def test_that_specified_temp_performs_same_as_constant_temperature(self):
temperatures = np.ones(len(self.times)) * self.temperature
outputs_specs = simulation.run_simulation(
self.solution,
self.times,
condition_type='specified-temperature-constant-volume',
output_species=False,
output_reactions=False,
output_directional_reactions=False,
output_rop_roc=False,
temperature_values=temperatures)
self.solution.TPX = self.temperature, self.pressure, self.mole_fractions
outputs_const = simulation.run_simulation(
self.solution,
self.times,
condition_type='constant-temperature-and-volume',
output_species=False,
output_reactions=False,
output_directional_reactions=False,
output_rop_roc=False)
print(outputs_specs)
print(outputs_const)
# make sure all are within 1000
self.compare_data_frames(outputs_const['conditions'],
outputs_specs['conditions'])
def test_specified_temp_over_different_time_steps(self):
"""compare the first iteration of one simulation with the last iteration
of one taking smaller steps"""
temperatures = np.ones(len(self.times)) * self.temperature
outputs1 = simulation.run_simulation(
self.solution,
self.times,
condition_type='specified-temperature-constant-volume',
output_species=False,
output_reactions=False,
output_directional_reactions=False,
output_rop_roc=False,
temperature_values=temperatures)
times2 = np.linspace(0, self.times[1])
temperatures2 = np.ones(len(times2)) * temperatures[0]
self.solution.TPX = self.temperature, self.pressure, self.mole_fractions
outputs2 = simulation.run_simulation(
self.solution,
times2,
condition_type='specified-temperature-constant-volume',
output_species=False,
output_reactions=False,
output_directional_reactions=False,
output_rop_roc=False,
temperature_values=temperatures2)
self.compare_data_frames(outputs1['conditions'].iloc[1, :],
outputs2['conditions'].iloc[-1, :])
def table_2(plot=False, verbose=False):
print '%table 2'
print '%', dt.datetime.now()
import os
os.system('git rev-parse HEAD')
for load in ['day', 'night', 'continuous', 'lighting', 'freezer']:
for battery in [lead_dict]:
sim.run_simulation(battery, inverter_type='typical', load_type=load, plot=plot, verbose=verbose)
def test_run_simulation():
"""Test that all particles are updated the correct amount of times."""
graph = network.new_network(100, 50)
simulation.init_network(graph, 0.5)
def update_particle(graph, node, timestep):
if "updates" not in graph.node[node]:
graph.node[node]["updates"] = -1
graph.node[node]["updates"] += 1
assert timestep == graph.node[node]["updates"]
simulation.run_simulation(graph, update_particle, 10, print_=False)
def get_data(args):
""" Read in a dataset and make sure it has fields
text, T_true, T_proxy, C_true, Y_sim
"""
if args.simulate:
# Add columns T_true T_proxy C_true Y_sim to the data
df = pd.read_csv(args.data, sep='\t', error_bad_lines=False)
df['text'] = df['text'].map(lambda x: x.lower()
if isinstance(x, str) else x)
df = simulation.run_simulation(
df,
propensities=[args.p1, args.p2] if args.p1 > 0 else None,
precision=args.pre,
recall=args.rec,
b0=args.b0,
b1=args.b1,
offset=args.off,
gamma=args.gamma,
accuracy=args.acc,
proxy_type=args.ptype,
size=args.size)
# df2 = df[['text', 'Y_sim', 'C_true', 'T_proxy']]
# df2.to_csv('music_complete.tsv', sep='\t'); quit()
else:
# use what's given without any changes
# (T_true, T_proxy, C_true, and Y should already be in there)
df = pd.read_csv(args.data, sep='\t', error_bad_lines=False)
df['text'] = df['text'].map(lambda x: x.lower()
if isinstance(x, str) else x)
df['Y_sim'] = df['Y']
df['C_true'] = df['C']
return df
def param_test(env, episodes, agent, params, verbose=False):
"""
Runs simulation under given conditions and tests specified range of
hyperparameters, tracks total reward from each episode and returns
list of those values.
"""
feat_count = env.env.observation_space.shape[0]
act_count = env.env.action_space.n
# track each params rewards and process a new learner per param
param_rewards = []
for param_value in params:
if verbose:
print('Processing param: ' + str(param_value))
# build the learner to utilize in the simulation, update the param value
# being adjust in the simulation here
learner = agent(num_feat=feat_count,
num_acts=act_count,
alpha=param_value,
gamma=0.99,
epsilon=0.99,
epsilon_decay=0.99)
# run simulation and track the returned rewards
rewards = run_simulation(env, episodes, learner, verbose=verbose)
# only track every 10th reward for cleaner plots
param_rewards.append(rewards[::10] + [rewards[-1]])
return param_rewards
def run_model(file_or_directory, param_map={}):
# Determine if input is single or multi run.
if os.path.isfile(file_or_directory):
# Single run mode, open file and run.
logger.debug("Importing file: {}".format(file_or_directory))
input_file = open(file_or_directory, "r")
# Parse into dataset.
dataset = parse(input_file)
logger.debug("Built dataset: {}".format(dataset.get("name")))
for attribute in sorted(dataset.to_dict().keys()):
logger.debug(" {} = {}".format(attribute, dataset.get(attribute)))
# Parse model parameters and call model.
result = None
if param_map:
logger.debug("Running model with custom parameters: {}".format(param_map))
result = run_simulation(dataset, **param_map)
else:
logger.debug("Running model with default parameters.")
result = run_simulation(dataset)
# Run complete, return results.
logger.info("Completed simulation run: {} ({} - {})".format(dataset.get("name"),
dataset.get("start_year"), dataset.get("end_year")))
return result
elif os.path.isdir(file_or_directory):
# Gather files an directories, and perform a recursive descent
# into the contents, collecting results from runs.
root_path = os.path.abspath(file_or_directory)
dir_files = os.listdir(file_or_directory)
dir_results = []
for dir_file in dir_files:
# Merge results, which may be lists of results, into a single list.
abs_path = "{}/{}".format(root_path, dir_file)
dir_results += [run_model(abs_path, param_map)]
logger.info("Processed {} dataset runs.".format(len(dir_results)))
return dir_results
else:
# Not a file or directory, exit.
raise Exception("Not a file or directory: {}".format(file_or_directory))
def evaluate_controller(self, controller, sleep_seconds=0, graphics=True, debug=False, print_results=False, _chromosome_controller=False):
""" Changes the state of the scenario.
