Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
STOCHASTIC PROCESSES: AN OVERVIEW M AT H 1 8 2 2 N D S E M AY 2 0 1 6 - 2 0 1 7 STOCHASTIC PROCESS ο Suppose we have an index set π» β [π, β) We usually call this βtimeβ ο πΏπ where π β π» is a stochastic or random process The set of random variables indexed by βtimeβ π» is called a stochastic or random process STOCHASTIC PROCESS οWhen π» is countable (e.g., π» = β βͺ {π}) πΏπ is a discrete-time process (e.g., πΏπ , πΏπ , πΏπ , πΏπ , β¦ ) οWhen π» is an interval in the real line (e.g., π» = π, ππ ) πΏπ is a continuous-time process οWhen π» is a singleton (e.g., π» = {π}) The process can be studied using a single random variable (MATH 181) STOCHASTIC PROCESS The value of the random variable πΏπ is the state of the system at time t. Note: πΏπ can be discrete or continuous (recall discrete and continuous probability distributions in MATH 181) οWhen πΏπ can only assume discrete values πΏπ is a discrete-state process (chain) οWhen πΏπ can assume continuous values πΏπ is a continuous-state process EXAMPLE Let the random process πΏπ , πΏπ , πΏπ , πΏπ represent the number of customers that have entered McDo Vega at time period 0 (8:00-8:59), 1 (9:00-9:59), 2 (10:00-10:59) and 3 (11:00-11:59). We have a discrete-time discrete-state stochastic process (or simply, a discrete-time chain). Stochastic processes usually model the evolution of a random system through time. STOCHASTIC PROCESS AS A FUNCTION OF TWO VARIABLES Recall, random variable: π β¦ πΏπ (π) Random variable is a function of outcome βa procedure assigning a number to an outcomeβ STOCHASTIC PROCESS AS A FUNCTION OF TWO VARIABLES We know that the outcome is uncertain. For now, let us consider a fixed outcome ππ .The random variable in a stochastic process with fixed outcome: π β¦ πΏπ (ππ ) Random variable as a function of time Actually, πΏπ (ππ ) is a function of time and outcome! SAMPLE PATH Again, the outcome is uncertain which means πΏπ (π) can have different possible states at time t. For example, at time 0, πΏπ (ππ ) and πΏπ (ππ ) are possible states where ππ and ππ are the possible outcomes. This means a stochastic process πΏπ (π) can have different βrealizationsβ. SAMPLE PATH A realization of the stochastic process is called a sample path (or sample trajectory). 1000 X_t 800 600 400 200 0 0 2 4 6 t sample path 1 sample path 2 8 10 TIME-SLICES OF A STOCHASTIC PROCESS 1000 600 30 runs 400 200 0 0 2 4 6 t 5sample path 1 4 3 2 1 0 8 10 Histogram of X_0 sample path 2 0 50 100 150 200 250 300 350 400 450 500 550 600 650 700 750 800 850 900 950 1000 X_t 800 TIME-SLICES OF A STOCHASTIC PROCESS 1000 600 30 runs 400 200 0 0 2 4 6 t 4sample path 1 8 10 Histogram of X_2 sample path 2 3 2 1 0 0 50 100 150 200 250 300 350 400 450 500 550 600 650 700 750 800 850 900 950 1000 X_t 800 UNCERTAINTIES Uncertainties does NOT mean there is no pattern! There are many cases where patterns exist, and mostly they are described by SPECIAL PROBABILITY DISTRIBUTIONS! and SPECIAL STOCHASTIC PROCESSES! UNCERTAINTIES SPECIAL PROBABILITY DISTRIBUTIONS and SPECIAL STOCHASTIC PROCESSES have well-studied properties and we can use these properties to describe systems with randomness. EXAMPLE Customers arrive RANDOMLY at a bakery at an average of 18 per hour on weekday mornings. The arrival distribution can be described by a POISSON distribution with a mean of 18. The saleslady can serve a customer in an average of three minutes; this time can be described by an exponential distribution with a mean of 3.0 minutes. EXAMPLE Using Queueing Theory, we can determine the Average number of customers waiting in line Average time customers wait in line The maximum expected number waiting in line The probability of zero customers in the system The probability of n customers in the system etc. SOME FAMOUS STOCHASTIC PROCESSES Bernoulli process Binomial process Branching process Markov chain Moran process Random walk Birth-death process Wiener or Brownian motion process Contact process SOME FAMOUS STOCHASTIC PROCESSES Diffusion process Interacting particle system Jump process Lévy process Ornstein-Uhlenbeck process Poisson process Gaussian process Martingale Renewal process