Open stephwon opened 4 months ago
Is this a dataset you can share? This is not an error I've seen with some of our existing data sets. The call is somehow getting more answers than should be possible. It's probably a bug that only occurs with a specific dataset.
What is your question?
I'm trying to conduct a Steiner Tree approximation with several terminal nodes (and subgraphs I need to analyze). However, whenever I try to iterate through, it fails at MST despite the graph being Undirected:
RuntimeError: RAFT failure at file=/scratch/user/anaconda3/envs/rapids-24.04/include/raft/sparse/solver/detail/mst_solver_inl.cuh line=152:
The code works if I run through terminal nodes for each subgraph (one at a time). I already checked for self-loops, isolated nodes, or disconnected components in the big graph data, and they don't exist. The data types are to the Dask cugraph standard, so there is no issue there. Even if I change `DiG = cugraph.Graph(directed=False)' problem persists with the same error message. I'm not sure how to address this error... or am I missing something here? Graph data (df_graph) has 46.6M edges, directed and unweighted. My code below:
Code of Conduct