Visualize

plot_parameter_distribution(fig, pd; constrained=true, n_sample=1e4, rng=Random.GLOBAL_RNG)

Plot histogram marginals of the parameter distribution pd on fig.

Each dimension of pd is drawn as a separate histogram panel. When constrained = true (default), samples are mapped from the unconstrained space to the constrained space before plotting.

Requires a Makie backend (e.g. using CairoMakie) to be loaded.

Arguments

• fig: a Makie.Figure, Makie.GridLayout, Makie.GridPosition, or Makie.GridSubposition that receives the plots.
• pd: the ParameterDistribution (or ParameterDistributionType) to plot.
• constrained: if true, transform samples to constrained space before plotting.
• n_sample: number of Monte Carlo samples used to build the histograms.
• rng: random-number generator.

plot_error_over_iters(gridposition, ekp; error_metric=nothing, kwargs...)

Plot the error metric of ekp against the iteration index on gridposition.

The x-axis shows iteration numbers; the y-axis shows the chosen error metric. error_metric selects which metric to display; when nothing the metric is chosen automatically based on the process type. Additional keyword arguments are forwarded to Makie.Lines.

Requires a Makie backend to be loaded.

plot_error_over_iters!(axis, ekp; error_metric=nothing, kwargs...)

Plot the error metric of ekp against the iteration index into an existing axis.

Mutating variant of plot_error_over_iters. The axis argument must be a Makie.AbstractAxis. Additional keyword arguments are forwarded to Makie.Lines.

Requires a Makie backend to be loaded.

plot_error_over_time(gridposition, ekp; error_metric=nothing, kwargs...)

Plot the error metric of ekp against algorithm time on gridposition.

The x-axis shows the accumulated pseudo-time (sum of step sizes Δt); the y-axis shows the chosen error metric. error_metric selects which metric to display; when nothing the metric is chosen automatically based on the process type. Additional keyword arguments are forwarded to Makie.Lines.

Requires a Makie backend to be loaded.

plot_error_over_time!(axis, ekp; error_metric=nothing, kwargs...)

Plot the error metric of ekp against algorithm time into an existing axis.

Mutating variant of plot_error_over_time. The axis argument must be a Makie.AbstractAxis. Additional keyword arguments are forwarded to Makie.Lines.

Requires a Makie backend to be loaded.

plot_ϕ_over_iters(gridposition, ekp, prior, dim_idx; kwargs...)
plot_ϕ_over_iters(gridpositions, ekp, prior, name; kwargs...)

Plot ensemble members of constrained parameter dim_idx (or all dimensions of distribution name) against the iteration index.

Each ensemble member is shown as a scatter point. The first form plots a single parameter dimension on gridposition; the second form iterates over gridpositions and plots every dimension of the named sub-distribution. Additional keyword arguments are forwarded to Makie.Scatter.

Requires a Makie backend to be loaded.

Arguments

• gridposition: a Makie.GridPosition (or compatible) receiving the plot.
• gridpositions: an iterable of grid positions, one per dimension of name.
• ekp: the EnsembleKalmanProcess providing ensemble trajectories.
• prior: the ParameterDistribution used to map unconstrained to constrained space.
• dim_idx: integer index into the full parameter vector.
• name: name string of a sub-distribution in prior.

plot_ϕ_over_iters!(axis, ekp, prior, dim_idx; kwargs...)

Plot ensemble members of constrained parameter dim_idx against the iteration index into an existing axis.

Mutating variant of plot_ϕ_over_iters. Additional keyword arguments are forwarded to Makie.Scatter.

Requires a Makie backend to be loaded.

plot_ϕ_over_time(gridposition, ekp, prior, dim_idx; kwargs...)
plot_ϕ_over_time(gridpositions, ekp, prior, name; kwargs...)

Plot ensemble members of constrained parameter dim_idx (or all dimensions of distribution name) against algorithm time.

Each ensemble member is shown as a scatter point. The first form plots a single parameter dimension on gridposition; the second form iterates over gridpositions and plots every dimension of the named sub-distribution. Additional keyword arguments are forwarded to Makie.Scatter.

Requires a Makie backend to be loaded.

Arguments

• gridposition: a Makie.GridPosition (or compatible) receiving the plot.
• gridpositions: an iterable of grid positions, one per dimension of name.
• ekp: the EnsembleKalmanProcess providing ensemble trajectories.
• prior: the ParameterDistribution used to map unconstrained to constrained space.
• dim_idx: integer index into the full parameter vector.
• name: name string of a sub-distribution in prior.

plot_ϕ_over_time!(axis, ekp, prior, dim_idx; kwargs...)

Plot ensemble members of constrained parameter dim_idx against algorithm time into an existing axis.

Mutating variant of plot_ϕ_over_time. Additional keyword arguments are forwarded to Makie.Scatter.

Requires a Makie backend to be loaded.

plot_ϕ_mean_over_iters!(axis, ekp, prior, dim_idx; plot_std=false, kwargs...)

Plot the ensemble mean of constrained parameter dim_idx against the iteration index into an existing axis.

When plot_std = true, a band spanning one standard deviation around the mean is also drawn. Additional keyword arguments are forwarded to Makie.Lines (and to Makie.Band when plot_std = true).

Requires a Makie backend to be loaded.

plot_ϕ_mean_over_iters(gridposition, ekp, prior, dim_idx; plot_std=false, kwargs...)
plot_ϕ_mean_over_iters(gridpositions, ekp, prior, name; kwargs...)

Plot the ensemble mean of constrained parameter dim_idx (or all dimensions of distribution name) against the iteration index.

The first form creates a new axis inside gridposition; the second form iterates over gridpositions and plots every dimension of the named sub-distribution. When plot_std = true, a band spanning one standard deviation around the mean is also drawn. Additional keyword arguments are forwarded to Makie.Lines (and to Makie.Band when plot_std = true).

Requires a Makie backend to be loaded.

plot_ϕ_mean_over_time!(axis, ekp, prior, dim_idx; plot_std=false, kwargs...)

Plot the ensemble mean of constrained parameter dim_idx against algorithm time into an existing axis.

When plot_std = true, a band spanning one standard deviation around the mean is also drawn. Additional keyword arguments are forwarded to Makie.Lines (and to Makie.Band when plot_std = true).

Requires a Makie backend to be loaded.

plot_ϕ_mean_over_time(gridposition, ekp, prior, dim_idx; plot_std=false, kwargs...)
plot_ϕ_mean_over_time(gridpositions, ekp, prior, name; kwargs...)

Plot the ensemble mean of constrained parameter dim_idx (or all dimensions of distribution name) against algorithm time.

The first form creates a new axis inside gridposition; the second form iterates over gridpositions and plots every dimension of the named sub-distribution. When plot_std = true, a band spanning one standard deviation around the mean is also drawn. Additional keyword arguments are forwarded to Makie.Lines (and to Makie.Band when plot_std = true).

Requires a Makie backend to be loaded.