[molpro-user] LCCSD amplitudes restart in optg

[Matthias Heger]

Hi everyone,

I have been doing geometry optimizations using local coupled-cluster
methods and I am wondering whether it is possible to start the
single-point calculations for the numerical gradients from saved
amplitudes. From the log files, it appears that no such thing happens,
as there is no "CCSD amplitudes read from record ..." line even when
defining an optimization procedure in which a start command is
explicitly given for LCCSD. Instead, the program restarts from scratch
at each point and thus takes a while to converge. This only seems to
happen when using local correlation, not in the canonical formulation;
but I think I could still save a considerable amount of time if picking
up the amplitudes from the last reference geometry was possible. Does
anyone have an idea how this could be realized?

Best regards
Matthias