GitHub

Deborah.Rahab

Deborah.Rahab — Module
module Rahab

Deborah.Rahab — Pre-scouting visualization & $T=0$ utilities for Deborah.DeborahCore/Deborah.Esther/Deborah.Miriam.

Deborah.Rahab provides exploratory analysis tools designed for pre-scouting an ensemble before committing to heavy Deborah.DeborahCore/Deborah.Esther/Deborah.Miriam calculations. Its functions help assess correlations among observables, check configuration histories, estimate autocorrelations and $\tau_{\text{int}}$, and probe how bootstrap/jackknife errors depend on block/bin sizes — thereby guiding the choice of near-optimal resampling windows. In addition, Deborah.Rahab includes $T=0$ scaling utilities and spectroscopy converters for baseline checks.

Scope & Responsibilities

  • Correlation heatmaps: assemble per-observable samples and render annotated correlation matrices to reveal inter-observable dependencies.
  • Original vs ML histograms: compare OG data (Y_tr $\oplus$ Y_bc $\oplus$ Y_ul) against ML predictions (Y_tr $\oplus$ Y_bc $\oplus$ YP_ul) with shared binning; outputs both plots and numerical bin files for further diagnostics.
  • Observable histories & $\tau_{\text{int}}$: visualize per-configuration time histories, compute autocorrelation functions $\rho(\Delta)$, and estimate $\tau_{\text{int}}$ with standard windows or truncations, highlighting regions of slow decorrelation.
  • Block/bin scans & NSR: sweep bootstrap block sizes or jackknife bin sizes, plot $\mu\pm\sigma$ bands with blocked averages, show NSR curves, and mark regimes with no discarded samples — useful for selecting stable resampling parameters.
  • Zero-temperature scaling: provide $g_0^2(\beta)$, $\kappa_c(g_0^2)$, and quark mass $m_q \, a$ relations; include weighted-least-squares fits, linear/quadratic extrapolations, and spectroscopy table I/O for $\beta$-slice analyses.

Role in the Ecosystem

While Deborah.DeborahCore/Deborah.Esther/Deborah.Miriam carry out the main ML, cumulant, and multi-ensemble reweighting pipelines, Rahab acts as reconnaissance: it allows researchers to examine the statistical structure of a single ensemble (correlations, histories, resampling behaviors) to inform downstream parameter choices and detect issues early.

Typical Workflow

julia> using Deborah.Rahab

# 1. Correlation scan
julia> Rahab.CorrPlot.plot_corr_matrix("Y_bundle.jld2"; savepdf=true)

# 2. Histogram of original vs ML
julia> Rahab.HistogramOrigML.plot_histogram_orig_vs_ml(traces, idx=3, nbins=50)

# 3. Observable history + autocorr
julia> Rahab.ObservableHistory.plot_history_and_tauint("obs.dat")

# 4. Block/bin scan with NSR
julia> Rahab.BlockBinScan.scan_block_jackknife(data; maxbin=50)

# 5. Zero-T scaling
julia> Rahab.ZeroTemperatureScale.fit_kappa_c_beta(beta_vals, kappa_vals)

See Also

source