search

snap-stanford / snap

Stanford Network Analysis Platform (SNAP) is a general purpose network analysis and graph mining library.
Other
2.18k stars 799 forks source link

Error trying to estimate community number #182

Closed oiwilliam closed 4 years ago

oiwilliam commented 4 years ago

I'm trying to estimate the -c parameter for my graph. For that, I've used the -mc and -xc parameters, but the system halts with "killed" error. Independently of the parameters specified in the -mc and -xc, when I use -c:-1, the application increases the memory consumption till there are no more memory and crashes down. Any Idea?

Command used: ./bigclam -mc:10 -xc:20 -nt:7 -nc:1 -o:out.txt i:input.csv -c:-1 Output:

ragm. build: 17:25:50, Oct 13 2019. Time: 12:52:05 [Oct 13 2019]
================================================================
Output Graph data prefix (-o:)=out.txt
Input edgelist file name (-i:)=input.csv
Input file name for node names (Node ID, Node label)  (-l:)=
The number of communities to detect (-1: detect automatically) (-c:)=-1
Minimum number of communities to try (-mc:)=10
Maximum number of communities to try (-xc:)=20
How many trials for the number of communities (-nc:)=1
Number of threads for parallelization (-nt:)=7
Alpha for backtracking line search (-sa:)=0.05
Beta for backtracking line search (-sb:)=0.3
Graph: 190608 Nodes 2436844 Edges
finding number of communities
generating hold out set
holding out 243684 edges...
243684 Edges hold out
holding out 243684 edges...
243684 Edges hold out
Killed
roks commented 4 years ago

It seems that there is insufficient RAM in the computer for this graph. Try to run the program on a machine with more RAM.

oiwilliam commented 4 years ago

It is using 64 Gb of RAM for this network. It really supposed to use so much memory? I’m able to use many communities detections algorithms in this machine, so I think it is strange...