Dear MRCC developers,
I'm trying to calculate the DF-CCSD analytical gradients but got the following error
Code:
Executing prop...
Allocation of 1800.0 Gbytes of memory...
forrtl: severe (39): error during read, unit 26, file /public/home/jxzou/vip_test/ccsd_test/rhf/riccsd_force_mrcc_FC/CCDENSITIES
Image PC Routine Line Source
prop 0000000008BC9C15 Unknown Unknown Unknown
prop 00000000035AD023 Unknown Unknown Unknown
prop 000000000040BEB8 Unknown Unknown Unknown
prop 000000000040B4D7 Unknown Unknown Unknown
prop 0000000000407929 Unknown Unknown Unknown
prop 00000000004054FD Unknown Unknown Unknown
prop 0000000008C71B7A Unknown Unknown Unknown
prop 0000000008C73A45 Unknown Unknown Unknown
prop 00000000004053D5 Unknown Unknown Unknown
Fatal error in prop.
It seems that the code wants to read an array whose dimensions are different from dimensions of data in CCDENSITIES. The input file is
Code:
mem=1800GB
charge=0
mult=1
calc=DF-CCSD
dfbasis_scf=def2-QZVPP-RI-JK
dfbasis_cor=def2-QZVPP-RI
core=1
dens=1
scftype=RHF
basis=PVTZ
ecp=none
scfiguess=mo
symm=0
qmmm=Amber
geom=xyz
3
O -0.11328527 -0.79814623 0.00000000
H 0.84671473 -0.79814623 0.00000000
H -0.43373986 0.10678960 0.00000000
pointcharges
0
This problem disappears when I turn off the density fitting (i.e. delete dfbasis_scf and dfbasis_cor, change calc=DF-CCSD to calc=CCSD). But the error remains when I change core=1 to core=0. Besides, I tried two versions (25.1.1, 26.1.3) and the error exists in both cases. You can find input and output files in riccsd_force_mrcc_FC.tar.gz, but I'm sorry that several large files are deleted such that it can be uploaded.
Are DF-CCSD analytical gradients not supported currently? Thanks.
Best wishes,
Jingxiang
