def post(self, request, format=None):
if request.user.is_authenticated:
try:
# Start connection with VASSAR
user_info = get_or_create_user_information(
request.session, request.user, 'EOSS')
port = user_info.eosscontext.vassar_port
client = VASSARClient(port)
client.startConnection()
inputType = request.data['inputType']
status = None
if inputType == 'binary':
status = client.client.isGABinaryInputRunning()
elif inputType == 'discrete':
status = client.client.isGADiscreteInputRunning()
else:
raise ValueError(
'Unrecognized input type: {0}'.format(inputType))
# End the connection before return statement
client.endConnection()
return Response({'ga_status': status})
except Exception:
logger.exception('Exception in starting the GA!')
client.endConnection()
return Response('')
else:
return Response('This is only available to registered users!')
class RunLocalSearch(APIView):
def post(self, request, format=None):
try:
# Start connection with VASSAR
user_info = get_or_create_user_information(request.session,
request.user, 'EOSS')
port = user_info.eosscontext.vassar_port
self.VASSARClient = VASSARClient(port)
self.VASSARClient.startConnection()
inputs = request.data['inputs']
inputs = json.loads(inputs)
architectures = self.VASSARClient.runLocalSearch(inputs)
for arch in architectures:
arch['id'] = user_info.eosscontext.last_arch_id
add_design(arch, user_info.eosscontext, False)
user_info.save()
# End the connection before return statement
self.VASSARClient.endConnection()
return Response(architectures)
except Exception:
logger.exception('Exception in evaluating an architecture')
self.VASSARClient.endConnection()
return Response('')
class CriticizeArchitecture(APIView):
def post(self, request, format=None):
try:
user_info = get_or_create_user_information(request.session, request.user, 'EOSS')
port = user_info.eosscontext.vassar_port
inputs = request.POST['inputs']
inputs = json.loads(inputs)
critique = self.get_history_critique(inputs)
critique += self.get_expert_critique(inputs, port)
critiques = json.dumps(critique)
return Response(critiques)
except Exception:
logger.exception('Exception in generating a critique of an architecture')
return Response('')
def get_history_critique(self, inputs):
try:
pass
# historian_critic = Critic()
#
# critique = historian_critic.criticizeArchitecture(inputs)
#
# return critique
except Exception:
logger.exception('Exc in generating a critique using historical database')
raise
def get_expert_critique(self, inputs, port):
try:
self.VASSARClient = VASSARClient(port)
# Start connection with VASSAR
self.VASSARClient.startConnection()
critique = self.VASSARClient.critiqueArchitecture(inputs, False)
# End the connection before return statement
self.VASSARClient.endConnection()
except Exception:
logger.exception('Exc in generating a critique using expert knowledge base')
self.VASSARClient.endConnection()
raise
def get_panel_scores(design_id, designs, panel, context: UserInformation):
port = context.eosscontext.vassar_port
client = VASSARClient(port)
try:
# Start connection with VASSAR
client.startConnection()
num_design_id = int(design_id)
stakeholders_to_excel = {
"atmospheric": "ATM",
"oceanic": "OCE",
"terrestrial": "TER",
"weather": "WEA",
"climate": "CLI",
"land and ecosystems": "ECO",
"water": "WAT",
"human health": "HEA"
}
panel_code = stakeholders_to_excel[panel.lower()]
panel_scores = client.client.getPanelScoreExplanation(
context.eosscontext.problem,
json.loads(designs[num_design_id].inputs), panel_code)
# End the connection before return statement
client.endConnection()
return panel_scores
except Exception:
logger.exception('Exception in loading panel score information')
client.endConnection()
return None
def get_instruments_for_stakeholder(stakeholder, context: UserInformation):
port = context.eosscontext.vassar_port
client = VASSARClient(port)
try:
# Start connection with VASSAR
client.startConnection()
stakeholders_to_excel = {
"atmospheric": "ATM",
"oceanic": "OCE",
"terrestrial": "TER",
"weather": "WEA",
"climate": "CLI",
"land and ecosystems": "ECO",
"water": "WAT",
"human health": "HEA"
}
panel_code = stakeholders_to_excel[stakeholder.lower()]
