search

qcscine / qcmaquis

Release-only repository for SCINE QCMaquis, the DMRG software from the Reiher group.
BSD 3-Clause "New" or "Revised" License
32 stars 15 forks source link

Energy does not depend on max_bond_dimension in 011.input? #19

Closed dmrg-learner closed 8 months ago

dmrg-learner commented 9 months ago

Hi,

I tried to run https://raw.githubusercontent.com/Molcas/OpenMolcas/dbb6ed0fcd4552d29700dd95d6896ed20c9e5857/test/qcmaquis/011.input from openmolcas v22.10

I got Total MC-PDFT energy for state 1 -150.37153838

when I changed the input as max_bond_dimension = 10 I got Total MC-PDFT energy for state 1 -150.37153838

Namely, the energy does not depends on max_bond_dimension. Anything wrong in modifying the input? Thank you very much

weymutht commented 9 months ago

Hi :smiley: I think that you don't see a change in the energy because the problem is very small (a small active space and a highly symmetric system). You should see an impact of the bond dimension when going to larger systems.

dmrg-learner commented 8 months ago

Thanks! I modified nactel = 6 and ras2 = 6 (is that (6,6) active space?), the energy becomes different.

      Total MC-PDFT energy for state     1                               -150.35137034

and

      Total MC-PDFT energy for state     1                               -150.35085025

for 10 and 100 max_bond_dimension, respectively.