I am a Spark GraphX newbie and I am trying to do triangle counts on a dataset based on your blogpost!

I get this error java.lang.IllegalArgumentException: requirement failed: Invalid initial capacity

Could you please let me know why?

Please refer to my question here : http://stackoverflow.com/questions/40337366/spark-graphx-requirement-failed-invalid-initial-capacity

Thanks

]]>Thanks a lot for sharing, I was looking for the explanation of PageRank in GraphX and your post helped me a lot.

There is one thing that frustrated me a lot though. Namely, initialMessage = 0.0.

I am comparing my Hadoop implementation’s results with Spark and guess what. Results were different whole time! We are using following formula for the calculations of PR for vertice i:

PR[i] = 0.15 + (1 – 0.15) * inNbrs[i].map(j => oldPR[j] / outDeg[j]).sum

where oldPR[j] =1, because you wrote “Now the vertices in initialGraph are assigned initial PageRank 1.0”. This was not true when initialValue =0, because then oldPR[j] gets 0 so whole equation gets 0.15. This means in next iteration, oldPR[j] / outDeg[j] is 0.15/outDeg[j] and not 1/outDeg[j]. Eureka!

Sorry for long comment and lot of my unreadable code, but it lasted me so long time to understand the differences in my soulution that I fet I need to write it down for further readers.

Thanks again for sharing!
]]>

Can you tell me how to do that ?
]]>