After a growing season is completed, gym-DSSAT provides a visualization interface both for raw state variables or rewards.
To vizualise the time of state variables of a trajectory, you can call:
env.render(type='ts', # time series mode feature_name_1='nstres', # mandatory first raw state variable, here the nitrogen stress factor (unitless) feature_name_2='grnwt') # optional second raw state variable, here the grain weight (kg/ha)
All variables in
env._state are available for rendering. You can find the list and agronomical meaning in gym environment’s yaml configuration file in the
For quick reward inspection, you can use:
env.render(type='reward') # plot reward time series (DOY for Day Of Year)
Simply adding the
cumsum=True argument, you can visualize the (undiscounted) cumulated reward.
env.render(type='reward', cumsum=True) # if you want cumulated rewards