I have now partitioned the graph into four parts of the reddit dataset by using the following code:
nmap, emap = dgl.distributed.partition_graph(g, ‘test’, 4, num_hops=1, part_method=‘metis’,
out_path=‘output/’,
balance_ntypes=g.ndata[‘train_mask’],
balance_edges=True, return_mapping=True)
At the same time I have loaded all the partitions, now I want to adjust some nodes of the partitions, for example I remove some nodes from the 0th partition and add those nodes to the 3rd partition, I want to ask how to call the function to update the value of gpb? (Because of deletion of nodes, addition of nodes caused a change in partition information, then gpb must also be updated, please ask how to update the gpb???