rumors: how can they work ? summer school math biology, 2007 nuno & sebastian

24
RUMORS: How can they work ? Summer School Math Biology, 2007 Nuno & Sebastian

Post on 19-Dec-2015

216 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: RUMORS: How can they work ? Summer School Math Biology, 2007 Nuno & Sebastian

RUMORS: How can they work ?

Summer School Math Biology, 2007

Nuno & Sebastian

Page 2: RUMORS: How can they work ? Summer School Math Biology, 2007 Nuno & Sebastian

Outline

• What is a rumor ?

• Deterministic vs Stochastic

• Simple models

• Not so simple models

• Summary

• Discrete & Galton-Watson Process

Page 3: RUMORS: How can they work ? Summer School Math Biology, 2007 Nuno & Sebastian

“unverified proposition of belief that bears topical relevance for persons actively involved in its dissemination”

“unauthenticated bits of information in that they are deprived of “secure standards of evidence”.”

What is a rumor ?

?

Page 4: RUMORS: How can they work ? Summer School Math Biology, 2007 Nuno & Sebastian
Page 5: RUMORS: How can they work ? Summer School Math Biology, 2007 Nuno & Sebastian

Deterministic vs Stochastic

• How can we model a rumor ?

• Is a deterministic or stochastic approach better ?

• How do these approaches differ ?

Page 6: RUMORS: How can they work ? Summer School Math Biology, 2007 Nuno & Sebastian

Deterministic vs Stochastic

Discrete

Galton-Watson Process

Markov Chain

Esteban likes it !!!!

WHY ?

Page 7: RUMORS: How can they work ? Summer School Math Biology, 2007 Nuno & Sebastian

tRtItItRttR

tRtItINtItIttI

Simple models

Deterministic

Natural recovered Forced recovered

Mass action interaction between Infectious and the total population

Natural recovered Forced recovered

Infectious after 1 time step

Infectious 1 time step before

Recovered 1 time step before

Recovered after 1 time step

Page 8: RUMORS: How can they work ? Summer School Math Biology, 2007 Nuno & Sebastian

tItRptIptRtIptRttR

ppptRtIptIttI

barbar

bar

0

021 2

Simple models

Probability of infected someone

Probability of doing nothing

Number of infected after forced recovery

Natural recoveredForced recovered

Infectious after 1 time step

Recovered 1 time step before

Recovered after 1 time step

Stochastic

Probability of forgetting the rumor

Page 9: RUMORS: How can they work ? Summer School Math Biology, 2007 Nuno & Sebastian

Simple model assumptions

1. Total population size is extremely large;

2. The number of susceptibles remains roughly constant;

3. The size of the epidemic remains quite small;

4. Mass action interaction (homogeneous population);

5. In the stochastic model, forced recovery precedes other events;

Page 10: RUMORS: How can they work ? Summer School Math Biology, 2007 Nuno & Sebastian

SIMPLE MODEL - Deterministic Results

• Infected 0

• 2 “types” fixed points

(I*,R*) = (0,0) and (I*,R*)=(0,R)

• eigenvalue of 1 ?

• “epidemic” if (αN/) > 1.

Page 11: RUMORS: How can they work ? Summer School Math Biology, 2007 Nuno & Sebastian

SIMPLE MODEL - Stochastic Results

Stochastic model extinction of the rumor!!!

Page 12: RUMORS: How can they work ? Summer School Math Biology, 2007 Nuno & Sebastian

Effect of I0 on rumor life-time (both models)

Rumor life-time is inversely proportional to I0

Page 13: RUMORS: How can they work ? Summer School Math Biology, 2007 Nuno & Sebastian

“Strange” Results: Effect of α on rumor lifetime

αN/ < 1 αN/ > 1

Page 14: RUMORS: How can they work ? Summer School Math Biology, 2007 Nuno & Sebastian

Simple model Extinction of the rumor

×

× ?

Page 15: RUMORS: How can they work ? Summer School Math Biology, 2007 Nuno & Sebastian

tRtRtItItRttR

tRtItItItStIttI

tRtItStSttS

Not so simple models models

DeterministicSusceptible after 1 time step

Recovered that become susceptible again

Page 16: RUMORS: How can they work ? Summer School Math Biology, 2007 Nuno & Sebastian

Not so simple models models

Stochastic

tRptItRptIptRtIptRttR

ptSpptRtIptIttI

tRptRtIptItSptSttS

barbar

bar

bar

40

021

42

2

Probability that recovered that become susceptible again

Page 17: RUMORS: How can they work ? Summer School Math Biology, 2007 Nuno & Sebastian

Not so simple model assumptions

1. Total population size is constant;

2. Mass action interaction (homogeneous population);

3. In the stochastic model, forced recovery precedes other events;

Page 18: RUMORS: How can they work ? Summer School Math Biology, 2007 Nuno & Sebastian

NOT SO SIMPLE MODEL - Deterministic results

ELVIS IS ALIVE!?!?!

• Model with model extinction and “endemic” rumors

• None of the fixed points are stable...

Page 19: RUMORS: How can they work ? Summer School Math Biology, 2007 Nuno & Sebastian

NOT SO SIMPLE MODEL - Stochastic results

Page 20: RUMORS: How can they work ? Summer School Math Biology, 2007 Nuno & Sebastian

Effect of population size – stochastic model

Page 21: RUMORS: How can they work ? Summer School Math Biology, 2007 Nuno & Sebastian

Effect of population size – deterministic model

• For the deterministic case population size only changes the scale of the epidemic

• In the stochastic model however, increasing the population size generates very different behaviour

Page 22: RUMORS: How can they work ? Summer School Math Biology, 2007 Nuno & Sebastian

Not so simple model – comparison of deterministic & stochastic results

• For large (~ p4) coexistence is observed in both deterministic and stochastic

• For small deterministic predicts repeated outbreaks of the rumor. This is not possible in the stochastic model (by varying p4)

• For the deterministic model the population size does not make any difference, but population size affects the predictions of the stochastic model

Page 23: RUMORS: How can they work ? Summer School Math Biology, 2007 Nuno & Sebastian

Summary

1. Rumors can be modelled similarly to infectious diseases;

2. Not so different models can give us very different predictions;

3. Under certain conditions, stochastic models predict very different results from deterministic ones

Page 24: RUMORS: How can they work ? Summer School Math Biology, 2007 Nuno & Sebastian

Acknowledgments

Julien Jungmin

Thank you very much !!

Group