Meeting societal challenges through cutting-edge computing
Complex challenges require complex solutions. One solution is the application of MSO methods, which has proven effective at solving such problems as predicting air pollution and climate change, improving the filtration process for drinking water, and optimising methods for intensity-modulated radiation therapy. These highly-complex methods are typically processed using the most cutting-edge ICT tools, including high performance computing (HPC) and big data. Although useful, due to their complexity, MSO methods require the support of skilled experts – experts who are in very short supply. To help fill this gap, the EU-funded MSO4SC project created an e-infrastructure to optimise the use of MSO methods. “Basically, we wanted to use complex pieces of software called mathematical frameworks, in combination with HPC and cloud technology, to optimise the resources being used,” says project coordinator Francisco Javier Nieto De Santos. “At the same time, we provided interfaces for using these maths frameworks that don’t require the user to have a deep knowledge of the technologies.” Simplifying simulations To ensure that the end frameworks matched real user needs, project researchers started by conferring with relevant stakeholders – namely mathematicians and pilot application users. Within the context of EU-MATHS-IN, an important European network of applied mathematicians, researchers organised workshops and a mini-symposium to let these stakeholders test and comment on the frameworks being developed. Within just 24 months, this process led to the improvement of three mathematical frameworks (FEniCS, Feel++ and OPM) and six applications addressing different domains (air quality prediction, design of wind turbines for clean energy, design of high field magnets, detection/treatment of neurodegenerative diseases, etc.). Each of these frameworks and applications was adapted to fit the specific needs of the project. “We set out to do a lot of things and the pressure was very high, but we managed to focus and deliver a good e-infrastructure that can be used by people with different interests and levels of knowledge,” explains Nieto De Santos. “We now have a solution that works quite well and that can be used to show good demonstrations.” All frameworks and applications can be deployed using containers and are available to be executed through the MSO4SC Portal. According to Nieto De Santos, this portal, integrated with a resource manager, is the project’s most important outcome. “The portal simplifies the execution of the simulations, meaning the end user just needs to take care of configuring the simulation inputs and let the orchestrator deal with the usage of resources,” he says. Pioneers The project’s main legacy is an operational e-infrastructure for mathematicians and people who want to run applications based on mathematical simulation. However, the MSO4SC project is also notable for being one of the first to use a combination of HPC and typical cloud resources to run containerised maths-based software. “We’re pioneers in the use of containers on top of HPC,” adds Nieto De Santos. “I think this is an important model of how we can achieve a fully integrated European Open Science Cloud where scientists can use any type of resource transparently.” With the project now closed, researchers are working to attract more users to the portal by improving current applications, adding new features, and marketing its benefits.
Keywords
MSO4SC, modelling simulation and optimisation, MSO, high performance computing, HPC, cloud technology, ITC, maths, mathematics