To use the same scenario for several controllers, make deepcopies of it.
"""
recorder = run_simulation(controller, self.floors, _sleep_seconds=sleep_seconds,
number_of_steps=self.number_of_steps, visualization=graphics, print_results=print_results, debug=debug,
_chromosome_controller=_chromosome_controller)
fitness = get_recorder_fitness(recorder)
return fitness
def test_detour_at_obstacle():
"""Test that the DO particle update is applied correctly."""
graph = nx.Graph()
graph.add_nodes_from([1, 2, 3, 4, 5, 6])
graph.add_edges_from([(1, 5), (1, 2), (1, 3), (1, 4), (4, 3), (3, 2),
(2, 6)])
simulation.init_network(graph, f=0)
graph.node[1]["particles"] = [simulation.Particle(1, 0, 6)]
graph.node[2]["particles"] = [simulation.Particle(2, 0, 5)]
graph.node[3]["particles"] = [simulation.Particle(3, 0, 5)]
expected_id = graph.node[1]["particles"][0].id
# Run the simulation on predictable order of nodes.
simulation.run_simulation(graph,
simulation.detour_at_obstacle,
1,
order=simulation.SimOrder.Increasing)
# We expect the particle at either node 4 or 5.
expected_at = list(
map(lambda p: p[0].id
if p else None, [graph.node[n]["particles"] for n in [4, 5]]))
assert expected_id in expected_at
def test_simulations(self):
try:
run_simulation(self.parameters, self.outfolder, solver="trajectory_recovery")
run_simulation(self.parameters, self.outfolder, solver="weighted_trajectory_recovery")
run_simulation(self.parameters, self.outfolder, solver="semidef_relaxation_noiseless")
except RuntimeError as e:
self.fail("run_simulation raised exception: " + str(e))
def test_constant_temperature_and_volume(self):
outputs = simulation.run_simulation(
self.solution,
self.times,
condition_type='constant-temperature-and-volume',
output_species=False,
output_reactions=False,
output_directional_reactions=False,
output_rop_roc=False)
conditions = outputs['conditions']
temperatures = conditions['temperature (K)'].values
pressures = conditions['pressure (Pa)'].values
volumes = conditions['volume (m3)'].values
int_energies = conditions['internal energy (J/kg)']
print(conditions)
self.assertNotAlmostEqual(min(pressures) / max(pressures), 1)
self.assertNotAlmostEqual(min(int_energies) / max(int_energies), 1)
self.assertAlmostEqual(min(volumes) / max(volumes), 1)
self.assertAlmostEqual(min(temperatures) / max(temperatures), 1)
self.assertAlmostEqual(temperatures[0], self.temperature)
def index(request):
form = CalculatorForm()
output = None
cycles = None
if request.method == 'POST':
submitted = True
form = CalculatorForm(request.POST)
print "posted"
print form.__dict__
if form.is_valid():
output = run_simulation(form)
cycles = len(output)
else:
submitted = False
return render(request, 'calculator.html', {
'output': output,
'cycles': cycles,
'form': form,
'submitted': submitted
})
def test_simulation():
stages = [
Stage("Build & Unit Test",
duration=timedelta(minutes=10),
failure_rate=0.02),
Stage("System Test", duration=timedelta(minutes=20), failure_rate=0.2),
Stage("Performance Test",
duration=timedelta(minutes=120),
failure_rate=0.1,
single_threaded=True),
]
start_time = datetime(year=2017, month=12, day=11, hour=8)
commits = generate_commits(100, start_time, offset=2000, max_interval=100)
deployer = Deployer(duration=timedelta(minutes=2),
deploy_policy=DeployPolicy.EveryPassing)
runs = run_simulation(start_time,
stages,
commits=commits,
deployer=deployer)
assert len(runs) <= 100
def table_5(plot=False, verbose=False):
print '% table 5'
print '%', dt.datetime.now()
import os
os.system('git rev-parse HEAD')
output = []
both_index = []
load_index = []
battery_index = []
for load in ['day', 'night', 'continuous', 'lighting', 'freezer']:
for battery in [lead_dict, lith_dict, pbc_dict]:
d = sim.run_simulation(battery, inverter_type='typical', load_type=load, plot=plot, verbose=verbose)
d['battery_type'] = battery['type']
d['load_type'] = load
output.append(d)
both_index.append(load + ' ' + battery['type'])
load_index.append(load)
battery_index.append(battery['type'])
print
df = p.DataFrame(output)
return df
def gossip_convergence(simParams):
"""Plot the convergence of gossip over different simulation parameters."""
