Professional Documents
Culture Documents
Introduction
In Classical Mechanics, motion is completely described by the least action
principle, is the Lagrangian,
q and are the generalized coordinates and velocity, respectively. In the
following derivation we shall see that starting with a probabilistic construction
of the time dependent state/wavefunction, we can actually recover the action
S of a path as its probability amplitude by insertion of the identity operator at
different times and application of the Trotter Product formula. We then show
that path integrals can be applied to solve for thermodynamic quantities and
the partition function by viewing temperature as imaginary time. At last the
Monte Carlo method of approximating the time-dependent path integral is
illustrated by two examples.
where
is the unitary propagator.
Hence
=
So far the theory is exact and the points integrated along the path does not
come into play yet. We will see that those N-1 points were taken to
approximate U(t, t0) as a product of U(ti+1, ti) .
Therefore =
which the Trotter formula tells us that each segment of the path has action
equivalent to the straight line action from to which is the result of
approximation when step size is small.
Although in some cases (Free particle and Harmonic Oscillator), the integral
can be exact, the infinite dimensional integral is impossible to evaluate
analytically for general potentials.
Free Particle
For free particles, the integral becomes one dimensional as potential term is 0.
For
Harmonic Oscillator
For the path that gives the extremum value of action from
to is
By expanding all paths around this extremal path, the path can be written
as sum of the extremal path and a perturbation , where
Thus
So that the hard evaluation of , the PDF of the path is not needed.
But for the time-dependent path integrals, we have no way to cancel the
probability density of any paths inside the integral. So we have to evaluate the
wavefunctions for each step, which is much less efficient than the former
method of direct sampling of entire paths.
The following are two demonstrations of the Monte Carlo Integration method
to calculate evolution of time-dependent wavefunctions.
The first one is a calculation for the free particle wavepacket in one step dt,
sampled with uniform distribution:
>
0.4015427855 (3.1)
>
where the blue asterisk is the one-step approximation of the red analytical
solution using Monte Carlo Integration. You can try to get more accurate
results by changing 6000 to a larger value, and see if the error increases at
different values of .
The former example showcases the one-step path integral to approximate
evolution of wavepackets in zero potential. For nonzero potentials we need to
break dt into multiple time steps to have accurate results.
The following part initializes samples of given normal distribution in a
Harmonic Oscillator potential with 3 steps:
>
>
>
>
The red curve is the initial wavepacket and the blue asterisks are the -step
approximations on the evolved wavepacket at T=1.
Another example is the minimal basis ground state electron wavefunction for
the Hydrogen atom, which is expected to remain the same no matter the
amount of time passed. This example could potentially serve to test the
accuracy of the minimal basis set in approximating the real ground state:
>
>
With the Quantum Chemistry Package we can show the density plot for the
minimal basis Hartree Fock solution of the H2 molecule:
>
The number of steps and the number of samples at each time step is very
small, so the result would not be accurate. But if you have a large enough
memory, try to modify the code above to include more steps and increase the
sample sizes. Have fun!
References
1. R. P. Feynman, "Space-Time Approach to Non-Relativistic Quantum
Mechanics," Rev. Mod. Phys 20(2), 367-387 (1948).
2. M. Nakahara. Geometry, Topology and Physics (Taylor & Francis Group,
Abingdon, 2003).
3. I. Ruokosenm aki, and T. T. Rantala, "Numerical path integral approach to
quantum dynamics and stationary quantum states," Commun. Comput. Phys.
18, 01 (2015).
4. M. J. E. Westbroek, and P. R.
for evaluating path integrals,"American Journal of Physics 86, 293 (2018).