Nothing
R/get_mixed_accepts.R
In MEIGOR: MEIGO - MEtaheuristics for bIoinformatics Global Optimization
Defines functions
get_mixed_accepts
#' @title A helper function that returns the thinned,
#' accepted positions after a user-specified burn-in period; also returns
#' the indices (step numbers) of each of the returned positions.
#'
#' @description Use this method instead of ``prune`` if you want to get the
#' mixed accepted steps without discarding all of the other ones.
#'
#'
#' @param options: list with entries as explained below.
#' Options set -- defines the problem and sets some
#' parameters to control the MCMC algorithm.
#' @param model: List of model parameters - to estimate..
#' The parameter objects must each have a
#' 'value' attribute containing the parameter's numerical value.
#' @param estimate_params: list.
#' List of parameters to estimate, all of which must also be listed
#' in 'options$model$parameters'.
#' @param initail_values: list of float, optional.
#' Starting values for parameters to estimate. If omitted, will use
#' the nominal values from 'options$model$parameters'
#' @param step_fn: callable f(output), optional.
#' User callback, called on every MCMC iteration.
#' @param likelihood_fn: callable f(output, position).
#' User likelihood function.
#' @param prior_fn: callable f(output, position), optional.
#' User prior function. If omitted, a flat prior will be used.
#' @param nsteps: int.
#' Number of MCMC iterations to perform.
#' @param use_hessian: logical, optional.
#' Wheter to use the Hessian to giude the walk. Defaults to false.
#' @param rtol: float or list of float, optional.
#' Relative tolerance for ode solver.
#' @param atol: float or list of float, optional.
#' Absolute tolerance for ode solver.
#' @param norm_step_size: float, optional.
#' MCMC step size. Defaults to a reasonable value.
#' @param hessian_period: int, optional.
#' Number of MCMC steps between Hessian recalculations. Defaults
#' to a reasonable but fairly large value, as Hessian calculation is expensive.
#' @param hessian_scale: float, optional.
#' Scaling factor used in generating Hessian-guided steps. Defaults to a
#' reasonable value.
#' @param sigma_adj_interval: int, optional.
#' How often to adjust 'output$sig_value' while annealing to meet
#' 'accept_rate_target'. Defaults to a reasonable value.
#' @param anneal_length: int, optional.
#' Length of initial "burn-in" annealing period. Defaults to 10% of
#' 'nsteps', or if 'use_hessian' is true, to 'hessian_period' (i.e.
#' anneal until first hessian is calculated)
#' @param T_init: float, optional.
#' Initial temperature for annealing. Defaults to a resonable values.
#' @param accept_rate_target: float, optional.
#' Desired acceptance rate during annealing. Defaults to a reasonable value.
#' See also 'sigma_adj_interval' above.
#' @param sigma_max: float, optional.
#' Maximum value for 'output$sig_value'. Defaults to a resonable value.
#' @param sigma_min: float, optional.
#' Minimum value for 'output$sig_value'. Defaults to a resonable value.
#' @param sigma_step: float, optional.
#' Increment for 'output$sig_value' adjustments. Defaults to a resonable
#' value. To eliminate adaptive step size, set sigma_step to 1.
#' @param thermo_temp: float in the range [0,1], optional.
#' Temperature for thermodynamic integration support. Used to scale
#' likelihood when calculating the posterior value. Defaults to 1,
#' i.e. no effect.
#'
#'
#'
#' @param output: List of output values with entries as explained below..
#' @param num_estimate: int.
#' Number of parameters to estimate.
#' @param estimate_idx: list of int.
#' Indices of parameters to estimate in the model's full parameter list.
#' @param initial_values : list of float.
#' Starting values for parameters to estimate, taken from the parameters'
#' nominal values in the model or explicitly specified in `options`.
#' @param initial_position : list of float.
#' Starting position of the MCMC walk in parameter space
#' (log10 of 'initial_values')
#' @param position: list of float.
#' Current position of MCMC walk in parameter space, i.e. the most
#' recently accepted move.
#' @param test_position: list of float.
#' Proposed MCMC mmove.
#' @param acceptance: int.
#' Number of accepted moves.
#' @param T: float.
#' Current value of the simulated annealing temperature.
#' @param T_decay: float.
