def __init__(self, fm_x, fm_y, dim_x, dim_y):
super(QMRegressor, self).__init__()
self.fm_x = fm_x
self.fm_y = fm_y
self.qm = layers.QMeasureClassif(dim_x=dim_x, dim_y=dim_y)
self.dmregress = layers.DensityMatrixRegression()
self.cp1 = layers.CrossProduct()
self.cp2 = layers.CrossProduct()
self.num_samples = tf.Variable(initial_value=0., trainable=False)
def __init__(self, input_dim, dim_x, dim_y, num_eig=0, gamma=1, random_state=None):
super(QMRegressorSGD, self).__init__()
self.fm_x = layers.QFeatureMapRFF(
input_dim=input_dim,
dim=dim_x, gamma=gamma, random_state=random_state)
self.qm = layers.QMeasureClassifEig(dim_x=dim_x, dim_y=dim_y, num_eig=num_eig)
self.dmregress = layers.DensityMatrixRegression()
self.dim_x = dim_x
self.dim_y = dim_y
self.gamma = gamma
self.random_state = random_state
qmc2.compile(optimizer, loss=tf.keras.losses.CategoricalCrossentropy()) train_y = tf.reshape(tf.keras.backend.one_hot(data_y, 2), (4, 2)) qmc2.fit(data_x, train_y, epochs=20) out2 = qmc2(data_x) print(out2) out = qmc.call_train(data_x, data_y) print(out) qmc.train_step((data_x, data_y)) inputs = tf.ones((2, 3)) qfm = layers.QFeatureMapRFF() out1 = qfm(inputs) print(out1) inputs = tf.ones((2, 3)) qfm = layers.QFeatureMapSmp(dim=2, beta=10) out1 = qfm(inputs) print(out1) qmeas = layers.QMeasureClassifEig(dim_x=8, dim_y=3, num_eig=0) out2 = qmeas(out1) print(out2) dmr = layers.DensityMatrixRegression() out3 = dmr(out2) print(out3) out3 = layers.CrossProduct()([out1, out2]) print(out3) out4 = layers.DensityMatrix2Dist()(out2) print(out4)