| 1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768 |
- '''
- Spatio-temporal animation
- '''
- from brian2 import ms
- from matplotlib.animation import FuncAnimation
- import numpy as np
- import matplotlib.pyplot as plt
- def create_neural_sheet_animation(times, activities, frame_rate=25):
- """
- :param times: array of time points
- :param activtivies: 3d array of spiking activity from zero no spike to one very recent spike with shape(#neuros,
- #neurons,
- time)
- :param frame_rate: in hertz
- :return: animation
- """
- number_of_frames = times.shape[0]
- fig, ax = plt.subplots(1,1)
- image = ax.imshow(activities[:,:,0], vmin=0, vmax=1, cmap="gray")
- timer = ax.set_title("t={:10.1f}ms".format(times[0]))
- def animate(frame):
- image.set_array(activities[:,:,frame])
- timer.set_text("t={:10.1f}ms".format(times[frame]))
- return image
- animation = FuncAnimation(fig, animate, number_of_frames, interval=1000/frame_rate)
- return animation
- #number_of_excitatory_neurons_per_row = 30
- #times = np.arange(0, 10, 0.5)
- #activities = np.random.choice(np.array([0,1]),(number_of_excitatory_neurons_per_row,
- # number_of_excitatory_neurons_per_row,
- # times.shape[0]))
- #animation = create_neural_sheet_animation(times, activities)
- #plt.show()
- #animation.save("out_2dgrid.mp4", dpi=number_of_excitatory_neurons_per_row, writer='imagemagick')
- def transform_spike_train_in_activity_array(spike_trains, dt_binning, length, activity_decay_time_constant):
- times = np.arange(0, length / ms, dt_binning / ms) * ms
- number_of_neurons = len(spike_trains)
- number_of_neurons_per_row = int(np.sqrt(number_of_neurons))
- spike_activities = [transform_spike_times_to_activity(spike_times, dt_binning, length,
- activity_decay_time_constant)
- for spike_times in spike_trains]
- spike_activities = np.reshape(np.array(spike_activities), (number_of_neurons_per_row,
- number_of_neurons_per_row,-1))
- return times, spike_activities
- def transform_spike_times_to_activity(spike_times, dt_binning, length, activity_decay_time_constant):
- kernel_times = np.arange(0, 4 * activity_decay_time_constant / ms, dt_binning / ms) * ms
- activity_kernel = np.exp(-kernel_times / activity_decay_time_constant)
- number_of_bins = int(length/dt_binning)+1
- indices_with_a_spike = np.floor(spike_times / dt_binning).astype(int)
- spike_train_in_binary_form = np.zeros((number_of_bins,))
- spike_train_in_binary_form[indices_with_a_spike] = 1
- spike_activity = np.convolve(spike_train_in_binary_form, activity_kernel, mode='same')
- return spike_activity
|
