def select_metric_value(m):
"""Interactive selection of a metric value
Input:
m : list of values that can be unpacked into valid
parameters for constructing a Metric
Return:
a valid index for the Metric
"""
m = Metric(*m)
default_metric_value = m.index
print("\n{0} {1} {2} {0}".format(10 * "+", m.name, m.short_name))
while True:
for v in m.values:
print(v, v.description)
idx = input('Select one [{0}]: '.format(default_metric_value)).upper()
if not idx:
idx = default_metric_value
print('Selected metric value ###|', idx, '|###')
try:
m.index = idx
except AssertionError:
print('Not valid')
else:
return m.index
def get_metrics_H1(datadir):
files = ['quality_WC_Unigram.label',
'quality_WC_Unigam_NonZero.label',
'quality_WC_Unigam_Content.label',
'quality_WC_Unigam_OrgAssign.label',
'quality_WC_Unigam_Speciteller.label',
'quality_WC_Unigam_Title.label',
'quality_DT.label',
'quality_New.label',
#'quality_New-Title.label',
'quality_firstnode.label'
]
metric = Metric()
body = []
for file in files:
labels, predicts, _ = load_label(datadir + file)
row = [file]
row.append(metric.accuracy(labels, predicts))
row.append(metric.kappa(labels, predicts))
row.append(metric.QWkappa(labels, predicts))
body.append(row)
output = datadir+'H1.txt'
print output
fio.WriteMatrix(output, body, header=None)
def get_metrics_H2b(datadir):
metric = Metric()
lectures = range(1, 9)
feature = 'quality_CrossTopic_Rubric_firstnode_'
feature_fixed = 'quality_CrossTopic_Rubric_firstnode_fixed_'
for fold in lectures:
input = datadir + feature + str(fold)+'_test.label'
zero_file = datadir + feature + str(fold)+'_test_0.txt'
output = datadir + feature_fixed + str(fold)+'_test.label'
fix_firstnode(input, zero_file, output)
body = []
for feature in [
'quality_CrossTopic_WC_Unigram_',
'quality_CrossTopic_Rubric_',
'quality_CrossTopic_Rubric_firstnode_fixed_',
'quality_CrossTopic_DT_',
]:
for lecture in lectures:
file = feature + str(lecture)+'_test.label'
print file
labels, predicts, _ = load_label(datadir + file)
row = [file]
row.append(metric.accuracy(labels, predicts))
row.append(metric.kappa(labels, predicts))
row.append(metric.QWkappa(labels, predicts))
body.append(row)
output = datadir+'H2b.txt'
print output
fio.WriteMatrix(output, body, header=None)
def get_metrics_H2c(datadir):
metric = Metric()
feature = 'quality_CrossCourse_Rubric_firstnode'
feature_fixed = 'quality_CrossCourse_Rubric_firstnode_fixed'
input = datadir + feature +'_test.label'
zero_file = datadir + feature +'_test_0.txt'
output = datadir + feature_fixed +'_test.label'
fix_firstnode(input, zero_file, output)
body = []
for feature in [
'quality_CrossCourse_WC_Unigram',
'quality_CrossCourse_Rubric',
'quality_CrossCourse_Rubric_firstnode_fixed',
'quality_CrossCourse_DT',
]:
file = feature+'_test.label'
print file
labels, predicts, _ = load_label(datadir + file)
row = [file]
row.append(metric.accuracy(labels, predicts))
row.append(metric.kappa(labels, predicts))
row.append(metric.QWkappa(labels, predicts))
body.append(row)
output = datadir+'H2c.txt'
print output
fio.WriteMatrix(output, body, header=None)
def select_metric_value(m):
"""Interactive selection of a metric value
Input:
m : list of values that can be unpacked into valid
parameters for constructing a Metric
Return:
a valid index for the Metric
"""
m = Metric(*m)
default_metric_value = m.index
print("\n{0} {1} {2} {0}".format(10 * "+", m.name, m.short_name))
while True:
for v in m.values:
print(v, v.description)
idx = input('Select one [{0}]: '.format(default_metric_value)).upper()
if not idx:
idx = default_metric_value
print('Selected metric value ###|', idx, '|###')
try:
m.index = idx
except AssertionError:
print('Not valid')
else:
return m.index
def test_pair_dist_starting_at_zero_dist(self):
sigma = 0.1
test_metric = Metric(sigma, edge=1)
positions = [[0.2, 0.2], [0.2 + 2 * sigma, 0.2]]
self.assertAlmostEqual(
test_metric.pair_dist(positions[0], positions[1], 10, [1, 0]), 0,
1, "Should collide")
def invert_metric(self, g):
""" Invert a given metric g and return the inverted metric
as a shorthand dictionary of sympy.core.symbols.Symbol objects.
"""
N = int(np.sqrt(len(g.elements)))
g_matrix = sp.MatrixSymbol('g_matrix', N, N)
g_matrix = sp.Matrix(g_matrix)
print('Converting the metric to sp.Matrix object...')
for i in range(N):
for j in range(N):
m = g.index_dict[i]
n = g.index_dict[j]
mn = m + n
g_matrix[i, j] = g.elements[mn]
print('Inverting the matrix...')
g_matrix_inv = g_matrix**(-1)
g_inv = Metric(index_dict=g.index_dict)
for i in range(N):
for j in range(N):
m = g_inv.index_dict[i]
n = g_inv.index_dict[j]
mn = m + n
g_inv.elements[mn] = g_matrix_inv[i, j]
print('Succesfully inverted the matrix.')
return g_inv
def metrics_lines(api_key):
if authenticator.valid(api_key):
metrics = bottle.request.body.readlines()
if type(metrics) == list:
for line in metrics:
try:
parts = line.split()
if len(parts) != 3:
continue
else:
metric = {
'metric': parts[0],
'value': parts[1],
'timestamp': parts[2]
}
m = Metric(metric)
m.enqueue(q)
except ValueError:
bottle.abort(400, "Invalid Metric Structure: %s" % (str(metric)))
except Exception:
bottle.abort(500, "Unable to store metric!")
else:
bottle.abort(400, "Metric structure must be lines of 'metric.path value timestamp'")
else:
bottle.abort(403, "API Key not valid")
def compute_suspicious_score(self):
scores = []
for i in xrange(1, self.total_stmt + 1):
cp, cf, np, nf = self.compute_statistics(i)
metric = Metric(cp, cf, np, nf)
scores.append(metric.odds_ratio())
return scores
def __init__(self, world):
# call base
Metric.__init__(self, world)
self.name = "compressibility"
self.value = 0.0
self.format = "%.2f%%"
self.co = zlib.compressobj()
def evaluation(data_type,trec_path,dev_query_embedding2id,passage_embedding2id,dev_I,dev_D,test_set="trec2019",dev_split_num=-1,split_idx=-1):
if data_type ==0 :
if test_set== "marcodev":
qrels="../data/raw_data/msmarco-docdev-qrels.tsv"
elif test_set== "trec2019":
qrels="../data/raw_data/2019qrels-docs.txt"
elif data_type ==1:
if test_set == "marcodev":
qrels="../data/raw_data/qrels.dev.small.tsv"
else:
logging.error("wrong data type")
exit()
trec_path=trec_path.replace(".trec",".formatted.trec")
met = Metric()
if split_idx >= 0:
split_file_path=qrels+f"{dev_split_num}_fold.split_dict"
with open(split_file_path,'rb') as f:
split=pickle.load(f)
else:
split=None
ndcg10 = met.get_metric(qrels, trec_path, 'ndcg_cut_10',split,split_idx)
mrr10 = met.get_mrr(qrels, trec_path, 'mrr_cut_10',split,split_idx)
mrr100 = met.get_mrr(qrels, trec_path, 'mrr_cut_100',split,split_idx)
print(f" evaluation for {test_set}, trec_file {trec_path}, split_idx {split_idx} \
ndcg_cut_10 : {ndcg10}, \
mrr_cut_10 : {mrr10}, \
mrr_cut_100 : {mrr100}"
)
return ndcg10
def __init__(self, world):
# call base
Metric.__init__(self, world)
self.name = "exploration"
self.value = 0.0
self.format = "%.2f%%"
# dictionary mapping hashes to ticks used for cycle detection
self.rule_count = np.zeros(self.world.rule.size, dtype=int)
def testAttributes(self):
metric = Metric()
metric.setName("foo")
metric.setValue(10)
metric.setSource("localhost")
self.assertEqual(metric.getName(), "foo", "Name does not match")
self.assertEqual(metric.getValue(), 10, "Values does nto match")
self.assertEqual(metric.getSource(), "localhost", "Source does not match")
def __init__(self, args):
"""
Create Evaluator object which handles the evaluation of a specified model,
the tracking of computed metrics, and saving of results.
:param args: ArgParse object which holds the path the experiment configuration file along
with other key experiment options.
