Scheidegger NetworksA Bonus First return random walk Calculation - - PowerPoint PPT Presentation

▶
scheidegger networks a bonus
SMART_READER_LITE
LIVE PREVIEW

Scheidegger NetworksA Bonus First return random walk Calculation - - PowerPoint PPT Presentation

Aug 29, 2022 331 likes •670 views

Scheidegger Networks Scheidegger NetworksA Bonus First return random walk Calculation References Complex Networks, Course 295A, Spring, 2008 Prof. Peter Dodds Department of Mathematics & Statistics University of Vermont Licensed

slide-1
SLIDE 1

Scheidegger Networks First return random walk References Frame 1/11

Scheidegger Networks—A Bonus Calculation

Complex Networks, Course 295A, Spring, 2008

  • Prof. Peter Dodds

Department of Mathematics & Statistics University of Vermont

Licensed under the Creative Commons Attribution-NonCommercial-ShareAlike 3.0 License.

slide-2
SLIDE 2

Scheidegger Networks First return random walk References Frame 2/11

Outline

First return random walk

slide-3
SLIDE 3

Scheidegger Networks First return random walk References Frame 2/11

Outline

First return random walk References

slide-4
SLIDE 4

Scheidegger Networks First return random walk References Frame 3/11

Random walks

◮ We’ve seen that Scheidegger networks have random

walk boundaries [1, 2]

slide-5
SLIDE 5

Scheidegger Networks First return random walk References Frame 3/11

Random walks

◮ We’ve seen that Scheidegger networks have random

walk boundaries [1, 2]

◮ Determining expected shape of a ‘basin’ becomes a

problem of finding the probability that a 1-d random walk returns to the origin after t time steps

slide-6
SLIDE 6

Scheidegger Networks First return random walk References Frame 3/11

Random walks

◮ We’ve seen that Scheidegger networks have random

walk boundaries [1, 2]

◮ Determining expected shape of a ‘basin’ becomes a

problem of finding the probability that a 1-d random walk returns to the origin after t time steps

◮ We solved this with a counting argument for the

discrete random walk the preceding Complex Systems course

slide-7
SLIDE 7

Scheidegger Networks First return random walk References Frame 3/11

Random walks

◮ We’ve seen that Scheidegger networks have random

walk boundaries [1, 2]

◮ Determining expected shape of a ‘basin’ becomes a

problem of finding the probability that a 1-d random walk returns to the origin after t time steps

◮ We solved this with a counting argument for the

discrete random walk the preceding Complex Systems course

◮ For fun and the constitution, let’s work on the

continuous time Wiener process version

slide-8
SLIDE 8

Scheidegger Networks First return random walk References Frame 3/11

Random walks

◮ We’ve seen that Scheidegger networks have random

walk boundaries [1, 2]

◮ Determining expected shape of a ‘basin’ becomes a

problem of finding the probability that a 1-d random walk returns to the origin after t time steps

◮ We solved this with a counting argument for the

discrete random walk the preceding Complex Systems course

◮ For fun and the constitution, let’s work on the

continuous time Wiener process version

◮ A classic, delightful problem

slide-9
SLIDE 9

Scheidegger Networks First return random walk References Frame 4/11

Random walks

The Wiener process (⊞)

slide-10
SLIDE 10

Scheidegger Networks First return random walk References Frame 5/11

Random walking on a sphere...

The Wiener process (⊞)

slide-11
SLIDE 11

Scheidegger Networks First return random walk References Frame 6/11

Random walks

◮ Wiener process = Brownian motion

slide-12
SLIDE 12

Scheidegger Networks First return random walk References Frame 6/11

Random walks

◮ Wiener process = Brownian motion ◮

x(t2) − x(t1) ∼ N(0, t2 − t1) where N(x, t) = 1 √ 2πt e−x2/2t

slide-13
SLIDE 13

Scheidegger Networks First return random walk References Frame 6/11

Random walks

◮ Wiener process = Brownian motion ◮

x(t2) − x(t1) ∼ N(0, t2 − t1) where N(x, t) = 1 √ 2πt e−x2/2t

◮ Continuous but nowhere differentiable

slide-14
SLIDE 14

Scheidegger Networks First return random walk References Frame 7/11

First return

◮ Objective: find g(t), the probability that Wiener

process first returns to the origin at time t.

slide-15
SLIDE 15

Scheidegger Networks First return random walk References Frame 7/11

First return

◮ Objective: find g(t), the probability that Wiener

process first returns to the origin at time t.

◮ Use what we know: the probability density for a

return (not necessarily the first) at time t is f(t) = 1 √ 2πt e−0/2t = 1 √ 2πt

slide-16
SLIDE 16

Scheidegger Networks First return random walk References Frame 7/11

First return

◮ Objective: find g(t), the probability that Wiener

process first returns to the origin at time t.

◮ Use what we know: the probability density for a

return (not necessarily the first) at time t is f(t) = 1 √ 2πt e−0/2t = 1 √ 2πt

◮ Observe that f and g are connected like this:

f(t) = t

τ=0

f(τ)g(t − τ)dτ + δ(t)

  • Dirac delta function
slide-17
SLIDE 17

Scheidegger Networks First return random walk References Frame 7/11

First return

◮ Objective: find g(t), the probability that Wiener

process first returns to the origin at time t.

