BchronDensity {Bchron} | R Documentation |
Non-parametric phase model
Description
This function runs a non-parametric phase model on 14C and non-14C ages via Gaussian Mixture density estimation
Usage
BchronDensity(
ages,
ageSds,
calCurves,
pathToCalCurves = system.file("data", package = "Bchron"),
dfs = rep(100, length(ages)),
numMix = 50,
iterations = 10000,
burn = 2000,
thin = 8,
updateAges = FALSE,
store_density = TRUE
)
Arguments
ages |
A vector of ages (most likely 14C) |
ageSds |
A vector of 1-sigma values for the ages given above |
calCurves |
A vector of values containing either |
pathToCalCurves |
File path to where the calibration curves are located. Defaults to the system directory where the 3 standard calibration curves are stored |
dfs |
Degrees-of-freedom values for the t-distribution associated with the calibration calculation. A large value indicates Gaussian distributions assumed for the 14C ages |
numMix |
The number of mixture components in the phase model. Might need to be increased if the data set is large and the phase behaviour is very complex |
iterations |
The number of iterations to run for |
burn |
The number of starting iterations to discard |
thin |
The step size of iterations to keep |
updateAges |
Whether or not to update ages as part of the MCMC run. Default is FALSE. Changing this to TRUE will improve performance but will fit a slightly invalid model |
store_density |
Whether or not to store the density and age grid. Useful for plotting the output in other packages |
Details
This model places a Gaussian mixture prior distribution on the calibrated ages and so estimates the density of the overall set of radiocarbon ages. It is designed to be a probabilistic version of the Oxcal SUM command which takes calibrated ages and sums the probability distributions with the aim of estimating activity through age as a proxy.
Value
An object of class BchronDensityRun
with the following elements:
- theta
The posterior samples of the restricted ages
- p
Posterior samples of the mixture proportions
- mu
Values of the means of each Gaussian mixture
- calAges
The calibrated ages from
BchronCalibrate
- G
The number of mixture components. Equal to numMix
- age_grid
A grid of ages used for the final density estimate
- density
The density estimate based on the above age grid
See Also
Bchronology
, BchronRSL
, BchronDensityFast
for a faster approximate version of this function
Examples
# Read in some data from Sluggan Moss
data(Sluggan)
# Run the model
SlugDens <- with(
Sluggan,
BchronDensity(
ages = ages,
ageSds = ageSds,
calCurves = calCurves
)
)
# plot it
plot(SlugDens)