Deborah.Rebekah

module Rebekah

Deborah.Rebekah — Plotting and I/O utilities for Deborah.DeborahCore/Deborah.Esther/Deborah.Miriam results.

Deborah.Rebekah is the lightweight plotting + I/O layer in the Deborah.jl ecosystem. It provides helpers to

1. load consolidated JLD2 result bundles,
2. set publication-grade PyPlot.jl/$\LaTeX$ styles, and
3. visualize grid scans and method comparisons (e.g., overlap/ERR heatmaps, Bhattacharyya-coefficient/JSD maps,

and P1/P2 vs. baseline curves).

Scope & Responsibilities

• Result loading
  • JLD2Loader.load_jld2 → (summary, labels, trains) for single-stage reports.
  • JLD2Loader.load_jld2_Miriam → rich multi-ensemble bundle with summary* dicts, kappa_list, rw_data, labels, trains.
• Plot style ($\LaTeX$)
  • PyPlotLaTeX.set_pyplot_latex_style and a dedicated correlation-matrix style preset for consistent paper figures.
• Heatmaps
  • Overlap/ERR pair: Heatmaps.render_overlap_and_error_heatmaps.
  • Bhattacharyya coefficient (BC): Heatmaps.render_bhattacharyya_heatmap.
  • Jensen-Shannon divergence (JSD): Heatmaps.render_jsd_heatmap.
• P1/P2 vs Baseline curves
  • At fixed LBP: PXvsBSPlotter.plot_PX_BS_vs_trains.
  • At fixed TRP: PXvsBSPlotter.plot_PX_BS_vs_labels.

Public API (selected)

• JLD2Loader.load_jld2
• JLD2Loader.load_jld2_Miriam
• PyPlotLaTeX.set_pyplot_latex_style
• Heatmaps.render_overlap_and_error_heatmaps Heatmaps.render_bhattacharyya_heatmap Heatmaps.render_jsd_heatmap
• PXvsBSPlotter.plot_PX_BS_vs_trains PXvsBSPlotter.plot_PX_BS_vs_labels

Minimal Usage (REPL)

julia> using Deborah
julia> using Deborah.Rebekah

# 1) Load a consolidated results bundle (e.g., from Miriam)
julia> b = Rebekah.JLD2Loader.load_jld2_Miriam("results_<overall_name>.jld2")

# 2) Set LaTeX plotting style (optional)
julia> Rebekah.PyPlotLaTeX.set_pyplot_latex_style(0.5)

# 3) Heatmaps for a chosen observable/method
julia> using Deborah.Rebekah.Heatmaps
julia> Heatmaps.render_bhattacharyya_heatmap(bc, N_lb_arr, N_tr_arr, :TrM1, :Y_P2, bname, "figs/"; save_file=true)

# 4) Compare P1/P2 vs baseline along TRP (fixed LBP)
julia> using Deborah.Rebekah.PXvsBSPlotter
julia> plot_PX_BS_vs_trains("TrM1", "Y_BS", "P1", "P2", 20, new_dict, trains, labels, bname, "figs/"; save_file=true)

Notes

• Heatmap utilities accept integer percentage grids for LBP/TRP axes and optionally annotate each cell with numeric values; PDFs are auto-cropped when pdfcrop exists.
• The Deborah.Rebekah.PXvsBSPlotter expects pre-aligned matrices in new_dict keyed by "KEY:Y_*:(avg|err)" (or "Y_*" when KEY == "Deborah").
• Style presets modify global matplotlib.rcParams; call them once per session.

See Also

• Deborah.DeborahDocument,
• Deborah.EstherDocument,
• Deborah.MiriamDocument (document/report layers).