scale_transform: Simple Preconditioning Transformation

In mpoll/scale: The Scaleable Langevin Exact Algorithm -- Reference Implementation

Description

Applies the ‘simple’ preconditioning transformation described within the ‘Implementation Details’ section of Pollock et al. (2016) to points in the original space (beta.pos), centred at the point at which the control variate is computed (beta.star). Outputs a vector giving the position within the transformed space (eta.pos).

Usage

scale_transform(beta.pos)

Arguments

beta.pos	Point within the original space to be transformed
beta.star	Point within the original space that the control variate is computed at
n.sigma.inv	Simple inverse preconditioning matrix

Value

eta.pos

Point within the transformed space

Note

As discussed within Pollock et al. (2016) the simple preconditioning procedure uses a diagonal matrix. Modification to this function would be required (along with associated package functions) if the full preconditioning matrix were to be used (precon / precon.inv).

Author(s)

Murray Pollock, Paul Fearnhead, Adam Johansen and Gareth Roberts

References

Pollock, Murray, Paul Fearnhead, Adam M. Johansen, and Gareth O. Roberts. "The scalable Langevin exact algorithm: Bayesian inference for big data." arXiv preprint arXiv:1609.03436 (2016).

See Also

un_scale_transform

mpoll/scale documentation built on Dec. 9, 2019, 7:15 a.m.