Stochastic method for estimating expectation values and thermal properties in quantum systems.Monte Carlo sampling is a computational method that uses random numbers to estimate the properties of systems with many interacting particles, especially in quantum physics where direct analytical solutions are impossible. It is particularly powerful for calculating expectation values of observables, such as energy or magnetization, at finite temperatures, and for simulating the behavior of systems in thermal equilibrium. The name comes from the Monte Carlo Casino in Monaco, reflecting the method's reliance on randomness.
The method was developed in the 1940s by Stanislaw Ulam, John von Neumann, and Nicholas Metropolis, initially for nuclear weapons research at Los Alamos. It was later adapted for quantum systems by D. M. Ceperley and others in the 1970s and 1980s.
The method generates a large number of random configurations of the system, weighted according to the probability distribution of interest (e.g., the Boltzmann distribution for thermal equilibrium). This is often done using a Markov chain algorithm like the Metropolis-Hastings algorithm, which generates a sequence of configurations where each new configuration is accepted or rejected based on the energy difference. By averaging the observable over many such configurations, the expectation value is approximated. The accuracy improves with the number of samples, but the computational cost can be high.
It is used to simulate quantum fluids like superfluid helium, to study the properties of strongly correlated electron systems (e.g., high-temperature superconductors), and to compute the equation of state of neutron stars.
Monte Carlo methods are also used in finance to price complex derivatives, and in computer graphics to render realistic lighting effects.