You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
@KaterinaGratsea and I found that generating the double factorized block encoding for large instances requires a lot of memory because the init method of ChemicalHamiltonian will perform a Jordan-Wigner transform on the Hamiltonian. For example, it was not possible to generate the block encoding for a 46 orbital Hamiltonian on a machine with 30 gb of memory.
Since the JW transformed Hamiltonian is not needed for the DF block encoding, I think it would be good to update ChemicalHamiltonian to not perform JW during initialization. Note however that this is not a blocker for us: we're able to work around the issue by just commenting out these lines.
The text was updated successfully, but these errors were encountered:
@KaterinaGratsea and I found that generating the double factorized block encoding for large instances requires a lot of memory because the init method of
ChemicalHamiltonian
will perform a Jordan-Wigner transform on the Hamiltonian. For example, it was not possible to generate the block encoding for a 46 orbital Hamiltonian on a machine with 30 gb of memory.Since the JW transformed Hamiltonian is not needed for the DF block encoding, I think it would be good to update
ChemicalHamiltonian
to not perform JW during initialization. Note however that this is not a blocker for us: we're able to work around the issue by just commenting out these lines.The text was updated successfully, but these errors were encountered: