Papers

Simulation-efficient marginal posterior estimation with swyft stop wasting your precious time

BK Miller, A Cole, G Louppe, C Weniger - arXiv preprint arXiv:2011.13951, 2020 - arxiv.org
Astrophysics paper astro-ph.IM Suggest

… We operate in the context of simulation-based inference where our simulator g is a nonlinear function mapping a vector of parameters θ = (θ1,…,θd) ∈ Rd and a stochastic latent state z …

Cited by Link to paper

BibTeX

@article{2011.13951v1,
Author = {Benjamin Kurt Miller and Alex Cole and Gilles Louppe and Christoph Weniger},
Title = {Simulation-efficient marginal posterior estimation with swyft: stop
wasting your precious time},
Eprint = {2011.13951v1},
ArchivePrefix = {arXiv},
PrimaryClass = {astro-ph.IM},
Abstract = {We present algorithms (a) for nested neural likelihood-to-evidence ratio
estimation, and (b) for simulation reuse via an inhomogeneous Poisson point
process cache of parameters and corresponding simulations. Together, these
algorithms enable automatic and extremely simulator efficient estimation of
marginal and joint posteriors. The algorithms are applicable to a wide range of
physics and astronomy problems and typically offer an order of magnitude better
simulator efficiency than traditional likelihood-based sampling methods. Our
approach is an example of likelihood-free inference, thus it is also applicable
to simulators which do not offer a tractable likelihood function. Simulator
runs are never rejected and can be automatically reused in future analysis. As
functional prototype implementation we provide the open-source software package
swyft.},
Year = {2020},
Month = {Nov},
Url = {http://arxiv.org/abs/2011.13951v1},
File = {2011.13951v1.pdf}
}

Share