Code: github.com/eriksen-lab/...

On my end, thanks to @villumfonden.bsky.social & Independent Research Fund Denmark for generous funding.

As such, we believe our work on the OpenTrustRegion library has massive potential in making genuine and robust second-order methods widely applicable throughout our community. This is the first paper in a new Nordic collaboration with @idamaho.bsky.social & Susi Lehtola. More to come!

Our permissibly licensed implementation can be included within both free and open-source, academically licensed closed-source, and commercial software packages. In the study, we present examples of interfaces to three different codes, and we bid future interfaces and extensions most welcome.

To that end, our implementation guarantees robust convergence onto a local minimum in both SCF calculations and other types of applications, and we demonstrate the difference in speed to conventional methods to be negligible, even for relatively simple systems.

In our opinion, the risk of false convergences onto saddle points in applications of HF and KS-DFT is undeniably real for first-order (and even many second-order) solvers to all but the most simple of molecular systems. These can only be safely avoided by employing robust second-order optimization.

As examples, our study presents applications to ground-state self-consistent field (SCF) calculations as well as the spatial localization and symmetrization of molecular orbitals, and we further highlight a variety of other interesting use cases towards the end.

The OpenTrustRegion library is designed with an intention to be more versatile, more robust, and more reusable than any previously reported implementation of the second-order trust region method for general electronic-structure calculations.

Please see the official call for more information. In case you have specific questions related to the position or research theme, then write me directly: janus [at] dtu [dot] dk.

The ideal candidate for the position is a driven early-career scientist with a deep and genuine interest in seeking to better influence and complement chemical synthesis by inferring mechanistic details and reaction rules.