This PR makes a few small additions and changes to the Monte Carlo simulation functionality.
It introduces a new class, MonteCarloSimulator, which is a lot like the AgentTypeMonteCarloSimulator, except that this doesn't make any assumptions about birth/mortality, and it doesn't have the 'read_shocks" functionality. This is a more stripped down version of what's already there.
It introduces some duplicate code, and maybe there's a way to make the implementation more efficient by thinking about what sort of abstract or super- class a Simulator should be. This will require some thinking about architecture which I think it would make sense to work with @mnwhite on
It makes a small change so that reward variables (i.e. u) get computed in simulations, and added into the history. This is useful when trying to compare the outcomes of agents across different models and parameterizatiosn.
The apply_fun_to_vals function compartmentalizes a coding pattern used widely in the Monte Carlo Simulator code. I didn't work it through the library yet since it replaces just one line of code with something arguably more readable.
[x] Tests for new functionality/models or Tests to reproduce the bug-fix in code.
[x] Updated documentation of features that add new functionality.
This PR makes a few small additions and changes to the Monte Carlo simulation functionality.
It introduces a new class,
MonteCarloSimulator, which is a lot like theAgentTypeMonteCarloSimulator, except that this doesn't make any assumptions about birth/mortality, and it doesn't have the 'read_shocks" functionality. This is a more stripped down version of what's already there.It makes a small change so that reward variables (i.e.
u) get computed in simulations, and added into thehistory. This is useful when trying to compare the outcomes of agents across different models and parameterizatiosn.The
apply_fun_to_valsfunction compartmentalizes a coding pattern used widely in the Monte Carlo Simulator code. I didn't work it through the library yet since it replaces just one line of code with something arguably more readable.