Monte Carlo Simulation Methods
Download
Report
Transcript Monte Carlo Simulation Methods
Monte Carlo Simulation Methods
QNVT
QNVT
1
N!h3N
H(p N ,r N )
dp dr exp kT
N
N
N
H(p ,r )
N
N
i 1
pi
2
V(r )
N
2m
2
N
1
p
V (r N )
N
dr exp
3N dp exp
N!h
kT
2mkT
2
3N
p
N
dp exp 2mkT 2mkT 2
3N
2
QNVT
1 2mkT
2
N!
h
QNVT
V 2mkT
N! h 2
N
N
V(r
)
N
dr
exp
kT
3N
2
- ideal gas
QNVT
ZNVT
V(r N )
dr exp
kT
N
N
V
N!3N
h
2mkT
2
1/ 2
QNVT Q
ideal
NVT
excess
NVT
ex ce ss
NVT
.Q
A kT ln QNVT
Q
A A
ideal
1
V(r N )
N
N dr exp
V
kT
A
ex ce ss
Calculating properties by integration
V (r ) dr V (r ) (r )
N
N
N
N
(r )
N
N
V (r ) / kT
e
Z
N tri al
V (r
i
V (r )
N
V (r N )
Z dr exp
kT
N
N
)expVi (r N )/ kT
i1
N tri al
N
exp
V
(r
) / kT
i
i 1
Theoretical background to Metropolis
Markov chain of events:
- the outcome of each trial depends only on the preceding trial
- each trial belongs to a finite set of possible outcomes
mn - probability of moving from state m to n
=(1, 2,…. m, n,…N) - probability that the system is in a particular
state
(2)= (1). (3)= (2). = (1). .
limit=limN (1) N - limiting (equilibrium) distribution
mn - probability to choose the two states m,n between which the move
is to be made (stochastic matrix).
mn = mn. pmn - where p is the probability to accept the move
mn = mn if n > m
mn = mn. (n/m) if n < m and if n=m
In practice if the energy of the n state is lower the move is accepted, if
not a random number between 0 and 1 is compared to the Boltzmann
factor exp(-∆V(rN)/kT). If the Boltzmann factor is greater then the
Random number the move is accepted.
Implementation
xne w x ol d (2 1)rmax
yne w yol d (2 1)rmax
zne w zol d (2 1)rmax
rand(0,1)≤ exp(-∆V(rN)/kT)
Random number generators
Linear congruential method
[1] seed
[i] mod{([i 1]* a b),m}