def get_system(name, args, schema=None, timed=False, model_path=None):
if name in ('rulebased', 'neural'):
lexicon = Lexicon(schema,
args.learned_lex,
stop_words=args.stop_words,
lexicon_path=args.lexicon)
if args.inverse_lexicon:
realizer = InverseLexicon.from_file(args.inverse_lexicon)
else:
realizer = DefaultInverseLexicon()
if name == 'rulebased':
templates = Templates.from_pickle(args.templates)
generator = Generator(templates)
manager = Manager.from_pickle(args.policy)
return RulebasedSystem(lexicon, generator, manager, timed)
elif name == 'neural':
assert args.model_path
return NeuralSystem(schema,
lexicon,
args.model_path,
args.fact_check,
args.decoding,
realizer=realizer)
elif name == 'cmd':
return CmdSystem()
else:
raise ValueError('Unknown system %s' % name)
def get_system(name, args, schema=None, timed=False, model_path=None):
lexicon = PriceTracker(args.price_tracker_model)
if name == 'rulebased':
templates = Templates.from_pickle(args.templates)
generator = Generator(templates)
manager = Manager.from_pickle(args.policy)
return RulebasedSystem(lexicon, generator, manager, timed)
#elif name == 'config-rulebased':
# configs = read_json(args.rulebased_configs)
# return ConfigurableRulebasedSystem(configs, lexicon, timed_session=timed, policy=args.config_search_policy, max_chats_per_config=args.chats_per_config, db=args.trials_db, templates=templates)
elif name == 'cmd':
return CmdSystem()
elif name.startswith('ranker'):
# TODO: hack
#retriever1 = Retriever(args.index+'-1', context_size=args.retriever_context_len, num_candidates=args.num_candidates)
#retriever2 = Retriever(args.index+'-2', context_size=args.retriever_context_len, num_candidates=args.num_candidates)
retriever = Retriever(args.index, context_size=args.retriever_context_len, num_candidates=args.num_candidates)
if name == 'ranker-ir':
return IRRankerSystem(schema, lexicon, retriever)
elif name == 'ranker-ir1':
return IRRankerSystem(schema, lexicon, retriever1)
elif name == 'ranker-ir2':
return IRRankerSystem(schema, lexicon, retriever2)
elif name == 'ranker-neural':
return NeuralRankerSystem(schema, lexicon, retriever, model_path, args.mappings)
else:
raise ValueError
elif name in ('neural-gen', 'neural-sel'):
assert model_path
return NeuralSystem(schema, lexicon, model_path, args.mappings, args.decoding, index=args.index, num_candidates=args.num_candidates, retriever_context_len=args.retriever_context_len, timed_session=timed)
else:
raise ValueError('Unknown system %s' % name)
def system_initialisation(l_pendulum=0.5,
m_pendulum=1.,
f_max=1500,
l_attach=0.17):
"""Generates a oscillatory system and its default initial conditions"""
# Neural parameters for oscillatory system
d = 1.
w = np.array([[0, -5, -5, 0], [-5, 0, 0, -5], [5, -5, 0, 0], [-5, 5, 0,
0]])
b = np.array([3., 3., -3., -3.])
tau = np.array([0.02, 0.02, 0.1, 0.1])
# Pendulum parameters
pendulum_params = PendulumParameters()
pendulum_params.L = l_pendulum
pendulum_params.m = m_pendulum
pendulum = PendulumSystem(pendulum_params)
# Muscles parameters
m1_param = MuscleParameters()
m1_param.f_max = f_max
m2_param = MuscleParameters()
m2_param.f_max = f_max
m1 = Muscle(m1_param)
m2 = Muscle(m2_param)
muscles = MuscleSytem(m1, m2)
# Muscle_attachment
m1_origin = np.array([-l_attach, 0.0])
m1_insertion = np.array([0.0, -l_attach])
m2_origin = np.array([l_attach, 0.0])
m2_insertion = np.array([0.0, -l_attach])
muscles.attach(np.array([m1_origin, m1_insertion]),
np.array([m2_origin, m2_insertion]))
# Neural network
n_params = NetworkParameters()
n_params.D = d
n_params.w = w
n_params.b = b
n_params.tau = tau
neural_network = NeuralSystem(n_params)
# System creation
sys = System() # Instantiate a new system
sys.add_pendulum_system(pendulum) # Add the pendulum model to the system
sys.add_muscle_system(muscles) # Add the muscle model to the system
sys.add_neural_system(
neural_network) # Add neural network model to the system
# Default initial conditions
x0_p = np.array([0, 0.]) # Pendulum initial condition
x0_m = np.array([0., m1.L_OPT, 0.,
m2.L_OPT]) # Muscle Model initial condition
x0_n = np.array([-0.5, 1, 0.5, 1]) # Neural Network initial condition
x0 = np.concatenate((x0_p, x0_m, x0_n)) # System initial conditions
return sys, x0
def exercise3():
""" Main function to run for Exercise 3.
