Convert to posterior::draws format

Description

Generic function to convert an object to a posterior::draws representation (e.g., draws_array, draws_matrix, or other formats).

Usage

as_draws(x, ...)

## S3 method for class 'rjuliabugs'
as_draws(x, ...)

## S3 method for class 'array'
as_draws(x, ...)

Arguments

Value

A draws object (e.g., draws_array) or a modified rjuliabugs object.

An object of class "rjuliabugs" (a named list) with the following elements:

params

Posterior samples converted to posterior::draws_array.

name

The name of the Julia sampler object (unchanged).

sampler

The sampler object returned by AbstractMCMC.sample in Julia (unchanged).

n_threads

Number of Julia threads detected (unchanged).

mcmc

MCMC configuration parameters; posterior_type is updated to "draws".

control

Control options passed to and used by the sampler (unchanged).

An object of class draws_array (from the posterior package). The 3D array (iterations × chains × parameters) is converted to a draws_array, preserving the chain structure and parameter names.

See Also

posterior::as_draws()

Convert to coda::mcmc or mcmc.list format

Description

Generic function to convert Markov Chain Monte Carlo (MCMC) output to coda::mcmc or coda::mcmc.list format.

Usage

as_mcmc(x, ...)

## S3 method for class 'rjuliabugs'
as_mcmc(x, ...)

## S3 method for class 'array'
as_mcmc(x, ...)

Arguments

Value

An object of class mcmc, mcmc.list, or a rjuliabugs object with updated params.

An object of class "rjuliabugs" (a named list) with the following elements:

params

Posterior samples converted to coda::mcmc if a single chain, or coda::mcmc.list if multiple chains.

name

The name of the Julia sampler object (unchanged).

sampler

The sampler object returned by AbstractMCMC.sample in Julia (unchanged).

n_threads

Number of Julia threads detected (unchanged).

mcmc

MCMC configuration parameters; posterior_type is updated to "mcmc".

control

Control options passed to and used by the sampler (unchanged).

Returns posterior samples converted to coda::mcmc if the array has one chain, or coda::mcmc.list if multiple chains. Input must be a 3D array (iterations × chains × parameters). Each column corresponds to a parameter, and each row corresponds to an iteration.

See Also

coda::as.mcmc(), coda::mcmc()

Convert to posterior::rvar format

Description

Generic function to convert an object to a posterior::rvar representation. This is typically used to convert Markov Chain Monte Carlo (MCMC) output into a more flexible and vectorized format.

Usage

as_rvar(x, ...)

## S3 method for class 'rjuliabugs'
as_rvar(x, ...)

## S3 method for class 'array'
as_rvar(x, n_mcmc = NULL, ...)

Arguments

Value

An object of class rvar, or an updated rjuliabugs object with params converted to rvar.

An object of class "rjuliabugs" (a named list) with the following elements:

params

Posterior samples, converted to the requested format: rvar for as_rvar, mcmc/mcmc.list for as_mcmc, draws_array for as_draws.

name

The name of the Julia sampler object.

sampler

The sampler object returned by AbstractMCMC.sample in Julia.

n_threads

Number of Julia threads detected.

mcmc

A list of MCMC configuration parameters, now including posterior_type indicating the format of params.

control

Control options passed to and used by the sampler.

An object of class rvar (from the posterior package). The input 3D array (iterations × chains × parameters) is converted into a posterior rvar object, where each parameter is represented as a random variable across iterations and chains.

See Also

posterior::rvar()

Convert Bayesian Updating for Gibbs Sampling (BUGS) Model to Julia's '@bugs' Macro Format

Description

This function formats a Bayesian Updating for Gibbs Sampling (BUGS) model string into Julia's '@bugs("""...""", convert_var_name, true)' macro syntax, used to run BUGS models in Julia. By default, R-style variable names (e.g., 'a.b.c') are converted to Julia-style ('a_b_c'). You can disable this behavior by setting 'convert_var_name = FALSE'.

Usage

bugs2juliaBUGS(model_code, convert_var_name = TRUE)

Arguments

Value

A character string representing a valid Julia '@bugs' macro call. This string can be evaluated in Julia to define the BUGS model with optional variable name conversion.

Examples

## Not run: 
model <- "
  for i in 1:N
    y[i] ~ dnorm(mu, tau)
  end
  mu ~ dnorm(0.0, 1.0E-6)
  tau ~ dgamma(0.001, 0.001)
"
bugs2juliaBUGS(model)
bugs2juliaBUGS(model, convert_var_name = FALSE)

## End(Not run)

Ensure Unique Name for Julia Sampler (Internal)

Description

Checks whether a Julia variable name is already defined in the Main module. If the name exists, appends random characters until a unique name is found.

Usage

check_sampler_is_defined(name)

Arguments

Value

A unique name not currently defined in Julia.

Convert Numeric Elements in a List to Integer or Float

Description

This function takes a list of numeric vectors and returns a new list where each numeric element is automatically converted to either an integer (if it is a whole number) or kept as a float (numeric). It also preserves the original names of vector elements, if any.