results = {} # <nodes: (expected_result, result_per_timestep)>
# For each given simulation parameters run the simulation.
for simParam in simParams:
# Setup the initial graph.
G = graph.new_network(simParam.network_size, simParam.avg_degree)
simulation.init_graph(
G,
max_node_particles=simParam.max_node_particles,
initial_data_f=lambda: random.uniform(1, sys.maxsize))
# Calculate expected result from initial graph state.
expected_result = simParam.expected_result_f(
[node_data["data"] for _, node_data in G.nodes(data=True)])
# Run the simulation and save collected data.
collected_data = simulation.run_simulation(
G,
update_f=lambda g, n, t: particle_dynamics.gossip(
g,
n,
t,
k_neighbors=simParam.gossip_k,
state_update=simParam.state_update_f),
timesteps=simParam.max_timesteps,
collect=collect.avg_node_data)
results[simParam.label] = (expected_result, collected_data)
# Finally plot the results for each simulation parameter.
for label, (expected_result, result_per_timestep) in results.items():
error_at_timestep = np.abs(expected_result - list(
result_per_timestep.values())) / expected_result
plt.semilogy(error_at_timestep, label=label)
plt.legend()
plt.show()
def test_specified_temperature_constant_volume_changing_T(self):
temperatures = np.linspace(900, 1000, len(self.times))
print temperatures
outputs = simulation.run_simulation(
self.solution,
self.times,
condition_type='specified-temperature-constant-volume',
output_species=False,
output_reactions=False,
output_directional_reactions=False,
output_rop_roc=False,
temperature_values=temperatures)
conditions = outputs['conditions']
temperatures = conditions['temperature (K)'].values
pressures = conditions['pressure (Pa)'].values
volumes = conditions['volume (m3)'].values
int_energies = conditions['internal energy (J/kg)']
print(conditions)
self.assertNotAlmostEqual(min(int_energies) / max(int_energies), 1)
self.assertNotAlmostEqual(min(temperatures) / max(temperatures), 1)
self.assertAlmostEqual(min(temperatures), 900)
self.assertAlmostEqual(max(temperatures), 1000)
self.assertAlmostEqual(min(volumes) / max(volumes), 1)
#agent_setup = [{"selfish":0.8, "ratio":0.6}]
agent_setup = []
trust_used = False
inbox_trust_sorted = False
trust_filter_on = False
# for (trust_used, trust_filter_on, inbox_trust_sorted) in \
# [(False,False,False), (True, True, False), (True, True, True)]:
#for (trust_used, trust_filter_on, inbox_trust_sorted) in \
# [(False,False,False)]:
for spamminess in [0]:
results = sim.run_simulation(NUM_FACTS=num_facts, \
NUM_NOISE=num_noise, \
NUM_AGENTS=num_agents, \
AGENT_PER_FACT=agent_per_fact, \
CONNECTION_PROBABILITY=connection_probability, \
NUM_STEPS=num_steps, \
WILLINGNESS=willingness, \
COMPETENCE=competence, \
NUM_TRIAL=num_trial,\
GRAPH_TYPE=graph_type, \
AGENT_SETUP=agent_setup, \
SPAMMINESS=spamminess, \
SELFISHNESS=selfishness, \
TRUST_USED=trust_used, \
INBOX_TRUST_SORTED=inbox_trust_sorted,\
TRUST_FILTER_ON=trust_filter_on)
print results
print_results(results)
prep_time is exp(40)
rec_time exp(40)
number of preparation rooms is either 4
number of recovery rooms is either 5
"""
firstExperiment = []
print("\n\nSimulation #1: \n")
for k in range(1, 11):
interarrival_time = randomize.exponential(25)
prep_time = randomize.exponential(40)
rec_time = randomize.exponential(40)
results = simulation.run_simulation(4, 5, interarrival_time, prep_time,
rec_time, simulation_length, k, False)
firstExperiment = firstExperiment + [results[0]]
#avg = stats.mean(results[0])
#conf = stats.confidence95(results[0])
#lower_bound = conf[0]
#upper_bound = conf[1]
#print("The average preparation queue length is %.2f with a 95%% confidence interval of [%.2f , %.2f]." % (avg, lower_bound, upper_bound))
"""
Simulation #2 - interarrival time exponentially distributed, others uniformly
interarrival_time is exp(25)
prep_time is unif(30,50)
rec_time unif(30,50)
top_node_percent = 0.01
# In[6]:
# prevent pickling error in multiprocessing
def defaultdict_using_list_func():
return defaultdict(list)
def earlist_date_func():
return datetime(1970, 1, 1)
# In[13]:
from simulation import run_simulation
# some test
result = run_simulation(retweets,
T_window=T_window,
top_node_percent=top_node_percent,
update_interval=update_interval,
incremental=True,
top_k=10,
top_k_computation_interval=timedelta(minutes=60),
min_rwc_score=0.85,
head_n=12 * 1e5,
return_graph=True)
pickle.dump(result, open('output/simulation_result.pkl', 'wb'))
def simulate():
try:
run_simulation(False)
except Exception as e:
print('oops ', e)
def f(params):
outfolder = 'results/{}/'.format(params['key'])
run_simulation(params, outfolder, solver=params['solver'], verbose=True)
for i in range(2):
office_2.add_worker(Worker('A'))