Parameters
----------
None
Returns
-------
None
"""
# Define and Setup your pendulum model here
# Check Pendulum.py for more details on Pendulum class
P_params = PendulumParameters() # Instantiate pendulum parameters
P_params.L = 0.5 # To change the default length of the pendulum
P_params.m = 1. # To change the default mass of the pendulum
pendulum = PendulumSystem(P_params) # Instantiate Pendulum object
#### CHECK OUT Pendulum.py to ADD PERTURBATIONS TO THE MODEL #####
pylog.info('Pendulum model initialized \n {}'.format(
pendulum.parameters.showParameters()))
# Define and Setup your pendulum model here
# Check MuscleSytem.py for more details on MuscleSytem class
M1_param = MuscleParameters() # Instantiate Muscle 1 parameters
M1_param.f_max = 1500 # To change Muscle 1 max force
M2_param = MuscleParameters() # Instantiate Muscle 2 parameters
M2_param.f_max = 1500 # To change Muscle 2 max force
M1 = Muscle(M1_param) # Instantiate Muscle 1 object
M2 = Muscle(M2_param) # Instantiate Muscle 2 object
# Use the MuscleSystem Class to define your muscles in the system
muscles = MuscleSytem(M1, M2) # Instantiate Muscle System with two muscles
pylog.info('Muscle system initialized \n {} \n {}'.format(
M1.parameters.showParameters(), M2.parameters.showParameters()))
# Define Muscle Attachment points
m1_origin = np.array([-0.17, 0.0]) # Origin of Muscle 1
m1_insertion = np.array([0.0, -0.17]) # Insertion of Muscle 1
m2_origin = np.array([0.17, 0.0]) # Origin of Muscle 2
m2_insertion = np.array([0.0, -0.17]) # Insertion of Muscle 2
# Attach the muscles
muscles.attach(np.array([m1_origin, m1_insertion]),
np.array([m2_origin, m2_insertion]))
##### Neural Network #####
# The network consists of four neurons
N_params = NetworkParameters() # Instantiate default network parameters
N_params.D = 2. # To change a network parameter
# Similarly to change w -> N_params.w = (4x4) array
# Create a new neural network with above parameters
neural_network = NeuralSystem(N_params)
pylog.info('Neural system initialized \n {}'.format(
N_params.showParameters()))
# Create system of Pendulum, Muscles and neural network using SystemClass
# Check System.py for more details on System class
sys = System() # Instantiate a new system
sys.add_pendulum_system(pendulum) # Add the pendulum model to the system
sys.add_muscle_system(muscles) # Add the muscle model to the system
# Add the neural network to the system
sys.add_neural_system(neural_network)
##### Time #####
t_max = 2.5 # Maximum simulation time
time = np.arange(0., t_max, 0.001) # Time vector
##### Model Initial Conditions #####
x0_P = np.array([0., 0.]) # Pendulum initial condition
# Muscle Model initial condition
x0_M = np.array([0., M1.L_OPT, 0., M2.L_OPT])
x0_N = np.array([-0.5, 1, 0.5, 1]) # Neural Network Initial Conditions
x0 = np.concatenate((x0_P, x0_M, x0_N)) # System initial conditions
##### System Simulation #####
# For more details on System Simulation check SystemSimulation.py
# SystemSimulation is used to initialize the system and integrate
# over time
sim = SystemSimulation(sys) # Instantiate Simulation object
# Add external inputs to neural network
# sim.add_external_inputs_to_network(np.ones((len(time), 4)))
# sim.add_external_inputs_to_network(ext_in)
sim.initalize_system(x0, time) # Initialize the system state
# Integrate the system for the above initialized state and time
sim.simulate()
# Obtain the states of the system after integration
# res is np.array [time, states]
# states vector is in the same order as x0
res = sim.results()
# Obtain the states of the system after integration
# res is np.array [time, states]
# states vector is in the same order as x0
res = sim.results()
# In order to obtain internal states of the muscle
# you can access the results attribute in the muscle class
muscle1_results = sim.sys.muscle_sys.Muscle1.results
muscle2_results = sim.sys.muscle_sys.Muscle2.results
# Plotting the results
plt.figure('Pendulum')
plt.title('Pendulum Phase')
plt.plot(res[:, 0], res[:, :2])
plt.xlabel('Position [rad]')
plt.ylabel('Velocity [rad.s]')
plt.grid()
if DEFAULT["save_figures"] is False:
plt.show()
else:
figures = plt.get_figlabels()
pylog.debug("Saving figures:\n{}".format(figures))
for fig in figures:
plt.figure(fig)
save_figure(fig)
plt.close(fig)
# To animate the model, use the SystemAnimation class
# Pass the res(states) and systems you wish to animate
simulation = SystemAnimation(res, sim.sys.pendulum_sys, sim.sys.muscle_sys,
sim.sys.neural_sys)
# To start the animation
simulation.animate()
def exercise3a():
""" Main function to run for Exercise 3.