#' Constant for exponential decay 'T', automatically calculated such
#' that T will decay from 'options$T_init' down to 1 over the first
#' 'options$anneal_length' setps.
#' @param sig_value: float.
#' Current value of sigma, the scaling factor for the proposal distribution.
#' Te MCMC algorithm dynamically tunes this to maintain the aaceptance
#' rate specified in 'options$accept_rate_target'.
#' @param iter: int.
#' Current MCMC step number.
#' @param start_iter: int.
#' Starting MCMC step number.
#' @param ode_options: list (dict in python).
#' Options for the ODE integrator, currently just 'rtol' for relative
#' tolerance and 'atol' for absolute tolerance.
#' @param initial_prior: float.
#' Starting prior value, i.e. the value at 'initial_position'.
#' @param initial_likelihood: float.
#' Starting likelihood value, i.e. the values at 'initial_position'.
#' @param initial_posterior: float.
#' Starting posterior value, i.e. the values at 'initial_position'.
#' @param accept_prior: float.
#' Current prior value i.e. the value at 'position'.
#' @param accept_likelihood: float.
#' Current likelihood value i.e. the value at 'position'.
#' @param accept_posterior: float.
#' Current posterior value i.e. the value at 'position'.
#' @param test_prior: float.
#' Prior value at 'test_position'.
#' @param test_likelihood: float.
#' Likelihood value at 'test_position'.
#' @param test_posterior: float.
#' Posterior value at 'test_position'.
#' @param hessian: array of float.
#' Current hessian of the posterior landscape. Size is
#' 'num_estimate' x 'num_estimate'.
#' @param positions: array of float.
#' Trace of all proposed moves. Size is 'num_estimate' x 'nsteps'.
#' @param priors: array of float.
#' Trace of all priors corresponding to positions'. Length is 'nsteps'.
#' @param likelihoods: array of float.
#' Trace of all likelihoods corresponding to positions'. Length is 'nsteps'.
#' @param posteriors: array of float.
#' Trace of all posteriors corresponding to positions'. Length is 'nsteps'.
#' @param alphas: array of float. Trace of 'alpha' parameter and calculated values. Length is 'nsteps'.
#' @param sigmas: array of float. Trace of 'sigma' parameter and calculated values. Length is 'nsteps'.
#' @param delta_posteriors: array of float. Trace of 'delta_posterior' parameter and calculated values. Length is 'nsteps'.
#' @param ts: array of float. Trace of 'T' parameter and calculated values. Length is 'nsteps'.
#' @param accepts: logical array.
#' Trace of wheter each proposed move was accepted or not.
#' Length is 'nsteps'.
#' @param rejects: logical array. Trace of wheter each proposed move was rejected or not. Length is 'nsteps'.
#' @param hessians: array of float.
#' Trace of all hessians. Size is 'num_estimate' x 'num_estimate' x 'num_hessians'
#' where 'num_hessians' is the actual number of hessians to be calculated.
#'
#'
#'
#' @param burn: int.
#' An integer specifying the number of steps to cut off
#' from the beginning of the walk.
#' @param thin: int.
#' An integer specifying how to thin the accepted steps of the walk.
#' If 1, it returns every step; if 2, every other step; if 5, every fifth step, etc.
#'
#'
#'
#' @return Returns a tuple of a array objects. The first element in the tuple contains
#' the array of accepted positions, burned and thined as required; the second
#' element contains a list of integers which indicate the indeces associated
#' with each of the steps returned.
#'
get_mixed_accepts = function(output, options, burn, thin=1){
mixed_steps = c(burn:options$nsteps)
mixed_positions = output$positions[burn:options$nsteps,]
mixed_accepts = mixed_positions[output$accepts[burn:options$nsteps],]
mixed_accept_steps = mixed_steps[output$accepts[burn:options$nsteps]]
thinned_accepts = mixed_accepts[seq(from=1, to=dim(mixed_accepts)[1], by=thin),]
thinned_accept_steps = mixed_accept_steps[seq(from=1, to=dim(mixed_accepts)[1], by=thin)]
return(list("thinned_accepts"=thinned_accepts, "thinned_accept_steps" = thinned_accept_steps))
}
Try the MEIGOR package in your browser
Any scripts or data that you put into this service are public.
