Using N as the number of orbitals, memory estimates on using a dense representation of the 2-rdm or two-electron integrals for real, double precision numbers are approximately
These numbers are amplified significantly for the storage of the corresponding JW operators for example (the additional bytes for storing the associated strings), which might effectively limit us to something like 75-100 spin orbitals in practice if we aren't careful. This may not be sufficient to cost out something like Fe2S2, though it is perhaps better to discover the bottlenecks by pushing to something like a real instance of Fe2S2 if possible.
Using N as the number of orbitals, memory estimates on using a dense representation of the 2-rdm or two-electron integrals for real, double precision numbers are approximately
N - Memory Cost 100 - 0.7 GB 150 - 3.78 GB 200 - 11.9 GB 250 - 29 GB
These numbers are amplified significantly for the storage of the corresponding JW operators for example (the additional bytes for storing the associated strings), which might effectively limit us to something like 75-100 spin orbitals in practice if we aren't careful. This may not be sufficient to cost out something like Fe2S2, though it is perhaps better to discover the bottlenecks by pushing to something like a real instance of Fe2S2 if possible.