office_2.add_worker(Worker('B'))
office_2.add_worker(Worker('C'))
office_2.add_worker(Worker('E'))
queue_1 = SinglyLinkedList()
queue_2 = SinglyLinkedList()
for i in range(30):
queue_1.add(Client())
queue_2.add(Client())
simulation.run_simulation(office_1, queue_1)
simulation.run_simulation(office_2, queue_2)
office_1.print_summary()
office_2.print_summary()
del office_1
del office_2
del queue_1
del queue_2
# Część 2ga
# Narysować histogram dla 100 kolejek
office_1_times = []
PATH = sys.argv[1] if len(sys.argv) == 2 else str(Path(__file__).parent.absolute())[:-3]+'imgs/'
NAME = 'huge'
fst = Param(
CLUSTERS = 12**2,
CLUST_SIDE_LEN = 7,
DAYS = 100,
INFECTION_TIME = 20,
DEATH_RATE = 0.4,
INFECT_RATE = 0.3,
MIGRATIONS_PER_DAY = 100,
FEAR_RATE = 0.4,
HEALTHCARE_CAPACITY = 2500
)
start = time()
cumulative, active, healed, dead, migrations, real_mig = run_simulation(fst, PATH)
print('time elapsed:', time() - start, flush=True)
make_gif(PATH, NAME)
# plt.plot(range(len(migrations)), migrations, label='migrations', color='blue')
plt.plot(range(len(real_mig)), real_mig, label='migrations', color='blue', ls=':')
make_plot(fst, cumulative, active, healed, dead, label='', color='red')
plt.savefig(PATH+NAME+'.png')
print("The visualization of the simulation is available at './imgs/'", flush=True)
#agent_setup = [{"selfish":0.8, "ratio":0.6}]
agent_setup = []
trust_used = False
inbox_trust_sorted = False
trust_filter_on = False
# for (trust_used, trust_filter_on, inbox_trust_sorted) in \
# [(False,False,False), (True, True, False), (True, True, True)]:
#for (trust_used, trust_filter_on, inbox_trust_sorted) in \
# [(False,False,False)]:
for spamminess in [0]:
results = sim.run_simulation(NUM_FACTS=num_facts, \
NUM_NOISE=num_noise, \
NUM_AGENTS=num_agents, \
AGENT_PER_FACT=agent_per_fact, \
CONNECTION_PROBABILITY=connection_probability, \
NUM_STEPS=num_steps, \
WILLINGNESS=willingness, \
COMPETENCE=competence, \
NUM_TRIAL=num_trial,\
GRAPH_TYPE=graph_type, \
AGENT_SETUP=agent_setup, \
SPAMMINESS=spamminess, \
SELFISHNESS=selfishness, \
TRUST_USED=trust_used, \
INBOX_TRUST_SORTED=inbox_trust_sorted,\
TRUST_FILTER_ON=trust_filter_on)
print results
print_results(results)
stages = [
Stage("Commit Stage", duration=timedelta(minutes=3), failure_rate=0.005),
Stage("Automated Acceptance Test",
duration=timedelta(minutes=20),
failure_rate=0.01),
Stage("Performance Test",
duration=timedelta(minutes=20),
failure_rate=0.01),
Stage("Internal Release",
duration=timedelta(minutes=4),
failure_rate=0.01,
single_threaded=True),
]
start_time = datetime(year=2017, month=12, day=11, hour=8)
commits = generate_commits(100, start_time, offset=2000, max_interval=100)
deployer = Deployer(duration=timedelta(minutes=4),
deploy_policy=DeployPolicy.OnceADay,
deploy_hour=17,
deploy_day=6)
runs = run_simulation(start_time, stages, commits=commits, deployer=deployer)
print_runs("simulation_farley", stages, runs)
metrics_calc = MetricsCalculator(runs)
metrics = metrics_calc.metrics()
print_metrics("simulation_farley", metrics)
print(metrics.pretty_print())
def f(params):
if params["outfolder"] != '':
outfolder = f'{params["outfolder"]}/{params["key"]}/'
else:
outfolder = None
run_simulation(params, outfolder, solver=params['solver'], verbose=True)
]
# Iterate over random seed
for i in range(5):
# bootstrap
np.random.seed(seed)
# Set a particular evaluation for the seed
seed_evaluation = evaluations[seed - seed_add]
# Iterate over trajectories of the given number to run simulation
for trajectory_index in range(
num_trajectories_list[num_trajectories_index - 1],
num_trajectories_list[num_trajectories_index]):
print(
f"seed:{seed - seed_add} / trajectory: {seed_evaluation.num_trajectories}",
end='\r')
# Run simulation and get trace & true performances of every target policy
total_latency_for_each_policy, trace = \
run_simulation(policies=policies, num_requests=num_requests, servers=servers)
# Get the estimate of every target policy on the trajectory.
# After finishing iterating over every random seed,
# display the estimate of each target policy for the given number of trajectories
for seed_evaluation in evaluations:
"""
Implement
"""
# ma być: 3xA, 3xB, 3xC, 1xE
for i in range(3):
office.add_worker(Worker('A'))
office.add_worker(Worker('B'))
office.add_worker(Worker('C'))
office.add_worker(Worker('E'))
for i in range(30):
c = Client()
print(c, 'lines up')
queue.add(c)
simulation.run_simulation(office, queue)
office.print_summary()
'''
# Propozycja usprawnienia:
Wykorzystanie stringów określających typ pracownika
i porównywanie tych stringów z typem zadania nie jest
wydajne, porównania łańcuchów znaków są z natury powolne,
nawet jesli mówimy o łańcuchach jednoznakowych
