Approximating discrete event simulation by batching Agent function calls

[EwoutH]

Basic idea approximating discrete event simulation by batching Agent function calls:

1. Agent gets scheduled to perform a method. This gets saved somewhere in a fast data structure (I.e. dict)
2. Every X time, these scheduled actions get grouped and applied on the relevant agents in the dataframe
3. All agents perform that function in batched

“If you have a fixed time resolution how is this different from just stepping?”

• You can directly port over your regular Mesa DES model for a performance boost
• It allows agent more natively to do nothing / skip steps
• You can make time steps adaptive, i.e. always running after N scheduled calls

CC @quaquel