MEIGOR documentation built on Nov. 8, 2020, 7:46 p.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Defines functions get_mixed_accepts
#' @title A helper function that returns the thinned,
#' accepted positions after a user-specified burn-in period; also returns
#' the indices (step numbers) of each of the returned positions.
#'
#' @description Use this method instead of ``prune`` if you want to get the
#' mixed accepted steps without discarding all of the other ones.
#'
#'
#' @param options: list with entries as explained below.
#' Options set -- defines the problem and sets some
#' parameters to control the MCMC algorithm.
#' @param model: List of model parameters - to estimate..
#' The parameter objects must each have a
#' 'value' attribute containing the parameter's numerical value.
#' @param estimate_params: list.
#' List of parameters to estimate, all of which must also be listed
#' in 'options$model$parameters'.
#' @param initail_values: list of float, optional.
#' Starting values for parameters to estimate. If omitted, will use
#' the nominal values from 'options$model$parameters'
#' @param step_fn: callable f(output), optional.
#' User callback, called on every MCMC iteration.
#' @param likelihood_fn: callable f(output, position).
#' User likelihood function.
#' @param prior_fn: callable f(output, position), optional.
#' User prior function. If omitted, a flat prior will be used.
#' @param nsteps: int.
#' Number of MCMC iterations to perform.
#' @param use_hessian: logical, optional.
#' Wheter to use the Hessian to giude the walk. Defaults to false.
#' @param rtol: float or list of float, optional.
#' Relative tolerance for ode solver.
#' @param atol: float or list of float, optional.
#' Absolute tolerance for ode solver.
#' @param norm_step_size: float, optional.
#' MCMC step size. Defaults to a reasonable value.
#' @param hessian_period: int, optional.
#' Number of MCMC steps between Hessian recalculations. Defaults
#' to a reasonable but fairly large value, as Hessian calculation is expensive.
#' @param hessian_scale: float, optional.
#' Scaling factor used in generating Hessian-guided steps. Defaults to a
#' reasonable value.
#' @param sigma_adj_interval: int, optional.
#' How often to adjust 'output$sig_value' while annealing to meet
#' 'accept_rate_target'. Defaults to a reasonable value.
#' @param anneal_length: int, optional.
#' Length of initial "burn-in" annealing period. Defaults to 10% of
#' 'nsteps', or if 'use_hessian' is true, to 'hessian_period' (i.e.
#' anneal until first hessian is calculated)
#' @param T_init: float, optional.
#' Initial temperature for annealing. Defaults to a resonable values.
#' @param accept_rate_target: float, optional.
#' Desired acceptance rate during annealing. Defaults to a reasonable value.
#' See also 'sigma_adj_interval' above.
#' @param sigma_max: float, optional.
#' Maximum value for 'output$sig_value'. Defaults to a resonable value.
#' @param sigma_min: float, optional.
#' Minimum value for 'output$sig_value'. Defaults to a resonable value.
#' @param sigma_step: float, optional.
#' Increment for 'output$sig_value' adjustments. Defaults to a resonable
#' value. To eliminate adaptive step size, set sigma_step to 1.
#' @param thermo_temp: float in the range [0,1], optional.
#' Temperature for thermodynamic integration support. Used to scale
#' likelihood when calculating the posterior value. Defaults to 1,
#' i.e. no effect.
#'
#'
#'
#' @param output: List of output values with entries as explained below..
#' @param num_estimate: int.
#' Number of parameters to estimate.
#' @param estimate_idx: list of int.
#' Indices of parameters to estimate in the model's full parameter list.
#' @param initial_values : list of float.
#' Starting values for parameters to estimate, taken from the parameters'
#' nominal values in the model or explicitly specified in `options`.
#' @param initial_position : list of float.
#' Starting position of the MCMC walk in parameter space
#' (log10 of 'initial_values')
#' @param position: list of float.
#' Current position of MCMC walk in parameter space, i.e. the most
#' recently accepted move.
#' @param test_position: list of float.
#' Proposed MCMC mmove.
#' @param acceptance: int.
#' Number of accepted moves.
#' @param T: float.
#' Current value of the simulated annealing temperature.
#' @param T_decay: float.