Parameters
----------
None
Returns
-------
None
"""
# Define and Setup your pendulum model here
# Check Pendulum.py for more details on Pendulum class
P_params = PendulumParameters() # Instantiate pendulum parameters
P_params.L = 0.5 # To change the default length of the pendulum
P_params.m = 1. # To change the default mass of the pendulum
P_params.PERTURBATION = True
pendulum = PendulumSystem(P_params) # Instantiate Pendulum object
#### CHECK OUT Pendulum.py to ADD PERTURBATIONS TO THE MODEL #####
pylog.info('Pendulum model initialized \n {}'.format(
pendulum.parameters.showParameters()))
# Define and Setup your pendulum model here
# Check MuscleSytem.py for more details on MuscleSytem class
M1_param = MuscleParameters() # Instantiate Muscle 1 parameters
M1_param.f_max = 1500 # To change Muscle 1 max force
M2_param = MuscleParameters() # Instantiate Muscle 2 parameters
M2_param.f_max = 1500 # To change Muscle 2 max force
M1 = Muscle(M1_param) # Instantiate Muscle 1 object
M2 = Muscle(M2_param) # Instantiate Muscle 2 object
# Use the MuscleSystem Class to define your muscles in the system
muscles = MuscleSytem(M1, M2) # Instantiate Muscle System with two muscles
pylog.info('Muscle system initialized \n {} \n {}'.format(
M1.parameters.showParameters(), M2.parameters.showParameters()))
# Define Muscle Attachment points
m1_origin = np.array([-0.17, 0.0]) # Origin of Muscle 1
m1_insertion = np.array([0.0, -0.17]) # Insertion of Muscle 1
m2_origin = np.array([0.17, 0.0]) # Origin of Muscle 2
m2_insertion = np.array([0.0, -0.17]) # Insertion of Muscle 2
# Attach the muscles
muscles.attach(np.array([m1_origin, m1_insertion]),
np.array([m2_origin, m2_insertion]))
##### Neural Network #####
# The network consists of four neurons
N_params = NetworkParameters() # Instantiate default network parameters
N_params.tau = [0.02, 0.02, 0.1, 0.1]
N_params.b = [3.0, 3.0, -3.0, -3.0]
N_params.D = 1.0 # To change a network parameter
N_params.w = np.asarray([[0.0, -5.0, -5.0, 0.0], [-5.0, 0.0, 0.0, -5.0],
[5.0, -5.0, 0.0, 0.0], [-5.0, 5.0, 0.0, 0.0]])
# Similarly to change w -> N_params.w = (4x4) array
print(N_params.w)
############################# Exercise 3A ######################
N_params.w = np.transpose(
np.asarray([[0, -1, 1, -1], [-1, 0, -1, 1], [-1, 0, 0, 0],
[0, -1, 0, 0]])) * 5
print(N_params.w, N_params.D, N_params.tau, N_params.b, N_params.exp)
# Create a new neural network with above parameters
neural_network = NeuralSystem(N_params)
pylog.info('Neural system initialized \n {}'.format(
N_params.showParameters()))
# Create system of Pendulum, Muscles and neural network using SystemClass
# Check System.py for more details on System class
sys = System() # Instantiate a new system
sys.add_pendulum_system(pendulum) # Add the pendulum model to the system
sys.add_muscle_system(muscles) # Add the muscle model to the system
sys.add_neural_system(
neural_network) # Add the neural network to the system
##### Time #####
t_max = 2. # Maximum simulation time
time = np.arange(0., t_max, 0.001) # Time vector
##### Model Initial Conditions #####
x0_P = np.array([[-0.5, 0], [-0.25, -0.25], [0., 0.],
[0.5, 0]]) # Pendulum initial condition
for i in x0_P:
# Muscle Model initial condition
x0_M = np.array([0., M1.L_OPT, 0., M2.L_OPT])
x0_N = np.array([-1.5, 1, 2.5, 1]) # Neural Network Initial Conditions
x0 = np.concatenate((i, x0_M, x0_N)) # System initial conditions
