Discussion:
fast mixing social graphs?
ianG
2014-06-24 19:31:05 UTC
Permalink
Hi all,

would anyone be able to explain what it means for a social graph to be
*fast mixing* ?

Preferably without needing to have read hundreds of papers... A pointer
to one nice descriptive paper would be ok.

iang
Valerio Schiavoni
2014-06-24 19:58:30 UTC
Permalink
Hello iang,
the "fast mixing” property of a social graph is an indicator of how quickly
a random walk on that graph approaches the stationary distribution.

You could read for example:
http://syssec.kaist.ac.kr/~yongdaek/doc/imc2010.pdf

best,
valerio
Post by ianG
Hi all,
would anyone be able to explain what it means for a social graph to be
*fast mixing* ?
Preferably without needing to have read hundreds of papers... A pointer
to one nice descriptive paper would be ok.
iang
_______________________________________________
p2p-hackers mailing list
http://lists.zooko.com/mailman/listinfo/p2p-hackers
Matteo Dell'Amico
2014-06-26 15:19:25 UTC
Permalink
Hello Ian and Valerio,

I also have worked on measuring mixing time on social networks; you can
find a paper I wrote with Yves Roudier where we suggest that networks that
the factor that has the most impact in determining mixing time in a social
graph is whether creating a link requires people to meet physically:
http://www.eurecom.fr/fr/publication/2900/download/rs-publi-2900.pdf .
Subsequent works, such as the one linked by Valerio (which, strangely
enough, does not cite us even if the first author had contacted me for
clarifications on our work), seem to corroborate that idea.

cheers
matteo
Post by Valerio Schiavoni
Hello iang,
the "fast mixing” property of a social graph is an indicator of how
quickly a random walk on that graph approaches the stationary distribution.
http://syssec.kaist.ac.kr/~yongdaek/doc/imc2010.pdf
best,
valerio
Post by ianG
Hi all,
would anyone be able to explain what it means for a social graph to be
*fast mixing* ?
Preferably without needing to have read hundreds of papers... A pointer
to one nice descriptive paper would be ok.
iang
_______________________________________________
p2p-hackers mailing list
http://lists.zooko.com/mailman/listinfo/p2p-hackers
_______________________________________________
p2p-hackers mailing list
http://lists.zooko.com/mailman/listinfo/p2p-hackers
--
matteo
Loading...