I have a question about the implementation of the paper. Diffusion convolution expressed in Eq.(2) is not found in the code. I only find ChebNet convolution in dcrnn_cell.py as followed:

            for support in self._supports:
                x1 = tf.sparse_tensor_dense_matmul(support, x0)
                x = self._concat(x, x1)

                for k in range(2, self._max_diffusion_step + 1):
                    x2 = 2 * tf.sparse_tensor_dense_matmul(support, x1) - x0
                    x = self._concat(x, x2)
                    x1, x0 = x2, x1

The paper pointed out that ChebNet is only for undirected graphs. So I wonder how to deal with directed graphs with ChebNet convolution. Or what important details am I missing? Could you kindly help me understand the codes?