samOptiPro — Package Overview • samOptiPro

Overview

This document provides a structured overview of all exported and internal functions of samOptiPro.

Constants

`.collect_lengths`

`.pad_mat_cols`

`.pad_vec`

`normalize_constants_generic`

Normalize constants in a model-agnostic way

`push`

Diagnostics

`.add_block`

`.add_block_family`

`.add_scalar`

`.add_scalar_family`

`.as_char_vec`

`.ask_yes_no_strict`

`.barrier_before_block`

Temporarily switch nimble buildDir (clear compiled & GC)

`.compile_and_run`

`.ensure_unsampled`

`.fam_metrics`

`.head6`

`.is_ll`

`.log`

`.metrics_for`

`.mkdir`

`.opt_integer`

`.opt_logical`

`.print_step`

`.prompt_info`

`.restore_builddir`

Restore previous nimble buildDir if it was changed

`.safe_scalar_targets`

`.sanitize_nodes`

`.sop_detect_nondiff_functions`

`.sop_supports_derivs`

Check if derivatives/HMC are supported for a nimble model

`.timed`

`.to_family`

`compute_diag_from_mcmc`

Build a standard diagnostics table from a

`compute_diag_from_mcmc_vect`

Compute Diagnostics from MCMC Samples (massively scalable)

`configure_hmc_safely_bis`

Configure and run HMC/NUTS safely (global / subset / auto)

`diagnose_model_structure`

Diagnose model structure, dependencies, and per-sampler time (parameter- and family-level)

`diagnostics_by_target`

Target-level diagnostics (time + optional step-size proxy)

`get_bound`

`get_deps`

`get_deps_memo`

`get_dist`

`get_family_nodes`

`get_varinfo_dim`

`is_ll`

`normalize_dist`

`profile_sampler_times`

Compile (once) and run a NIMBLE MCMC configured in , with

`profile_sampler_times_uncompiled`

Uncompiled per-sampler timing (R-level)

`proxy_step_sd`

Step proxy: sd(diff(chain)) on columns matching a pattern

`rebuild_nodes_df`

`root_of`

`run_structure_and_hmc_test`

Run structural diagnostics and (optional) HMC/NUTS smoke test

`safe_time_check`

`say`

`strip_index`

`summarise_showstoppers`

`support_of`

`test_strategy`

Test and compare MCMC strategies on selected bottleneck nodes

`test_strategy_family`

Family-based sampler strategy: full HMC if allowed, else surgical on bottlenecks

`test_strategy_family_fast`

Fast strategy testing for sampler plans (family-level, with optional full-model HMC/NUTS)

`to_logical`

`vmsg`

HMC Wrappers

`configure_hmc_safely`

Configure and run HMC/NUTS safely

Inits / Robustness

`checkInits`

Robustly check initial values against a compiled NIMBLE model

`checkInitsAndRun`

Check inits then run MCMC

Misc

`.avail_vars`

Available variable roots present in a model

`.compile_mcmc_with_build`

Compile a nimble MCMC for a given build object

`.configure_with_monitors`

Build a configureMCMC with sanitized/expanded monitors

`.default_sanitize_roots`

Discover default monitors from a nimble model (variable-level)

`.derive_sampler_params_auto`

`.derive_sampler_params_from_conf`

`.family_of`

`.is_ignored`

`.plot_rhat_bar`

Save a bar chart of R-hat for selected nodes

`.plot_traces`

Save trace plots for a set of nodes

`.print_monitors`

Pretty-print selected monitors (optional)

`.reorder_cols_family`

`.reorder_cols_param`

`.sample_cols`

`.sop_cbind_align`

Bind two matrices by rows count (truncate to common minimum)

`.sop_detect_chain_prefix`

Detect chain prefix “chainK.” in column names

`.sop_expand_roots_to_nodes`

Expand many roots to nodes

`.sop_expand_var_nodes`

Expand one root to nodes via getVarInfo()

`.sop_get_uncompiled_model`

Return the R-level (uncompiled) nimble model when given either R or compiled model

`.sop_has_model_api`

Heuristic check that an object exposes nimble model API

`.sop_is_chol_ok`

Cholesky check

`.sop_is_model`

Is this a nimble model?

