[OT] Best algorithm for extremely large hashtable?

[bearophile]

H. S. Teoh:

> I'm hoping for a kind of compressed representation where 
> multiple
> adjacent nodes can be represented in a very compact way, by 
> somehow
> taking advantage of the fact that they lie on an n-dimensional 
> grid and
> the fact that graph edges are always a subset of grid edges (as 
> opposed
> to edges between two arbitrary nodes).

Look for Succinct Data-Structures, Graphs. You can impose further 
memory savings in your problem.

Bye,
bearophile