Spatial Agent-Based Model
In the agent-based simulation model, there are r regions (‘land’ cells as opposed to ‘sea’ cells), m knowledge sections, n agents per region, and a bipartite matrix assigning knowledge sections to agents (as their specialization). In the present simulation study, each agent is specialized in only one of eight knowledge sections, each associated with one of the eight patent IPC sections. Each firm starts with one of the root knowledge units (i.e. one of the ‘nodes’ at the bottom in Fig. 5. In each period of the simulation, each agent conducts knowledge search as described in Sect. 3.2. There is no economic principle (e.g. a demand market) that rewards or paces knowledge discovery. Moreover, there is no entry or exit of agents. Finally, the circular layout of regions is fixed for all simulations. The model is deliberately kept this simple to get undistorted observations of the role of the network structure and knowledge distribution on the advancement of technological knowledge.