Hamiltonian Monte Carlo
Augments parameters with momentum variables and simulates Hamiltonian dynamics using the gradient of the log density, proposing distant states along energy-conserving trajectories that are accepted with high probability — suppressing random-walk behaviour and scaling far better with dimension.
This Concept is waiting for its first lesson!
Augments parameters with momentum variables and simulates Hamiltonian dynamics using the gradient of the log density, proposing distant states along energy-conserving trajectories that are accepted with high probability — suppressing random-walk behaviour and scaling far better with dimension.
Are you a teacher? Sign in to start contributing.
Sign In