Scientific Collaboration and Knowledge Production 1.0.0
A holistic, agent-based model of scientific knowledge production to simulate large-scale collaborative research processes. The model represents the scientific process as a stochastic, cooperative task-completion game where agents repeatedly collaborate on projects that can yield publications.
The model is implemented in Python as an environment in the popular Gymnasium reinforcement learning library.
Release Notes
Scientific Knowledge Production Simulation - Initial Release 0.0.1
Features
- Run simulations of collaborative scientific projects.
- Test three predefined policies based on realistic behavioral archetypes found in the literature.
- Calibrate the simulation to real world data
- Train multi-agent reinforcement learning policy functions.
- Inspect resulting outputs