AdvancedHMC.jl
AdvancedHMC.jl copied to clipboard
TuringLang
Robust, modular and efficient implementation of advanced Hamiltonian Monte Carlo algorithms
It would be nice if the adapted mass matrix could be retrieved. @sethaxen pointed out that giving `save_state=true` to `sample` is possible, but that it currently only gives the metric...
The following paper presented a novel approach for adapting the metric: Ben Bales, Arya Pourzanjani, Aki Vehtari, Linda Petzold, Selecting the Metric in Hamiltonian Monte Carlo, 2019. https://arxiv.org/abs/1905.11916 The idea...
We need 4 things from a Euclidean metric: - a size - left-multiplication by the metric `M⁻¹` - a quadratic form `dot(v, M⁻¹, v)` - left-division by `W`, such that...
@stanniszhou gave a talk by this title at PROBPROG 2020. Could AdvancedHMC be adapted to do this? HMC for discrete variables could be a game changer :) https://stanniszhou.github.io/papers/mixed_hmc.pdf
Came across http://proceedings.mlr.press/v108/zhang20e.html which seems to guarantee asymptotic convergence with the goodness of efficiency with SGHMC. Adding here to see if there is interest from users or plans from the...
Pathfinder is a variational method for initializing MCMC methods, implemented at https://github.com/sethaxen/Pathfinder.jl. When it works well, it returns draws from the typical set so that the warm-up period can be...
Not sure if this has been reported. Setting `n_samples = 1` in [the minimal example](https://github.com/TuringLang/AdvancedHMC.jl#a-minimal-example---sampling-from-a-multivariate-gaussian-using-nuts) stops the sampler from working properly. ```julia using AdvancedHMC, Distributions, ForwardDiff using LinearAlgebra function debug()...
Add the Explicit RMHMC algorithm from @article{cobb2019introducing, title={Introducing an Explicit Symplectic Integration Scheme for Riemannian Manifold Hamiltonian Monte Carlo}, author={Cobb, Adam D and Baydin, At{\i}l{\i}m G{\"u}ne{\c{s}} and Markham, Andrew and...
Dropping this paper on autocorrelation-reducing techniques for HMC by sampling with anticorrelated proposals here: http://proceedings.mlr.press/v108/piponi20a/piponi20a.pdf Closely related to the QMC suggestion in another issue earlier, but more likely to yield...
The current implementation of the AbstractMCMC.jl-interface uses the https://github.com/TuringLang/AdvancedHMC.jl/blob/7cad9f02e6eb4095ad8b2e7ef06d3474dfb19135/src/abstractmcmc.jl#L47-L52 as the `AbstractMCMC.AbstractModel` type for this package. But such a model is useful elsewhere too, hence it's unfortunate that we tie...
Metadata
Owner
Metadata
Robust, modular and efficient implementation of advanced Hamiltonian Monte Carlo algorithms