Szybszą metodą byłoby wykorzystanie zestawu flag bitowych
określających typ zadania: na przykład
```
A = 0b001
def main():
# Points for graphing data
points_g2_a = []
points_g2_b = []
points_g3 = [(0, 0)]
# Run independent matching trial
print('\nMATCHING WITHOUT PARTNER DEPENDENCE')
player1, player2, num_turns, debug = setup(0)
m0_p1_a, m0_p1_b, m0_p2_a, m0_p2_b = run_simulation(player1, player2, num_turns, debug,
points_g2_a, points_g2_b, 0)
show_info(player1, player2, num_turns)
''' GRAPH SETUP '''
# Set up graph 1a
plt.figure(1)
plt.title('Probability of choosing A as a function\nPlayer 1 (0.6, 0.6)')
plt.xlabel('Fractional income of A')
plt.ylabel('Ratio for A choices')
plt.xlim(0, 1)
plt.ylim(0, 1)
t1_patch = mpatches.Patch(color='red', label='Independent')
t2_patch = mpatches.Patch(color='blue', label=str(player2.weight) + ' Weighting')
t3_patch = mpatches.Patch(color='green', label='Joint')
plt.legend(handles=[t1_patch, t2_patch, t3_patch])
# Set up graph 1b
plt.figure(2)
plt.title('Probability of choosing A as a function\nPlayer 1 (0.2, 0.8)')
plt.xlabel('Fractional income of A')
plt.ylabel('Ratio for A choices')
plt.xlim(0, 1)
plt.ylim(0, 1)
t1_patch = mpatches.Patch(color='red', label='Independent')
t2_patch = mpatches.Patch(color='blue', label=str(player2.weight) + ' Weighting')
t3_patch = mpatches.Patch(color='green', label='Joint')
plt.legend(handles=[t1_patch, t2_patch, t3_patch])
# Set up graph 2a
plt.figure(3)
plt.title('Change in Income over time for Player 2, A' +
'\nPlayer 1 (' + str(player1.prob_a) + ', ' + str(player1.prob_b) + '); ' +
'Player 2 (' + str(player2.prob_a) + ', ' + str(player2.prob_b) + ')')
plt.xlabel('Time')
plt.ylabel('Income')
plt.xlim(0, num_turns)
plt.ylim(0, 1)
t1_patch = mpatches.Patch(color='red', label='Independent')
t2_patch = mpatches.Patch(color='blue', label=str(player2.weight) + ' Weighting')
t3_patch = mpatches.Patch(color='green', label='Joint Decision History')
true_patch = mpatches.Patch(color='black', label='P(A)/[P(A) + P(B)]: '
+ str(player2.prob_a / (player2.prob_a + player2.prob_b)))
plt.legend(handles=[t1_patch, t2_patch, t3_patch, true_patch])
# Set up graph 2b
plt.figure(4)
plt.title('Change in income over time for Player 2, B' +
'\nPlayer 1 (' + str(player1.prob_a) + ', ' + str(player1.prob_b) + '); ' +
'Player 2 (' + str(player2.prob_a) + ', ' + str(player2.prob_b) + ')')
plt.xlabel('Time')
plt.ylabel('Income')
plt.xlim(0, num_turns)
plt.ylim(0, 1)
t1_patch = mpatches.Patch(color='red', label='Independent')
t2_patch = mpatches.Patch(color='blue', label=str(player2.weight) + ' Weighting')
t3_patch = mpatches.Patch(color='green', label='Joint')
true_patch = mpatches.Patch(color='black', label='P(B)/[P(A) + P(B)]: ' +
str(player2.prob_b / (player2.prob_a + player2.prob_b)))
plt.legend(handles=[t1_patch, t2_patch, t3_patch, true_patch])
# Set up graph 3
plt.figure(5)
plt.title('Cumulative choices' +
'\nPlayer 1 (' + str(player1.prob_a) + ', ' + str(player1.prob_b) + '); ' +
'Player 2 (' + str(player2.prob_a) + ', ' + str(player2.prob_b) + ')')
plt.xlabel('Cumulative A choices')
plt.ylabel('Cumulative B choices')
t1_patch = mpatches.Patch(color='red', label='Independent')
t2_patch = mpatches.Patch(color='blue', label=str(player2.weight) + ' Weighting')
t3_patch = mpatches.Patch(color='green', label='Joint')
slope_patch = mpatches.Patch(color='black', label='y = x')
plt.legend(handles=[t1_patch, t2_patch, t3_patch, slope_patch])
plt.plot(range(0, num_turns), range(0, num_turns), color='black', linewidth=3)
''' FEED DATA INTO GRAPHS '''
# 2a
plt.figure(3)
plt.plot([point[1] for point in points_g2_a], [point[0] for point in points_g2_a], color='red', linewidth=2)
plt.hlines(player2.prob_a / (player2.prob_a + player2.prob_b), 0, num_turns, colors='black', linewidth=4)
points_g2_a = [] # Reset
# 2b
plt.figure(4)
plt.plot([point[1] for point in points_g2_b], [point[0] for point in points_g2_b], color='red', linewidth=2)
plt.hlines(player2.prob_b / (player2.prob_a + player2.prob_b), 0, num_turns, colors='black', linewidth=4)
points_g2_b = [] # Reset
# Get choices for 3
for i in range(2, len(player2.history)):
if player2.history[i][0] == 'A':
points_g3.append((points_g3[-1][0] + 1, points_g3[-1][1]))
else:
points_g3.append((points_g3[-1][0], points_g3[-1][1] + 1))
# 3
plt.figure(5)
graph3_max_x = max(point[0] for point in points_g3)
graph3_max_y = max(point[1] for point in points_g3)
plt.plot([point[0] for point in points_g3], [point[1] for point in points_g3], color='red')
points_g3 = [(0, 0)]
# 1a
graph1_points = []
iterations = np.linspace(0.0, 1.0, num=21)
plt.figure(1)
# Trial 1
for step in iterations:
player1 = Player(1, 0, 0, 0.6, 0.6)