##### System Simulation #####
# For more details on System Simulation check SystemSimulation.py
# SystemSimulation is used to initialize the system and integrate
# over time
sim = SystemSimulation(sys) # Instantiate Simulation object
# sim.add_external_inputs_to_network(np.ones((len(time), 4)))
# wave_h1 = np.sin(time*3)*2 #makes a sinusoidal wave from 'time'
# wave_h2 = np.sin(time*3 + np.pi)*1 #makes a sinusoidal wave from 'time'
#
# wave_h1[wave_h1<0] = 0 #formality of passing negative values to zero
# wave_h2[wave_h2<0] = 0 #formality of passing negative values to zero
#
# act1 = wave_h1.reshape(len(time), 1) #makes a vertical array like act1
# act2 = wave_h2.reshape(len(time), 1) #makes a vertical array like act1
# column = np.ones((len(time), 1))
# ext_in = np.hstack((act1, column, act2, column))
# sim.add_external_inputs_to_network(ext_in)
sim.initalize_system(x0, time) # Initialize the system state
sim.sys.pendulum_sys.parameters.PERTURBATION = False
# Integrate the system for the above initialized state and time
sim.simulate()
# Obtain the states of the system after integration
# res is np.array [time, states]
# states vector is in the same order as x0
res = sim.results()
# In order to obtain internal states of the muscle
# you can access the results attribute in the muscle class
muscle1_results = sim.sys.muscle_sys.Muscle1.results
muscle2_results = sim.sys.muscle_sys.Muscle2.results
# Plotting the results: Position(phase) vs time
plt.figure('Pendulum Phase')
plt.title('Pendulum Phase')
plt.plot(res[:, 0], res[:, 1]) #to plot pendulum Position (phase)
# plt.plot(res[:, 0], time) #to plot position
# plt.plot(res[:, 0], res[:, -5:-1]) # to Plot neurons' states
plt.xlabel('time [s]')
plt.ylabel('Position [rad]')
plt.grid()
# Plotting the results: Velocity vs Position (phase)
plt.figure('Pendulum Vel v.s. Phase')
plt.title('Pendulum Vel v.s. Phase')
plt.plot(res[:, 1], res[:, 2]) #to plot Velocity vs Position (phase)
plt.xlabel('Position [rad]')
plt.ylabel('Velocity [rad.s]')
plt.grid()
# Plotting the results: Velocity vs time
plt.figure('Pendulum Velocity')
plt.title('Pendulum Velocity')
plt.plot(res[:, 0], res[:, 2]) #to plot Velocity vs Position
plt.xlabel('time [s]')
plt.ylabel('Velocity [rad.s]')
plt.grid()
# Plotting the results: Output of the network
plt.figure('Network output')
plt.title('Network output')
plt.plot(res[:, 0], res[:, -1],
label='neuron1') #to plot Velocity vs Position
plt.plot(res[:, 0], res[:, -2], label='neuron2')
plt.plot(res[:, 0], res[:, -3], label='neuron3')
plt.plot(res[:, 0], res[:, -4], label='neuron4')
plt.xlabel('time [s]')
plt.ylabel('Stimulation ')
plt.legend(loc='upper right')
plt.grid()
if DEFAULT["save_figures"] is False:
plt.show()
else:
figures = plt.get_figlabels()
pylog.debug("Saving figures:\n{}".format(figures))
for fig in figures:
plt.figure(fig)
save_figure(fig)
plt.close(fig)
# To animate the model, use the SystemAnimation class
# Pass the res(states) and systems you wish to animate
simulation = SystemAnimation(res, sim.sys.pendulum_sys, sim.sys.muscle_sys,
sim.sys.neural_sys)
# To start the animation
simulation.animate()
def system_init():
""" Use this function to create a new default system. """
########## PENDULUM ##########
# Define and Setup your pendulum model here
# Check Pendulum.py for more details on Pendulum class
P_params = PendulumParameters() # Instantiate pendulum parameters
P_params.L = 1.0 # To change the default length of the pendulum
