RESEARCH HIGHLIGHTS

When it comes to inspiration for innovations, humans have long looked to the natural world when designing new technologies. While birds and fish may have provided early strategies for new ways of transporting ourselves across air and sea, many researchers focused on new ways of transport have now set their sights lower—perhaps more accurately, they have set their sights smaller.

In an effort to design new ways to clean up pollutants in the environment or deliver drugs in a more targeted manner, for instance, scientists now study microorganisms like bacteria, algae, and sperm cells. By recording their behaviours and properties, scientists can gain insights for developing artificial microdevices that can effectively move with high precision in specific environments.

This research area has become so important, in fact, that the German Research Foundation (German: Deutsche Forschungsgemeinschaft, DFG) in 2014 funded an expansive initiative focused on better understanding “microswimmers,” or microorganisms able to effectively propel themselves in liquids.

While experimental techniques have helped scientists to better understand and design artificial microdevices that can mimic their natural counterparts, researchers may not be able to observe some of the smallest-scale interactions between organism and environment that play a major role in effective movements.

As a result, a team led by Prof. Dr. Jens Harting at the Helmholtz Institute Erlangen-Nürnberg for Renewable Energy (HI ERN) has linked up with the experimental group of Prof. Nicolas Vandewalle at the University of Liege in Belgium as well as the “Physics underlying life science” group led by Prof. Ana-Sunčana Smith at the Friedrich-Alexander-Universität Erlangen-Nürnberg.

The researchers set out to combine experimental studies and theoretical modelling with state-of-the-art computer simulations to better understand a particular subset of artificial microswimmers called magnetocapillary swimmers. The HI ERN group looked to the power of high-performance Performance Computing Center Stuttgart (HLRS) and Jülich Supercomputing Centre (JSC).

“The motivation for our research was to create as highly complicated a system as numerically possible without multiple approximations and to study its dynamics close to the experimental conditions and parameters,” Harting said. “Furthermore, our numerical method serves as an independent tool for verifying some theoretical predictions and open questions appearing in the experiments.”

Surface-level sophistication

Simply put, magnetocapillary swimmers can be just a few micrometre-sized beads of magnetic material floating on the surface of water. While these relatively simple systems might seem straightforward to study, the challenge comes from the subtle changes they make to their immediate surroundings and, as a result, the influence they exert on one another.

For a clearer picture, think of pouring breakfast cereal in milk. While the individual flakes may be light enough to float “on top” of the milk, they are still deforming the surface of the milk by pushing it down. Anyone who has tried to sleep on an underinflated air mattress knows what happens next—this deformation creates a gradient of sorts that causes other flakes to drift toward one another and cluster, further deforming the surface area.

If the magnetocapillary swimmers’ particles were just floating on water, a similar process would play out. Unlike cereal, though, these swimmers can be put under the influence of a magnetic field, and when properly configured, this magnetic field can not only oppose the particles’ attraction to ultimately offset the water deformation, it can also guide a swimmer to travel where the researchers want it to go. Understanding how to efficiently and accurately guide swimmers’ movements is a key component to developing artificial structures capable of assisting with precision tasks such as drug delivery.

When beginning their investigation, the researchers did not initially assume they would need some of Germany’s most power computing resources to simulate these interactions—this relatively simple system moves slowly enough not to create large-scale turbulent motions in the water. The swimmers’ slow movements actually wound up being among the most computationally intensive challenges for the team’s simulations, though.

The swimmers are guided by the subtle oscillations of a magnetic field, so to accurately simulate this process, the researchers must ensure that individual oscillations are represented within the simulation, meaning that they must advance time in their computations very slowly while also modelling small-scale hydrodynamic interactions. And while these small particles are not moving fast enough to create large-scale turbulent motion in the liquid in which they are floating, the individual particles’ movements still have subtle yet significant influence on other nearby particles’ movements. Taken together, if the researchers want a realistic view of swimmers’ movements and the constituent particles’ interactions with one another, they must have a huge range of time and size scales in their simulations.

With access to GCS computing resources, the team is already capable of simulating these interactions at the level of detail necessary to help verify experimentalists’ hypotheses. Unlike experiment, though, the researchers can also make slight modifications to inputs to speed-up the time-consuming, trial-and-error process necessary when designing these systems purely experimentally.

“Designing these materials is always an iterative process, but access to HPC resources allows us to speed up the iteration necessary to achieve desired outcomes significantly,” Harting said.

Swimming and simulating toward the horizon

The team noted that current generation machines such as Hawk at HLRS and JUWELS at JSC have allowed them to push their simulations to a point where simply adding more computer cores no longer helps the team achieve its results any faster. “For our simulations in particular, we are already today able to reach sufficient system sizes,” said Dr. Alexander Sukhov, a HI ERN researcher and collaborator on the project. “To further develop these simulations, we require faster cores in order to be able to achieve more time steps in less wall clock time.”

To achieve maximum performance on current generation compute cores, the team has worked closely with user support specialists at HLRS and JSC to address issues arising from how their code runs scripts on machines after updates as well as ensuring that they are getting the most out of the machine memory.

Moving forward, supercomputing resources will continue to play an increasingly important role in understanding the motion of magnetocapillary swimmers. In their interactions with experimentalists, Harting, Sukhov, and their collaborators already discovered that prior simulations did not fully account for the influence of particles on their nearby partners, and as the DFG Priority Programme continues, these types of exchanges between experimentalists and computational scientists will only become more essential.

For researchers to fully understand the complex mechanisms that influence biological swimmers’ motions, though, researchers will have to continue to focus on specific environmental and physics-based factors that play a role in propelling swimmers, whether that is light, gravity, chemical interactions, or other mechanisms. With the help of supercomputing resources, this cross-disciplinary collaboration aims to use HPC to take experimental data and run rapid-successions of simulations with slight modifications to input data, and ultimately design a new class of microdevices capable of helping clean up our environments and fight off illness.

 -Eric Gedenk

Funding for Hawk was provided by the Baden-Württemberg Ministry for Science, Research, and Art (MWK) and the German Federal Ministry of Education and Research (BMBF) through the Gauss Centre for Supercomputing (GCS).