Upload
zarifa
View
70
Download
4
Embed Size (px)
DESCRIPTION
The basics of quantum Monte Carlo. Lucas K. Wagner Computational Nanosciences Group University of California, Berkeley In collaboration with Jeffrey C. Grossman. Introduction. Introduction. The quantum many-body problem. Fundamental object in quantum mechanics: - PowerPoint PPT Presentation
Citation preview
WagnerNCN
nanoHUB.org
The basics of quantum Monte Carlo
Lucas K. WagnerComputational Nanosciences Group
University of California, Berkeley
In collaboration with Jeffrey C. Grossman
nanoHub.org
Wagner NCN
Introduction
Introduction
nanoHub.org
Wagner NCN
The quantum many-body problem
• Fundamental object in quantum mechanics:
• Want to find special wave functions such that
where
• This is a fundamentally many-body equation!
• We want to find an accurate method that will work in general to solve this
(r1,r2,...,rN )
1 2 1 2( , ,..., ) ( , ,..., )n N n NH r r r E r r r
21 1
2I
ii iI i jiI ij
ZH
r r
nanoHub.org
Wagner NCN
Fluctuation minimization
• Eigenvalue problem
•
• Non-eigenfunction andeigenfunction for simple harmonic osc.
• One way to approximatethe eigenfunction: minimizedeviation from a constant
( )
( )n
nn
H xE
x
( )( )
( )
H xE x
x
nanoHub.org
Wagner NCN
Monte Carlo
• Monte Carlo is solving a problem using random numbers!
• Evaluate integrals: – expectation value of a random variable is just the integral over its
probability distribution– generate a bunch of random numbers and average to get the
integral
• Simulate random processes--random walks sample configuration space
• Number of dimensions doesn’t matter
nanoHub.org
Wagner NCN
Quantum Monte Carlo
• Comes in many flavors
• Deals with the many body wave function directlywork with R=[r1,r2,…,rn]
• We’ll cover two main flavors: variational Monte Carlo and diffusion Monte Carlo
nanoHub.org
Wagner NCN
Results
• On 55 molecules, mean absolute deviation of atomization energy is 2.9 kcal/mol
• Successfully applied to organic molecules, transition metal oxides, solid state silicon, systems up to ~1000 electrons
• Can calculate accurate atomization energies, phase energy differences, excitation energies, one particle densities, correlation functions, etc..
• Scaling is from O(1) to O(N3), depending on implementation and quantity.
nanoHub.org
Wagner NCN
VMC
Variational Monte Carlo
nanoHub.org
Wagner NCN
Variational Monte Carlo
• Rewrite expectationvalue
2
2
( , ) ( , )( )
( , )( , )
R P H R PE P dR
R PR P dR
Probability distribution function
• Idea: generate random walkers with probability equal to the above pdf and average (Metropolis method)
• Minimize the variance of the local energy with respect to the parameters
nanoHub.org
Wagner NCN
Trial wave functions in the tool
• General form of wave function
– Slater determinant (Hartree-Fock)
– Two-body Jastrow
– Three-body Jastrow
• We optimize only the c coefficients
T (R) Det[ i(rj )]exp(U)
U 0
U ckeiak (riI )
k
iI
ckeebk (rij )
k
ij
U two body cklmeei [ak (riI )
klm
ijI
al (rjI ) ak (rjI )al (riI )]bk (rij )
nanoHub.org
Wagner NCN
DMC
Diffusion Monte Carlo
nanoHub.org
Wagner NCN
• General strategy: stochastically simulate a differential equation that converges to the eigenstate
• Equation:
• Must propagate an entire function forward in time <=> distribution of walkers
Diffusion Monte Carlo
( , )( ) ( , )
d R tH E R t
dt
nanoHub.org
Wagner NCN
Diffusion Monte Carlo: imaginary time dynamics
• We want to find a wave function so
• Our differential equation
• Suppose that
• decreases
• Kinetic energy (curvature)decreases, potential energystays the same
• Time derivative is zero when
t=0
t=infinity
( , )( ) ( , )
d R tH E R t
dt
H E
H E
1
22 V (R)
E
nanoHub.org
Wagner NCN
Diffusion Monte Carlo: Harmonic Oscillator
d(R,t)
dt
1
22(R, t) (V (R) E)(R, t)
Diffusion Birth/death
•Generate walkers with a guess distribution
•Each time step:•Take a random step (diffuse)•A walker can either die, give birth, or just keep going
•Keep following rules, and we find the groundstate!
•Works in an arbitrary number of dimensions
t
Initial
Final
nanoHub.org
Wagner NCN
Diffusion Monte Carlo: importance of a good trial
function
• Importance sampling: multiply the differential equation by a trial wave function– Converges to instead of– The better the trial function, the faster DMC is-- feed it
a wave function from VMC
• Fixed node approximation: for fermions, ground state has negative and positive parts– Not a pdf--can’t sample it– Approximation:
T0
0
T0 0
nanoHub.org
Wagner NCN
Typical calculation
• Choose system and get one-particle orbitals (we’ve already prepared the orbitals for you)
• Optimize wave function using VMC, evaluate energy and properties of wave function
• Use optimized wave function in DMC for most accurate, lowest energy calculations
• Check the tooltips for explanations of the few settings
nanoHub.org
Wagner NCN
Interesting things to look at
• The pair correlation function for the Slater determinant and optimized two-body wave functions, also compare with DMC
• The relative energies of Slater determinant/two-body wave function/DMC.
• The fluctuations in the trace of the Slater determinant versus two-body wave function.
nanoHub.org
Wagner NCN
References
• Hammond, Lester, and Reynolds. Monte Carlo Methods in Quantum Chemistry (book)
• Wikipedia, www.qwalk.org, www.qmcwiki.org
• Foulkes, Mitas, Needs, and Rajagopal. Rev. Mod. Phys. 73, 33 (2001)
• Umrigar, Nightingale, and Runge. J. Chem. Phys. 99, 2865 (1993)