mbakenhus / rumba_sampler

R implementation of the RUMBA algorithm

rumba_sampler

This is the R implementation of the RUMBA algorithm prsented in the paper ``Sampling lattice points in a polytope: a Bayesian biased algorithm with random updates" by Miles Bakenhus and Sonja Petrović. The folder "output" contains data and figures contained in the Simulations section of the paper.

Function implementaiton of RUMBA in samplers.R:

movingBayes(init_x, basis,
            init_alpha_pos,init_alpha_neg,
            init_beta_pos, init_beta_neg,
            n_moves, n_itr, 
            min_per_itr, max_per_itr,
            verbose=TRUE,
            sample_frequency=FALSE,
            move_method='runif',
            track_inits=FALSE,
            track_moves=FALSE,
            upper_bds=NULL, 
            weightNew=0,
            epsilon = 0, writeToFile=NULL)

init_x: (numeric matrix) row vector representing the initial feasible point

basis: (numeric matrix) matrix with rows that span the lattice

init_alpha_pos: (numeric vector) shape params for positive prior

init_alpha_neg: (numeric vector) shape params for negative prior

init_beta_pos: (numeric vector) rate params for positive prior

init_beta_neg: (numeric vector) rate params for negative prior

n_moves: (numeric value) indicates the number of moves

min_per_itr: (numeric value) minimum number of samples per iteration

max_per_itr: (numeric value) maximum number of samples per iteration

verbose: (logical value) where indicating verbose output

sample_frequency NOT IMPLEMENTED: (logical value) return the sample frequency

move_method: (character value in c("runif")) denotes the method for deciding the next initial point

track_initsNOT IMPLEMENTED: (logical value) return the initial points

track_movesNOT IMPLEMENTED: (logical value) return the list of moves

upper_bds: (numeric vector) upper bounds for fiber element values

weightNew: (numeric value) weight between 0 and 1 for sampling next move from new points

epsilonNOT IMPLEMENTED: (numeric value) Rate by which to increase the variance