◮ Use what we know: the probability density for a

return (not necessarily the first) at time t is f(t) = 1 √ 2πt e−0/2t = 1 √ 2πt

◮ Observe that f and g are connected like this:

f(t) = t

τ=0

f(τ)g(t − τ)dτ + δ(t)

  • Dirac delta function

◮ In words: Probability of returning at time t equals the

integral of the probability of returning at time τ and then not returning until exactly t − τ time units later.

slide-18
SLIDE 18

Scheidegger Networks First return random walk References Frame 8/11

First return

◮ Next see that right hand side of

f(t) = t

τ=0 f(τ)g(t − τ)dτ + δ(t) is a juicy

convolution.

slide-19
SLIDE 19

Scheidegger Networks First return random walk References Frame 8/11

First return

◮ Next see that right hand side of

f(t) = t

τ=0 f(τ)g(t − τ)dτ + δ(t) is a juicy

convolution.

◮ So we take the Laplace transform:

L[f(t)] = F(s) = ∞

t=0− f(t)e−stdt

slide-20
SLIDE 20

Scheidegger Networks First return random walk References Frame 8/11

First return

◮ Next see that right hand side of

f(t) = t

τ=0 f(τ)g(t − τ)dτ + δ(t) is a juicy

convolution.

◮ So we take the Laplace transform:

L[f(t)] = F(s) = ∞

t=0− f(t)e−stdt ◮ and obtain

F(s) = F(s)G(s) + 1

slide-21
SLIDE 21

Scheidegger Networks First return random walk References Frame 8/11

First return

◮ Next see that right hand side of

f(t) = t

τ=0 f(τ)g(t − τ)dτ + δ(t) is a juicy

convolution.

◮ So we take the Laplace transform:

L[f(t)] = F(s) = ∞

t=0− f(t)e−stdt ◮ and obtain

F(s) = F(s)G(s) + 1

◮ Rearrange:

G(s) = 1 − 1/F(s)

slide-22
SLIDE 22

Scheidegger Networks First return random walk References Frame 9/11

First return

◮ We are here: G(s) = 1 − 1/F(s)

slide-23
SLIDE 23

Scheidegger Networks First return random walk References Frame 9/11

First return

◮ We are here: G(s) = 1 − 1/F(s) ◮ Now we want to invert G(s) to find g(t)

slide-24
SLIDE 24

Scheidegger Networks First return random walk References Frame 9/11

First return

◮ We are here: G(s) = 1 − 1/F(s) ◮ Now we want to invert G(s) to find g(t) ◮ Use calculation that F(s) = (2s)−1/2

slide-25
SLIDE 25

Scheidegger Networks First return random walk References Frame 9/11

First return

◮ We are here: G(s) = 1 − 1/F(s) ◮ Now we want to invert G(s) to find g(t) ◮ Use calculation that F(s) = (2s)−1/2 ◮

G(s) = 1 − (2s)1/2

slide-26
SLIDE 26

Scheidegger Networks First return random walk References Frame 9/11

First return

◮ We are here: G(s) = 1 − 1/F(s) ◮ Now we want to invert G(s) to find g(t) ◮ Use calculation that F(s) = (2s)−1/2 ◮

G(s) = 1 − (2s)1/2 ≃ e−(2s)1/2

slide-27
SLIDE 27

Scheidegger Networks First return random walk References Frame 10/11

First return

Groovy aspects of g(t) ∼ t−3/2:

slide-28
SLIDE 28

Scheidegger Networks First return random walk References Frame 10/11

First return

Groovy aspects of g(t) ∼ t−3/2:

◮ Variance is infinite (weird but okay...)

slide-29
SLIDE 29

Scheidegger Networks First return random walk References Frame 10/11

First return

Groovy aspects of g(t) ∼ t−3/2:

◮ Variance is infinite (weird but okay...) ◮ Mean is also infinite (just plain crazy...)

slide-30
SLIDE 30

Scheidegger Networks First return random walk References Frame 10/11

First return

Groovy aspects of g(t) ∼ t−3/2:

◮ Variance is infinite (weird but okay...) ◮ Mean is also infinite (just plain crazy...) ◮ Distribution is normalizable so process always

returns to 0.

slide-31
SLIDE 31

Scheidegger Networks First return random walk References Frame 10/11

First return

Groovy aspects of g(t) ∼ t−3/2:

◮ Variance is infinite (weird but okay...) ◮ Mean is also infinite (just plain crazy...) ◮ Distribution is normalizable so process always

returns to 0.

◮ For river networks: P(ℓ) ∼ ℓ−γ so γ = 3/2 for

Scheidegger networks.

slide-32
SLIDE 32

Scheidegger Networks First return random walk References Frame 11/11

References I

  • A. E. Scheidegger.

A stochastic model for drainage patterns into an intramontane trench.

  • Bull. Int. Assoc. Sci. Hydrol., 12(1):15–20, 1967.

.

  • A. E. Scheidegger.

Theoretical Geomorphology. Springer-Verlag, New York, third edition, 1991. .