Stochastic processes in natural and social phenomena

Term: 
Winter
Credits: 
2.0
Course Description: 
Course code: CNSC 6003
Level:  Doctoral or MA
Course Status:  Elective

e-learning.ceu.edu

Office hours: upon agreement

Course Description

Very few processes are completely deterministic – an element of randomness is almost always present. The adequate mathematical framework to treat this randomness  is the theory of stochastic processes. They are abundant in practically all aspects of nature and society. The theory of stochastic processes is part of the mathematical theory of probability, however, in this course the pragmatic approach will be taken that the basic laws will be introduced to the students through the analysis of natural and social phenomena.  Relatively limited mathematical prerequisites will be sufficient for attending the course (elementary calculus and some basic knowledge about differential equations). After a short overview of the fundamentals of probability theory we will discuss random processes from physics, population dynamics, epidemiology and finance. We will discuss the basic stochastic processes and the tools to handle them always in a heuristic, application-oriented manner.

COURSE SCHEDULE

Week no.  Session title

1. Introduction

What are stochastic processes? Their abundance. Examples from physics, population dynamics, epidemics and finance.

2. Probability theory

Random variables, distributions. Independent variables. Conditional probability. Averages and moment. Central limit theorem. Dependencies. Multivariate problems.

3. Poisson process

Radioactivity, random queuing problem, birth-and-death processes. Waiting time distribution. Properties of the Poisson process.

4. Renewal processes

Problems of insurance policy. Generalization of the Poisson process to arbitrary waiting time distributions. Waiting time paradox. Renewal equation.

5. Markov processes

The Markov property. Brownian motion and random walk. Discrete random walk. Further examples (population dynamics, economics).

Markov chains, transition matrix, representation as directed graphs.

6. Branching processes

Examples: Genealogy, cascading. The tree topology. The problem of extinction. Tree-approximation to networks with loops.

Master (rate) equation

Ergodicity. Continuous time Markov processes. Transition probabilities from empirical data.  Stationary solution and relaxation to it.

7. Fokker-Planck (forward Kolmogorov) equation

Diffusion and migration in space. Derivation of the FP equation, interpretation  of the terms. Simple solutions.

8. Continuous time random walks

Generalization of the Poisson process. Montroll-Weiss equation. Fat tailed distributions.  Applications to financial time series.

9. First passage time problems

Expected lifetimes of patients, the Gambler’s ruin problem. The problem of stability in a noisy environment. Relation to risk estimation. First passage time of the one-dimensional Brownian motion. Characteristic times in the limit order book.

10. Extreme value statistics

The definition of the problem. Applications: reliability and financial risk estimation. The three basic cases and distributions. Consequences for stochastic processes.

11. Stochastic differential equations

The Langevin equation of the Brownian particle. Types of noise, white noise. The Wiener and the Ornstein Uhlenbeck processes. Geometric Brownian motion and the Black-Scholes equation of option pricing.

12. Non-Markovian processes

Colored noise, memory effects. Power spectrum, 1/fα noise. Water level fluctuations of the Nile, Hurst exponent. Fractional Brownian motion. White to colored noise transformation. ARCH-GARCH models of return fluctuations. 

5

Small world model

Watts-Strogatz model: rewiring and adding shortcuts. Calculation of the shortest path, degree distribution and clustering coefficient. Scaling.

6

Power law degree distribution

Barabási-Albert model. Temporal evolution. Degree distribution. Clustering coefficient. Small and ultrasmall worlds. Vertex copying models. Geographical models.

7

Mesoscopic structures

Motifs and their calculation. Communities. Basic detection methods. Benchmarks.

Midterm test

8

Weighted network models

Different sources of weights. Min cut max flow theorem. Intensity and coherence. Weighted motifs and communities.

9

Diffusion and spreading on networks

The diffusion problem. Diffusion on the ER, WS, BA graphs. Stationary solutions, relaxation. Searching on networks.

10

Basic epidemic spreading models. Threshold phenomena. Mean field solutions. Epidemic models on graphs. The dynamics of spreading.

11

Bursty signals and temporal networks

Burstiness as basic feature of human behavior. Communication networks are temporal. Characterization. Temporal motifs.

12

Project presentation

Reading:

A. Barrat, M. Barthélemy and A. Vespignani: Dynamic Processes on Networks (Cambridge, 2008)

N.G. van Kampen: Stochastic Processes (North Holland, 1992)

M.E.J. Newman: Networks. An introduction (Oxford UP, 2010)

W. Paul and J. Baschnagel: Stochastic Processes – From Physics to Finance (Springer, 1999)

Further literature will be suggested during the course.

Learning Outcomes: 
  • Students will learn, what stochastic processes are and will get acquainted with a broad set of examples from the fields of physics, population dynamics, epidemiology and  finance;
  • They will learn the basics of the theory of stochastic processes;
  • They will be able to recognize the need of applying tools from the theory of stochastic processes in solving real world problems;
  • They will acquire skills to solve simple problems in the field of stochastic processes.
Assessment: 

(1) Assessment 1 (20% of final grade): Students home works. Simple exercises to deepen understanding.

(2) Assessment 2 (35% of final grade): Midterm test.

(3) Assessment 3 (45% of final grade): Final test. Answering questions to account for understanding the basic concepts.

Prerequisites: 

Elements of calculus, differential equations.