v1.1

Updates to support the paper "Playing Risky Games."

Changes that affect both versions of the Hawk/Dove model:

• Revise default random play odds to 0.1 ; add to default batch run parameters
• Add boolean parameter to control whether to include extreme risk-attitudes 0 and 9 (include_endpoints, default is True)
• Add include_endpoints parameter to interactive simulation controls
• Update labels and legends in the interactive simulation and analysis notebooks to match language in the paper.
• Add a marimo notebook to serve as the interface for running the Hawk/Dove simulation interactively; supports export to html+wasm for publication via static site.

Model-specific changes:

Sim 1. Multiple Risk-Attiudes (no adjustment)

• Significance testing for payoff results from previous batch run data
• Set minimum number of rounds to 50 before checking for convergence

Sim 2. Evolving Risk-Attiudes (with adjustment)

• Tie-breaker: chose randomly when there is a tie for best payoff when selecting neighbor to emulate
• Set minimum number of rounds to 300 before checking for convergence
• Improved logic for sum of agents with changed risk attitudes in the last adjustment round (part of convergence check)

Improvements to custom batch run script

• Improve multiprocessing logic
  • Add a timeout on the wait so main loop doesn't wait forever
  • Cycle workers to avoid memory leaks on long runs
• Shift data collection schedule logic to the model: only collect data needed for convergence checks and requested reporting (agent data optional; end of round or adjustment rounds only)

Miscellaneous

• Jupyter notebooks from prior analysis have been moved to notebooks/archive/
• Static html exports from jupyter notebooks have to docs/analysis/prev/

---

What's Changed. (auto-generated)

• Add citation CFF file for the software by @rlskoeser in #92
• Update charts labels and legends; test cumulative payoff significance by @rlskoeser in #106
• Simulation updates by @rlskoeser in #107
• Use marimo as interface for interactive hawk/dove simulation by @rlskoeser in #108
• Feature/rerun batch analysis by @rlskoeser in #110

Full Changelog: 1.0...1.1.0

v1.0

Update Binder badge link