Supplementary Material On Provable Benefits of Depth in Training Graph Convolutional Networks Organization