player2 = Player(2, 0, 0.25, step, 1.0 - step)
player1.other_player = player2
player2.other_player = player1
run_simulation(player1, player2, num_turns, debug, [], [], 1)
fractional_income = player2.prob_a / (player2.prob_a + player2.prob_b)
prob_choice = player2.times_picked_a / (player2.times_picked_a + player2.times_picked_b)
graph1_points.append((fractional_income, prob_choice))
plt.plot([point[0] for point in graph1_points], [point[1] for point in graph1_points], color='red')
# Trial 2
graph1_points = []
for step in iterations:
player1 = Player(1, 0, 0, 0.6, 0.6)
player2 = Player(2, 1, 0.25, step, 1.0 - step)
player1.other_player = player2
player2.other_player = player1
run_simulation(player1, player2, num_turns, debug, [], [], 1)
fractional_income = player2.prob_a / (player2.prob_a + player2.prob_b)
prob_choice = player2.times_picked_a / (player2.times_picked_a + player2.times_picked_b)
graph1_points.append((fractional_income, prob_choice))
plt.plot([point[0] for point in graph1_points], [point[1] for point in graph1_points], color='blue')
# Trial 3
graph1_points = []
for step in iterations:
player1 = Player(1, 0, 0, 0.6, 0.6)
player2 = Player(2, 2, 0.25, step, 1.0 - step)
player1.other_player = player2
player2.other_player = player1
run_simulation(player1, player2, num_turns, debug, [], [], 1)
fractional_income = player2.prob_a / (player2.prob_a + player2.prob_b)
prob_choice = player2.times_picked_a / (player2.times_picked_a + player2.times_picked_b)
graph1_points.append((fractional_income, prob_choice))
plt.plot([point[0] for point in graph1_points], [point[1] for point in graph1_points], color='green')
# 1b
graph1_points = []
plt.figure(2)
# Trial 1
for step in iterations:
player1 = Player(1, 0, 0, 0.2, 0.8)
player2 = Player(2, 0, 0.25, step, 1.0 - step)
player1.other_player = player2
player2.other_player = player1
run_simulation(player1, player2, num_turns, debug, [], [], 1)
fractional_income = player2.prob_a / (player2.prob_a + player2.prob_b)
prob_choice = player2.times_picked_a / (player2.times_picked_a + player2.times_picked_b)
graph1_points.append((fractional_income, prob_choice))
plt.plot([point[0] for point in graph1_points], [point[1] for point in graph1_points], color='red')
# Trial 2
graph1_points = []
for step in iterations:
player1 = Player(1, 0, 0, 0.2, 0.8)
player2 = Player(2, 1, 0.25, step, 1.0 - step)
player1.other_player = player2
player2.other_player = player1
run_simulation(player1, player2, num_turns, debug, [], [], 1)
fractional_income = player2.prob_a / (player2.prob_a + player2.prob_b)
prob_choice = player2.times_picked_a / (player2.times_picked_a + player2.times_picked_b)
graph1_points.append((fractional_income, prob_choice))
plt.plot([point[0] for point in graph1_points], [point[1] for point in graph1_points], color='blue')
# Trial 3
graph1_points = []
for step in iterations:
player1 = Player(1, 0, 0, 0.2, 0.8)
player2 = Player(2, 2, 0.25, step, 1.0 - step)
player1.other_player = player2
player2.other_player = player1
run_simulation(player1, player2, num_turns, debug, [], [], 1)
fractional_income = player2.prob_a / (player2.prob_a + player2.prob_b)
prob_choice = player2.times_picked_a / (player2.times_picked_a + player2.times_picked_b)
graph1_points.append((fractional_income, prob_choice))
plt.plot([point[0] for point in graph1_points], [point[1] for point in graph1_points], color='green')
# Run weighted matching trial
print('\nPLAYER 2 TAKES PLAYER 1\'S DECISIONS INTO ACCOUNT. WEIGHS OWN DECISIONS MORE')
print(str(player2.weight))
player1, player2, num_turns, debug = setup(1)
m1_p1_a, m1_p1_b, m1_p2_a, m1_p2_b = run_simulation(player1, player2, num_turns, debug, points_g2_a, points_g2_b, 0)
show_info(player1, player2, num_turns)
''' FEED DATA INTO GRAPHS '''
# 2a
plt.figure(3)
plt.plot([point[1] for point in points_g2_a], [point[0] for point in points_g2_a], color='blue', linewidth=2)
points_g2_a = []
# 2b
plt.figure(4)
plt.plot([point[1] for point in points_g2_b], [point[0] for point in points_g2_b], color='blue', linewidth=2)
points_g2_b = []
# Get choices for 3
for i in range(2, len(player2.history)):
if player2.history[i][0] == 'A':
points_g3.append((points_g3[-1][0] + 1, points_g3[-1][1]))
else:
points_g3.append((points_g3[-1][0], points_g3[-1][1] + 1))
# 3
plt.figure(5)
graph3_max_x_2 = max(point[0] for point in points_g3)
graph3_max_x = max(graph3_max_x_2, graph3_max_x)
graph3_max_y_2 = max(point[1] for point in points_g3)
graph3_max_y = max(graph3_max_y_2, graph3_max_y)
plt.plot([point[0] for point in points_g3], [point[1] for point in points_g3], color='blue')
points_g3 = [(0, 0)]
# Run unweighted matching trial
print('\nPLAYER 2 JOINTLY TAKES PLAYER 1\'S DECISIONS INTO ACCOUNT.')