stakeholder_instruments = client.client.getInstrumentsForPanel(
context.eosscontext.problem, panel_code)
# End the connection before return statement
client.endConnection()
return stakeholder_instruments
except Exception:
logger.exception('Exception in loading related instruments to a panel')
client.endConnection()
return None
class GetOrbitList(APIView):
def post(self, request, format=None):
try:
# Start connection with VASSAR
user_info = get_or_create_user_information(request.session,
request.user, 'EOSS')
port = user_info.eosscontext.vassar_port
self.VASSARClient = VASSARClient(port)
self.VASSARClient.startConnection()
list = self.VASSARClient.getOrbitList(request.data['problem_name'])
# End the connection before return statement
self.VASSARClient.endConnection()
return Response(list)
except Exception:
logger.exception('Exception in getting the orbit list')
self.VASSARClient.endConnection()
return Response('')
def get_instruments_for_objective(objective, context: UserInformation):
port = context.eosscontext.vassar_port
client = VASSARClient(port)
try:
# Start connection with VASSAR
client.startConnection()
instruments = client.client.getInstrumentsForObjective(
context.eosscontext.problem, objective)
# End the connection before return statement
client.endConnection()
return instruments
except Exception:
logger.exception(
'Exception in loading related instruments to an objective')
client.endConnection()
return None
def get_architecture_scores(design_id, designs, context: UserInformation):
port = context.eosscontext.vassar_port
client = VASSARClient(port)
try:
# Start connection with VASSAR
client.startConnection()
num_design_id = int(design_id)
scores = client.client.getArchitectureScoreExplanation(
context.eosscontext.problem,
json.loads(designs[num_design_id].inputs))
# End the connection before return statement
client.endConnection()
return scores
except Exception:
logger.exception('Exception in loading architecture score information')
client.endConnection()
return None
class EvaluateArchitecture(APIView):
def post(self, request, format=None):
try:
user_info = get_or_create_user_information(request.session,
request.user, 'EOSS')
port = user_info.eosscontext.vassar_port
self.VASSARClient = VASSARClient(port)
# Start connection with VASSAR
self.VASSARClient.startConnection()
inputs = request.data['inputs']
inputs = json.loads(inputs)
architecture = self.VASSARClient.evaluateArchitecture(
user_info.eosscontext.problem, inputs)
is_same = True
for old_arch in user_info.eosscontext.design_set.all():
is_same = True
old_arch_outputs = json.loads(old_arch.outputs)
for i in range(len(old_arch_outputs)):
if old_arch_outputs[i] != architecture['outputs'][i]:
is_same = False
if is_same:
break
if not is_same:
architecture['id'] = user_info.eosscontext.last_arch_id
print(user_info.eosscontext.last_arch_id)
add_design(architecture, user_info.eosscontext, False)
user_info.save()
# End the connection before return statement
self.VASSARClient.endConnection()
return Response(architecture)
except Exception:
logger.exception('Exception in evaluating an architecture')
self.VASSARClient.endConnection()
return Response('')
def post(self, request, format=None):
if request.user.is_authenticated:
try:
user_info = get_or_create_user_information(
request.session, request.user, 'EOSS')
# Start connection with VASSAR
port = user_info.eosscontext.vassar_port
client = VASSARClient(port)
client.startConnection()
problem = request.data['problem']
inputType = request.data['inputType']
# Call the GA stop function on Engineer
if inputType == 'binary':
client.client.stopGABinaryInput(request.user.username)
while client.client.isGABinaryInputRunning():
time.sleep(0.1)
elif inputType == 'discrete':
client.client.stopGADiscreteInput(request.user.username)
while client.client.isGADiscreteInputRunning():
time.sleep(0.1)
else:
raise ValueError(
'Unrecognized input type: {0}'.format(inputType))
# End the connection before return statement
client.endConnection()
return Response('GA stopped correctly!')
except Exception:
logger.exception('Exception in stopping the GA!')