P_params.m = 0.25 # To change the default mass of the pendulum
pendulum = PendulumSystem(P_params) # Instantiate Pendulum object
#### CHECK OUT Pendulum.py to ADD PERTURBATIONS TO THE MODEL #####
pylog.info('Pendulum model initialized \n {}'.format(
pendulum.parameters.showParameters()))
########## MUSCLES ##########
# Define and Setup your muscle model here
# Check MuscleSystem.py for more details on MuscleSystem class
m1_param = MuscleParameters() # Instantiate Muscle 1 parameters
m1_param.f_max = 200. # To change Muscle 1 max force
m1_param.l_opt = 0.4
m1_param.l_slack = 0.45
m2_param = MuscleParameters() # Instantiate Muscle 2 parameters
m2_param.f_max = 200. # To change Muscle 2 max force
m2_param.l_opt = 0.4
m2_param.l_slack = 0.45
m1 = Muscle('m1', m1_param) # Instantiate Muscle 1 object
m2 = Muscle('m2', m2_param) # Instantiate Muscle 2 object
# Use the MuscleSystem Class to define your muscles in the system
# Instantiate Muscle System with two muscles
muscles = MuscleSystem(m1, m2)
pylog.info('Muscle system initialized \n {} \n {}'.format(
m1.parameters.showParameters(), m2.parameters.showParameters()))
# Define Muscle Attachment points
m1_origin = np.asarray([0.0, 0.9]) # Origin of Muscle 1
m1_insertion = np.asarray([0.0, 0.15]) # Insertion of Muscle 1
m2_origin = np.asarray([0.0, 0.8]) # Origin of Muscle 2
m2_insertion = np.asarray([0.0, -0.3]) # Insertion of Muscle 2
# Attach the muscles
muscles.attach(np.asarray([m1_origin, m1_insertion]),
np.asarray([m2_origin, m2_insertion]))
########## Network ##########
# The network consists of four neurons
N_params = NetworkParameters() # Instantiate default network parameters
N_params.D = 1 # To change a network parameter
# Similarly to change w -> N_params.w = (4x4) array
N_params.tau = [0.02, 0.02, 0.1, 0.1]
N_params.w = [[0, -5, 5, -5], [-5, 0, -5, 5], [-5, 0, 0, 0], [0, -5, 0, 0]]
N_params.b = [3.0, 3.0, -3.0, -3.0]
N_params.w = np.transpose(N_params.w)
# Create a new neural network with above parameters
neural_network = NeuralSystem(N_params)
pylog.info('Neural system initialized \n {}'.format(
N_params.showParameters()))
########## ADD SYSTEMS ##########
# Create system of Pendulum, Muscles and neural network using SystemClass
# Check System.py for more details on System class
sys = System() # Instantiate a new system
sys.add_pendulum_system(pendulum) # Add the pendulum model to the system
sys.add_muscle_system(muscles) # Add the muscle model to the system
# Add the neural network to the system
sys.add_neural_system(neural_network)
##### Time #####
t_max = 2.5 # Maximum simulation time
time = np.arange(0., t_max, 0.001) # Time vector
##### Model Initial Conditions #####
x0_P = np.asarray([np.pi / 2, 0.]) # Pendulum initial condition
# Muscle Model initial condition
l_ce_0 = sys.muscle_sys.initialize_muscle_length(np.pi / 2)
x0_M = np.asarray([0.05, l_ce_0[0], 0.05, l_ce_0[1]])
x0_N = np.asarray([-0.5, 1, 0.5, 1]) # Neural Network Initial Conditions
x0 = np.concatenate((x0_P, x0_M, x0_N)) # System initial conditions
##### System Simulation #####
# For more details on System Simulation check SystemSimulation.py
# SystemSimulation is used to initialize the system and integrate
# over time
sim = SystemSimulation(sys) # Instantiate Simulation object
sim.initalize_system(x0, time) # Initialize the system state
return sim
def exercise3():
""" Main function to run for Exercise 3.
