def method1_subroutine(model, layer_size, mode):
result = list()
query = 'python3 /home/changseok/layerwise_quantization/slave/main.py -b 1 -m ' + model
for i in reversed(range(23)):
if 0 in mode:
query += ' -wb ' + init_parameter(i + 1, layer_size)
if 1 in mode:
query += ' -fb ' + init_parameter(i + 1, layer_size)
if 2 in mode:
query += ' -pt ' + init_parameter(i + 1, layer_size)
print('running command is : ', query)
result.append(run(query))
print(result[-1])
return result
def method2_subroutine(model, layer_size, mode, order):
result = list()
query = 'python3 /home/changseok/layerwise_quantization/slave/main.py -b 1 -m ' + model
processing_factor = [0 for i in range(layer_size)]
for i in order:
for j in reversed(range(23)):
processing_factor[i] = j + 1
if 0 in mode:
query += ' -wb ' + init_parameter(processing_factor,
layer_size)
if 1 in mode:
query += ' -fb ' + init_parameter(processing_factor,
layer_size)
if 2 in mode:
query += ' -pt ' + init_parameter(processing_factor,
layer_size)
print('running command is : ', query)
result.append(run(query))
print(result[-1])
return result
start_time = time.time() # time the beginning of the simulation
configs_list = []
number_of_runs = 25 #define the number of replications to run each configuration
bounds_list = [[1, 3], [2, 4], [3, 5], [4, 6]]
for bounds in bounds_list:
config = {
"number_of_agents": 1000,
"rate_of_infection_per_contact": 0.03,
"recovery_rate": 0.08,
"incubation_period": 3,
"number_of_hubs": 10,
"degree_of_homophily": 0.89,
"hub_densities": [8, 8, 8, 8, 8, 12, 12, 12, 12, 12],
"hub_sizes": [80, 80, 80, 80, 80, 120, 120, 120, 120, 120],
"infection_costs": [bounds] * 10,
"infection_cost_key": "uniform",
"starting_vaccination_rate": 0.15,
"number_of_seasons": 25,
"vax_choice_key": "seasonal_learning",
"vax_choice_params": {
"discount_factor": 1
},
"log_time_period_data": False
}
configs_list.append(config)
if __name__ == '__main__': # this line of code is needed to make concurrent.futures work on Windows
run(configs_list, number_of_runs)