client.endConnection()
return Response('')
else:
return Response('This is only available to registered users!')
def expert_critic(self, design):
# Criticize architecture (based on rules)
port = self.context.eosscontext.vassar_port
problem = self.context.eosscontext.problem
inputs = json.loads(design.inputs)
client = VASSARClient(port)
client.startConnection()
result_list = None
if problem in self.assignation_problems:
result_list = client.client.getCritiqueBinaryInputArch(
problem, inputs)
elif problem in self.partition_problems:
result_list = client.client.getCritiqueDiscreteInputArch(
problem, inputs)
client.endConnection()
result = []
for advice in result_list:
result.append({"type": "Expert", "advice": advice})
return result
def post(self, request, format=None):
try:
# Start connection with VASSAR
user_info = get_or_create_user_information(request.session,
request.user, 'EOSS')
port = user_info.eosscontext.vassar_port
client = VASSARClient(port)
client.startConnection()
assignation_problems = [
'SMAP', 'SMAP_JPL1', 'SMAP_JPL2', 'ClimateCentric'
]
partition_problems = ['Decadal2017Aerosols']
# Get the correct architecture
this_arch = None
arch_id = int(request.data['arch_id'])
problem = request.data['problem']
for arch in user_info.eosscontext.design_set.all():
if arch.id == arch_id:
if problem in assignation_problems:
this_arch = BinaryInputArchitecture(
arch.id, json.loads(arch.inputs),
json.loads(arch.outputs))
elif problem in partition_problems:
this_arch = DiscreteInputArchitecture(
arch.id, json.loads(arch.inputs),
json.loads(arch.outputs))
break
subobjective_explanation = None
if problem in assignation_problems:
subobjective_explanation = client.client.getSubscoreDetailsBinaryInput(
problem, this_arch, request.data['subobjective'])
elif problem in partition_problems:
subobjective_explanation = client.client.getSubscoreDetailsDiscreteInput(
problem, this_arch, request.data['subobjective'])
# End the connection before return statement
client.endConnection()
def explanation_to_json(explanation):
json_exp = {
'subobjective': request.data['subobjective'],
'param': explanation.param,
'attr_names': explanation.attr_names,
'attr_values': explanation.attr_values,
'scores': explanation.scores,
'taken_by': explanation.taken_by,
'justifications': explanation.justifications
}
return json_exp
return Response({
'subobjective':
explanation_to_json(subobjective_explanation)
})
except Exception:
logger.exception(
'Exception when retrieving information from the current architecture!'
)
client.endConnection()
return Response('')
def post(self, request, format=None):
try:
# Start connection with VASSAR
user_info = get_or_create_user_information(request.session,
request.user, 'EOSS')
port = user_info.eosscontext.vassar_port
client = VASSARClient(port)
client.startConnection()
assignation_problems = [
'SMAP', 'SMAP_JPL1', 'SMAP_JPL2', 'ClimateCentric'
]
partition_problems = ['Decadal2017Aerosols']
# Get the correct architecture
this_arch = None
arch_id = int(request.data['arch_id'])
problem = request.data['problem']
for arch in user_info.eosscontext.design_set.all():
if arch.id == arch_id:
if problem in assignation_problems:
this_arch = BinaryInputArchitecture(
arch.id, json.loads(arch.inputs),
json.loads(arch.outputs))
elif problem in partition_problems:
this_arch = DiscreteInputArchitecture(
arch.id, json.loads(arch.inputs),
json.loads(arch.outputs))
break
score_explanation = None
cost_explanation = None
if problem in assignation_problems:
score_explanation = client.client.getArchScienceInformationBinaryInput(
problem, this_arch)
cost_explanation = client.client.getArchCostInformationBinaryInput(
problem, this_arch)
elif problem in partition_problems:
score_explanation = client.client.getArchScienceInformationDiscreteInput(
problem, this_arch)
cost_explanation = client.client.getArchCostInformationDiscreteInput(
problem, this_arch)
# End the connection before return statement
client.endConnection()
def score_to_json(explanation):
json_list = []
for exp in explanation:
json_exp = {
'name': exp.name,
'description': exp.description,
'value': exp.value,
'weight': exp.weight
}
if exp.subscores is not None:
json_exp['subscores'] = score_to_json(exp.subscores)
json_list.append(json_exp)
return json_list
def budgets_to_json(explanation):
json_list = []
for exp in explanation:
json_exp = {
'orbit_name': exp.orbit_name,
'payload': exp.payload,
'launch_vehicle': exp.launch_vehicle,
'total_mass': exp.total_mass,
'total_power': exp.total_power,
'total_cost': exp.total_cost,
'mass_budget': exp.mass_budget,
'power_budget': exp.power_budget,
'cost_budget': exp.cost_budget
}
json_list.append(json_exp)
return json_list
return Response({
'score': score_to_json(score_explanation),
'budgets': budgets_to_json(cost_explanation)
})
except Exception:
logger.exception(
'Exception when retrieving information from the current architecture!'