Parameters
----------
None
Returns
-------
None
"""
# Define and Setup your pendulum model here
# Check Pendulum.py for more details on Pendulum class
P_params = PendulumParameters() # Instantiate pendulum parameters
P_params.L = 0.5 # To change the default length of the pendulum
P_params.m = 1. # To change the default mass of the pendulum
pendulum = PendulumSystem(P_params) # Instantiate Pendulum object
#### CHECK OUT Pendulum.py to ADD PERTURBATIONS TO THE MODEL #####
pylog.info('Pendulum model initialized \n {}'.format(
pendulum.parameters.showParameters()))
# Define and Setup your pendulum model here
# Check MuscleSytem.py for more details on MuscleSytem class
M1_param = MuscleParameters() # Instantiate Muscle 1 parameters
M1_param.f_max = 1500 # To change Muscle 1 max force
M2_param = MuscleParameters() # Instantiate Muscle 2 parameters
M2_param.f_max = 1500 # To change Muscle 2 max force
M1 = Muscle(M1_param) # Instantiate Muscle 1 object
M2 = Muscle(M2_param) # Instantiate Muscle 2 object
# Use the MuscleSystem Class to define your muscles in the system
muscles = MuscleSytem(M1, M2) # Instantiate Muscle System with two muscles
pylog.info('Muscle system initialized \n {} \n {}'.format(
M1.parameters.showParameters(),
M2.parameters.showParameters()))
# Define Muscle Attachment points
m1_origin = np.array([-0.17, 0.0]) # Origin of Muscle 1
m1_insertion = np.array([0.0, -0.17]) # Insertion of Muscle 1
m2_origin = np.array([0.17, 0.0]) # Origin of Muscle 2
m2_insertion = np.array([0.0, -0.17]) # Insertion of Muscle 2
# Attach the muscles
muscles.attach(np.array([m1_origin, m1_insertion]),
np.array([m2_origin, m2_insertion]))
##### Neural Network #####
# The network consists of four neurons
N_params = NetworkParameters() # Instantiate default network parameters
# Similarly to change w -> N_params.w = (4x4) array
# From lecture 4, slide 85 -> Generate oscillations !!
N_params.D = 2.
N_params.tau = [0.02,0.02,0.1,0.1]
N_params.b = [3.0,3.0,-3.0,-3.0]
N_params.w = [[0,-5,-5,0], # 1 <- 2
[-5,0,0,-5],
[5,-5,0,-5],
[-5,5,0,0]]
# Create a new neural network with above parameters
neural_network = NeuralSystem(N_params)
pylog.info('Neural system initialized \n {}'.format(
N_params.showParameters()))
# Create system of Pendulum, Muscles and neural network using SystemClass
# Check System.py for more details on System class
sys = System() # Instantiate a new system
sys.add_pendulum_system(pendulum) # Add the pendulum model to the system
sys.add_muscle_system(muscles) # Add the muscle model to the system
# Add the neural network to the system
sys.add_neural_system(neural_network)
##### Time #####
t_max = 2.5 # Maximum simulation time
time = np.arange(0., t_max, 0.001) # Time vector
##### Model Initial Conditions #####
x0_P = np.array([0., 0.]) # Pendulum initial condition
# Muscle Model initial condition
x0_M = np.array([0., M1.L_OPT, 0., M2.L_OPT])
x0_N = np.array([-0.5, 1, 0.5, 1]) # Neural Network Initial Conditions
x0 = np.concatenate((x0_P, x0_M, x0_N)) # System initial conditions
##### System Simulation #####
# For more details on System Simulation check SystemSimulation.py
# SystemSimulation is used to initialize the system and integrate
# over time
sim = SystemSimulation(sys) # Instantiate Simulation object
# Add external inputs to neural network
# sim.add_external_inputs_to_network(np.ones((len(time), 4)))
#ext_in = np.ones((len(time), 4))
#ext_in[:,2] = 0.2
#ext_in[:,3] = 0.2
#sim.add_external_inputs_to_network(ext_in)
sim.initalize_system(x0, time) # Initialize the system state
# Integrate the system for the above initialized state and time
sim.simulate()
# Obtain the states of the system after integration
# res is np.array [time, states]
# states vector is in the same order as x0
res = sim.results()
# Obtain the states of the system after integration
# res is np.array [time, states]
# states vector is in the same order as x0
#res = sim.results()
# In order to obtain internal states of the muscle
# you can access the results attribute in the muscle class
muscle1_results = sim.sys.muscle_sys.Muscle1.results