"""
# get experiment configuration file
osj = os.path.join
oss = os.path.split
base_path = os.path.dirname(os.path.abspath(__file__))
config = get_config(base_path, args.config)
self.config = config
config_path = osj(base_path, 'config')
# create experiment, logging, and model checkpoint directories
model_name = os.path.split(args.model)[1].split('.')[0]
self.eval_dir = osj(base_path, oss(args.config)[0], 'eval_' + args.data + '_' + model_name)
self.visual_path = osj(self.eval_dir, 'visuals')
if os.path.exists(self.eval_dir):
print("Error: This evaluation already exists")
print("Cancel Session: 0")
print("Overwrite: 1")
if input() == str(1):
shutil.rmtree(self.eval_dir)
else:
exit(1)
os.mkdir(self.eval_dir)
os.mkdir(self.visual_path)
# Device
self.device = torch.device("cpu" if args.gpuid == 'cpu' else "cuda:{}".format(args.gpuid))
# Model - load pre-trained
model = get_model(config)
self.model = model
self.load_model(args.model)
self.model.to(self.device)
# Loss metric
self.criterion = nn.MSELoss()
# Dataset and DataLoader
self.data_type = args.data
self.dataset = ShapeNet(config, config_path, args.data)
self.tracked_samples = extract_categories(self.dataset.samples, 5)
self.loader = DataLoader(self.dataset, batch_size=config['batch_size'],
shuffle=True, num_workers=1, drop_last=False)
print("Commencing evaluation with {} model on {} split".format(config['model'], args.data))
# Metrics
self.metrics = Metric(config)
self.metric_tracker = MetricTracker(config, self.eval_dir, args.data)
self.epoch = 0
def __init__(self, world):
# call base
Metric.__init__(self, world)
self.name = "cyclic"
self.value = False
# dictionary mapping hashes to ticks used for cycle detection
self.hist = {}
self.cycle_start = None
self.cycle_period = None
def test_wall_dist(self):
test_metric = Metric(sigma=0.1, edge=1)
pos = [0.5, 0.5]
for wall in [1, -1, 2, -2]:
v_hat = [0, 0]
v_hat[np.abs(wall) - 1] = np.sign(wall)
dist_to_wall, wall_type = test_metric.wall_dist(pos, v_hat, 1)
self.assertEqual(dist_to_wall, 0.4, "Wrong distance to wall")
self.assertEqual(wall_type, wall, "Wrong Classification of wall")
def evaluate(self):
test_featureset = self._get_featuresets(self.test_file)
labels = [int(x[1]) for x in test_featureset]
featureset = [x[0] for x in test_featureset]
predicts = [int(x) for x in self._model.classify_many(featureset)]
metric = Metric()
return metric.accuracy(labels, predicts), metric.kappa(labels, predicts), metric.QWkappa(labels, predicts)
def test_metric_improvement(self):
params = {"name": "logloss"}
m = Metric(params)
y_true = np.array([0, 0, 1, 1])
y_predicted = np.array([0, 0, 0, 1])
score_1 = m(y_true, y_predicted)
y_true = np.array([0, 0, 1, 1])
y_predicted = np.array([0, 0, 1, 1])
score_2 = m(y_true, y_predicted)
self.assertTrue(m.improvement(score_1, score_2))
def test_step_size(self):
sigma = 0.1
test_metric = Metric(sigma, edge=1)
positions = [[0.2, 0.2], [0.7, 0.2]]
current_step_size, step_type, sphere_or_wall_ind = \
test_metric.step_size(positions, 0, [1, 1], .2)
self.assertEqual(current_step_size, .2, "Problem with free step")
current_step_size, step_type, sphere_or_wall_ind = \
test_metric.step_size(positions, 0, [1, 0], 1)
self.assertAlmostEqual(current_step_size, .3, 1,
"Problem with pair collision")
def test_wall_dist_starting_at_the_wall(self):
test_metric = Metric(sigma=0.1, edge=1)
pos = [0.1, 0.5]
dist_to_wall, wall_type = test_metric.wall_dist(pos, [-1, 0], 1)
self.assertEqual(dist_to_wall, 0, "Wrong distance to wall")
self.assertEqual(wall_type, -1, "Wrong Classification of wall")
dist_to_wall, wall_type = test_metric.wall_dist(pos, [0, 1], 1)
self.assertEqual(dist_to_wall, 0.4, "Wrong distance to wall")
self.assertEqual(wall_type, 2, "Wrong Classification of wall")
def __init__(self, init_list, goal_list):
"""Initialise Solver object. Raise ValueError if solution not possible."""
self.initial_state = copy.deepcopy(self.list_to_grid(init_list))
self.goal_state = copy.deepcopy(self.list_to_grid(goal_list))
self.frontier = []
self.explored = set()
self.metrics = Metric(self.frontier)
def test_pair_dist(self):
sigma = 0.1
test_metric = Metric(sigma, edge=1)
positions = [[0.2, 0.2], [0.7, 0.2]]
self.assertAlmostEqual(
test_metric.pair_dist(positions[0], positions[1], 10, [1, 0]),
0.5 - 2 * sigma, 1, "Wrong distance between pairs")
positions = [[0.2, 0.2], [0.7, 0.7]]
self.assertAlmostEqual(
test_metric.pair_dist(positions[0], positions[1], 10, [0, 1]),
float('inf'), 1, "Should not collide")
def get_QWkappa(input):
head,body = fio.ReadMatrix(input, True)
metric = Metric()
data = {}
for i,row in enumerate(body):
for coder, label in enumerate(row):
if label == 'a': label = '0'
label = int(label)
if head[coder] not in data:
data[ head[coder] ] = []
data[ head[coder] ].append(label)
print 'annototor 1', '\t','annototor 2', '\t', 'accuracy', '\t', 'kappa', '\t', 'QWkappa'
print head[0], '\t', head[1], '\t', metric.accuracy(data[head[0]], data[head[1]]), '\t', metric.kappa(data[head[0]], data[head[1]]), '\t', metric.QWkappa(data[head[0]], data[head[1]])
print head[0], '\t', head[2], '\t', metric.accuracy(data[head[0]], data[head[2]]), '\t', metric.kappa(data[head[0]], data[head[2]]), '\t', metric.QWkappa(data[head[0]], data[head[2]])
print head[1], '\t', head[2], '\t', metric.accuracy(data[head[1]], data[head[2]]), '\t', metric.kappa(data[head[1]], data[head[2]]), '\t', metric.QWkappa(data[head[1]], data[head[2]])
print '', '\t', 'Average', '\t', np.mean([metric.accuracy(data[head[0]], data[head[1]]), metric.accuracy(data[head[0]], data[head[2]]), metric.accuracy(data[head[1]], data[head[2]])]), '\t',\
np.mean([metric.kappa(data[head[0]], data[head[1]]), metric.kappa(data[head[0]], data[head[2]]), metric.kappa(data[head[1]], data[head[2]])]), '\t',\
np.mean([metric.QWkappa(data[head[0]], data[head[1]]), metric.QWkappa(data[head[0]], data[head[2]]), metric.QWkappa(data[head[1]], data[head[2]])])
print metric.confusion_matrix(data[head[0]], data[head[1]])
return 0
def test_eval(self):
iris = load_iris()
X_train, y_train = iris.data[:120], iris.target[:120]
X_test, y_test = iris.data[120:], iris.target[120:]
rf = Model('rf', RandomForestClassifier, {'random_state':0})
rf.run(X_train, y_train)
metrics = rf.eval(X_test, y_test, metrics=['acc'])
rf_metric = Metric('rf')
rf_metric.addValue('acc', 0.7666666666666667)
expected_metrics = rf_metric.getValues()
self.assertEqual( metrics, expected_metrics )
def list_metrics(self, start_date, end_date):
# Create CloudWatch client
cloudwatch = boto3.client('cloudwatch')
metrics = []
statistics_type = 'Average'
period = 3600 #hour
# List metrics through the pagination interface
paginator = cloudwatch.get_paginator('list_metrics')
for response in paginator.paginate(Namespace=self.namespace):
for row in response['Metrics']:
metric_name = row['MetricName']
for row2 in row['Dimensions']:
dimension_name = row2['Name']
dimension_value = row2['Value']
if (self.namespace == 'AWS/EC2' and metric_name in ["CPUUtilization","NetworkOut","NetworkIn","EBSWriteBytes","EBSReadBytes","DiskReadBytes","DiskWriteBytes","DiskWriteOps","DiskReadOps"]) \
or self.namespace != 'AWS/EC2':
metric = Metric(metric_name, self.namespace,
dimension_name, dimension_value,
statistics_type, period, start_date,
end_date)
metrics.append(metric)
return metrics
def get_info(self):
distance = None
try:
date = datetime.datetime.utcnow()
GPIO.output(self.pin_trigger, GPIO.HIGH)
time.sleep(0.00001)
GPIO.output(self.pin_trigger, GPIO.LOW)
while GPIO.input(self.pin_echo) == 0:
pulse_start_time = time.time()
while GPIO.input(self.pin_echo) == 1:
pulse_end_time = time.time()
pulse_duration = pulse_end_time - pulse_start_time
distance = round(pulse_duration * 17150, 2)
except Exception as e:
logger.error(f'Error in distance: {e}')
self.reset()
if distance is not None:
name = self.metric_prefix + 'distance'
if self.output == 'WF':
name = 'Distance'
if len(self.metric_prefix) > 0:
name = self.metric_prefix + '.' + name
if self.format == 'f':
distance = distance / 30.48
if self.format == 'i':
distance = distance / 30.48 * 12
self.metrics.append(Metric(name, distance, date))
def soft_nms(bboxes, threshold=0.75):
"""
soft non-max suppression implementation for object detection. takes a list of bounding boxes with scores,
decreases score of bounding boxes with respect to their IoU with the highest score bounding box
:param bboxes: (array) list of bounding boxes in format [x1, y1, x2, y2, score]
:param threshold: (scalar, float) minimum acceptable score for bounding box
:return: filtered list of bounding boxes after soft non-max suppression
"""
bboxes_nms = []
while bboxes:
# filter all the bboxes with score below the threshold
bboxes = [bbox for bbox in bboxes if bbox[4] >= threshold]
# grab the highest score bbox, remove it from the list and add it to output list
maxscore_bbox = max(bboxes, key=lambda x: x[4])
bboxes.remove(maxscore_bbox)
bboxes_nms.append(maxscore_bbox)
# loop over all the possible overlaps with the selected bbox and update their scores
bbox_pairs = [(maxscore_bbox, bbox) for bbox in bboxes]
for maxscore_bbox, bbox2 in bbox_pairs:
bbox2[4] -= Metric.iou(maxscore_bbox, bbox2)
return bboxes_nms
def test1(n):
from numpy import array
# One equilibrium of Ar4 LJ cluster (in coordinates of
# c2v_tetrahedron1 Func):
w = 0.39685026
A = array([w, w, +w])
# Another equilibrium:
B = array([w, w, -w])
# Halfway between A and B:
C = (A + B) / 2.0
C = array([w + 0.01, w - 0.01, 0.0])
xs = array([A, C, B])
from test.testfuns import c2v_tetrahedron1, diagsandhight
from path import MetricPath
from metric import Metric
from numpy import linspace
z = c2v_tetrahedron1()
