Probabilistic Decomposition of Archaeological Palimpsests.

palimpsestr

Probabilistic decomposition of archaeological palimpsests using Stratigraphic Entanglement Fields.

Installation

# install.packages("remotes")
remotes::install_github("enzococca/palimpsestr")

What it does

palimpsestr models each archaeological find as a probabilistic member of latent depositional phases by integrating spatial proximity, stratigraphic depth, chronological overlap, and cultural similarity via diagonal Gaussian mixture EM.

Three statistics quantify the deposit:

• SEI — Stratigraphic Entanglement Index (pairwise coherence)
• ESE — Excavation Stratigraphic Energy (local disruption)
• PDI — Palimpsest Dissolution Index (global separability, 0--1)

Quick example

library(palimpsestr)

# Simulate a 3-phase deposit with 30% mixing
x <- archaeo_sim(n = 200, k = 3, mixing = 0.30, seed = 42)

# Fit the SEF model
fit <- fit_sef(x, k = 3, context = "context")

print(fit)
summary(fit)

# Visualisation (base R)
plot_phasefield(fit)
plot_entropy(fit)

# ggplot2 equivalents (requires ggplot2)
gg_phasefield(fit)
gg_entropy(fit)
gg_energy(fit)
gg_intrusions(fit)

# Intrusion detection
detect_intrusions(fit)

# Compare multiple K values
compare_k(x, k_values = 2:6, context = "context")

Demo datasets

data(demo_easy)        # 3 phases, 5% mixing
data(demo_moderate)    # 3 phases, 30% mixing
data(demo_compressed)  # 4 phases, 50% mixing
data(villa_romana)     # Real data: Poggio Gramignano (VRPG 3004)

Features

• Gaussian Mixture EM with taphonomic weighting and stratigraphic penalties
• Harris Matrix integration as stratigraphic constraint
• Bootstrap confidence intervals (bootstrap_sef())
• Cross-validation for model selection (cv_sef(), optimize_weights())
• GIS export via sf (as_sf_phase(), as_sf_links())
• Geometry overlay maps (gg_map())
• Interpretive reports in English and Italian (report_sef())
• Database import from any DBI source (read_db())

License

MIT.