muscle2_results = sim.sys.muscle_sys.Muscle2.results
# Plotting the phase
fig = plt.figure('Pendulum')
plt.title('Pendulum Phase')
plt.plot(res[:, 1], res[:, 2])
plt.xlabel('Position [rad]')
plt.ylabel('Velocity [rad/s]')
plt.grid()
fig.tight_layout()
fig.savefig('graphs/PendulumPhase.png')
# Plotting the neuronal activation
# Access the neurons outputs:
# [t] theta theta. A1 lCE1 A2 lCE2 m1 m2 m3 m4
fig = plt.figure('Neuron output')
plt.title('Membrane potentials')
plt.plot(res[:, 0], res[:, 7],label='m1')
plt.plot(res[:, 0], res[:, 8],label='m2')
plt.plot(res[:, 0], res[:, 9],label='m3')
plt.plot(res[:, 0], res[:, 10],label='m4')
plt.xlabel('Time [s]')
plt.ylabel('Potential')
plt.legend()
plt.grid()
fig.tight_layout()
fig.savefig('graphs/MembranePotentials.png')
if DEFAULT["save_figures"] is False:
plt.show()
else:
figures = plt.get_figlabels()
pylog.debug("Saving figures:\n{}".format(figures))
for fig in figures:
plt.figure(fig)
save_figure(fig)
plt.close(fig)
# To animate the model, use the SystemAnimation class
# Pass the res(states) and systems you wish to animate
simulation = SystemAnimation(
res,
sim.sys.pendulum_sys,
sim.sys.muscle_sys,
sim.sys.neural_sys)
# To start the animation
simulation.animate()
# 3.b
ext_in = np.ones((len(time), 4))*0.0
plotExternalDrive(sys,x0,ext_in,typ='low')
ext_in = np.ones((len(time), 4))
plotExternalDrive(sys,x0,ext_in,typ='high')
ext_in = np.ones((len(time), 4))
ext_in[:,0] *= 0.1
ext_in[:,1] *= 0.1
plotExternalDrive(sys,x0,ext_in,typ='asymmetric')
def exercise3():
""" Main function to run for Exercise 3.
Parameters
----------
None
Returns
-------
None
"""
'''
# Create system
sim = system_init()
# Add external inputs to neural network
sim.add_external_inputs_to_network(np.ones((len(sim.time), 4)))
# Integrate the system for the above initialized state and time
sim.simulate()
# Obtain the states of the system after integration
# res is np.asarray [time, states]
# states vector is in the same order as x0
res = sim.results()
# Obtain the states of the system after integration
# res is np.asarray [time, states]
# states vector is in the same order as x0
res = sim.results()
# In order to obtain internal states of the muscle
# you can access the results attribute in the muscle class
muscle_1_results = sim.sys.muscle_sys.muscle_1.results
muscle_2_results = sim.sys.muscle_sys.muscle_2.results
# Plotting the results
plt.figure('Pendulum')
plt.title('Pendulum Phase')
plt.plot(res[:, 1], res[:, 2])
plt.xlabel('Position [rad]')
plt.ylabel('Velocity [rad.s]')
plt.grid()
'''
######################################################
# initialization
########## PENDULUM ##########
P_params = PendulumParameters()
P_params.L = 1.0
P_params.m = 0.25
pendulum = PendulumSystem(P_params)
pylog.info('Pendulum model initialized \n {}'.format(
pendulum.parameters.showParameters()))
########## MUSCLES ##########
m1_param = MuscleParameters()
m1_param.f_max = 200.
m1_param.l_opt = 0.4
m1_param.l_slack = 0.45
m2_param = MuscleParameters()
m2_param.f_max = 200.
m2_param.l_opt = 0.4
m2_param.l_slack = 0.45
m1 = Muscle('m1', m1_param)
m2 = Muscle('m2', m2_param)
muscles = MuscleSystem(m1, m2)
pylog.info('Muscle system initialized \n {} \n {}'.format(
m1.parameters.showParameters(), m2.parameters.showParameters()))
######## Define Muscle Attachment points
m1_origin = np.asarray([0.0, 0.9])
m1_insertion = np.asarray([0.0, 0.15])
m2_origin = np.asarray([0.0, 0.8])
m2_insertion = np.asarray([0.0, -0.3])
muscles.attach(np.asarray([m1_origin, m1_insertion]),
np.asarray([m2_origin, m2_insertion]))
##### Time #####
t_max = 2.5
time = np.arange(0., t_max, 0.001)
###########################################################
###########################################################
###########################################################
###########################################################
###########################################################
### code for 3a
pylog.info("3a")
d = 1.
tau = np.array([0.02, 0.02, 0.1, 0.1])
b = np.array([3., 3., -3., -3.])
