Over the past several years, researchers at Google and a nuclear fusion startup called Tri Alpha Energy have been quietly working on an algorithm to advance nuclear fusion research.

On Tuesday morning, the researchers published a report in the journal Scientific Reports describing the “Optometrist Algorithm,” a machine-learning tool that aids in choosing parameters for a nuclear fusion experiment. The algorithm asks researchers to select between pairs of outcomes, allowing them to narrow down a complex set of parameters.

"To increase the speed of learning and optimization of plasma, we developed the Optometrist Algorithm. Just as in a visit to an optometrist, the algorithm offers a pair of choices to a human, and asks which one is preferable. Given the choice, the algorithm proceeds to offer another choice," write the researchers.

According to the authors, the tool is necessary to help with guiding experiments: "The highly nonlinear and temporally varying interaction between the plasma, its environment and external controls presents a considerable complexity in these experiments. A further difficulty arises from the fact that there is no single objective metric that fully captures both plasma quality and equipment constraints," they write.

Using the tool, Tri Alpha Energy researchers were able to hold hotter nuclear plasma for longer periods, one of the keys to cracking the complex code of nuclear fusion. Tri Alpha Energy president and CTO Michl Binderbauer described the research as having “huge promise for all of us” that would be able to “speed up...physics research.” 

Nuclear fusion technology smashes atomic nuclei together to release energy. It can generate much more energy than nuclear fission, the process of splitting atoms apart, which is used in today’s nuclear reactors.

However, no one has been able to make fusion reactors commercially viable due to high energy and temperature requirements.

The researchers’ algorithm combines the best of computer intelligence and human intelligence, and uses them both to advance an ultra complex experiment. A nuclear fusion experiment can have thousands of individual factors that could affect the outcome, and a computer on its own might optimize an experiment to make it unsafe.

"The inclusion of expert oversight in parameter exploration is especially important given the large number of simultaneously-perturbed MPs. For instance, a priori identification of safe operating regimes cannot always be made for all combinations of machine parameters. As such, it is possible to set the machine to an unsafe state due to unanticipated nonlinear interactions between settings. This is not clear until the unsafe shot is actually run," write the researchers.

The Optometrist Algorithm could be used to work on any type of complicated problem, not just energy.

“The creativity and ingenious spark in the human mind is what drives progress. When combined with the tireless, agnostic and data-driven work of an AI algorithm,” there can be major progress, said Binderbauer. “The fusion field in particular has not done good job of exploiting these things on a strategic level,” he added.

Tri Alpha Energy -- a nearly two-decade-old company that has raised $500 million in funding -- developed the Optometrist Algorithm while working on its C-2U plasma generator. The algorithm enabled the team to optimize the generator, but the company plans to use it on other aspects of its research such as its new reactor, called Norman.

Tri Alpha Energy uses a nuclear fusion design that shoots beams of plasma into a vessel where it’s held in place, spinning, by a magnetic field. The design shares some properties with particle accelerators.

Traditional nuclear fusion reactors tend to look more like the donut-shaped rings of “the tokamak.” This technology turns hydrogen fuel into plasma under intense heat and pressure, shaped by magnetic coils. At high heat, the plasma particles can fuse together, releasing fusion energy. 

Tri Alpha Energy's investors include Microsoft co-founder Paul Allen, Goldman Sachs, Wellcome Trust, and Silicon Valley’s NEA and Venrock. Former Energy Secretary and physicist Ernie Moniz joined the company’s board of directors earlier this year.

The work was done in collaboration with machine-learning researchers at Google Research.