Hey, coders: Want to do nuclear fusion research? This crowdsource challenge gives you a chance

Geekwire ALAN BOYLE on December 11, 2015

General Fusion’s Brendan Cassidy shows off a test reactor in Burnaby, B.C. (Photo by Alan Boyle)

It’s not clear when fusion power will pay off, but there’s a way to earn a cool $20,000 in fusion research. And you don’t even have to be a plasma physicist or an energy entrepreneur.

All you have to do is make perfect sense out of the data generated by the plasma experiments being conducted by General Fusion in Burnaby, B.C.

“The challenge is basically to come up with a metric for predicting the performance of a plasma shot,” Brendan Cassidy, the company’s crowdsourcing project leader, told GeekWire.

General Fusion is a private venture that’s attracted tens of millions of dollars in venture capital, including investments from Amazon billionaire Jeff Bezos. Over the past five years or so, the company has conducted about 100,000 experiments. Those experiments, or shots, involve injecting blobs of super-heated hydrogen gas into plasma chambers and studying how they behave. A single shot lasts somewhere around a thousandth of a second.

“Our shot data includes signals from nearly 100 probes measuring things like magnetic field strength, plasma density and the spectral composition of plasma light,” Cassidy explained in a blog post outlining the challenge. “There are also configuration settings for each shot, and calculated single point, or scalar, metrics.”

The quality of the plasma varies from shot to shot, and General Fusion’s researchers don’t fully understand why. It’d be nice to distill the shot data into algorithms that predict which settings will produce the best shots.

Toward that end, hundreds of gigabytes of data from previous shots are being made available for a challenge titled “Data-Driven Prediction of Plasma Performance.” After signing up, competitors can download the data, look for correlations and patterns, devise their algorithms and send them in for evaluation by March 9.

Competitors can submit a new algorithm every week. To stir the pot, General Fusion will post weekly standings that rate the competitors. At the end of the contest, the best algorithm will win the $20,000 first prize. There’ll also be several $2,000 awards for runners-up.

Head on over to the InnoCentive website for rules and registration. The contest went live on Thursday, and 100 “active solvers” have signed up already.

We’re not looking for participants who are plasma physicists.

This is the second crowdsourcing challenge that General Fusion has sponsored. The first challenge, posed this spring, asked competitors to come up with an improved design for a fusion reactor seal. That contest attracted more than 60 submissions from 17 countries.

The winner of the seal competition’s $20,000 prize was Kirby Meacham, an MIT-trained mechanical engineer from Cleveland with more than 30 years of experience in the field and more than 35 issued U.S. patents.

That level of experience probably won’t be necessary this time around.

“We’re not looking for participants who are plasma physicists and aren’t looking for a physics analysis of the data,” Cassidy said. “Rather, we’re looking for participants who can find statistical patterns in large amounts of difficult-to-digest data, just like is done with data of all kinds, all over the Internet, every day.”

Improved plasma performance could hasten the day when General Fusion’s plasma shots achieve nuclear fusion in such a way that the energy going into the system is exceeded by the energy coming out. And that’s the real prize.