On the convergence of dynamic implementations of Hamiltonian Monte Carlo and No U-Turn Samplers
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2023
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866910654870323200 |
|---|---|
| author | Durmus, Alain Gruffaz, Samuel Kailas, Miika Saksman, Eero Vihola, Matti |
| author_facet | Durmus, Alain Gruffaz, Samuel Kailas, Miika Saksman, Eero Vihola, Matti |
| contents | There is substantial empirical evidence about the success of dynamic implementations of Hamiltonian Monte Carlo (HMC), such as the No U-Turn Sampler (NUTS), in many challenging inference problems but theoretical results about their behavior are scarce. The aim of this paper is to fill this gap. More precisely, we consider a general class of MCMC algorithms we call dynamic HMC. We show that this general framework encompasses NUTS as a particular case, implying the invariance of the target distribution as a by-product. Second, we establish conditions under which NUTS is irreducible and aperiodic and as a corrolary ergodic. Under conditions similar to the ones existing for HMC, we also show that NUTS is geometrically ergodic. Finally, we improve existing convergence results for HMC showing that this method is ergodic without any boundedness condition on the stepsize and the number of leapfrog steps, in the case where the target is a perturbation of a Gaussian distribution. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2307_03460 |
| institution | arXiv |
| publishDate | 2023 |
| record_format | arxiv |
| spellingShingle | On the convergence of dynamic implementations of Hamiltonian Monte Carlo and No U-Turn Samplers Durmus, Alain Gruffaz, Samuel Kailas, Miika Saksman, Eero Vihola, Matti Computation Probability Statistics Theory Machine Learning 62 There is substantial empirical evidence about the success of dynamic implementations of Hamiltonian Monte Carlo (HMC), such as the No U-Turn Sampler (NUTS), in many challenging inference problems but theoretical results about their behavior are scarce. The aim of this paper is to fill this gap. More precisely, we consider a general class of MCMC algorithms we call dynamic HMC. We show that this general framework encompasses NUTS as a particular case, implying the invariance of the target distribution as a by-product. Second, we establish conditions under which NUTS is irreducible and aperiodic and as a corrolary ergodic. Under conditions similar to the ones existing for HMC, we also show that NUTS is geometrically ergodic. Finally, we improve existing convergence results for HMC showing that this method is ergodic without any boundedness condition on the stepsize and the number of leapfrog steps, in the case where the target is a perturbation of a Gaussian distribution. |
| title | On the convergence of dynamic implementations of Hamiltonian Monte Carlo and No U-Turn Samplers |
| topic | Computation Probability Statistics Theory Machine Learning 62 |
| url | https://arxiv.org/abs/2307.03460 |