Usage

convert_numeric_types(data)

Arguments

Value

A list of the same structure where each numeric element is coerced to the appropriate type: integers for whole numbers, and floats otherwise. Names are preserved.

Examples

## Not run: 
input_list <- list(
  a = c(x = 1.0, y = 2.5, z = 3.0),
  b = c(foo = 4.0, bar = 5.1),
  c = c(6, 7, 8)
)
convert_numeric_types(input_list)

## End(Not run)

Delete an object from the Julia Main environment

Description

This function removes a variable or object from the Julia Main module using JuliaCall. It is useful for cleaning up or resetting objects defined in the Julia environment from R.

Usage

delete_julia_obj(obj_name)

Arguments

Value

No return value, called for side effects.

Examples

## Not run: 
JuliaCall::julia_command("x = 10")  # Define a Julia variable
delete_julia_obj("x")               # Delete it

## End(Not run)

Extract Posterior Samples from an 'rjuliabugs' S3 Object

Description

Extracts posterior samples for specified parameters from a fitted 'rjuliabugs' object. Output can be returned in several formats depending on downstream analysis requirements.

Usage

extract(rjuliabugs, pars = NULL, type = "array", include = TRUE)

Arguments

Value

The posterior samples in the specified format:

• a 3D 'array' [iterations × chains × parameters],

• a 'posterior::rvar' object,

• a 'coda::mcmc' or 'coda::mcmc.list',

• or a 'posterior::draws_array'/'draws_list'.

Extract Parameter Samples from a Turing.jl MCMCChains Object (internal use only)

Description

This function extracts posterior samples for specified parameters from a Julia 'Chains' object (from the 'Turing.jl' package) using 'JuliaCall'.

Usage

get_params_from_name_raw(params, name)

Arguments

Details

It supports extracting one or more parameters and returns the result as a numeric matrix in R.

This function builds a Julia expression of the form 'Array(chains[:, [:param1, :param2], :])' to extract values for the specified parameters from the sampler object. The result is evaluated in Julia and returned to R as a numeric matrix.

The function reshapes the output if only one parameter is extracted to ensure a consistent matrix format.

Value

A numeric matrix where rows represent samples and columns represent parameters.

Run a Julia HMC Sampler for a BUGS-like Probabilistic Model

Description

Executes a Hamiltonian Monte Carlo (HMC) sampler in Julia from R, using a model specified in Julia or in BUGS syntax. It compiles the model, converts data, sets sampler parameters, and returns posterior samples in various formats. The setup for the HMC sampler uses Not-U-Turn Sampler (NUTS) with the target acceptance probability \delta=0.8) for step size adaptation.

Usage

juliaBUGS(
  data,
  model_def,
  params_to_save,
  initializations = NULL,
  name = "sampler_juliaBUGS",
  n_iter = 2000,
  n_warmup = floor(n_iter/2),
  n_discard = n_warmup,
  n_thin = 1,
  n_chain = 1,
  use_parallel = TRUE,
  posterior_type = "array",
  force_setup_juliaBUGS = FALSE,
  control = NULL,
  progress = TRUE,
  verbose = TRUE,
  ...
)

Arguments

Details

This function relies on Julia packages LogDensityProblems, AdvancedHMC, and AbstractMCMC. Gradients are computed via ReverseDiff. The model is compiled before sampling.

The posterior_type argument determines the return format:

• "array": 3D numeric array (iterations × chains × parameters).

• "rvar": posterior::rvar object.

• "mcmc": coda::mcmc (single chain) or mcmc.list (multiple chains).

• "draws": posterior::draws_array.

Value

An object of class "rjuliabugs" (a named list) with the following elements:

params

Posterior samples, in the format specified by posterior_type ("array", "rvar", "mcmc", or "draws").

name

A character string giving the name of the Julia sampler object.

sampler

The sampler object returned by AbstractMCMC.sample in Julia.

n_threads

An integer giving the number of Julia threads detected.

mcmc

A list of MCMC configuration parameters used in the run.

control

The list of control options passed to and used by the sampler.

Note

You must call setup_juliaBUGS() at least once before using this function. If parallel sampling is requested but only one Julia thread is available, a warning is issued and sampling will run serially.

Examples

## Not run: 
model_def <- "model = @model ... end"
data <- list(N = 10, x = rnorm(10))
result <- juliaBUGS(
  data = data,
  model_def = model_def,
  params_to_save = c("mu"),
  name = "my_sampler"
)

## End(Not run)

Assign an Integer Value to a Julia Variable

Description

This function wraps 'JuliaCall::julia_assign' to assign a value from R to a Julia variable, and then explicitly casts the variable to the 'Integer' type in Julia.

Usage

julia_assign_int(x, value)

Arguments

Details

The function first assigns the value from R to a variable named 'x' in the Julia session using 'JuliaCall::julia_assign'. It then forces Julia to cast the variable to the ‘Integer' type using Julia’s 'Integer()' constructor.

Note: This function assumes that the ‘value' provided is compatible with Julia’s 'Integer' type. If it is not, an error will be thrown by Julia.

