Download Monte Carlo for Partial Differential Equations

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts
Transcript
Green’s Function Monte Carlo
Fall 2012
By Yaohang Li, Ph.D.
Review
• Last Class
– Solution of Linear Operator Equations
• Ulam-von Neumann Algorithm
• Adjoin Method
• Fredholm integral equation
• Dirichlet Problem
• Eigenvalue Problem
• This Class
– PDE
• Green’s Function
• Next Class
– Random Number Generation
Green’s Function (I)
• Consider a PDE written in a general form
– L(x)u(x)=f(x)
• L(x) is a linear differential operator
• u(x) is unknown
• f(x) is a known function
– The solution can be written as
• u(x)=L-1(x)f(x)
• L-1L=I
Green’s Function
•The inverse operator
L1 f   G ( x; x' ) f ( x' )dx'
– G(x; x’) is the Green’s Function
– kernel of the integral
– two-point function depends on x and x’
•Property of the Green’s Function
•Solution to the PDE
Dirac Delta Function
Green’s Function in Monte Carlo
• Green’s Function
– G(x;x’) is a complex expression depending on
• the number of dimensions in the problem
• the distance between x and x’
• the boundary condition
– G(x;x’) is interpreted as a probability of “walking” from x’ to x
• Each walker at x’ takes a step sampled from G(x;x’)
Green’s Function for Laplacian
• Laplacian
• Green’s Function
– where
Solution to Laplace Equation using
Green’s Function Monte Carlo
•Random Walk on a Mesh
– G is the Green’s Function
• The number of times that a walker from the point (x,y) lands at the
boundary (xb,yb)
u ( x, y) 
1
G( x, y, xb , yb )u ( xb , yb )

n b
Poisson’s Equation
•Poisson’s Equation
– u(r)=-4(r)
•Approximation
1
1
u ( x, y )  [u ( x  x, y )  u ( x  x, y )  u ( x, y  y )  u ( x, y  y ))  xy 4( x, y )
4
4
•Random Walk Method
E (u ( x, y )) 
1
1
f
(
x
,
y
)

 ( xi , , yi , )
   n xy 
n 
i ,
– n: walkers
– i: the points visited by the walker
– The second term is the estimation of the path integral
Summary
• Green’s Function
• Laplace’s Equation
• Poisson’s Equation
What I want you to do?
• Review Slides
• Review basic probability/statistics concepts
• Work on your Assignment 4