player1, player2, num_turns, debug = setup(2)
m2_p1_a, m2_p1_b, m2_p2_a, m2_p2_b = run_simulation(player1, player2, num_turns, debug, points_g2_a, points_g2_b, 0)
show_info(player1, player2, num_turns)
''' FEED DATA INTO GRAPHS '''
# 2a
plt.figure(3)
plt.plot([point[1] for point in points_g2_a], [point[0] for point in points_g2_a], color='green', linewidth=2)
# 2b
plt.figure(4)
plt.plot([point[1] for point in points_g2_b], [point[0] for point in points_g2_b], color='green', linewidth=2)
# Get choices for 3
for i in range(2, len(player2.history)):
if player2.history[i][0] == 'A':
points_g3.append((points_g3[-1][0] + 1, points_g3[-1][1]))
else:
points_g3.append((points_g3[-1][0], points_g3[-1][1] + 1))
# 3
plt.figure(5)
graph3_max_x_2 = max(point[0] for point in points_g3)
graph3_max_x = max(graph3_max_x_2, graph3_max_x)
graph3_max_y_2 = max(point[1] for point in points_g3)
graph3_max_y = max(graph3_max_y_2, graph3_max_y)
plt.plot([point[0] for point in points_g3], [point[1] for point in points_g3], color='green')
plt.xlim(0, graph3_max_x)
plt.ylim(0, graph3_max_y)
# Prints out summary of income values
print('\n\n***** SUMMARY (' + str(num_turns) + ' RUNS) *****')
print('\t\tP2 true values:')
print('\t\t\t' + 'a: ' + str(player2.prob_a / (player2.prob_a + player2.prob_b)))
print('\t\t\t' + 'b: ' + str(player2.prob_b / (player2.prob_a + player2.prob_b)))
print('\nTRIAL 1: Player 2 ignores Player 1\'s actions:')
print('\tP2\'s Button A Income Rate: ' + str(round(m0_p2_a, 2)))
print('\tP2\'s Button B Income Rate: ' + str(round(m0_p2_b, 2)))
if player2.weight < 0.5:
keyword = 'less'
else:
keyword = 'more'
print('TRIAL 2: Player 2 weighs (' + str(player2.weight) + ') Player 1\'s actions ' + keyword)
print('\tP2\'s Button A Income Rate: ' + str(round(m1_p2_a, 2)))
print('\tP2\'s Button B Income Rate: ' + str(round(m1_p2_b, 2)))
print('TRIAL 3: Player 2 jointly uses Player 1\'s actions:')
print('\tP2\'s Button A Income Rate: ' + str(round(m2_p2_a, 2)))
print('\tP2\'s Button B Income Rate: ' + str(round(m2_p2_b, 2)))
# Show plots
plt.show(1)
plt.show(2)
plt.show(3)
plt.show(4)
plt.show(5)
return
d = -10.0
# probability that game repeats
w = .895
# population size
pop = 100
# probability that any one player will be mutated each generation
# or, the expected percentage of the population that will be mutated each generation
mutate_prob = .05
generations = 10000
milestone = 10
silent = True
selection_strength = .4
# options: dwol, dwl, cwl, cwol, onlyl
player1_seed = 'dwol'
# options: alle, allc, only_cwol
player2_seed = 'alle'
# where to store trial data
# if set to None, it will store in ./trials/trial{x}/ where {x} is the next available number
file_prefix = './demo/'
simulation.run_simulation(p=p, a=a, c_l=c_l, c_h=c_h, b=b, d=d, w=w, pop=pop, selection_strength=selection_strength, mutate_prob=mutate_prob, generations=generations, milestone=milestone, file_prefix=file_prefix, silent=False, player1_seed=player1_seed, player2_seed=player2_seed)
def run(config_file, output_loc, is_slave, identity):
random.seed(10)
## Read input configuration
f = open(config_file)
config = sj.loads(f.read())
f.close()
## Set output configuration
output(output_loc, is_slave, identity, sj.dumps(config) + '\n')
num_steps = config['num_steps']
num_trials = config['num_trials']
trust_used = config['trust_used']
inbox_trust_sorted = config['inbox_trust_sorted']
if 'trust_filter_on' in config.keys():
trust_filter_on = config['trust_filter_on']
else:
trust_filter_on = True
i = 1
for num_fact in config['num_facts']:
for num_noise in config['num_noise']:
for num_agents in config['num_agents']:
for agent_per_fact in config['agent_per_fact']:
for graph in config['graph_description']:
graph_type = graph['type']
radius = graph['radius']
for agent_setup in config['agent_setup']:
for w in config['willingness']:
for c in config['competence']:
for spam in config['spamminess']:
for selfish in config['selfishness']:
print "Case", i, "being executed"
print "running for %d/%d facts %d agents"\
%(num_fact, num_noise, num_agents)
print "\t%d facts per agent "\
%agent_per_fact
print "\t%s/%.1f graph for %s steps" \
%(graph_type, radius, num_steps)
print "\tw:%.1f/c:%.1f for %d trials"\
%(w,c,num_trials)
print "\tagent setup", agent_setup
i += 1
results = sim.run_simulation(num_fact, \
num_noise,\
num_agents, \
agent_per_fact,\
radius, \
num_steps, \
w, c, \
num_trials, \
graph_type,\
agent_setup,\
spam, selfish,\
trust_used,\
inbox_trust_sorted, \
trust_filter_on)
output(output_loc, is_slave, identity, sj.dumps(results) + '\n' )
# This is a sample Python script. from simulation import Simulation, run_simulation simulation = Simulation() run_simulation()
def run(config_file, output_loc, is_slave, identity):
random.seed(10)
## Read input configuration
f = open(config_file)
config = sj.loads(f.read())
f.close()
## Set output configuration
output(output_loc, is_slave, identity, sj.dumps(config) + "\n")
num_steps = config["num_steps"]
num_trials = config["num_trials"]
trust_used = config["trust_used"]
inbox_trust_sorted = config["inbox_trust_sorted"]
if "trust_filter_on" in config.keys():
trust_filter_on = config["trust_filter_on"]
else:
trust_filter_on = True
i = 1
for num_fpro in config["num_fpro"]:
for num_fcon in config["num_fcon"]:
for num_npro in config["num_npro"]:
for num_ncon in config["num_ncon"]:
for num_groups in config["num_groups"]:
for num_agents in config["num_agents"]:
for agent_per_fact in config["agent_per_fact"]:
for graph in config["graph_description"]:
graph_type = graph["type"]
radius = graph["radius"]
for agent_setup in config["agent_setup"]:
for w in config["willingness"]:
for c in config["competence"]:
for spam in config["spamminess"]:
for selfish in config["selfishness"]:
for e in config["engagement"]:
for u in config["uncertainty_handling"]:
print "Case", i, "being executed"
print "running for %d/%d facts per group %d groups %d agents" % (
num_fpro + num_fcon,
num_npro + num_ncon,
num_groups,
num_agents,
)
print "\t%d agents per fact " % agent_per_fact
print "\t%s/%.1f graph for %s steps" % (
graph_type,
radius,
num_steps,
)
print "\tw:%.1f/c:%.1f/e:%.1f/u:%.1f for %d trials" % (
w,
c,
e,
u,
num_trials,
)
print "\tagent setup", agent_setup
i += 1
results = sim.run_simulation(
num_fpro,
num_fcon,
num_npro,
num_ncon,
num_groups,
num_agents,
agent_per_fact,
radius,
num_steps,
w,
c,
e,
u,
num_trials,
graph_type,
agent_setup,
spam,
selfish,
trust_used,
inbox_trust_sorted,
trust_filter_on,
)
output(
output_loc,
is_slave,
identity,
sj.dumps(results) + "\n",
)
stiffness=STIFFNESS,
unstretched_length=UNSTRETCHED_LENGTH,
viscosity=VISCOSITY)
robot = TensegrityRobot()
robot.add_rods([rod1, rod2, rod3, rod4, rod5, rod6, rod7, rod8, rod9])
robot.add_cables([cab1, cab3, cab4, cab5, cab7, cab8, cab9, cab10])
robot.add_cables([cab13, cab14, cab15, cab16, cab21, cab22, cab23, cab24])
robot.add_cables([
cab25, cab26, cab27, cab28, cab29, cab30, cab31, cab32, cab33, cab34,
cab35, cab36
])
robot.add_cables([cab37, cab38, cab39, cab40])
rob_vis.plot_cur_state(robot)
hist_states = run_simulation(robot, time=10, dt=0.01)
rob_vis.animate_historical_states(robot=robot,
states=hist_states,
interval=0.01)
'''
pos0 = robot.get_rods()[4].get_endpoint_a().get_position()
K = []
for i in robot.get_cables():
print("1", i.get_unstretched_length())
len = i.get_unstretched_length()
i.set_unstretched_length(len /2)
hist_states = run_simulation( robot, time=2, dt=0.01 )
rob_vis.animate_historical_states( robot=robot, states=hist_states, interval=0.01 )
print("2", i.get_unstretched_length() )
print(robot.get_rods()[4].get_endpoint_a().get_position() - pos0)
i.set_unstretched_length(len)
def test_all_system():
money = 0
betcount = 0
won = 0
count = 0
allodds = pinnacle.execute("SELECT * FROM csgo").fetchall()
for i in allodds:
if (float(i[4]) - 1) * (float(i[5]) - 1) > 1:
continue
i = list(i)
if i[2] == 'Virtus Pro':
i[2] = 'Virtus.Pro'
elif i[3] == 'Virtus Pro':
i[3] = 'Virtus.Pro'
times = [(datetime.strptime(i[1], '%Y-%m-%d') -
timedelta(days=1)).isoformat()[:10],
(datetime.strptime(i[1], '%Y-%m-%d') +
timedelta(days=1)).isoformat()[:10]]
matches = statsdb.execute("SELECT * FROM Match WHERE date = '%s'" %
(i[1]))
for match in matches:
team1 = statsdb.execute(
"SELECT hltv_alias FROM Team WHERE team_id = %s" %
(str(match[2]))).fetchone()[0]
team2 = statsdb.execute(
"SELECT hltv_alias FROM Team WHERE team_id = %s" %
(str(match[3]))).fetchone()[0]
if set([team1.lower(),
team2.lower()]) == set([i[2].lower(), i[3].lower()]):
if team1.lower() == i[2].lower():
team1 = match[6:11]
team2 = match[11:16]
else:
team1 = match[11:16]
team2 = match[6:11]
winner = get_winner(match[0])
if winner:
try:
winrate = run_simulation('all', team1, team2, True,
i[1])
except Exception:
traceback.print_exc()
continue
if winrate[0] != 0:
try:
print i[2], get_players(team1), winrate[0], i[4]
print i[3], get_players(team2), winrate[1], i[5]
except:
pass
if profit(winrate[0], i[4]):
betcount += 1
money -= 200
if winner == match[2]:
won += 1
money += 200 * float(i[4])
elif profit(winrate[1], i[5]):
betcount += 1
money -= 200
if winner == match[3]:
won += 1
money += 200 * float(i[5])
break
print betcount, money
print float(money) / (betcount * 200)
import scraper import myconstants import forecast import squad_selection import simulation # Scrape data # players = scraper.scrape_player_data(1) # Populate database # myconstants.PLAYER_COLLECTION.remove() # new_players = scraper.populate_collection(myconstants.PLAYER_COLLECTION, players) # Run forecasts # myconstants.FORECASTS_COLLECTION.remove() # forecast.run_forecasts() # # Run simulation of season myconstants.SQUADS_COLLECTION.remove() simulation.run_simulation()
unstretched_length=UNSTRETCHED_LENGTH_h,
viscosity=VISCOSITY)
# Side cables
cab7 = Cable(end_point1=rod1.get_endpoint_a(),
end_point2=rod2.get_endpoint_b(),
stiffness=STIFFNESS,
unstretched_length=UNSTRETCHED_LENGTH_v,
viscosity=VISCOSITY)
cab8 = Cable(end_point1=rod2.get_endpoint_a(),
end_point2=rod3.get_endpoint_b(),
stiffness=STIFFNESS,
unstretched_length=UNSTRETCHED_LENGTH_v,
viscosity=VISCOSITY)
cab9 = Cable(end_point1=rod3.get_endpoint_a(),
end_point2=rod1.get_endpoint_b(),
stiffness=STIFFNESS,
unstretched_length=UNSTRETCHED_LENGTH_v,
viscosity=VISCOSITY)
robot = TensegrityRobot()
robot.add_rods([rod1, rod2, rod3])
robot.add_cables([cab1, cab2, cab3, cab4, cab5, cab6, cab7, cab8, cab9])
rob_vis.plot_cur_state(robot)
hist_states = run_simulation(robot, time=3, dt=0.005)
rob_vis.plot_cur_state(robot)
rob_vis.plot_com_graphs(hist_states)
#rob_vis.animate_historical_states(robot=robot, states=hist_states, interval=0.01)