#' Constant for exponential decay 'T', automatically calculated such
#' that T will decay from 'options$T_init' down to 1 over the first
#' 'options$anneal_length' setps.
#' @param sig_value: float.
#' Current value of sigma, the scaling factor for the proposal distribution.
#' Te MCMC algorithm dynamically tunes this to maintain the aaceptance
#' rate specified in 'options$accept_rate_target'.
#' @param iter: int.
#' Current MCMC step number.
#' @param start_iter: int.
#' Starting MCMC step number.
#' @param ode_options: list (dict in python).
#' Options for the ODE integrator, currently just 'rtol' for relative
#' tolerance and 'atol' for absolute tolerance.
#' @param initial_prior: float.
#' Starting prior value, i.e. the value at 'initial_position'.
#' @param initial_likelihood: float.
#' Starting likelihood value, i.e. the values at 'initial_position'.
#' @param initial_posterior: float.
#' Starting posterior value, i.e. the values at 'initial_position'.
#' @param accept_prior: float.
#' Current prior value i.e. the value at 'position'.
#' @param accept_likelihood: float.
#' Current likelihood value i.e. the value at 'position'.
#' @param accept_posterior: float.
#' Current posterior value i.e. the value at 'position'.
#' @param test_prior: float.
#' Prior value at 'test_position'.
#' @param test_likelihood: float.
#' Likelihood value at 'test_position'.
#' @param test_posterior: float.
#' Posterior value at 'test_position'.
#' @param hessian: array of float.
#' Current hessian of the posterior landscape. Size is
#' 'num_estimate' x 'num_estimate'.
#' @param positions: array of float.
#' Trace of all proposed moves. Size is 'num_estimate' x 'nsteps'.
#' @param priors: array of float.
#' Trace of all priors corresponding to positions'. Length is 'nsteps'.
#' @param likelihoods: array of float.
#' Trace of all likelihoods corresponding to positions'. Length is 'nsteps'.
#' @param posteriors: array of float.
#' Trace of all posteriors corresponding to positions'. Length is 'nsteps'.
#' @param alphas: array of float. Trace of 'alpha' parameter and calculated values. Length is 'nsteps'.
#' @param sigmas: array of float. Trace of 'sigma' parameter and calculated values. Length is 'nsteps'.
#' @param delta_posteriors: array of float. Trace of 'delta_posterior' parameter and calculated values. Length is 'nsteps'.
#' @param ts: array of float. Trace of 'T' parameter and calculated values. Length is 'nsteps'.
#' @param accepts: logical array.
#' Trace of wheter each proposed move was accepted or not.
#' Length is 'nsteps'.
#' @param rejects: logical array. Trace of wheter each proposed move was rejected or not. Length is 'nsteps'.
#' @param hessians: array of float.
#' Trace of all hessians. Size is 'num_estimate' x 'num_estimate' x 'num_hessians'
#' where 'num_hessians' is the actual number of hessians to be calculated.
#'
#'
#'
#' @param burn: int.
#' An integer specifying the number of steps to cut off
#' from the beginning of the walk.
#' @param thin: int.
#' An integer specifying how to thin the accepted steps of the walk.
#' If 1, it returns every step; if 2, every other step; if 5, every fifth step, etc.
#'
#'
#'
#' @return Returns a tuple of a array objects. The first element in the tuple contains
#' the array of accepted positions, burned and thined as required; the second
#' element contains a list of integers which indicate the indeces associated
#' with each of the steps returned.
#'
get_mixed_accepts = function(output, options, burn, thin=1){
mixed_steps = c(burn:options$nsteps)
mixed_positions = output$positions[burn:options$nsteps,]
mixed_accepts = mixed_positions[output$accepts[burn:options$nsteps],]
mixed_accept_steps = mixed_steps[output$accepts[burn:options$nsteps]]
thinned_accepts = mixed_accepts[seq(from=1, to=dim(mixed_accepts)[1], by=thin),]
thinned_accept_steps = mixed_accept_steps[seq(from=1, to=dim(mixed_accepts)[1], by=thin)]
return(list("thinned_accepts"=thinned_accepts, "thinned_accept_steps" = thinned_accept_steps))
}
Try the MEIGOR package in your browser
Any scripts or data that you put into this service are public.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.