w = np.zeros((4, 4))
w[0, 1] = w[0, 3] = w[1, 0] = w[1, 2] = -5
w[0, 2] = w[1, 3] = 5
w[2, 0] = w[3, 1] = -5
w = w.T
N_params = NetworkParameters()
N_params.D = d
N_params.tau = tau
N_params.b = b
N_params.w = w
neural_network = NeuralSystem(N_params)
sys = System()
sys.add_pendulum_system(pendulum)
sys.add_muscle_system(muscles)
sys.add_neural_system(neural_network)
x0_P = np.asarray([np.pi / 2, 0.])
l_ce_0 = sys.muscle_sys.initialize_muscle_length(np.pi / 2)
x0_M = np.asarray([0.05, l_ce_0[0], 0.05, l_ce_0[1]])
x0_N = np.asarray([-0.5, 1, 0.5, 1])
x0 = np.concatenate((x0_P, x0_M, x0_N))
sim = SystemSimulation(sys)
sim.initalize_system(x0, time)
sim.simulate()
res = sim.results()
positions = res[:, 1]
vels = res[:, 2]
plt.figure('3a. Activation with time ')
plt.title('Activation with time')
plt.plot(res[:, 0], res[:, 3], label="Activation 1")
plt.plot(res[:, 0], res[:, 5], label="Activation 2")
plt.xlabel('Time [s]')
plt.ylabel('Activation')
plt.legend()
plt.grid()
plt.show()
# Plotting the results
plt.figure('3a. Pendulum state with time')
plt.title('Pendulum state with time')
plt.plot(res[:, 0], positions)
plt.xlabel('Time [s]')
plt.ylabel('Position [rad]')
plt.grid()
plt.show()
# Plotting the results
plt.figure('3a. Pendulum phase plot')
plt.title('Pendulum phase plot')
plt.plot(positions, vels)
plt.xlabel('Position [rad]')
plt.ylabel('Velocity [rad/s]')
plt.grid()
plt.show()
###########################################################
###########################################################
###########################################################
###########################################################
###########################################################
### code for 3b
pylog.info("3b")
all_positions = []
all_vels = []
all_time = []
all_act_1 = []
all_act_2 = []
external_drives = np.array([0, 0.2, 0.5, 1., 2., 5.])
for temp_drive in external_drives:
sim = SystemSimulation(sys)
sim.initalize_system(x0, time)
sim.add_external_inputs_to_network(
np.ones((len(sim.time), 4)) * temp_drive)
sim.simulate()
res = sim.results()
all_time = all_time + [res[:, 0]]
all_positions = all_positions + [res[:, 1]]
all_vels = all_vels + [res[:, 2]]
all_act_1 = all_act_1 + [res[:, 3]]
all_act_2 = all_act_2 + [res[:, 5]]
plt.figure('3a. Activation with time by different external drives')
plt.title('Activation with time by different external drives')
for i in range(len(external_drives)):
plt.plot(all_time[i], all_act_1[i])
plt.plot(all_time[i], all_act_2[i])
plt.xlabel('Time [s]')
plt.ylabel('Activation')
temp_legends = [
'external drive: ' + format((temp_drive), '.2f')
for temp_drive in external_drives
]
plt.legend(temp_legends)
plt.grid()
plt.show()
plt.figure('3b. Pendulum state with time by different external drives')
plt.title('Pendulum state with time by different external drives')
for i in range(len(external_drives)):
plt.plot(all_time[i], all_positions[i])
plt.xlabel('Time [s]')
plt.ylabel('Position [rad]')
temp_legends = [
'external drive: ' + format((temp_drive), '.2f')
for temp_drive in external_drives
]
plt.legend(temp_legends)
plt.grid()
plt.show()
plt.figure('3a. Pendulum phase plot by different external drives')
plt.title('Pendulum phase plot by different external drives')
for i in range(len(external_drives)):
plt.plot(all_positions[i], all_vels[i])
plt.xlabel('Position [rad]')
plt.ylabel('Velocity [rad/s]')
temp_legends = [
'external drive: ' + format((temp_drive), '.2f')
for temp_drive in external_drives
]
plt.legend(temp_legends)
plt.grid()
plt.show()
# To animate the model, use the SystemAnimation class
# Pass the res(states) and systems you wish to animate
simulation = SystemAnimation(res, sim.sys.pendulum_sys, sim.sys.muscle_sys,
sim.sys.neural_sys)
if DEFAULT["save_figures"] is False:
# To start the animation
simulation.animate()
plt.show()
else:
figures = plt.get_figlabels()
pylog.debug("Saving figures:\n{}".format(figures))
for fig in figures:
plt.figure(fig)
save_figure(fig)
plt.close(fig)