The clustering coefficient for erdos renyi model $G(n,p) = p$.

Now i have been studying in various papers that it cannot model real world networks which has high clustering coefficient. My question is that if we use a higher probability value for the model then we our clustering coefficient should increase then why cant we use this model for such real world networks like small world phenomenon ?

  • $\begingroup$ Side remark: for Geometric Random Graph clustering coefficient is known and depend only on dimension of space: en.wikipedia.org/wiki/… $\endgroup$ May 22, 2020 at 16:34
  • $\begingroup$ What could that mean intuatively? $\endgroup$ May 22, 2020 at 16:39

1 Answer 1


In an ER graph, density and clustering coefficient are equal. In most "real-world networks", they differ by orders of magnitude.

Therefore, if an ER graph has a realistic density, then it has not a realistic clustering coefficient; and if it has a realistic clustering coefficient, then it has not a realistic density.

This is actually a key element that makes ER graphs poor models for most real-world networks: in practice, despite a low density, the clustering coefficient if high. You are right in pointing the fact that a high clustering coefficient alone is easy to obtain.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.