# z = diagsandhight()
# r = 1.12246195815
# A = array([r, r, r / sqrt(2.)])
# B = array([r, r, -r / sqrt(2.)])
# C = array([r, r * sqrt(2.), 0.])
# xs = array([A, C, B])
p = MetricPath(xs, Metric(z).norm_up)
x0 = map(p, linspace(0., 1., n))
from ase import Atoms
from qfunc import QFunc
from func import compose
pes = compose(QFunc(Atoms("Ar4")), z)
from rc import Volume
vol = compose(Volume(), z)
def callback(x, e, g, t):
# from pts.tools.jmol import jmol_view_path
print "energies=", e # map(pes, x)
print "volume=", map(vol, x)
# jmol_view_path(map(z, x), syms=["Ar"]*4, refine=1)
pass
print "BEFORE:"
callback(x0, map(pes, x0), map(pes.fprime, x0), None)
x1, info = soptimize(pes, x0, tangent1, rc=vol, callback=callback)
# print "info=", info
print "AFTER:"
callback(x1, map(pes, x1), map(pes.fprime, x1), None)
def convertDF(self, data, date):
data = self.convertStringToList(data)
length = len(data)
for i in range(length):
if i > 0:
colLength = len(data[i])
for c in range(colLength):
if c > 0 and c < 4:
name = 'disk.' + data[0][c]
tag = {}
tag['filesystem'] = data[i][0]
tag['mount'] = data[i][5]
value = int(data[i][c])
metric = Metric(name, value, date)
metric.tags = tag
self.metrics.append(metric)
def train(self, data, prm):
best_corr_de = -1e100
best_net = None
patient = 0
for it in range(prm.max_epochs_num):
Logger.write_log('iter = ' + str(it))
self.net.fit(x=data.Xtr,
y=data.Ytr,
batch_size=prm.batch_size,
epochs=1,
verbose=1,
shuffle=True)
data_pred = self.evaluate(data, prm)
corr_de = Metric.report_all_metrics(data_pred)
if corr_de > best_corr_de:
Logger.write_log('nice, model gets better ...')
Logger.write_log('###############################')
best_corr_de = corr_de
best_net = copy.deepcopy(self.net)
patient = 0
else:
Logger.write_log('oops, model gets worse ...')
Logger.write_log('###############################')
patient += 1
if patient >= prm.max_patience:
break
self.net = best_net
def get_metrics_cv(datadir, files):
metric = Metric()
_, _, folds = load_label(datadir + files[0])
body = []
for file in files:
labels, predicts, _ = load_label(datadir + file)
row = [file]
row += metric.cv_accuracy(labels, predicts, folds)
row += metric.cv_kappa(labels, predicts, folds)
row += metric.cv_QWkappa(labels, predicts, folds)
body.append(row)
output = datadir+'H1_p.txt'
print output
fio.WriteMatrix(output, body, header=None)
def worker(self, train, test):
'''
:parmas: train, 训练数据集
:params: test, 测试数据集
:return: 各指标的值
'''
print('Starting train data with function: {}'.format(self.rt))
if self.rt == 'LFM':
getRecommendation, P, Q = self.alg[self.rt](train, self.ratio, self.K_latent, \
self.lr, self.step, self.lmbda, self.N)
else:
getRecommendation = self.alg[self.rt](train, self.K,
self.N) # train
print('Starting cacluate evaluation metrics with function: {}'.format(
self.rt))
metric = Metric(train, test, getRecommendation)
return metric.eval()
def testString(self):
metric = Metric()
metric.setName("bar")
metric.setValue(100)
metric.setSource("snafu")
print(str(self))
self.assertEqual(str(metric), "bar 100 snafu", "String does not match")
def sample_sectors(self):
"""
Runs the process to sample the repository.
:return:
"""
self.__sectors = self.__generate_sectors()
for sector in self.__sectors:
commits_in_sector = sector.get_objects()
commits_count = len(commits_in_sector)
if commits_count:
m = Metric()
m.commit_count = commits_count
last_commit = commits_in_sector[commits_count - 1]
m.commit = commits_in_sector[0]
score = self.__score(commits_in_sector[0].id.hex,
last_commit.id.hex, commits_count)
m.activity = score[0]
m.additions = score[1]
m.deletions = score[2]
m.timestamp = datetime.datetime.fromtimestamp(
last_commit.commit_time)
self.__metrics.append(m)
def get_info(self):
current_value = GPIO.input(self.pin)
while current_value == self.initial_value:
#do nothing, wait for a change
current_value = GPIO.input(self.pin)
time.sleep(0.00001)
# changed
self.initial_value = current_value
self.metrics.append(Metric(self.name, current_value, datetime.datetime.utcnow()))
def get_info(self):
date = datetime.datetime.utcnow()
humidity, temp = Adafruit_DHT.read_retry(self.code, self.pin)
if temp is not None:
name = self.metric_prefix + 'humidity'
if self.output == 'WF':
name = 'Humidity'
if len(self.metric_prefix) > 0:
name = self.metric_prefix + '.' + name
self.metrics.append(Metric(name, humidity, date))
if self.format == 'f':
temp = (temp * 9 / 5) + 32
name = self.metric_prefix + 'temperature'
if self.output == 'WF':
name = 'Temperature'
if len(self.metric_prefix) > 0:
name = self.metric_prefix + '.' + name
self.metrics.append(Metric(name, temp, date))
def createMetric(category, metric, value):
# Build a metric type name for the given metric
m = METRIC_TYPE + ' / ' + category + ' / ' + metric
# Return a metric description object
return Metric(source=METRIC_SOURCE,
metric=m,
resource=METRIC_RESOURCE,
node=METRIC_NODE,
value=value)
def test_generate_normal(self):
sys = SystemDataGenerator(name='test00')
metric_vals = {'name': 'cpu_busy',
'min_val': 0,
'max_val': 10,
'mean': 10,
'sd': 3}
metric = Metric(**metric_vals)
metric_value = sys.generate_metric_value_normal(metric)
assert metric_value[metric.name] > 0
def convertUptime(self, data, date):
hasUser = False
if "user" in data.lower().strip():
hasUser = True
data = list(data.replace(",", "").split(' '))
if len(data) > 1:
for i in range(len(data)):
data[i] = list(filter(None, data[i].split(' ')))
else:
data = list(filter(None, data[0].split(' ')))
data = [[data[0], data[1]], [data[2]], [data[4], data[5]],
[data[6], data[7], data[8], data[9], data[10]]]
#check if uptime has days
uptime = 0
i = 0
if (hasUser and len(data) > 3) or (len(data) > 2 and not hasUser):
if (len(data[0]) > 2):
uptime += int(data[i][2]) * 24 * 60
i += 1
uptime += self.convertDurationToInt(data[i][0])
else:
if len(data[i]) == 5:
uptime = int(data[i][2]) * 24 * 60
uptime += self.convertDurationToInt(data[i][4])
elif len(data[i]) == 6:
uptime = int(data[i][2]) * 24 * 60
uptime += int(data[i][4])
else:
uptime += self.convertDurationToInt(data[i][2])
self.metrics.append(Metric("uptime", uptime, date))
i += 1
#users
if hasUser:
users = int(data[i][0])
self.metrics.append(Metric("users", users, date))
i += 1
#load
minMean = float(data[i][2])
fiveMinMean = float(data[i][3])
fifteenMean = float(data[i][4])
self.metrics.append(Metric("load1min.mean", minMean, date))
self.metrics.append(Metric("load5min.mean", fiveMinMean, date))
self.metrics.append(Metric("load15min.mean", fifteenMean, date))
def test_create(self):
params = {"name": "logloss"}
m = Metric(params)
y_true = np.array([0, 0, 1, 1])
y_predicted = np.array([0, 0, 1, 1])
score = m(y_true, y_predicted)
self.assertTrue(score < 0.1)
y_true = np.array([0, 0, 1, 1])
y_predicted = np.array([1, 1, 0, 0])
score = m(y_true, y_predicted)
self.assertTrue(score > 1.0)
def __init__(self):
self.con = Container()
self.ret = self.con.runContainerId()
self.dt,self.dt1,self.dt2,self.dt3,self.dt4 = {},{},{},{},{}
for i in self.ret:
self.con.setFilePath(i)
self.data = Metric(self.con.container_id,self.con.cpuacct_path,self.con.memStat_path,self.con.blkio_path,self.con.netStat_path)
self.data.getNetns()
self.dt_cpu,self.dt_mem,self.dt_blk,self.dt_net = {},{},{},{}
self.dt_cpu['cpuAcct'] = self.data.cpuAcct()
self.dt_mem['memStat'] = self.data.memStat(i)
self.dt_blk['blkio'] = self.data.blkio()
self.dt_net['netStat'] = self.data.netStat()
# we set the trunc id for the metric
_i = i[:12]
