MCMCStreamTempering

NamedNTupleOrComponentVector{T}

Shorthand for Union of NamedNTuple{T} or ComponentVector{T} of eltype(T).

MHLogDensityFunction

A log density function for the MHSampler sampler.

This variant uses the set_namedtuple! function to update the VarInfo.

NamedNTuple{T}

Shorthand for NamedTuple with Tuple type being an NTuple with eltype T`.

Context that multiplies each log-prior by mod used to test whether varwise_logpriors respects child-context.

Context that multiplies each log-prior by mod used to test whether varwise_logpriors respects child-context.

Compute temperature of data streams for given ratio of r = model_discrepancy / stddev(obs_error).

dist_val_tuple(spl::Sampler{<:MHSampler}, vi::VarInfo)

Return two NamedTuples.

The first NamedTuple has symbols as keys and distributions as values. The second NamedTuple has model symbols as keys and their stored values as values.

estimate_obs_unc(x,y; span=0.8)

Estimate std-deviation of observation error in a series by inspecting the residuals of a loess model y ~ x. Keyword argument span controls the smoothmess of the fit. A value of 1 corresponds to a straight line, 0.8 is adequate for a single maximum. The fit can be inspected by plotting pred ~ x in the same plot as y ~ x.

Return value is a NamedTuple

• std_obsunc: estimated standard deviation of observation uncertainty
• SS_obsunc: sum of squared residuals of observation uncertainty
• pred: predictions by the loess model, useful to inspect smoothness.

estimate_prop_weights_model_unc(obs_streams, std_eff_streams)

Estimate ratio of spread / stddev(obs_error), where spread is the 95% interval of observed value.

This is helpful for allowing larger relative model discrepancy (per observation uncertainty) for the streams with a larger spread per observation uncertainty.

Arguments

• obs_streams: ComponentVector or NamedTuple stream -> observations
• std_eff_streams: ComponentVector or NamedTuple stream -> std_dev(obs_error)

Result NamedTuple with ComponentVectors

• w: multiplicators of allowed model discrepancy. Specifically w0 ./ minimum(w0)
• w0: original ratios spread / stddev(obs_error)

set_namedtuple!(vi::VarInfo, nt::NamedTuple)

Places the values of a NamedTuple into the relevant places of a VarInfo.