdocuments.net/reader033/viewer/2022060116/557d7828d8b42a58788b45d2/html5/thumbnails/23.jpg)
Weighted and unweighted real-world nets:
(a) astro-phys (upper) and cond-mat-2005 (lower) weighted
networks. (b) Karate-club network (unweighted). The lower curve
accounts for only the eigenstate Λ1. The most upper curve is the
“exact” ρ.(arXiv:1207.0448) 23 / 28
![Page 24: 1 dorogovtsev](https://reader033.vdocuments.net/reader033/viewer/2022060116/557d7828d8b42a58788b45d2/html5/thumbnails/24.jpg)
An uncorrelated scale-free net:
A scale-free network of 105 vertices generated by the static model
with γ = 4, 〈q〉 = 10. (b) Zoom of the prevalence at λ near
λc = 1/Λ1. I, II eigenvectors are localized, III is delocalized.(arXiv:1207.0448) 24 / 28
![Page 25: 1 dorogovtsev](https://reader033.vdocuments.net/reader033/viewer/2022060116/557d7828d8b42a58788b45d2/html5/thumbnails/25.jpg)
A Bethe lattice with a hub:
Find Λ1 if B = k − 1 is the branching and then Bsubstitute with 〈B〉.The localization of the principal eigenvector at a hubwith qmax occurs if
Λ1 = qmax/√
qmax − B ≥ Λd >≈ ΛMF
(arXiv:1207.0448) 25 / 28
![Page 26: 1 dorogovtsev](https://reader033.vdocuments.net/reader033/viewer/2022060116/557d7828d8b42a58788b45d2/html5/thumbnails/26.jpg)
Uncorrelated scale-free networks:
The principal eigenvector is localized if γ > 5/2
The principal eigenvector is delocalized if 2 < γ ≤ 5/2(but notice loops!)
(arXiv:1207.0448) 26 / 28
![Page 27: 1 dorogovtsev](https://reader033.vdocuments.net/reader033/viewer/2022060116/557d7828d8b42a58788b45d2/html5/thumbnails/27.jpg)
Conclusion
If the principal eigenvector is localized,then below the endemic epidemic threshold λc ,
the disease localized on a finite number of verticesslowly decays with time.
The decay time diverges at the epidemic threshold.
(arXiv:1207.0448) 27 / 28
![Page 28: 1 dorogovtsev](https://reader033.vdocuments.net/reader033/viewer/2022060116/557d7828d8b42a58788b45d2/html5/thumbnails/28.jpg)
Lectures on
Complex Networks
(arXiv:1207.0448) 28 / 28