self.dt[_i]= [self.dt_cpu,self.dt_mem,self.dt_blk,self.dt_net]
self.dt1[_i] = [self.dt_cpu]
self.dt2[_i] = [self.dt_mem]
self.dt3[_i] = [self.dt_blk]
self.dt4[_i] = [self.dt_net]
def get_metrics_NewCourse(datadir):
files = ['DT.txt',
'DT_NoneZero.txt',
'Rubric.txt',
'Rubric_NoneZero.txt',
]
metric = Metric()
body = []
for file in files:
labels, predicts, _ = load_label(datadir + file)
row = [file]
row.append(metric.accuracy(labels, predicts))
row.append(metric.kappa(labels, predicts))
row.append(metric.QWkappa(labels, predicts))
body.append(row)
output = datadir+'H_NewCourse.txt'
print output
fio.WriteMatrix(output, body, header=None)
def get_metrics_H1_CV(datadir):
metric = Metric()
folds = range(0, 10)
#fix firstnode
feature = 'quality_rubric_firstnode_'
feature_fixed = 'quality_rubric_firstnode_fixed_'
for fold in folds:
input = datadir + feature + str(fold)+'_test.label'
zero_file = datadir + feature + str(fold)+'_test_0.txt'
output = datadir + feature_fixed + str(fold)+'_test.label'
fix_firstnode(input, zero_file, output)
body = []
for feature in ['quality_WC_Unigram_',
'quality_WC_Unigam_NonZero_',
'quality_WC_Unigam_Content_',
'quality_WC_Unigam_OrgAssign_',
'quality_WC_Unigam_Speciteller_',
'quality_WC_Unigam_Title_',
'quality_rubric_',
'quality_rubric_firstnode_fixed_',
'quality_DT_',
]:
for fold in folds:
file = feature + str(fold)+'_test.label'
print file
labels, predicts, _ = load_label(datadir + file)
row = [file]
row.append(metric.accuracy(labels, predicts))
row.append(metric.kappa(labels, predicts))
row.append(metric.QWkappa(labels, predicts))
body.append(row)
output = datadir+'H1_cv.txt'
print output
fio.WriteMatrix(output, body, header=None)
def publish(api_key):
if authenticator.valid(api_key):
try:
body = bottle.request.body.read()
metrics = json.loads(body)
except:
log.error("Request unparsable: %s" % (body))
bottle.abort(400, "Unable to successfully parse JSON")
if type(metrics) == list:
for metric in metrics:
try:
m = Metric(metric)
m.enqueue(q)
except ValueError:
bottle.abort(400, "Invalid Metric Structure: %s" % (str(metric)))
except Exception:
bottle.abort(500, "Unable to store metric!")
else:
bottle.abort(400, "Metric structure must be <list> containing n <dict> items")
else:
bottle.abort(403, "API Key not valid")
def get_metrics_H0(datadir):
metric = Metric()
body = []
for feature in ['quality_rubric', 'quality_binary_model', 'quality_New', 'quality_firstnode']:
file = feature+'.label'
print file
labels, predicts, _ = load_label(datadir + file)
fio.WriteMatrix(datadir + file + '.cm', metric.confusion_matrix(labels, predicts), None)
row = [file]
row.append(metric.accuracy(labels, predicts))
row.append(metric.kappa(labels, predicts))
row.append(metric.QWkappa(labels, predicts))
# row += metric.cv_accuracy(labels, predicts)
# row += metric.cv_kappa(labels, predicts)
# row += metric.cv_QWkappa(labels, predicts)
body.append(row)
output = datadir+'H0.txt'
print output
fio.WriteMatrix(output, body, header=None)
def __init__(self, user, password, ip, port, metric_labels):
"""Generate metrics list, connect to RabbitMQ server"""
# Add valid metrics to list
self.metric_list = []
for label in metric_labels:
metric = Metric.create(label)
if metric is not None:
self.metric_list.append(metric)
# Connect to RabbitMQ server
credentials = PlainCredentials(user,
password)
basicConfig(format='%(levelname)s:%(message)s',
level=CRITICAL)
connection_parameters = ConnectionParameters(ip,
port,
'/',
credentials)
self.connection = BlockingConnection(connection_parameters)
self.channel = self.connection.channel()
self.host = gethostname()
# Prepare request queue
self.processing_request = False
self.channel.exchange_declare(exchange='request',
type='fanout')
result = self.channel.queue_declare(exclusive=True)
request_queue = result.method.queue
self.channel.queue_bind(exchange='request',
queue=request_queue)
self.channel.basic_consume(self.request_metric,
queue=request_queue,
no_ack=True)
# Prepare reply queue
self.channel.queue_declare(queue='reply')
from metric import Metric
import sympy as sp
x=sp.symbols(['z','t','x','y'])
gs=[sp.Symbol('g'+str(i))(x[0]) for i in x]
M=Metric(x,sp.diag(*gs))
A=[sp.Symbol(t)(M.x[0]) for t in ['Az','At','Ax','Ay']]
der=M.covariantVectorDerivative(A)
assert(M.g.is_diagonal())
sp.pprint(sp.simplify(sum(der[i][i]*M.ginv[i,i] for i in range(len(M.x)))))
def __init__(self, ministry, zbins=None, magbins=None,
catalog_type=['galaxycatalog'], tag=None,
appmag=True, lower_limit=True, cutband=None,
normed=True, selection_dict=None, **kwargs):
"""
Angular Number density of objects as a function of redshift.
inputs
--------
ministry - Ministry
keywords
------
zbins - np.array
An array containing the edges of the redshift bins to use for dn/dz
lower_limit - boolean
Whether or not the magnitudes in magbins should be interpreted
as bin edges or lower limits of cuts (i.e. mag<magbins[i])
magbins - np.array
A list of magnitudes to use as bin edges or lower limits of cuts
catalog_type - list
A list of catalog types, ususally not used
tag - string
A name for the metric to be used when making comparisons using bb-compare
appmag - boolean
True if we want to use apparent magnitude for cuts, false for absolute magnitudes
cutband - int
The index of the column of a vector of magnitudes to use
normed - boolean
Whether the metric integrates to N/deg^2 or not. Usually want True.
"""
Metric.__init__(self, ministry, tag=tag, **kwargs)
self.catalog_type = catalog_type
if zbins is None:
self.zbins = np.linspace(0, 2.0, 61)
else:
self.zbins = zbins
self.zbins = np.array(self.zbins)
self.nzbins = len(self.zbins)-1
if appmag:
self.mkey = 'appmag'
defmbins = np.array([18, 19, 20, 21])
else:
self.mkey = 'luminosity'
defmbins = np.array([-24, -23, -22, -21])
if cutband is None:
self.cutband = 0
else:
self.cutband = cutband
self.lower_limit = lower_limit
if magbins is None:
self.magbins = [None]
self.nmagbins = 0
self.nomags = True
else:
self.nomags = False
self.magbins = magbins
if self.lower_limit:
self.nmagbins = len(self.magbins)
else:
self.nmagbins = len(self.magbins) - 1
self.normed = normed
self.aschema = 'galaxyonly'
if self.nmagbins > 0:
self.mapkeys = [self.mkey, 'redshift']
self.unitmap = {self.mkey :'mag'}
else:
self.mapkeys = ['redshift']
self.unitmap = {}
#Make selection dict here
if (selection_dict is None) & lower_limit:
selection_dict = {'mag':{'selection_type':'cut1d',
'mapkeys':['appmag'],
'bins':self.magbins,
'selection_ind':self.cutband,
'lower':True}}
elif (selection_dict is None):
selection_dict = {'mag':{'selection_type':'binned1d',
'mapkeys':['appmag'],
'bins':self.magbins,
'selection_ind':self.cutband}}
nmkeys = []
for s in selection_dict:
if 'mapkeys' in selection_dict[s]:
ss = selection_dict[s]
for m in ss['mapkeys']:
if m not in self.mapkeys:
self.mapkeys.append(m)
if m not in self.unitmap:
self.unitmap[m] = self.defaultUnits(m)
self.zcounts = None
self.selector = Selector(selection_dict)
class Traffic:
"""
Traffic class definition : Traffic data
Methods:
Traffic(tmx) : constructor
create(acid,actype,aclat,aclon,achdg,acalt,acspd) : create aircraft
delete(acid) : delete an aircraft from traffic data
update(sim) : do a numerical integration step
findnearest(lat,lon) : find nearest a/c to lat/lon position
trafperf () : calculate aircraft performance parameters
Members: see create
Created by : Jacco M. Hoekstra
"""
def __init__(self, tmx):
self.tmx = tmx # tmx object contains sim, scr and other main objects
self.dts = []
