Sanibel Symposium

The reliable description of molecular phenomena, such as excitation energies and
magnetic properties, often requires an accurate electronic wavefunction that goes beyond the
scope of single-reference coupled-cluster techniques. Internally contracted multireference
coupled-cluster (icMRCC) can be used to provide these accurate benchmark results and, when
combined with embedding schemes, investigate large molecular complexes.

However, the standard icMRCC equations can become computationally expensive to
solve, par-ticularly if the embedding region in a molecule needs to be increased to capture
the relevant chemistry. This is primarily due to the large number of terms that arise in the
energy and am-plitude equations which can far exceed the number of corresponding terms in
single-reference coupled-cluster.

In response to this to this problem, linear and quadratic approximations have been
developed which aim to decrease the computational expense and increase the scope and usage of
icMRCC. These have been inspired by older methods, like the well known CEPA(0)
approximation, based upon the unlinked coupled-cluster formalism. The approximations are
benchmarked against full configuration interaction, to investigate the accuracy and shed light
on interesting features of these methods. The quintet-triplet splitting energy of an FeC₇₂N₂H₁₀₀
complex is also inves-tigated, suggesting a future use of these approximations.