Purdue's nanoHUB@Home project lets anyone contribute to cutting-edge nanotechnology research – no PhD required

The next time you close your computer’s lid and put it to sleep, you could instead be putting it to work on research projects in computational nanotechnology.

That’s the idea behind nanoHUB@Home, the latest addition to the Berkeley Open Infrastructure for Network Computing (BOINC), an open-source project that lets users donate idle computing cycles to a variety of scientific research projects available on the site.

The nanoHUB@Home project is based at nanoHUB.org, a computational nanotechnology community built on cyberinfrastructure developed at Purdue. NanoHUB allows any researcher to put their simulation tools online, instantly enabling access to those tools from anywhere in the world, instructors, students and researchers can access these tools without having to download or install any software.

“It’s a win for us because we get access to a whole lot more computers than we would have access to otherwise,” says Ben Haley, an ITaP software engineer who works on nanoHUB, of the nanoHUB@Home project. “And it’s a win for the people volunteering their time because they’re passionate about scientific research and this is an easy way for them contribute.”

More than 600 individuals are already working on nanotechnology research through nanoHUB@Home. Users who want to donate their idle computer time can get started by downloading the BOINC client here.

In addition to helping individual users of nanoHUB get results faster, the nanoHUB@Home project is part of a larger effort to use collective knowledge to develop machine learning models that will be used to give users quick approximations of simulation results, potentially accelerating the rate of discoveries.

Even though nanoHUB hosts over 17,000 users performing millions of simulations per year, those simulations all involve different tools and parameters. This means that for certain parameters, there wasn’t enough data to train the machine learning models that can be used to make decisions, explains Alejandro Strachan, professor of materials engineering, who is leading the machine learning aspect of the project.

Using the BOINC system and the nanoHUB@Home project, he’s been able to automatically take advantage of all the donated computing time to run additional simulations to fill in the gaps.

“Having nanoHUB@Home really enabled us to do this,” says Strachan. “We’ve been able to very effectively distribute the jobs to the volunteer computers around the world and get back the answers that we needed.”

The eventual goal is to use the machine learning tools to give users approximate simulation results very quickly. Many of the simulation tools available on nanoHUB are computationally intensive and take days or weeks to return results. Knowing an approximate result immediately can better help users allocate their resources and decide whether waiting for precise results on that particular simulation is worth their while.

“The way simulations traditionally work is that you run one simulation and you get one answer,” says Strachan. “What this will allow us to do is connect all these simulations and with these models we can very quickly interpolate and learn from the collective work of a lot of people. That way, we get much more out of the simulation than what each individual would do with it on their own.”

To learn more about using nanoHUB@Home, contact Jared Gray West, communications specialist for nanoHUB, jgwest@purdue.edu.

Writer:  Adrienne Miller, technology writer, Information Technology at Purdue (ITaP), 765-496-8204, mill2027@purdue.edu

Last updated: September 3, 2019