self.ntraf = 0
# model-specific parameters.
# Default: BlueSky internal performance model.
# Insert your BADA files to the folder "BlueSky/data/coefficients/BADA"
# for working with EUROCONTROL`s Base of Aircraft Data revision 3.12
# Check for BADA OPF file
path = os.path.dirname(__file__) + '/../../data/coefficients/BADA/'
files = os.listdir(path)
self.bada = False
for f in files:
if f.upper().find(".OPF")!=-1:
self.bada=True
break
# Initialize correct performance models
if self.bada:
self.perf = PerfBADA(self)
else:
self.perf = Perf(self)
self.dts = []
self.ntraf = 0
# Create datalog instance
self.log = Datalog()
# Traffic list & arrays definition
# !!!IMPORTANT NOTE!!!
# Anny variables added here should also be added in the Traffic
# methods self.create() (append) and self.delete() (delete)
# which can be found directly below __init__
# Traffic basic flight data
# Traffic basic flight data
self.id = [] # identifier (string)
self.type = [] # aircaft type (string)
self.lat = np.array([]) # latitude [deg]
self.lon = np.array([]) # longitude [deg]
self.trk = np.array([]) # track angle [deg]
self.tas = np.array([]) # true airspeed [m/s]
self.gs = np.array([]) # ground speed [m/s]
self.cas = np.array([]) # callibrated airspeed [m/s]
self.M = np.array([]) # mach number
self.alt = np.array([]) # altitude [m]
self.fll = np.array([]) # flight level [ft/100]
self.vs = np.array([]) # vertical speed [m/s]
self.rho = np.array([]) # atmospheric air density [m/s]
self.temp = np.array([]) # atmospheric air temperature [K]
self.dtemp = np.array([]) # delta t for non-ISA conditions
# Traffic performance data
self.avsdef = np.array([]) # [m/s]default vertical speed of autopilot
self.aphi = np.array([]) # [rad] bank angle setting of autopilot
self.ax = np.array([]) # [m/s2] absolute value of longitudinal accelleration
self.bank = np.array([]) # nominal bank angle, [radian]
self.bphase = np.array([]) # standard bank angles per phase
self.hdgsel = np.array([]) # determines whether aircraft is turning
# Crossover altitude
self.abco = np.array([])
self.belco = np.array([])
# Traffic autopilot settings
self.ahdg = [] # selected heading [deg]
self.aspd = [] # selected spd(eas) [m/s]
self.aptas = [] # just for initializing
self.ama = [] # selected spd above crossover altitude (Mach) [-]
self.aalt = [] # selected alt[m]
self.afll = [] # selected fl [ft/100]
self.avs = [] # selected vertical speed [m/s]
# limit settings
self.lspd = [] # limit speed
self.lalt = [] # limit altitude
self.lvs = [] # limit vertical speed due to thrust limitation
# Traffic navigation information
self.orig = [] # Four letter code of origin airport
self.dest = [] # Four letter code of destination airport
# LNAV route navigation
self.swlnav = np.array([]) # Lateral (HDG) based on nav?
self.swvnav = np.array([]) # Vertical/longitudinal (ALT+SPD) based on nav info
self.actwplat = np.array([]) # Active WP latitude
self.actwplon = np.array([]) # Active WP longitude
self.actwpalt = np.array([]) # Active WP altitude to arrive at
self.actwpspd = np.array([]) # Active WP speed
self.actwpturn = np.array([]) # Distance when to turn to next waypoint
# VNAV cruise level
self.crzalt = np.array([]) # Cruise altitude[m]
# Route info
self.route = []
# ASAS info per aircraft:
self.iconf = [] # index in 'conflicting' aircraft database
self.asasactive = np.array([]) # whether the autopilot follows ASAS or not
self.asashdg = np.array([]) # heading provided by the ASAS [deg]
self.asasspd = np.array([]) # speed provided by the ASAS (eas) [m/s]
self.asasalt = np.array([]) # speed alt by the ASAS [m]
self.asasvsp = np.array([]) # speed vspeed by the ASAS [m/s]
self.desalt = np.array([]) #desired altitude [m]
self.deshdg =np.array([]) #desired heading
self.desvs =np.array([]) #desired vertical speed [m/s]
self.desspd =np.array([]) #desired speed [m/s]
# Display information on label
self.label = [] # Text and bitmap of traffic label
self.trailcol = [] # Trail color: default 'Blue'
# Area
self.inside = []
# Transmitted data to other aircraft due to truncated effect
self.adsbtime=np.array([])
self.adsblat=np.array([])
self.adsblon=np.array([])
self.adsbalt=np.array([])
self.adsbtrk=np.array([])
self.adsbtas=np.array([])
self.adsbgs=np.array([])
self.adsbvs=np.array([])
#-----------------------------------------------------------------------------
# Not per aircraft data
# Scheduling of FMS and ASAS
self.t0fms = -999. # last time fms was called
self.dtfms = 1.01 # interval for fms
self.t0asas = -999. # last time ASAS was called
self.dtasas = 1.00 # interval for ASAS
# Flight performance scheduling
self.perfdt = 0.1 # [s] update interval of performance limits
self.perft0 = -self.perfdt # [s] last time checked (in terms of sim.t)
self.warned2 = False # Flag: Did we warn for default engine parameters yet?
# ASAS objects: Conflict Database
self.dbconf = Dbconf(self,300., 5.*nm, 1000.*ft) # hard coded values to be replaced
# Import navigation data base
self.navdb = Navdatabase("global") # Read nav data from global folder
# Traffic area: delete traffic when it leaves this area (so not when outside)
self.swarea = False
self.arealat0 = 0.0 # [deg] lower latitude defining area
self.arealat1 = 0.0 # [deg] upper latitude defining area
self.arealat0 = 0.0 # [deg] lower longitude defining area
self.arealat1 = 0.0 # [deg] upper longitude defining area
self.areafloor = -999999.0 # [m] Delete when descending through this h
self.areadt = 5.0 # [s] frequency of area check (simtime)
self.areat0 = -100. # last time checked
# Taxi switch
self.swtaxi = False # Default OFF: delete traffic below 1500 ft
# Research Area ("Square" for Square, "Circle" for Circle area)
self.area = ""
# Metrics
self.metricSwitch = 0
self.metric = Metric()
# Bread crumbs for trails
self.lastlat = []
self.lastlon = []
self.lasttim = []
self.trails = Trails()
self.swtrails = False # Default switched off
# ADS-B Coverage area
self.swAdsbCoverage = False
# Noise (turbulence, ADBS-transmission noise, ADSB-truncated effect)
self.setNoise(False)
self.eps = np.array([])
return
def create(self, acid, actype, aclat, aclon, achdg, acalt, acspd):
"""Create an aircraft"""
# Check if not already exist
if self.id.count(acid.upper()) > 0:
return # already exists do nothing
# Increase number of aircraft
self.ntraf = self.ntraf + 1
# Process input
self.id.append(acid.upper())
self.type.append(actype)
self.lat = np.append(self.lat, aclat)
self.lon = np.append(self.lon, aclon)
self.trk = np.append(self.trk, achdg) # TBD: add conversion hdg => trk
self.alt = np.append(self.alt, acalt)
self.fll = np.append(self.fll, (acalt)/100)
self.vs = np.append(self.vs, 0.)
self.rho = np.append(self.rho, density(acalt))
self.temp = np.append(self.temp, temp(acalt))
self.dtemp = np.append(self.dtemp, 0) # at the moment just ISA conditions
self.tas = np.append(self.tas, acspd)
self.gs = np.append(self.gs, acspd)
self.cas = np.append(self.cas, tas2cas(acspd, acalt))
self.M = np.append (self.M, tas2mach(acspd, acalt))
# AC is initialized with neutral max bank angle
self.bank = np.append(self.bank, 25.)
if self.ntraf<2:
self.bphase = np.deg2rad(np.array([15,35,35,35,15,45]))
self.hdgsel = np.append(self.hdgsel, False)
#------------------------------Performance data--------------------------------
# Type specific data
#(temporarily default values)
self.avsdef = np.append(self.avsdef, 1500. * fpm) # default vertical speed of autopilot
self.aphi = np.append(self.aphi, radians(25.)) # bank angle setting of autopilot
self.ax = np.append(self.ax, kts) # absolute value of longitudinal accelleration
# Crossover altitude
self.abco = np.append(self.abco, 0)
self.belco = np.append(self.belco, 1)
# performance data
self.perf.create(actype)
# Traffic autopilot settings: hdg[deg], spd (CAS,m/s), alt[m], vspd[m/s]
self.ahdg = np.append(self.ahdg, achdg) # selected heading [deg]
self.aspd = np.append(self.aspd, tas2eas(acspd, acalt)) # selected spd(eas) [m/s]
self.aptas = np.append(self.aptas, vcas2tas(self.aspd, self.alt)) # [m/s]
self.ama = np.append(self.ama, 0.) # selected spd above crossover (Mach) [-]
self.aalt = np.append(self.aalt, acalt) # selected alt[m]
self.afll = np.append(self.afll, (acalt/100)) # selected fl[ft/100]
self.avs = np.append(self.avs, 0.) # selected vertical speed [m/s]
# limit settings: initialize with 0
self.lspd = np.append(self.lspd, 0.0)
self.lalt = np.append(self.lalt, 0.0)
self.lvs = np.append(self.lvs, 0.0)
# Traffic navigation information
self.dest.append("")
self.orig.append("")
# LNAV route navigation
self.swlnav = np.append(self.swlnav, False) # Lateral (HDG) based on nav
self.swvnav = np.append(self.swvnav, False) # Vertical/longitudinal (ALT+SPD) based on nav info
self.actwplat = np.append(self.actwplat, 89.99) # Active WP latitude
self.actwplon = np.append(self.actwplon, 0.0) # Active WP longitude
self.actwpalt = np.append(self.actwpalt, 0.0) # Active WP altitude
self.actwpspd = np.append(self.actwpspd, -999.) # Active WP speed
self.actwpturn = np.append(self.actwpturn, 1.0) # Distance to active waypoint where to turn
# VNAV cruise level
self.crzalt = np.append(self.crzalt,-999.) # Cruise altitude[m] <0=None
# Route info
self.route.append(Route(self.navdb)) # create empty route connected with nav databse
# ASAS info: no conflict => -1
self.iconf.append(-1) # index in 'conflicting' aircraft database
self.asasactive = np.append(self.asasactive, False)
self.asashdg = np.append(self.asashdg, achdg)
self.asasspd = np.append(self.asasspd, tas2eas(acspd, acalt))
self.asasalt = np.append(self.asasalt, acalt)
self.asasvsp = np.append(self.asasvsp, 0.)
