10. Thermal path integrals, ring polymers, and exchange
Position in the course: Lesson 10 of 10. Complete the preceding derivation and use the explained exercises to check understanding.
Prerequisites: Operator exponentials; Monte Carlo; fermion signs
Learning goal: explain the mathematical steps, reproduce the analytical examples, and state the conditions under which the conclusions hold.
1. Begin with a quantum partition function
Finite-temperature equilibrium uses \(Z=\operatorname{Tr}e^{-\beta H}\). It cannot generally be replaced by a classical integral of \(e^{-\beta V}\) because kinetic and potential operators do not commute. Divide inverse temperature into \(P\) slices, \(\epsilon=\beta/P\), and use a symmetric short-time factorization
under suitable operator conditions. Insert coordinate resolutions between slices, and take the trace by closing the path. Singular interactions may require better short-time actions than this primitive construction. Increasing bead count controls a discretization approximation, distinct from Monte Carlo sampling uncertainty.
2. Derive the spring action
For one distinguishable particle of mass \(m\), the free kinetic kernel is
Combining all kernels and potential factors gives a weight proportional to \(e^{-S_P}\) with
Beads form a closed imaginary-time ring, with harmonic coupling between neighbors. They are not simultaneously occupied classical particles and bead index is not physical time. An equivalent ring-polymer energy can be written with spring frequency \(P/(\beta\hbar)\) at inverse temperature \(\beta/P\); state that convention to avoid mixing coefficients.
3. Why ordinary bead moves can be slow
At large \(P\) the springs are stiff, so individual-bead displacements have poor scale separation. Staging or normal-mode coordinates separate high-frequency internal fluctuations from the centroid. Collective moves can improve efficiency without altering the target. Thermodynamic energy estimators may have large variance because kinetic energy comes from derivatives of normalization and spring factors; virial estimators exploit coordinate transformations to improve it. Their implementation must match constraints, dimensionality, and the action used.
4. Exchange and the sign return
For identical particles the trace includes permutations of end points. Bosonic terms add positively; fermionic permutations carry \((-1)^\mathcal P\). Ignoring exchange gives distinguishable-particle statistics, not an exact fermion simulation. Exchange cancellation leads to the sign problem; restricted paths and other approximations require stated constraints. A ring-polymer representation of quantum nuclei in a molecular system does not automatically solve the electronic fermion problem.
Equilibrium path integrals and ring-polymer dynamical approximations are different. Correct bead-converged thermal sampling does not imply that classical bead evolution gives exact real-time quantum correlations. Thermostats, fictitious masses, and propagation rules can change dynamical approximations while preserving equilibrium averages. Specify whether the desired quantity is a thermal expectation or a real-time observable before choosing the method.
5. Worked oscillator limits and exercises
For a quantum harmonic oscillator, \(Z=[2\sinh(\beta\hbar\omega/2)]^{-1}\) and \(E=(\hbar\omega/2)\coth(\beta\hbar\omega/2)\). At high temperature, \(\coth x\sim1/x\) gives \(E\sim k_BT\); at low temperature it approaches \(\hbar\omega/2\). These exact limits check thermal estimators and show zero-point energy surviving when classical energy would vanish.
Exercise: Does \(P=1\) generally give exact quantum equilibrium?
Solution
No. It collapses the internal ring representation and leaves a crude short-time approximation. Convergence in \(P\) must be demonstrated or an exact kernel supplied for a special solvable system.
Exercise: Does more sampling remove primitive-action error?
Solution
No. More samples estimate the chosen discretized action more precisely. Increase \(P\) or improve the action while separately controlling sampling and exchange approximations.
Original teaching schematic of the mathematics or algorithm; it is not simulation or experimental data.
6. Sources and connections
Related theory: Classical Monte Carlo · Molecular methods · Density functional theory
Software connection: CP2K · Quantum ESPRESSO · Gaussian
These software courses provide related background on energies, orbitals, or convergence management; they do not imply that the Monte Carlo or QMC examples on this page were executed there.
7. Related theory and practice
Quantum mechanics · Molecular methods · Monte Carlo · Molecular dynamics · QMCPACK