Developing intuitive tools for molecular simulations on powerful computers
To deliver open-source code that anyone worldwide can use, the Parallel Computing Centre (PDC) recently made a long-awaited breakthrough in creating code for quantum chemistry on graphics processing units (GPUs). "What we see is proof that the developed algorithm works," says Patrick Norman, Director of the PDC.
It all started many years ago when Patrick Norman was a doctoral student working in a Dalton program, where he wrote code to simulate spectroscopy. Shortly after moving to KTH, he and the core VeloxChem team decided to start the work from scratch.
"We brought our knowledge and previous experience into this new project and were able to choose to implement algorithms selectively, knowing they worked and had wide usage. We started with modern and strictly object-oriented programming in a Python/C++ hybrid," he says.
Aim is to develop easy programmes
The aim is to have modules and objects that are independent of each other and can interact reliably. VeloxChem has created a platform that can be run on CPU-based computers, and now, a breakthrough has been made to perform quantum chemical calculations on GPUs.
"We want to develop programmes that are so easy to use that research groups can use them independently without special training. This includes an ambition to provide almost seamless access to supercomputing resources like Dardel. "We want to blur the difference in the user interface between the laptop or desktop and the supercomputer so that the latter can work in the background without the user noticing it, in other words, an accelerated experience," says Patrick Norman.
The software developed by the VeloxChem team can be used in several areas of chemistry research, including energy, spectroscopy and materials. The overall goal of the project is to help researchers with characterisation and rational molecular design. As a rule of thumb, the simpler and more intuitive a program is to install and use, the more is required in terms of design and development. In this work, PDC, and by extension KTH, has played a crucial role with its investment in HPC programme development.
"In the borderland where researchers meet professional software developers, we see an exciting development that takes both their work to a higher level," he says.