def plot_prediction(t, V, data_indexes, kernel_times, delay_indexes, image_indexes, input_to_image, kernel_to_input, h0, h1, h2, ims, ims2): prediction = calculate_prediction(data_indexes, kernel_times, delay_indexes, image_indexes, input_to_image, kernel_to_input, h0, h1, h2, ims, ims2) # Plot data plt.plot(t,V[data_indexes], '-*', label='Data') plt.hold('on') # Plot prediction plt.plot(t,prediction, '-*', label='Prediction') plt.legend()
h1_sparse = np.load(filename_h1kernel_sparse) h2_sparse = np.load(filename_h2kernel_sparse) #### Load the dense part filename_h1kernel_dense = kernels_folder + 'h1' + str(cell_number) + stimuli_type_dense + kernel_format filename_h2kernel_dense = kernels_folder + 'h2' + str(cell_number) + stimuli_type_dense + kernel_format h1_dense = np.load(filename_h1kernel_dense) h2_dense = np.load(filename_h2kernel_dense) # Calculate the convolutions aux_zeros = np.zeros(np.shape(h1_sparse)) h1_sparse_convoluted = calculate_prediction(training_indexes, kernel_times, delay_indexes, image_indexes, input_to_image, kernel_to_input, h0, h1_sparse, aux_zeros, ims_sparse, ims_sparse**2) h2_sparse_convoluted = calculate_prediction(training_indexes, kernel_times, delay_indexes, image_indexes, input_to_image, kernel_to_input, h0, aux_zeros, h2_sparse, ims_sparse, ims_sparse**2) h1_dense_convoluted = calculate_prediction(training_indexes, kernel_times, delay_indexes, image_indexes, input_to_image, kernel_to_input, h0, h1_dense, aux_zeros, ims_dense, ims_dense**2) h2_dense_convoluted = calculate_prediction(training_indexes, kernel_times, delay_indexes, image_indexes, input_to_image, kernel_to_input, h0, aux_zeros, h2_dense, ims_dense, ims_dense**2) ### Calculate the SI's SI_sparse[cell_number] = np.sum(h1_sparse_convoluted**2) / np.sum( h2_sparse_convoluted**2 + h1_sparse_convoluted**2) SI_dense[cell_number] = np.sum(h1_dense_convoluted**2 ) / np.sum( h2_dense_convoluted**2 + h1_dense_convoluted**2)
filename_h2kernel_sparse = kernels_folder + 'h2' + str(cell_number) + stimuli_type_sparse + kernel_format h1_sparse = np.load(filename_h1kernel_sparse) h2_sparse = np.load(filename_h2kernel_sparse) #### Load the dense part filename_h1kernel_dense = kernels_folder + 'h1' + str(cell_number) + stimuli_type_dense + kernel_format filename_h2kernel_dense = kernels_folder + 'h2' + str(cell_number) + stimuli_type_dense + kernel_format h1_dense = np.load(filename_h1kernel_dense) h2_dense = np.load(filename_h2kernel_dense) # Calculate the convolutions dense_over_sparsefield_convoluted = calculate_prediction(training_indexes, kernel_times, delay_indexes, image_indexes, input_to_image, kernel_to_input, h0, h1_sparse, h2_sparse, ims_dense, ims_dense**2) sparse_over_densefield_convoluted = calculate_prediction(training_indexes, kernel_times, delay_indexes, image_indexes, input_to_image, kernel_to_input, h0, h1_dense, h2_dense, ims_sparse, ims_sparse**2) # cut a window of time_window new_time_trace_SD = dense_over_sparsefield_convoluted[1: time_window] new_time_trace_DS = sparse_over_densefield_convoluted[1: time_window] print 'PLOT new time traces' figure = plt.figure plt.subplot(211) plt.plot(new_time_trace_SD)
#### Load the dense part filename_h1kernel_dense = kernels_folder + 'h1' + str( cell_number) + stimuli_type_dense + kernel_format filename_h2kernel_dense = kernels_folder + 'h2' + str( cell_number) + stimuli_type_dense + kernel_format h1_dense = np.load(filename_h1kernel_dense) h2_dense = np.load(filename_h2kernel_dense) # Calculate the convolutions aux_zeros = np.zeros(np.shape(h1_sparse)) h1_sparse_convoluted = calculate_prediction(training_indexes, kernel_times, delay_indexes, image_indexes, input_to_image, kernel_to_input, h0, h1_sparse, aux_zeros, ims_sparse, ims_sparse**2) h2_sparse_convoluted = calculate_prediction(training_indexes, kernel_times, delay_indexes, image_indexes, input_to_image, kernel_to_input, h0, aux_zeros, h2_sparse, ims_sparse, ims_sparse**2) h1_dense_convoluted = calculate_prediction(training_indexes, kernel_times, delay_indexes, image_indexes, input_to_image, kernel_to_input, h0, h1_dense, aux_zeros, ims_dense, ims_dense**2)