33
Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford University Based on joint work with Jose Blanchet, Henry Lam, Denis Saure, and Assaf Zeevi Presented at Stochastic Networks Conference, Cambridge, UK March 23, 2010

Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford

Embed Size (px)

Citation preview

Page 1: Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford

Numerical Methods for Stochastic Networks

Peter W. GlynnInstitute for Computational and Mathematical Engineering

Management Science and Engineering Stanford University

Based on joint work with Jose Blanchet, Henry Lam, Denis Saure, and Assaf Zeevi

Presented at Stochastic Networks Conference, Cambridge, UKMarch 23, 2010

Page 2: Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford

Stochastic models:

Descriptive

Prescriptive

Predictive

Page 3: Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford

Today’s Talk:

• an LP based algorithm for computing the stationary distribution of RBM (Saure / Zeevi)

• a Lyapunov bound for stationary expectations (Zeevi)

• Rare-event simulation for many-server queues (Blanchet / Lam)

Page 4: Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford

Computing Steady-State Distributions for Markov Chains

Page 5: Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford

One Approach

Page 6: Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford
Page 7: Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford

An LP Alternative

Page 8: Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford

( ) ( )n

nx K

x h x c

( ) ("tightness")inf ( )c c

n n

nx K x K

cx

h x

Page 9: Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford

• Where does constraint

come from?

• We assume that we can obtain a computable bound on

i.e.

• This will come from Lyapunov bounds (later).

( ) ( )n

nx K

x h x c

( )h XE( )h X c E

Page 10: Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford

Application to RBM

Reflected Brownian Motion (RBM):

For some stochastic models, the LP algorithm is particularly natural and powerful

Page 11: Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford

dR

( :1 , )ijb i j d

Page 12: Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford

d ( ) d d ( ) d ( )X t t B t R Y t

jY 0jX

2

1 , 1

1, =

2

d d

i ij j iji i j ii i j i

b Rx x x x

DL

Page 13: Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford
Page 14: Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford

1int ( )

int ( )

min

s/t ( )( ) ( )( ) u

1, 0, 0

| |d

k jk

dk

d

k i k i i k jkj x Fx

k k jkx

u

p f x f x

p p

R

R

L D

Page 15: Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford
Page 16: Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford

(d )( )( )d

x f x c

R L

(d )( )( ).j

j jFx f x D

j

Page 17: Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford

Theorem: This algorithm converges as , in the sense that

as .

,m n

n

n

j

n

Page 18: Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford

Numerical Results

Smoothed marginal distribution estimates for the two-dimensional diffusion. The dotted line is computed via Monte Carlo simulation, and the solid line represents the algorithm estimates based on n = 50 and m = 4, incorporating smoothness constraints.

Page 19: Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford

• - valued Markov process with cadlag paths

• We say if there exists such that

is a –local martingale for each

( ( ) : 0)X X t t S

( )g AD k

0( ) ( ( )) ( ( ))d

tM t g X t k X s s

xP x S

0( ( )) ( ) ( ( ))d

h

x xg X h g x k X s s E E

Page 20: Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford

Computing Bounds on Stationary Expectations

• Main Theorem:

Suppose is non–negative and satisfies

Then,

( )g AD

sup( )( ) .x S

Ag x

(d )( )( ) 0.S

x Ag x

Page 21: Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford

Diffusion Upper Bound

• Suppose is and satisfies for , where and

where . Then,

d ( ) ( ( ))d ( ( ))d ( )X t X t t X t B t

0g 2C ( )( ) ( )g x h x c Ldx R 0h

2

1 ,

1( ) ( )

2

d

i iji i ji i j

x b xx x x

L

T( ) ( ) ( )b x x x

.h c

Page 22: Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford

Many-Server Loss Systems

Page 23: Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford

Many-Server Asymptotic Regime

Page 24: Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford

Simplify our model (temporarily):

• using slotted time• eliminate Markov modulation• discrete service time distributions with finite

support

Consider equilibrium fraction of customers lost in the network.

Page 25: Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford

Key Idea:

Many server loss systems behave identically

to infinite-server systems up to the time of

the first loss.

Page 26: Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford

Step 1:

Page 27: Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford

Step 2:

Page 28: Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford

Step 3:

Page 29: Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford

Step 3:

Page 30: Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford

Step 3:

Page 31: Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford

Crude Monte Carlo : 3.7 days

I.S. : A few seconds

Page 32: Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford

Network Extension

• Estimate loss at a particular station

• If the most likely path to overflow a given station does not involve upstream stations, previous algorithm if efficient

• If an upstream station does hit its capacity constraint, we have “constrained Poisson statistics” that need to be sampled

Page 33: Numerical Methods for Stochastic Networks Peter W. Glynn Institute for Computational and Mathematical Engineering Management Science and Engineering Stanford

Questions?