default

default: object

Defined in: mc/src/index.js:135

Type Declaration

diagnostics

diagnostics: object

diagnostics.effectiveSampleSize

effectiveSampleSize: (samples) => number

Compute effective sample size (ESS) using autocorrelation

Parameters

samples

number[]

Array of samples

Returns

number

Effective sample size

diagnostics.gelmanRubin

gelmanRubin: (chains) => number

Compute the Gelman-Rubin diagnostic (R-hat) for convergence Requires multiple chains

Parameters

chains

number[][]

Array of chains (each chain is an array of samples)

Returns

number

R-hat statistic

diagnostics.printSummary

printSummary: (trace) => void

Print trace summary for all variables

Parameters

trace

any

Trace object from sampling

Returns

void

diagnostics.summarize

summarize: (samples) => object

Compute summary statistics for a trace

Parameters

samples

number[]

Array of samples

Returns

object

Summary statistics: the mean, median, standard deviation, variance, 2.5%/97.5% interval bounds, and the sample count

hdi_2_5

hdi_2_5: number

hdi_97_5

hdi_97_5: number

mean

mean: number

median

median: number

n

n: number

std

std: number

variance

variance: number

diagnostics.traceToCSV

traceToCSV: (samples) => string

Save trace to CSV format (for a single variable)

Parameters

samples

number[]

Array of samples

Returns

string

CSV string

diagnostics.traceToJSON

traceToJSON: (trace) => string

Export trace to JSON format

Parameters

trace

any

Trace object

Returns

string

JSON string

distributions

distributions: object

distributions.Bernoulli

Bernoulli: typeof Bernoulli

distributions.Beta

Beta: typeof Beta

distributions.Distribution

Distribution: typeof Distribution

distributions.Gamma

Gamma: typeof Gamma

distributions.HalfNormal

HalfNormal: typeof HalfNormal

distributions.Lognormal

Lognormal: typeof Lognormal

distributions.Normal

Normal: typeof Normal

distributions.Uniform

Uniform: typeof Uniform

getRng

getRng: () => any

Get the current package RNG ({float, int, normal}).

Returns

any

Model

Model: typeof Model

plot

plot: object

plot.autocorrPlot

autocorrPlot: (trace, variables, maxLag, options) => any

Generate autocorrelation plot specification Shows autocorrelation to assess mixing

Parameters

trace

any

MCMC trace object

variables?

string[] = null

Variable names to plot

maxLag?

number = 50

Maximum lag to compute

options?

any = {}

Plot options

Returns

any

Plot specification with .show() method

plot.forestPlot

forestPlot: (trace, variables, hdi, options) => any

Generate forest plot specification Shows posterior summaries with credible intervals

Parameters

trace

any

MCMC trace object

variables?

string[] = null

Variable names to plot

hdi?

number = 0.95

Highest Density Interval (default 0.95)

options?

any = {}

Plot options

Returns

any

Plot specification with .show() method

plot.pairPlot

pairPlot: (trace, variables, options) => any

Generate pair plot specification (scatter plot matrix) Shows relationships between parameters

Parameters

trace

any

MCMC trace object

variables?

string[] = null

Variable names to plot

options?

any = {}

Plot options

Returns

any

Plot specification with .show() method

plot.posteriorPlot

posteriorPlot: (trace, variables, options) => any

Generate posterior distribution plot specification Shows histograms and KDE of posterior samples

Parameters

trace

any

MCMC trace object

variables?

string[] = null

Variable names to plot

options?

any = {}

Plot options

Returns

any

Plot specification with .show() method

plot.rankPlot

rankPlot: (trace, variables, options) => any

Generate rank plot specification (for convergence diagnostics) Useful for detecting non-stationarity and comparing chains

Parameters

trace

any

MCMC trace object

variables?

string[] = null

Variable names to plot

options?

any = {}

Plot options

Returns

any

Plot specification with .show() method

plot.tracePlot

tracePlot: (trace, variables, options) => any

Generate trace plot specification Shows the sampled values over iterations to assess convergence

Parameters

trace

any

MCMC trace object

variables?

string[] = null

Variable names to plot (null = all)

options?

any = {}

Plot options

Returns

any

Plot specification with .show() method

samplers

samplers: object

samplers.HamiltonianMC

HamiltonianMC: typeof HamiltonianMC

samplers.HMC

HMC: typeof HMC

samplers.MetropolisHastings

MetropolisHastings: typeof MetropolisHastings

samplers.NUTS

NUTS: typeof NUTS

samplers.summary

summary: (chainsOrResults, opts?) => any[]

ArviZ-style posterior summary across one or more chains.

Parameters

chainsOrResults

any

Array of chain results ({trace} from HMC#sample), an array of raw trace dicts, or a single trace dict.

opts?

hdi?

number = 0.94

HDI mass (e.g. 0.94 → hdi_3%/hdi_97%).

Returns

any[]

One row per scalar parameter component with { param, mean, sd, hdi_lo, hdi_hi, ess, rhat }.

setRandomSeed

setRandomSeed: (seed) => void

Seed the package RNG for reproducible sampling.

Parameters

seed

number

Any finite number

Returns

void