2010PA1404 โ€“ Optimization of the code Octopus

PRACE

Status: finished project
Contract Number:
2010PA1404
Starting date 
1 March 2013
Ending date 
31 July 2013

Octopus is a computer code to calculate excitations of electronic systems, i.e., to simulate the dynamics of electrons and nuclei under the influence of external time-dependent fields. The code relies on Time-Dependent Density Functional Theory (TDDFT) to accurately describe the electronic structure of finite 1-, 2- and 3-dimensional systems, like e.g. quantum dots, molecules and clusters. The initial implementation relied on finite-system boundary conditions, but the most recent version also enables the user to specify periodic boundary conditions, opening the door to the simulation of 1, 2 and 3D infinite systems.

The main mode of operation of the code consists in the time propagation of the ground state wave functions upon disturbance by an external perturbation. This run mode is extremely efficient, scaling to tens of thousands of processors. However, in order to start these runs, a well-converged ground-state is required. For this, octopus has to be run in ground-state run mode, that is substantially different from the time-propagation run mode. Parallelization of the ground-state run mode is much more complicated, and as a result, octopus is not as well optimized as for the time-propagation runs.

With this project, we aim to improve significantly the efficiency of the ground-state run mode, in order to allow for
calculations of the excitations of systems of biological interest.

Objectives

Time-Dependent Density-Functional Theory (TDDFT) has proven to be quite accurate for predicting, ab-initio, the absorption spectrum of biological systems. Low-energy transitions between bound states are usually quite well described with TDDFT, the error being usually smaller than 40-50 nm. A few problems remain, most notably charge-transfer excitations, and transitions to weakly bound states like Rydberg states. Nevertheless, when properly validated, TDDFT calculations can be quite reliable, and are increasingly used by non-experts to support and interpret experimental results. Important reasons for the success of TDDFT in photochemistry are its cost/performance ratio which is unmatched by traditional methods and the relatively wide applicability range.

In many biological systems of interest, light absorbers (the chromophore) are not very large structures. However, the absorption profile is severely affected by the environment, that is, by the remaining structures of the biological system surrounding the chromophore. A predictive study of light absorption by a biological molecule has to take these effects into account. The most accurate way of doing it is by including nearby residues in the TDDFT
simulation. The drawback is the increase in size and complexity of the simulation, particularly at the level of the determination of the ground-state of the compound system (chromophore + residues).

We intend to use octopus to study light-absorption in a class of proteins that are relevant for optogenetic studies:
the channel-rhodopsin variants.

Participants

Fernando Nogueira
Micael JT Oliveira
Jose Rui Faustino de Sousa
Joseba Alberdi-Rodriguez

Related Research Areas