[molpro-user] Printed orbitals accuracy

[Kaveh Haghighi Mood]

Dear all,

I want to use wave function generated by Molpro for Quantum Monte Carlo 
calculations. My interface works fine, if I generate wave function
using HF method and I had same results from Molpro orbitals and other 
packages like Gaussian and GAMESS.
The problem starts when I want to use multi-determinant wave function 
from CASSCF calculation. Gaussian and Molpro energies are exactly same 
but my QMC results (that only depends on orbitals, determinants and ci 
coeffs. I set Jastrow factor off)  are a bit higher but not negligible.

I extract information from Molden format file. I want to know how 
accurate are printed pseudo canonical (or natural) orbitals in Molden 
file and if they are not accurate enough how can I increase the accuracy?

Any clarification  appreciated,


Regards,

Kaveh Haghighi Mood