Value

Invisibly returns 'NULL'. The main effect is side-effects in the Julia session.

Examples

## Not run: 
julia_assign_int("x", 3.5)  # Will be cast to 3L in Julia
JuliaCall::julia_eval("x")  # Returns 3 (as Integer in Julia)

## End(Not run)

Load an rjuliabugs Object and Restore the Julia State

Description

Loads an object of class rjuliabugs from an .rds file and restores the corresponding Julia sampler object using Julia’s Serialization.deserialize. The path linking the Chains object from Julia is defined in the when the function save_rjuliabugs() is called.

Usage

load_rjuliabugs(file)

Arguments

Details

If the original sampler name (name) already exists in the active Julia session, a new unique name is generated to avoid overwriting it. A warning will be issued to indicate that the name has changed.

The .rds file must contain a valid rjuliabugs object with both the name and chains_file fields defined. The function checks if the sampler name is already defined in Julia. If so, a unique name is generated using check_sampler_is_defined, and the Julia object is loaded under that name.

Value

An object of class rjuliabugs, with the Julia sampler object loaded into the current session. If the name was changed to avoid conflict, the returned object reflects the updated name.

See Also

check_sampler_is_defined

Examples

## Not run: 
model <- load_rjuliabugs("my_model.rds")
# model$name now contains the (possibly updated) name used in Julia

## End(Not run)

Save an rjuliabugs Object and Its Julia State

Description

Serializes the Julia object contained in an rjuliabugs object and saves the entire object as an .rds file. The Julia object is saved separately using Julia's Serialization.serialize. The file path can be passed manually, or retrieved from the chains_file slot in the object.

Usage

save_rjuliabugs(rjuliabugs_model, file, chains_file = NULL)

Arguments

Value

No return value, called for saving both the Julia object and the R rjuliabugs object to disk.

Examples

## Not run: 
save_rjuliabugs(my_model, file = "my_model", chains_file = "chains.jls")

## End(Not run)

Setup Julia Environment for JuliaBUGS

Description

Installs and loads the required Julia packages to use JuliaBUGS via JuliaCall in R.

Usage

setup_juliaBUGS(
  extra_packages = NULL,
  verify_package = TRUE,
  install_from_dev = FALSE,
  verbose = TRUE,
  ...
)

Arguments

Details

This function checks whether the core Julia packages needed for running JuliaBUGS are installed, installs any missing ones, and loads them into the current Julia session. Optionally, additional Julia packages can be installed and loaded by specifying them via extra_packages.

The core Julia packages installed (if needed) are:

• Serialization

• LogDensityProblemsAD

• ReverseDiff

• AdvancedHMC

• AbstractMCMC

• LogDensityProblems

• MCMCChains

• DataFrames

• JuliaBUGS

After installation, all these packages are loaded in the Julia session using using. Any additional packages provided via extra_packages are also installed and loaded.

Value

Invisibly returns NULL. The function is called for its side effects.

See Also

julia_install_package_if_needed, julia_eval

Examples

## Not run: 
# Setup Julia with core packages only
setup_juliaBUGS()

# Setup Julia with additional packages
setup_juliaBUGS(extra_packages = c("Distributions", "Turing"))

## End(Not run)

Summary Method for JuliaBUGS Sampler Output

Description

Provides a summary of the results from a JuliaBUGS sampler object, including Markov Chain Monte Carlo (MCMC) settings, summary statistics, and optionally quantiles.

Usage

## S3 method for class 'rjuliabugs'
summary(object, ...)

Arguments

Details

This method wraps Julia's MCMCChains.summarystats and MCMCChains.quantile to extract and display results in R using the JuliaCall interface. It also extracts key MCMC settings like number of chains, iterations, and samples per chain. The printed summary is truncated to n_display rows.

Value

If julia_summary_only = TRUE, no value is called for the print message. Otherwise, returns a list possibly containing:

• summary: Data frame of summary statistics (if get_summary = TRUE).

• quantiles: Data frame of quantiles (if get_quantiles = TRUE).

Wrap Bayesian Updating for Gibbs Sampling (BUGS) Model for Julia

Description

Wraps a Bayesian Updating for Gibbs Sampling (BUGS) model string with 'model = @bugs begin' and 'end', if it is not already wrapped. This is useful for preparing BUGS models for use with Julia packages that expect this specific block structure.

Usage

wrap_model_to_juliaBUGS(model_code)

Arguments

Value

A character string with the BUGS model wrapped properly in Julia-compatible syntax.

Examples

## Not run: 
model_body <- "
  for i in 1:N
    r[i] ~ dbin(p[i], n[i])
    b[i] ~ dnorm(0.0, tau)
    p[i] = logistic(alpha0 + alpha1 * x1[i] + alpha2 * x2[i] + alpha12 * x1[i] * x2[i] + b[i])
  end
  alpha0 ~ dnorm(0.0, 1.0E-6)
  tau ~ dgamma(0.001, 0.001)
  sigma = 1 / sqrt(tau)
"
wrap_model_to_juliaBUGS(model_body)

## End(Not run)