emulator_plot: Plot Emulator Outputs

In hmer: History Matching and Emulation Package

Plot Emulator Outputs

Description

A function for plotting emulator expectations, variances, and implausibilities

Usage

emulator_plot(
  ems,
  plot_type = "exp",
  ppd = 30,
  targets = NULL,
  cb = FALSE,
  params = NULL,
  fixed_vals = NULL,
  nth = 1,
  imp_breaks = NULL,
  include_legend = TRUE
)

Arguments

ems	An Emulator object, or a list thereof.
plot_type	The statistic to plot (see description or examples).
ppd	The number of points per plotting dimension
targets	If required, the targets from which to calculate implausibility
cb	A boolean representing whether a colourblind-friendly plot is produced.
params	Which two input parameters should be plotted?
fixed_vals	For fixed input parameters, the values they are held at.
nth	If plotting nth maximum implausibility, which level maximum to plot.
imp_breaks	If plotting nth maximum implausibility, defines the levels at which to draw contours.
include_legend	For multiple plots, should a combined legend be appended?

Details

Given a single emulator, or a set of emulators, the emulator statistics can be plotted across a two-dimensional slice of the parameter space. Which statistic is plotted is determined by plot_type: options are ‘exp’, ‘var’, ‘sd’, ‘imp’, and ‘nimp’, which correspond to expectation, variance, standard deviation, implausibility, and nth-max implausibility.

By default, the slice varies in the first two parameters of the emulators, and all other parameters are taken to be fixed at their mid-range values. This behaviour can be changed with the params and fixed_vals parameters (see examples).

If the statistic is ‘exp’, ‘var’ or ‘sd’, then the minimal set of parameters to pass to this function are ems (which can be a list of emulators or a single one) and plot_type. If the statistic is ‘imp’ or ‘nimp’, then the targets must be supplied - it is not necessary to specify the individual target for a single emulator plot. If the statistic is ‘nimp’, then the level of maximum implausibility can be chosen with the parameter nth.

Implausibility plots are typically coloured from green (low implausibility) to red (high implausibility): a colourblind-friendly option is available and can be turned on by setting cb = TRUE.

The granularity of the plot is controlled by the ppd parameter, determining the number of points per dimension in the grid. For higher detail, at the expense of longer computing time, increase this value. The default is 30.

Value

A ggplot object, or collection thereof.

See Also

Other visualisation tools: behaviour_plot(), diagnostic_wrap(), effect_strength(), hit_by_wave(), output_plot(), plot_actives(), plot_lattice(), plot_wrap(), simulator_plot(), space_removed(), validation_pairs(), wave_dependencies(), wave_points(), wave_values()

Examples

 # Reducing ppd to 10 for speed.
 emulator_plot(SIREmulators$ems, ppd = 10)
 emulator_plot(SIREmulators$ems$nS, ppd = 10)
 emulator_plot(SIREmulators$ems, plot_type = 'var', ppd = 10, params = c('aIR', 'aSR'))
  # Excessive runtime
    emulator_plot(SIREmulators$ems, plot_type = 'imp', ppd = 10,
     targets = SIREmulators$targets,
     fixed_vals = list(aSR = 0.02))
    emulator_plot(SIREmulators$ems, plot_type = 'nimp', cb = TRUE,
     targets = SIREmulators$targets, nth = 2, ppd = 10)
 

hmer documentation built on June 22, 2024, 9:22 a.m.