Example of output from DLW-paper

Below is an example from the DLW-paper, referred to as the base case.

First, we need to create an object of ezclimate.tree.TreeModel with desicion times at 0, 15, 45, 85, 185, and 285 years from now.

from ezclimate.tree import TreeModel

t = TreeModel(decision_times=[0, 15, 45, 85, 185, 285, 385])

Next we create an ezclimate.bau.DLWBusinessAsUsual object and set up the business as usual emission using the tree structure given by t.

from ezclimate.bau import DLWBusinessAsUsual

bau_default_model = DLWBusinessAsUsual()
bau_default_model.bau_emissions_setup(tree=t)

We move on to create an ezclimate.cost.DLWCost object using the base case parameters.

from ezclimate.cost import DLWCost

c = DLWCost(tree=t, emit_at_0=bau_default_model.emit_level[0], g=92.08, a=3.413, join_price=2000.0,
            max_price=2500.0, tech_const=1.5, tech_scale=0.0, cons_at_0=30460.0)

After this we are ready to create an ezclimate.damage.DLWDamage object and simulate damages using the damage_simulation() method, again using the base case parameters.

from ezclimate.damage import DLWDamage

df = DLWDamage(tree=t, bau=bau_default_model, cons_growth=0.015, ghg_levels=[450, 650, 1000], subinterval_len=5)
df.damage_simulation(draws=4000000, peak_temp=6.0, disaster_tail=18.0, tip_on=True, 
                     temp_map=1, temp_dist_params=None, maxh=100.0, cons_growth=0.015)

We are now ready to initiate the ezclimate.utility.EZUtility object using the above created objects.

from ezclimate.utility import EZUtility

u = EZUtility(tree=t, damage=df, cost=c, period_len=5.0, eis=0.9, ra=7.0, time_pref=0.005)

Next step is to find the optimial mitigation plan using the optimization algorithms found in ezclimate.optimization, and print the Social Cost of Carbon (SCC) given by this mitigation plan.

from ezclimate.optimization import GeneticAlgorithm, GradientSearch
import numpy as np

ga_model = GeneticAlgorithm(pop_amount=150, num_generations=75, cx_prob=0.8, mut_prob=0.5, 
                            bound=2.0, num_feature=63, utility=u, print_progress=True)
gs_model = GradientSearch(var_nums=63, utility=u, accuracy=1e-8, 
                          iterations=200, print_progress=True)
final_pop, fitness = ga_model.run()
sort_pop = final_pop[np.argsort(fitness)][::-1]
m_opt, u_opt = gs_model.run(initial_point_list=sort_pop, topk=1)

print("SCC: ", c.price(0, m_opt[0], 0))

Putting it all together

from ezclimate.tree import TreeModel
from ezclimate.bau import DLWBusinessAsUsual
from ezclimate.cost import DLWCost
from ezclimate.damage import DLWDamage
from ezclimate.utility import EZUtility
from ezclimate.optimization import GeneticAlgorithm, GradientSearch
import numpy as np

def base_case():
	t = TreeModel(decision_times=[0, 15, 45, 85, 185, 285, 385])

	bau_default_model = DLWBusinessAsUsual()
	bau_default_model.bau_emissions_setup(tree=t)

	c = DLWCost(t, bau_default_model.emit_level[0], g=92.08, a=3.413, join_price=2000.0, max_price=2500.0,
					tech_const=1.5, tech_scale=0.0, cons_at_0=30460.0)

	df = DLWDamage(tree=t, bau=bau_default_model, cons_growth=0.015, ghg_levels=[450, 650, 1000], subinterval_len=5)
	df.damage_simulation(draws=4000000, peak_temp=6.0, disaster_tail=18.0, tip_on=True, 
							 temp_map=1, temp_dist_params=None, maxh=100.0)

	u = EZUtility(tree=t, damage=df, cost=c, period_len=5.0, eis=0.9, ra=7.0, time_pref=0.005)

	ga_model = GeneticAlgorithm(pop_amount=150, num_generations=75, cx_prob=0.8, mut_prob=0.5, 
	                            bound=1.5, num_feature=63, utility=u, print_progress=True)
	gs_model = GradientSearch(var_nums=63, utility=u, accuracy=1e-8, 
	                          iterations=200, print_progress=True)

	final_pop, fitness = ga_model.run()
	sort_pop = final_pop[np.argsort(fitness)][::-1]
	m_opt, u_opt = gs_model.run(initial_point_list=sort_pop, topk=1)

	print("SCC: ", c.price(0, m_opt[0], 0))

if __name__ == "__main__":
	base_case()