We’re excited to be welcoming around half of our current TMCS cohort of students to the CCC in Bristol on Monday, and the other half the following week.
The first batch of students will be with us for the week, talking science with academics and group members, and trying to decide what they’re going to work on for their PhD project.
TMCS 2015 cohort
The second group of students will be working on 5-week short projects with all of the groups in the CCC.
The latest release of Molpro provides a range of new functionality, including methods developed in the Manby group
A key new addition is wavefunction-in-DFT embedding through the projector method with basis-set truncation. This enables straightforward embedding of practically any electronic structure method implemented in Molpro in a chemical environment modelled by DFT. Unlike many embedding methods of this kind
there are no issues with partitioning across covalent bonds.
We have performed many calculations using the combination CCSD(T)-in-DFT, with the expensive coupled-cluster calculation only on a few, chemically active atoms. The exciting opportunity here is that the method all but eliminates dependence on choice of exchange-correlation functional.
The research behind this functionality was carried out by Martina Stella, Simon Bennie and Rob Pennifold in the Manby group; and by Jason Goodpaster and Taylor Barnes in Tom Miller’s group at Caltech.
Distinguishable cluster theory
Distinguishable cluster theory with singles and doubles (DCSD) is a simple modification of CCSD, but with remarkable properties, including the ability to dissociate molecules correctly even with a closed-shell reference state.
The method was developed by Daniel Kats while he was a DFG-funded postdoc in the Manby group. There is much discussion about what exactly this approximation means (but see Daniel’s interesting paper about a screened Coulomb derivation). Whatever it turns out to mean, DCSD is already a useful tool and a powerful addition to the set of available methods. Now in Molpro running it is as simple as replacing ccsd with dcsd in the input (and the computational cost is the same as CCSD). Gradients as well as F12, orbital-optimized and Brueckner variants are available.
Thanks to the efforts of Dr James Womack, we now have Intception, an incredibly powerful framework for automatic implementation of Gaussian integrals from equations that define the recurrence relations.
More on this later – but feel free to look at the information online and use Intception to generate integrals for your project.
Intception is open source, but as with a compiler, you can generate inputs and outputs under any licensing model you like.