# Area variable set to False to avoid deletion upon creation outside
self.inside.append(False)
# Display information on label
self.label.append(['', '', '', 0])
# Bread crumbs for trails
self.trailcol.append(self.trails.defcolor)
self.lastlat = np.append(self.lastlat, aclat)
self.lastlon = np.append(self.lastlon, aclon)
self.lasttim = np.append(self.lasttim, 0.0)
# ADS-B Coverage area
self.swAdsbCoverage = False
# Transmitted data to other aircraft due to truncated effect
self.adsbtime=np.append(self.adsbtime,np.random.rand(self.trunctime))
self.adsblat=np.append(self.adsblat,aclat)
self.adsblon=np.append(self.adsblon,aclon)
self.adsbalt=np.append(self.adsbalt,acalt)
self.adsbtrk=np.append(self.adsbtrk,achdg)
self.adsbtas=np.append(self.adsbtas,acspd)
self.adsbgs=np.append(self.adsbgs,acspd)
self.adsbvs=np.append(self.adsbvs,0.)
self.eps = np.append(self.eps, 0.01)
return
def delete(self, acid):
"""Delete an aircraft"""
try: # prevent error due to not found
idx = self.id.index(acid)
except:
return False
del self.id[idx]
del self.type[idx]
# Traffic basic data
self.lat = np.delete(self.lat, idx)
self.lon = np.delete(self.lon, idx)
self.trk = np.delete(self.trk, idx)
self.alt = np.delete(self.alt, idx)
self.fll = np.delete(self.fll, idx)
self.vs = np.delete(self.vs, idx)
self.tas = np.delete(self.tas, idx)
self.gs = np.delete(self.gs, idx)
self.cas = np.delete(self.cas, idx)
self.M = np.delete(self.M, idx)
self.T = np.delete(self.T, idx)
self.p = np.delete(self.p, idx)
self.rho = np.delete(self.rho, idx)
self.temp = np.delete(self.temp, idx)
self.dtemp = np.delete(self.dtemp, idx)
self.hdgsel = np.delete(self.hdgsel, idx)
self.bank = np.delete(self.bank, idx)
# Crossover altitude
self.abco = np.delete(self.abco, idx)
self.belco = np.delete(self.belco, idx)
# Type specific data (temporarily default values)
self.avsdef = np.delete(self.avsdef, idx)
self.aphi = np.delete(self.aphi, idx)
self.ax = np.delete(self.ax, idx)
# performance data
self.perf.delete(idx)
# Traffic autopilot settings: hdg[deg], spd (CAS,m/s), alt[m], vspd[m/s]
self.ahdg = np.delete(self.ahdg, idx)
self.aspd = np.delete(self.aspd, idx)
self.ama = np.delete(self.ama, idx)
self.aptas = np.delete(self.aptas, idx)
self.aalt = np.delete(self.aalt, idx)
self.afll = np.delete(self.afll, idx)
self.avs = np.delete(self.avs, idx)
# limit settings
self.lspd = np.delete(self.lspd, idx)
self.lalt = np.delete(self.lalt, idx)
self.lvs = np.delete(self.lvs, idx)
# Traffic navigation variables
del self.dest[idx]
del self.orig[idx]
self.swlnav = np.delete(self.swlnav, idx)
self.swvnav = np.delete(self.swvnav, idx)
self.actwplat = np.delete(self.actwplat, idx)
self.actwplon = np.delete(self.actwplon, idx)
self.actwpalt = np.delete(self.actwpalt, idx)
self.actwpspd = np.delete(self.actwpspd, idx)
self.actwpturn = np.delete(self.actwpturn, idx)
# VNAV cruise level
self.crzalt = np.delete(self.crzalt, idx)
# Route info
del self.route[idx]
# ASAS info
del self.iconf[idx]
self.asasactive=np.delete(self.asasactive, idx)
self.asashdg=np.delete(self.asashdg, idx)
self.asasspd=np.delete(self.asasspd, idx)
self.asasalt=np.delete(self.asasalt, idx)
self.asasvsp=np.delete(self.asasvsp, idx)
self.desalt=np.delete(self.desalt, idx)
self.desvs=np.delete(self.desvs, idx)
self.desspd=np.delete(self.desspd, idx)
self.deshdg=np.delete(self.deshdg, idx)
# Metrics, area
del self.inside[idx]
# Traffic display data: label
del self.label[idx]
# Delete bread crumb data
self.lastlat = np.delete(self.lastlat, idx)
self.lastlon = np.delete(self.lastlon, idx)
self.lasttim = np.delete(self.lasttim, idx)
del self.trailcol[idx]
# Transmitted data to other aircraft due to truncated effect
self.adsbtime=np.delete(self.adsbtime,idx)
self.adsblat=np.delete(self.adsblat,idx)
self.adsblon=np.delete(self.adsblon,idx)
self.adsbalt=np.delete(self.adsbalt,idx)
self.adsbtrk=np.delete(self.adsbtrk,idx)
self.adsbtas=np.delete(self.adsbtas,idx)
self.adsbgs=np.delete(self.adsbgs,idx)
self.adsbvs=np.delete(self.adsbvs,idx)
# Decrease number fo aircraft
self.ntraf = self.ntraf - 1
self.eps = np.delete(self.eps, idx)
return True
def deleteall(self):
"""Clear traffic buffer"""
ndel = self.ntraf
for i in range(ndel):
self.delete(self.id[-1])
self.ntraf = 0
self.dbconf.reset()
self.perf.reset
return
def update(self):
"""Sim and command objects quick access"""
sim = self.tmx.sim
cmd = self.tmx.cmd
if (sim.mode == sim.op and sim.dt > 0.0 and self.ntraf > 0):
self.dts.append(sim.dt)
#---------------- Atmosphere ----------------
self.T, self.rho, self.p = vatmos(self.alt)
#-------------- Performance limits autopilot settings --------------
# Check difference with AP settings for trafperf and autopilot
self.delalt = self.aalt - self.alt # [m]
# below crossover altitude: CAS=const, above crossover altitude: MA = const
# aptas hast to be calculated before delspd
self.aptas = vcas2tas(self.aspd, self.alt)*self.belco + vmach2tas(self.ama, self.alt)*self.abco
self.delspd = self.aptas - self.tas
###############################################################################
# Debugging: add 10000 random aircraft
# if sim.t>1.0 and self.ntraf<1000:
# for i in range(10000):
# acid="KL"+str(i)
# aclat = random.random()*180.-90.
# aclon = random.random()*360.-180.
# achdg = random.random()*360.
# acalt = (random.random()*18000.+2000.)*0.3048
# self.create(acid,'B747',aclat,aclon,achdg,acalt,350.)
#
#################################################################################
#-------------------- ADSB update: --------------------
self.adsbtime+=sim.dt
ADSB_update=np.where(self.adsbtime>self.trunctime)
if not self.ADSBtrunc:
ADSB_update=range(self.ntraf)
for i in ADSB_update:
self.adsbtime[i]-=self.trunctime
self.adsblat[i]=self.lat[i]
self.adsblon[i]=self.lon[i]
self.adsbalt[i]=self.alt[i]
self.adsbtrk[i]=self.trk[i]
self.adsbtas[i]=self.tas[i]
self.adsbgs[i]=self.gs[i]
self.adsbvs[i]=self.vs[i]
#------------------- ASAS update: ---------------------
# Scheduling: when dt has passed or restart:
if self.t0asas+self.dtasas<sim.t or sim.t<self.t0asas \
and self.dbconf.swasas:
self.t0asas = sim.t
# Save old result
iconf0 = np.array(self.iconf)
# Call with traffic database and sim data
self.dbconf.cd_state(sim)
self.dbconf.cr_eby(sim)
self.dbconf.APorASAS(sim)
# Reset label because of colour change
chnged = np.where(iconf0!=np.array(self.iconf))[0]
for i in chnged:
self.label[i]=[" "," ", ""," "]
#----------------- FMS GUIDANCE & NAVIGATION ------------------
# Scheduling: when dt has passed or restart:
if self.t0fms+self.dtfms<sim.t or sim.t<self.t0fms:
self.t0fms = sim.t
# FMS LNAV mode:
qdr, dist = qdrdist(self.lat, self.lon, self.actwplat, self.actwplon) #[deg][nm]
# Check whether shift based dist [nm] is required, set closer than WP turn distance
iwpclose = np.where(self.swlnav*(dist < self.actwpturn))[0]
# Shift for aircraft i where necessary
for i in iwpclose:
# Get next wp (lnavon = False if no more waypoints)
lat, lon, alt, spd, xtoalt, toalt, lnavon = \
self.route[i].getnextwp() # note: xtoalt,toalt in [m]
# End of route/no more waypoints: switch off LNAV
if not lnavon:
self.swlnav[i] = False # Drop LNAV at end of route
# In case of no LNAV, do not allow VNAV mode on it sown
if not self.swlnav[i]:
self.swvnav[i] = False
self.actwplat[i] = lat
self.actwplon[i] = lon
# User entered altitude
if alt >= 0.:
self.actwpalt[i] = alt
# VNAV calculated altitude is available and active
if toalt >= 0. and self.swvnav[i]:
# Descent VNAV mode (T/D logic)
if self.alt[i]>toalt+10.*ft: # Descent part is in this range of waypoints:
# Flat earth distance to next wp
dy = (lat-self.lat[i])
dx = (lon-self.lon[i])*cos(radians(lat))
dist2wp = 60.*nm*sqrt(dx*dx+dy*dy)
steepness = 3000.*ft/(10.*nm) # 1:3 rule of thumb for now
maxaltwp = toalt + xtoalt*steepness # max allowed altitude at next wp
self.actwpalt[i] = min(self.alt[i],maxaltwp) #To descend now or descend later?