)
client.endConnection()
return Response('')
def post(self, request, format=None):
if request.user.is_authenticated:
try:
# Start listening for redis inputs to share through websockets
connection = pika.BlockingConnection(
pika.ConnectionParameters(host='localhost'))
channel = connection.channel()
channel.queue_declare(queue=request.user.username + '_gabrain')
channel.queue_purge(queue=request.user.username + '_gabrain')
def callback(ch, method, properties, body):
thread_user_info = get_or_create_user_information(
request.session, request.user, 'EOSS')
message = json.loads(body)
if message['type'] == 'new_arch':
print('Processing some new archs!')
nonlocal inputs_unique_set
# Archs are added one by one
new_archs = [message['data']]
send_back = []
# Add archs to the context data before sending back to user
for arch in new_archs:
hashed_input = hash(tuple(arch['inputs']))
if hashed_input not in inputs_unique_set:
full_arch = {
'id':
thread_user_info.eosscontext.last_arch_id,
'inputs': arch['inputs'],
'outputs': arch['outputs']
}
if thread_user_info.eosscontext.activecontext.show_background_search_feedback:
add_design(full_arch,
thread_user_info.eosscontext,
False)
else:
add_design(full_arch,
thread_user_info.eosscontext,
True)
send_back.append(full_arch)
inputs_unique_set.add(hashed_input)
thread_user_info.save()
# Look for channel to send back to user
channel_layer = get_channel_layer()
background_queue_qs = Design.objects.filter(
activecontext_id__exact=thread_user_info.
eosscontext.activecontext.id)
if background_queue_qs.count() >= 10:
async_to_sync(
channel_layer.send
)(thread_user_info.channel_name, {
'type': 'active.notification',
'notification': {
'title': 'Background search results',
'message':
'The background search has found more than 10 architectures, but you have chosen to not show them. Do you want to see them now?',
'setting':
'show_background_search_feedback'
}
})
if thread_user_info.eosscontext.activecontext.show_background_search_feedback:
back_list = send_archs_from_queue_to_main_dataset(
thread_user_info)
send_back.extend(back_list)
send_archs_back(channel_layer,
thread_user_info.channel_name,
send_back)
if message['type'] == 'ga_started':
# Look for channel to send back to user
channel_layer = get_channel_layer()
async_to_sync(channel_layer.send)(
thread_user_info.channel_name, {
'type': 'ga.started'
})
if message['type'] == 'ga_done':
channel_layer = get_channel_layer()
async_to_sync(channel_layer.send)(
thread_user_info.channel_name, {
'type': 'ga.finished'
})
print('Ending the thread!')
channel.stop_consuming()
channel.basic_consume(callback,
queue=request.user.username + '_gabrain',
no_ack=True)
thread = threading.Thread(target=channel.start_consuming)
thread.start()
# Start connection with VASSAR
user_info = get_or_create_user_information(
request.session, request.user, 'EOSS')
port = user_info.eosscontext.vassar_port
client = VASSARClient(port)
client.startConnection()
problem = request.data['problem']
inputType = request.data['inputType']
# Restart archs queue before starting the GA again
Design.objects.filter(activecontext__exact=user_info.
