By Gillian Law.
Physicist and new RSE fellow Phil Hasnip specialises in software to predict materials properties, and in making that software accessible to all researchers.
Hasnip believes that most physics problems end up being materials problems. “You want a better battery? You need a better battery material. A better turbine? You need a stronger material for the blades. Wherever you look, materials are key.” says Hasnip. Running experiments on these potential new materials is expensive and difficult, so using computational methods to either predict what a material might do, or to explain what is going in on experiments, is incredibly useful.
Having completed his PhD at the University of Cambridge followed by several post doc positions, Hasnip is now at the University of York where he is developing tools including the CASTEP program which uses quantum mechanics to compute the properties of materials and chemicals.
There is a constant tension in research between taking the time to improve the tools used and getting research results, Hasnip says, and he hopes to use his fellowship as an opportunity to focus on improving the tools available and making them more accessible to all researchers, rather than just those with computational skills. “Many of the tools being used aren’t really high enough quality. They’ve been developed by researchers who are good scientists but aren’t always good software engineers. So I’ve been developing new tools to do similar things but with a proper planning and design process,” he says. “My goal in the fellowship is to focus on making the tools easier to use, so that you don’t have to be a computational person – you can run it if you’re an experimentalist who never liked computers very much, in the same way that you can use a web browser, or Microsoft Word. You just have to understand the science, not the programming.” This means talking to users to understand what they need, and creating software that they can control. “They might want, for example, to choose a temperature range to use, but they don’t need to know what algorithm I used to solve the equation,” Hasnip says.
To date, Hasnip has mainly worked with expert modellers who run simulation software, including CASTEP on the Archer national supercomputer. “Many of them develop software themselves, and even those who don’t do development have been using this for years. In my fellowship I’d like to make contact with people whose primary focus is experiments, and make sure the tools are useful to everybody.” The software also needs to be easy to maintain. “That’s another strand of my fellowship, to make sure that the software can basically look after itself, and the users don’t have to worry about it.”
A further aspect of Hasnip’s fellowship will involve work on BOUT++, a framework for writing fluid and plasma simulations in curvilinear geometry. “I’ve never been involved with this package before, but there is a big group at York using it to solve equations, and my role is to take the framework that they’re using and make it more generally usable. For example, we have a project with a food manufacturer to model liquid chocolate, and how bubbles form, to make chocolate bars better!” says Hasnip, “You can get somewhere with the program as it is, but the system is set up to expect certain things about the geometry, such as that it’s a sort of ‘doughnut’ shape, as that’s the shape of the reactor. So I’m working with their team to generalise the model”.
The University of York’s maths department is also looking to use Bout ++ for modelling how bacteria swim. “And they also have an extremely good way of testing their programs, and I’d like to learn how they do that in order to improve CASTEP.”
More generally, Hasnip hopes to use his position as a fellow to develop the research software skills of graduate students and early career researchers, and generally show researchers how software development skills can help their own work. “Some of the tools that we use for software, like version control to manage multiple versions, can actually be really useful for scientific papers, or just research papers generally. I’m keen to go to other departments within York and show them these ideas: if they can track who contributed what and when, it can make their lives a bit easier.”
Hasnip is keen to spread the word on the power of computing to help all researchers. “I’ve always been interested in both computing and in science, but I didn’t realise when I was younger that they could be combined. I think many people think of them as separate interests, but computing can help everyone,” he says. “The important thing is that they don’t need to understand how it works. I’d struggle to explain in detail how a combustion engine works, but I can drive! We need to make tools available to researchers so that they can focus on their work but benefit from computation.”