if maxaltwp<self.alt[i]: # if descent is necessary with maximum steepness
self.aalt[i] = self.actwpalt[i] # dial in altitude of next waypoint as calculated
t2go = max(0.1,dist2wp)/max(0.01,self.gs[i])
self.avs[i] = (self.actwpalt[i] - self.alt[i])/t2go
else:
print "else 1"
pass # TBD
# Climb VNAV mode: climb as soon as possible (T/C logic)
elif self.swvnav[i] and self.alt[i]<toalt-10.*ft:
# Flat earth distance to next wp
dy = (lat-self.lat[i])
dx = (lon-self.lon[i])*cos(radians(lat))
dist2wp = 60.*nm*sqrt(dx*dx+dy*dy)
self.aalt[i] = self.actwpalt[i] # dial in altitude of next waypoint as calculated
t2go = max(0.1,dist2wp)/max(0.01,self.gs[i])
self.avs[i] = (self.actwpalt[i] - self.alt[i])/t2go
if spd>0. and lnavon and self.swvnav[i]:
if spd<2.0:
self.actwpspd[i] = mach2cas(spd,self.alt[i])
else:
self.actwpspd[i] = cas2tas(spd,self.alt[i])
else:
self.actwpspd[i] = -999.
# Calculate distance before waypoint where to start the turn
# Turn radius: R = V2 tan phi / g
# Distance to turn: wpturn = R * tan (1/2 delhdg) but max 4 times radius
turnrad = self.tas[i]*self.tas[i]/tan(self.bank[i]) /g0 /nm # [nm] default bank angle per flight phase
# print turnrad
dy = (self.actwplat[i]-self.lat[i])
dx = (self.actwplon[i]-self.lon[i])*cos(radians(self.lat[i]))
qdr[i] = degrees(atan2(dx,dy))
self.actwpturn[i] = max(3.,abs(turnrad*tan(radians(0.5*degto180(qdr[i]- \
self.route[i].wpdirfrom[self.route[i].iactwp]))))) # [nm]
# Set headings based on swlnav
self.ahdg = np.where(self.swlnav, qdr, self.ahdg)
# NOISE: Turbulence
if self.turbulence:
timescale=np.sqrt(sim.dt)
trkrad=np.radians(self.trk)
#write turbulences in array
turb=np.array(self.standardturbulence)
turb=np.where(turb>1e-6,turb,1e-6)
#horizontal flight direction
turbhf=np.random.normal(0,turb[0]*timescale,self.ntraf) #[m]
#horizontal wing direction
turbhw=np.random.normal(0,turb[1]*timescale,self.ntraf) #[m]
#vertical direction
turbalt=np.random.normal(0,turb[2]*timescale,self.ntraf) #[m]
#latitudinal, longitudinal direction
turblat=np.cos(trkrad)*turbhf-np.sin(trkrad)*turbhw #[m]
turblon=np.sin(trkrad)*turbhf+np.cos(trkrad)*turbhw #[m]
else:
turbalt=np.zeros(self.ntraf) #[m]
turblat=np.zeros(self.ntraf) #[m]
turblon=np.zeros(self.ntraf) #[m]
# ASAS AP switches
#--------- Select Autopilot settings to follow: destination or ASAS ----------
# desired autopilot settings due to ASAS
self.deshdg = self.asasactive*self.asashdg+(1-self.asasactive)*self.ahdg
self.desspd = self.asasactive*self.asasspd+(1-self.asasactive)*self.aspd
self.desalt = self.asasactive*self.asasalt+(1-self.asasactive)*self.aalt
self.desvs = self.asasactive*self.asasvsp+(1-self.asasactive)*self.avs
# check for the flight envelope
self.perf.limits()
# update autopilot settings with values within the flight envelope
# speed
self.aspd = (self.lspd ==0)*self.desspd + (self.lspd!=0)*self.lspd
# altitude
self.aalt = (self.lalt ==0)*self.desalt + (self.lalt!=0)*self.lalt
# hdg
self.ahdg = self.deshdg
# vs
self.avs = (self.lvs==0)*self.desvs + (self.lvs!=0)*self.lvs
# below crossover altitude: CAS=const, above crossover altitude: MA = const
#climb/descend above crossover: Ma = const, else CAS = const
#ama is fixed when above crossover
check = self.abco*(self.ama == 0.)
swma = np.where(check==True)
self.ama[swma] = cas2mach(self.aspd[swma], self.alt[swma])
# ama is deleted when below crossover
check2 = self.belco*(self.ama!=0.)
swma2 = np.where(check2==True)
self.ama[swma2] = 0.
#---------- Basic Autopilot modes ----------
# SPD HOLD/SEL mode: aspd = autopilot selected speed (first only eas)
# for information:
self.aptas = (self.actwpspd>0.)*self.actwpspd + \
(self.actwpspd<=0.)*self.aptas
self.delspd = self.aptas - self.tas
swspdsel = np.abs(self.delspd) > 0.4 # <1 kts = 0.514444 m/s
ax = np.minimum(abs(self.delspd / sim.dt), self.ax)
self.tas = swspdsel * (self.tas + ax * np.sign(self.delspd) * \
sim.dt) + (1. - swspdsel) * self.aptas
# print "DELSPD", self.delspd/sim.dt, "AX", self.ax, "SELECTED", ax
# without that part: non-accelerating ac would have TAS = 0
# print "1-sw", (1. - swspdsel) * self.aptas
# print "NEW TAS", self.tas
# Speed conversions
self.cas = vtas2cas(self.tas, self.alt)
self.gs = self.tas
self.M = vtas2mach(self.tas, self.alt)
# Update performance every self.perfdt seconds
if abs(sim.t - self.perft0) >= self.perfdt:
self.perft0 = sim.t
self.perf.perf()
# update altitude
self.eps = np.array(self.ntraf * [0.01]) # almost zero for misc purposes
swaltsel = np.abs(self.aalt-self.alt) > \
np.maximum(3.,np.abs(2. * sim.dt * np.abs(self.vs))) # 3.[m] = 10 [ft] eps alt
# print swaltsel
self.vs = swaltsel*((1-self.swvnav)*np.abs(1500./60.*ft) + \
self.swvnav*np.abs(self.avs)*np.sign(self.aalt-self.alt))
self.alt = swaltsel * (self.alt + self.vs * sim.dt) + \
(1. - swaltsel) * self.aalt + turbalt
# HDG HOLD/SEL mode: ahdg = ap selected heading
delhdg = (self.ahdg - self.trk + 180.) % 360 - 180. #[deg]
# print delhdg
# omega = np.degrees(g0 * np.tan(self.aphi) / \
# np.maximum(self.tas, self.eps))
# nominal bank angles per phase from BADA 3.12
omega = np.degrees(g0 * np.tan(self.bank) / \
np.maximum(self.tas, self.eps))
self.hdgsel = np.abs(delhdg) > np.abs(2. * sim.dt * omega)
self.trk = (self.trk + sim.dt * omega * self.hdgsel * np.sign(delhdg)) % 360.