eosscontext.activecontext).delete()
user_info.eosscontext.last_arch_id = user_info.eosscontext.design_set.count(
)
user_info.eosscontext.save()
# Convert the architecture list and wait for threads to be available (ask for stop again just in case)
thrift_list = []
inputs_unique_set = set()
if inputType == 'binary':
for arch in user_info.eosscontext.design_set.all():
thrift_list.append(
BinaryInputArchitecture(arch.id,
json.loads(arch.inputs),
json.loads(arch.outputs)))
hashed_input = hash(tuple(json.loads(arch.inputs)))
inputs_unique_set.add(hashed_input)
client.client.stopGABinaryInput(request.user.username)
while client.client.isGABinaryInputRunning():
time.sleep(0.1)
client.client.startGABinaryInput(problem, thrift_list,
request.user.username)
elif inputType == 'discrete':
for arch in user_info.eosscontext.design_set.all():
thrift_list.append(
DiscreteInputArchitecture(arch.id,
json.loads(arch.inputs),
json.loads(
arch.outputs)))
hashed_input = hash(tuple(json.loads(arch.inputs)))
inputs_unique_set.add(hashed_input)
client.client.stopGADiscreteInput(request.user.username)
while client.client.isGADiscreteInputRunning():
time.sleep(0.1)
client.client.startGADiscreteInput(problem, thrift_list,
request.user.username)
else:
raise ValueError(
'Unrecognized input type: {0}'.format(inputType))
# End the connection before return statement
client.endConnection()
return Response('GA started correctly!')
except Exception:
logger.exception('Exception in starting the GA!')
return Response('')
else:
return Response('This is only available to registered users!')
def explorer_critic(self, design):
def get_advices_from_bit_string_diff(difference):
out = []
ninstr = len(self.instruments_dataset)
for i in range(len(difference)):
advice = []
if difference[i] == 1:
advice.append("add")
elif difference[i] == -1:
advice.append("remove")
else:
continue
orbit_index = i // ninstr # Floor division
instr_index = i % ninstr # Get the remainder
advice.append("instrument {}".format(
self.instruments_dataset[instr_index]['name']))
if difference[i] == 1:
advice.append("to")
elif difference[i] == -1:
advice.append("from")
advice.append("orbit {}".format(
self.orbits_dataset[orbit_index]['name']))
advice = " ".join(advice)
out.append(advice)
out = ", and ".join(out)
out = out[0].upper() + out[1:]
return out
original_outputs = json.loads(design.outputs)
original_inputs = json.loads(design.inputs)
problem = self.context.eosscontext.problem
port = self.context.eosscontext.vassar_port
client = VASSARClient(port)
client.startConnection()
archs = None
advices = []
if problem in self.assignation_problems:
archs = client.client.runLocalSearchBinaryInput(
problem, json.loads(design.inputs))
for arch in archs:
new_outputs = arch.outputs
new_design_inputs = arch.inputs
diff = [
a - b for a, b in zip(new_design_inputs, original_inputs)
]
advice = [get_advices_from_bit_string_diff(diff)]
# TODO: Generalize the code for comparing each metric. Currently it assumes two metrics: science and cost
if new_outputs[0] > original_outputs[0] and new_outputs[
1] < original_outputs[1]:
# New solution dominates the original solution
advice.append(
" to increase the science benefit and lower the cost.")
elif new_outputs[0] > original_outputs[0]:
advice.append(
" to increase the science benefit (but cost may increase!)."
)
elif new_outputs[1] < original_outputs[1]:
advice.append(
" to lower the cost (but science may decrease too!).")
else:
continue
advice = "".join(advice)
advices.append(advice)
elif problem in self.partition_problems:
archs = client.client.runLocalSearchDiscreteInput(
problem, json.loads(design.inputs))
# TODO: Add the delta code for discrete architectures
client.endConnection()
result = []
for advice in advices:
result.append({"type": "Explorer", "advice": advice})
return result