Best way to store graph 'history'


I am not sure what exactly you want to achieve and for what reasons you
need the "trajectories" (e.g. whether they have to be stored in a text-file
format for some other processing step), but having a "dynamic network
object", with
a property map as you described, sounds like a good idea. Having 5000
nodes, with each having an associated 350 element double list will require
roughly 5000*350*8bytes ~ 14MB of memory .... that should fit. For storing,
you could pickle the whole graph object.... or/and, as you said, write the
property map into a text-file using your format of choice ...


attachment.html (2.99 KB)