#--------- Kinematics: update lat,lon,alt ----------
ds = sim.dt * self.gs
self.lat = self.lat + np.degrees(ds * np.cos(np.radians(self.trk)+turblat) \
/ Rearth)
self.lon = self.lon + np.degrees(ds * np.sin(np.radians(self.trk)+turblon) \
/ np.cos(np.radians(self.lat)) / Rearth)
# Update trails when switched on
if self.swtrails:
self.trails.update(sim.t, self.lat, self.lon,
self.lastlat, self.lastlon,
self.lasttim, self.id, self.trailcol)
else:
self.lastlat = self.lat
self.lastlon = self.lon
self.lattime = sim.t
# Update metrics
if self.metricSwitch == 1:
self.metric.update(self, sim, cmd)
# ----------------AREA check----------------
# Update area once per areadt seconds:
if self.swarea and abs(sim.t - self.areat0) > self.areadt:
# Update loop timer
self.areat0 = sim.t
# Chekc all aicraft
for i in xrange(self.ntraf):
# Current status
if self.area == "Square":
inside = self.arealat0 <= self.lat[i] <= self.arealat1 and \
self.arealon0 <= self.lon[i] <= self.arealon1 and \
self.alt[i] >= self.areafloor and \
(self.alt[i] >= 1500 or self.swtaxi)
elif self.area == "Circle":
## Average of lat
latavg = (radians(self.lat[i]) + radians(self.metric.fir_circle_point[0])) / 2
cosdlat = (cos(latavg))
# Distance x to centroid
dx = (self.lon[i] - self.metric.fir_circle_point[1]) * cosdlat * 60
dx2 = dx * dx
# Distance y to centroid
dy = self.lat[i] - self.metric.fir_circle_point[0]
dy2 = dy * dy * 3600
# Radius squared
r2 = self.metric.fir_circle_radius * self.metric.fir_circle_radius
# Inside if smaller
inside = (dx2 + dy2) < r2
# Compare with previous: when leaving area: delete command
if self.inside[i] and not inside:
cmd.stack("DEL " + self.id[i])
# Update area status
self.inside[i] = inside
return
def findnearest(self, lat, lon):
"""Find nearest aircraft"""
if self.ntraf > 0:
d2 = (lat - self.lat) ** 2 + cos(radians(lat)) * (lon - self.lon) ** 2
idx = np.argmin(d2)
del d2
return idx
else:
return -1
def id2idx(self, acid):
"""Find index of aircraft id"""
try:
return self.id.index(acid.upper())
except:
return -1
def changeTrailColor(self, color, idx):
"""Change color of aircraft trail"""
# print color
# print idx
# print " " + str(self.trails.colorsOfAC[idx])
self.trailcol[idx] = self.trails.colorList[color]
# print " " + str(self.trails.colorsOfAC[idx])
return
def setNoise(self,A):
"""Noise (turbulence, ADBS-transmission noise, ADSB-truncated effect)"""
self.noise=A
self.trunctime=1 # seconds
self.transerror = [1,100, 100 * ft] #[degree,m,m] standard bearing, distance, altitude error
self.standardturbulence = [0,0.1,0.1] #m/s standard turbulence (nonnegative)
# in (horizontal flight direction, horizontal wing direction, vertical)
if self.noise:
self.turbulence=True
self.ADSBtransnoise=True
self.ADSBtrunc=True
else:
self.turbulence=False
self.ADSBtransnoise=False
self.ADSBtrunc=False
return
def engchange (self, acid, engid):
"""Change of engines"""
self.perf.engchange(acid, engid)
return
def __init__(self, tmx):
self.tmx = tmx # tmx object contains sim, scr and other main objects
self.dts = []
self.ntraf = 0
# model-specific parameters.
# Default: BlueSky internal performance model.
# Insert your BADA files to the folder "BlueSky/data/coefficients/BADA"
# for working with EUROCONTROL`s Base of Aircraft Data revision 3.12
# Check for BADA OPF file
path = os.path.dirname(__file__) + '/../../data/coefficients/BADA/'
files = os.listdir(path)
self.bada = False
for f in files:
if f.upper().find(".OPF")!=-1:
self.bada=True
break
# Initialize correct performance models
if self.bada:
self.perf = PerfBADA(self)
else:
self.perf = Perf(self)
self.dts = []
self.ntraf = 0
# Create datalog instance
self.log = Datalog()
# Traffic list & arrays definition
# !!!IMPORTANT NOTE!!!
# Anny variables added here should also be added in the Traffic
# methods self.create() (append) and self.delete() (delete)
# which can be found directly below __init__
# Traffic basic flight data
# Traffic basic flight data
self.id = [] # identifier (string)
self.type = [] # aircaft type (string)
self.lat = np.array([]) # latitude [deg]
self.lon = np.array([]) # longitude [deg]
self.trk = np.array([]) # track angle [deg]
self.tas = np.array([]) # true airspeed [m/s]
self.gs = np.array([]) # ground speed [m/s]
self.cas = np.array([]) # callibrated airspeed [m/s]
self.M = np.array([]) # mach number
self.alt = np.array([]) # altitude [m]
self.fll = np.array([]) # flight level [ft/100]
self.vs = np.array([]) # vertical speed [m/s]
self.rho = np.array([]) # atmospheric air density [m/s]
self.temp = np.array([]) # atmospheric air temperature [K]
self.dtemp = np.array([]) # delta t for non-ISA conditions
# Traffic performance data
self.avsdef = np.array([]) # [m/s]default vertical speed of autopilot
self.aphi = np.array([]) # [rad] bank angle setting of autopilot
self.ax = np.array([]) # [m/s2] absolute value of longitudinal accelleration
self.bank = np.array([]) # nominal bank angle, [radian]
self.bphase = np.array([]) # standard bank angles per phase
self.hdgsel = np.array([]) # determines whether aircraft is turning
# Crossover altitude
self.abco = np.array([])
self.belco = np.array([])
# Traffic autopilot settings
self.ahdg = [] # selected heading [deg]
self.aspd = [] # selected spd(eas) [m/s]
self.aptas = [] # just for initializing
self.ama = [] # selected spd above crossover altitude (Mach) [-]
self.aalt = [] # selected alt[m]
self.afll = [] # selected fl [ft/100]
self.avs = [] # selected vertical speed [m/s]
# limit settings
self.lspd = [] # limit speed
self.lalt = [] # limit altitude
self.lvs = [] # limit vertical speed due to thrust limitation
# Traffic navigation information
self.orig = [] # Four letter code of origin airport
self.dest = [] # Four letter code of destination airport
# LNAV route navigation
self.swlnav = np.array([]) # Lateral (HDG) based on nav?
self.swvnav = np.array([]) # Vertical/longitudinal (ALT+SPD) based on nav info
self.actwplat = np.array([]) # Active WP latitude
self.actwplon = np.array([]) # Active WP longitude
self.actwpalt = np.array([]) # Active WP altitude to arrive at
self.actwpspd = np.array([]) # Active WP speed
self.actwpturn = np.array([]) # Distance when to turn to next waypoint
# VNAV cruise level
self.crzalt = np.array([]) # Cruise altitude[m]
# Route info
self.route = []
# ASAS info per aircraft:
self.iconf = [] # index in 'conflicting' aircraft database
self.asasactive = np.array([]) # whether the autopilot follows ASAS or not
self.asashdg = np.array([]) # heading provided by the ASAS [deg]
self.asasspd = np.array([]) # speed provided by the ASAS (eas) [m/s]
self.asasalt = np.array([]) # speed alt by the ASAS [m]
self.asasvsp = np.array([]) # speed vspeed by the ASAS [m/s]
self.desalt = np.array([]) #desired altitude [m]
self.deshdg =np.array([]) #desired heading
self.desvs =np.array([]) #desired vertical speed [m/s]
self.desspd =np.array([]) #desired speed [m/s]
# Display information on label
self.label = [] # Text and bitmap of traffic label
self.trailcol = [] # Trail color: default 'Blue'
# Area
self.inside = []
# Transmitted data to other aircraft due to truncated effect
self.adsbtime=np.array([])
self.adsblat=np.array([])
self.adsblon=np.array([])
self.adsbalt=np.array([])
self.adsbtrk=np.array([])
self.adsbtas=np.array([])
self.adsbgs=np.array([])
self.adsbvs=np.array([])
#-----------------------------------------------------------------------------
# Not per aircraft data
# Scheduling of FMS and ASAS
self.t0fms = -999. # last time fms was called
self.dtfms = 1.01 # interval for fms
self.t0asas = -999. # last time ASAS was called
self.dtasas = 1.00 # interval for ASAS
# Flight performance scheduling
self.perfdt = 0.1 # [s] update interval of performance limits
self.perft0 = -self.perfdt # [s] last time checked (in terms of sim.t)
self.warned2 = False # Flag: Did we warn for default engine parameters yet?
# ASAS objects: Conflict Database
self.dbconf = Dbconf(self,300., 5.*nm, 1000.*ft) # hard coded values to be replaced
# Import navigation data base
self.navdb = Navdatabase("global") # Read nav data from global folder
# Traffic area: delete traffic when it leaves this area (so not when outside)
self.swarea = False
self.arealat0 = 0.0 # [deg] lower latitude defining area
self.arealat1 = 0.0 # [deg] upper latitude defining area
self.arealat0 = 0.0 # [deg] lower longitude defining area
self.arealat1 = 0.0 # [deg] upper longitude defining area
self.areafloor = -999999.0 # [m] Delete when descending through this h
self.areadt = 5.0 # [s] frequency of area check (simtime)
self.areat0 = -100. # last time checked
# Taxi switch
self.swtaxi = False # Default OFF: delete traffic below 1500 ft
# Research Area ("Square" for Square, "Circle" for Circle area)
self.area = ""
# Metrics
self.metricSwitch = 0
self.metric = Metric()
# Bread crumbs for trails
self.lastlat = []
self.lastlon = []
self.lasttim = []
self.trails = Trails()
self.swtrails = False # Default switched off
# ADS-B Coverage area
self.swAdsbCoverage = False
# Noise (turbulence, ADBS-transmission noise, ADSB-truncated effect)
self.setNoise(False)
self.eps = np.array([])
return
def __init__(self, name):
Metric.__init__(self, name)
self._count = 0
return
def __init__(self, name, fn):
Metric.__init__(self, name)
self._fn = fn
return