[gvegayon]

Follow-up to #28. Docs only; no runner, record, or figure changes.

The README figures. The figures on main already include EpiModel: README_files/figure-commonmark/summary-time-plot-1.png as served by raw.githubusercontent.com matches the committed blob (EpiModel is the bottom row, 7 s at 100k and 79.3 s at 1M). GitHub serves repository images with max-age=300, so a page opened shortly after the merge can show the earlier, twelve-engine figure until a hard refresh.

What this fixes

• docs/results.qmd
  • The memory paragraph listed the largest peaks at 1,000,000 agents (FRED, epydemic, EoN) but left out EpiModel, whose peak (1,893 MiB) is above EoN's (1,671).
  • The "Environment" bullet said every scenario's replicates ran in one recorded environment. ABM's (Add the ABM R package as a twelfth engine #26) and EpiModel's (Add EpiModel as a thirteenth engine #28) records came from later, separate runs, so every scenario report shows a mixed-environment warning. The bullet now says so.
• docs/methods.qmd: after the environment table, a note that the table describes the run behind every engine except ABM and EpiModel, and how those two ran.
• scenario_02/README.qmd: the notes said extraction is small for every engine but epiworldR and FRED. EpiModel's is about 45 ms, nearly all of it get_transmat(). That function keeps one infector per new case with a grouped dplyr sample_n(), about 45 µs per transmission. A synthetic test confirmed it is linear: 28 ms for 1,000 rows and 0.27 s for 10,000.
• scenario_00/README.qmd
  • ABM was missing from the list of uncalibrated engines; scenario.toml already says it needs no factor.
  • The model-lines note now mentions EpiModel's hand-written modules.
• scenarios.md: the runner layout and the checklist now list epimodel.R, plus agents.jl and run_FRED.py, which were also missing.

Reports re-rendered with make container-report. The scenario simulation-time PNGs were left as they were: those plots use unseeded jitter, so a re-render changes their bytes but not their data.

Not changed

• The mixed environments themselves. Recording one environment for every engine would mean rerunning the full design in a single run.

🤖 Generated with Claude Code