the walk result dimension seems to be incorrect about fastwalk HOT 6 CLOSED

The link: https://pan.baidu.com/s/13Hua7ZtRVS_XC2YSMFQTrA ,and password: be5h

from fastwalk.

The result's shape should be (V*num_walks, walk_length), what's the shape you got?

from fastwalk.

The row should be 7115*10, the column should be 80, but i get 194749 * 2.
And, I find your vertex count is not correct,The vertex of my graph is 7115 , howerver, your code says 8491,I don't konw why. The edges number is correct.

from fastwalk.

Can you offer some data to debug?

from fastwalk.

There are Windows newline symbols in your input data, which results in the incorrect output, I will add string trimming during loading graph in the future, before that please preprocess the input data yourself.

You can use sed to remove those symbols first:

cat edges.txt | sed -e 's/\r//' > edges_trimmed.txt

it shows that there are 10312 vertexes:

~/workspace/fastwalk(master*) » cat edges_trimmed.txt | awk '{printf("%s\n%s\n", $1, $2);}' | sort -u | wc -l

10312

and the program will output as expection:

~/workspace/fastwalk(master*) » ./fastwalk --edge_list edges_trimmed.txt --output edges_out.txt --delimiter space --p 10 --q 0.01 --max_nodes 50000 --max_edges 50000 --workers 4
[2021-02-06 13:38:45] Loading file: edges_trimmed.txt
[2021-02-06 13:38:45] Done reading edgelist files, starting to build graph...
[2021-02-06 13:38:46] Done building graph.
[2021-02-06 13:38:46] Graph node num: 10312 edge num: 333983
[2021-02-06 13:38:46] Graph loaded.
[2021-02-06 13:38:47] Starting to walk...
[2021-02-06 13:38:47] Done feeding nodes.
[2021-02-06 13:39:12] All done.

Notice that the input data is loaded as directed graph, which means there can be vertex whose out-degree is 0, the walk will stop when reached those vertexes, so each row may have columns less than walk_length.

from fastwalk.

OK,thanks.

from fastwalk.