`.sop_make_propCov_PD`

Try to make a symmetric matrix positive definite (nearPD/shrinkage/jitter)

`.sop_max`

`.sop_sanitize_roots`

Sanitize monitor roots (drop numerics & empties)

`.sop_split_into_blocks`

Correlation-based clustering into compact blocks

`.sop_strip_chain_prefix`

Strip “chainK.” prefix from column names

`agg`

`as_mcmc_list`

Convert an object to mcmc.list

`as_mcmc_list_sop`

Convert various sample formats to coda::mcmc.list (optionally merging samples2)

`assess_performance`

Compute per-parameter and global performance diagnostics from an MCMC

`compute_WAIC`

Compute WAIC from coda samples that include logLik[] columns

`customize_samplers`

`default_monitors`

`ensure_monitors_exist`

Ensure monitors exist on the model (variable-aware)

`family_of`

`ggh`

`has_fun`

`identify_bottlenecks`

`identify_bottlenecks_family`

Identify bottlenecks by parameter stats::family(medians within families)

`merge_mcmc_samples`

Merge res$samples and res$samples2 into a single mcmc.list

`mk_top3`

`plot_family_ess_bar`

Family-level ESS bar plot (harmonised style)

`plot_family_rhat_bar`

Family-level Rhat bar plot (harmonised style)

`plot_mcmc_histograms`

Histograms for ESS / CE / AE / Rhat (ggplot2)

`strip_loglik_cols`

`take`

`time_it`

`to_mlist`

Plots

`.derive_sampler_targets_fast`

`.family_from_fast`

`.ggsave`

`.mk_dir`

`agg_safe`

`as_num`

`enrich_hmc_diag_tbl_for_plots`

`flatten_chr`

`get_param_df`

`get_sampler_targets`

`headn`

`mk_plot_all`

`mk_plot_step`

`mk_trace_density`

`pick_rhat`

`pick_vec`

`plot_bottlenecks`

Plot MCMC Bottlenecks by Node or Family

`plot_bottlenecks_fast`

Fast bottleneck plots (samplers-only) for large models

`plot_bottlenecks_index`

Plot diagnostics for stochastic sampler targets (strict) + all-nodes panels

`plot_convergence_checks`

Plot convergence diagnostics (R-hat & traces) with family-level R-hat bars

`plot_convergence_rhat_sampled_fast`

Fast convergence plots for R-hat (samplers-only), thresholded

`plot_strategies_from_test_result_fast`

Plot strategy comparisons from test_strategy_family_fast results (fast path)

`plot_strategies_from_test_result_hmc_fast`

@export

`save_fig`

Runners

`.fresh_build`

`.merge_mcmc`

`.run_and_collect`

`.safe_as_mcmc_list`

`build_conf_with_monitors`

Build a fresh MCMC configuration with automatic monitors if missing

`grab_mcpar`

`run_baseline_coda`

Run baseline et retourne un mcmc.list fusionne (samples + samples2)

`run_baseline_config`

Run baseline RW/Slice with robust compile/run

`run_hmc_all_nodes`

Run HMC/NUTS where possible; fallback samplers for uncovered nodes

Samplers Utilities

`.as_list_safe`

`.get_targets`

`.get_type`

`.us_configure_with_monitors`

`.us_expand_compact_node`

Expand a compact node Character scalar, e.g. p[1, 1:3] into explicit indices (numeric only)

`.us_expand_vars_to_nodes`

Expand variable roots (e.g. “beta”) to explicit node names using model metadata

`.us_sanitize_monitors`

Drop numeric junk & control tokens from monitors

`.us_tokenize_monitors`

Tokenize a monitors specification into clean symbols

`sampler_df`

Return a data.frame (name, type, target, scale) from a sampler container.

`sampler_df_from_conf`

Convenience: directly return sampler_df from an existing conf.

`sampler_env_dump`

Long data.frame of sampler env numeric fields

`sampler_env_numeric_fields`

Numeric scalar fields in the environment of each compiled sampler function

`sampler_functions_from_conf`

Build and compile an MCMC from a conf to access samplerFunctions.

`sampler_scale`

Extract a numeric scale proxy from a sampler object.

`sampler_scales_after_run`

Extract sampler scales after a short adaptive run.

`sampler_scales_from_run`

Preferred: read scales from a fresh samplerConf list (pre-run).

`sampler_targets`

Targets (node names) for each sampler in a MCMC configuration