Split Hamiltonian Monte Carlo revisited
comunitat-uji-handle:10234/9
comunitat-uji-handle2:10234/7037
comunitat-uji-handle3:10234/8635
comunitat-uji-handle4:
INVESTIGACIONMetadatos
Título
Split Hamiltonian Monte Carlo revisitedFecha de publicación
2022Editor
SpringerCita bibliográfica
Casas, F., Sanz-Serna, J.M. & Shaw, L. Split Hamiltonian Monte Carlo revisited. Stat Comput 32, 86 (2022). https://doi.org/10.1007/s11222-022-10149-4Tipo de documento
info:eu-repo/semantics/articleVersión
info:eu-repo/semantics/publishedVersionPalabras clave / Materias
Resumen
We study Hamiltonian Monte Carlo (HMC) samplers based on splitting the Hamiltonian H as H0(θ , p)+U1(θ ), where H0 is
quadratic and U1 small. We show that, in general, such samplers suffer from stepsize stability ... [+]
We study Hamiltonian Monte Carlo (HMC) samplers based on splitting the Hamiltonian H as H0(θ , p)+U1(θ ), where H0 is
quadratic and U1 small. We show that, in general, such samplers suffer from stepsize stability restrictions similar to those of
algorithms based on the standard leapfrog integrator. The restrictions may be circumvented by preconditioning the dynamics.
Numerical experiments show that, when the H0(θ , p) + U1(θ ) splitting is combined with preconditioning, it is possible to
construct samplers far more efficient than standard leapfrog HMC. [-]
Publicado en
Statistics and Computing (2022) 32:86Derechos de acceso
info:eu-repo/semantics/openAccess
Aparece en las colecciones
- MAT_Articles [765]