def run_large_experiment(): num_dcs = 10 num_customers = 50 num_commodities = 5 orders_per_day = 10 dcs_per_customer = 3 demand_mean = 200 demand_var = 20 num_steps = 30 num_episodes = 5 runner_random = experiment_runner.create_random_experiment_runner( num_dcs, num_customers, dcs_per_customer, demand_mean, demand_var, num_commodities, orders_per_day, num_steps) runner_dqn = experiment_runner.create_dqn_experiment_runner( num_dcs, num_customers, dcs_per_customer, demand_mean, demand_var, num_commodities, orders_per_day, num_steps) runner_zero = experiment_runner.create_alwayszero_experiment_runner( num_dcs, num_customers, dcs_per_customer, demand_mean, demand_var, num_commodities, orders_per_day, num_steps) runner_bestfit = experiment_runner.create_bestfit_experiment_runner( num_dcs, num_customers, dcs_per_customer, demand_mean, demand_var, num_commodities, orders_per_day, num_steps) #runner_random.run_episodes(num_steps, num_episodes, orders_per_day, experiment_name='dumb_agent') runner_dqn.run_episodes(num_steps, num_episodes, orders_per_day, experiment_name='dqn_agent')
def run_bestfit(): print("===RUNNING BESTFIT===") runner_bestfit = experiment_runner.create_bestfit_experiment_runner( num_dcs, num_customers, dcs_per_customer, demand_mean, demand_var, num_commodities, orders_per_day, num_steps) runner_bestfit.run_episodes(num_steps, num_episodes, orders_per_day, experiment_name=f"bestfit_few_warehouses_v2") print("===DONE BESTFIT===")
def run_bestfit(): print("===RUNNING BESTFIT===") reproducibility.set_seeds(0) print("Check this array to ensure reproducibility") print("Reproducibility BESTFIT", np.random.randint(0, 500, size=(5, 1))) runner_bestfit = experiment_runner.create_bestfit_experiment_runner( num_dcs, num_customers, dcs_per_customer, demand_mean, demand_var, num_commodities, orders_per_day, num_steps) runner_bestfit.run_episodes(num_steps, num_episodes, orders_per_day, experiment_name=f"bestfit_few_warehouses_v3") print("===DONE BESTFIT===")
def small_validation_experiments(): num_dcs = 3 num_customers = 2 num_commodities = 4 orders_per_day = 2 dcs_per_customer = 3 demand_mean = 100 demand_var = 20 num_steps = 50 num_episodes = 5 # runner_random = experiment_runner.create_random_experiment_runner(num_dcs, num_customers, dcs_per_customer, # demand_mean, demand_var, num_commodities, # orders_per_day, num_steps) # runner_dqn = experiment_runner.create_dqn_experiment_runner(num_dcs, num_customers, dcs_per_customer, demand_mean, # demand_var, num_commodities, orders_per_day, num_steps) # runner_zero = experiment_runner.create_alwayszero_experiment_runner(num_dcs, num_customers, dcs_per_customer, # demand_mean, # demand_var, num_commodities, orders_per_day, # num_steps) runner_bestfit = experiment_runner.create_bestfit_experiment_runner( num_dcs, num_customers, dcs_per_customer, demand_mean, demand_var, num_commodities, orders_per_day, num_steps) runner_randomvalid = experiment_runner.create_randomvalid_experiment_runner( num_dcs, num_customers, dcs_per_customer, demand_mean, demand_var, num_commodities, orders_per_day, num_steps) # runner = experiment_runner.create_always_first_dc_agent(num_dcs, num_customers,dcs_per_customer,demand_mean,demand_var,num_commodities,orders_per_day,num_steps) # runner_random.run_episodes(num_steps, num_episodes, orders_per_day, experiment_name='dumb_agent_unittest') #runner_dqn.run_episodes(num_steps, num_episodes, orders_per_day, experiment_name='dqn_agent_unittest') # runner_zero.run_episodes(num_steps, num_episodes, orders_per_day, experiment_name='zero_agent_unittest') #runner_bestfit.run_episodes(num_steps, num_episodes, orders_per_day, experiment_name='bestfit_agent_unittest') runner_randomvalid.run_episodes(num_steps, num_episodes, orders_per_day, experiment_name='bestfit_agent_unittest')
num_customers, dcs_per_customer, demand_mean, demand_var, num_commodities, orders_per_day, num_steps, )) # runner_random = experiment_runner.create_random_experiment_runner( # num_dcs, # num_customers, # dcs_per_customer, # demand_mean, # demand_var, # num_commodities, # orders_per_day, # num_steps, # ) runner_bestfit = experiment_runner.create_bestfit_experiment_runner( num_dcs, num_customers, dcs_per_customer, demand_mean, demand_var, num_commodities, orders_per_day, num_steps) # runner_random.run_episodes( # num_steps, num_episodes, orders_per_day, experiment_name=f"dumb_agent_experiment_{i}" # ) runner_bestfit.run_episodes( num_steps, num_episodes, orders_per_day, experiment_name=f"bestfit_agent_experiment_{i}")