Google and Tri Alpha Energy partnered and started working together in 2014, and this paper is the first publicly-announced result from the collaboration. Their objective was to optimise the performance of plasma fusion experiments - an incredibly difficult task using traditional testing methods.
Ted Baltz, a Senior Staff Software Engineer at Google explains why the problem is so complicated:
"The ion temperature is three times larger than the electron temperature, so the plasma is far out of thermal equilibrium, also, the fluid approximation is totally invalid, so you have to track at least some of the trillion+ individual particles, so the whole thing is beyond what we know how to do even with Google-scale compute resources."
There appears to be no single metric which allows for both limitations with equipment and quality of the plasma, and scientists disagree on what the definition of 'good' performance is when relating to plasma experiments - how can you improve something when you can't agree where the base line is? To help them, the scientists developed the Optometrist Algorithm. (Named after Optometrist because when you get your eyes tested, the optometrist will ask which of the two options is better then then he will decide what to do next.)
This algorithm offers a user two alternative experimental scenarios along with the associated outcomes, then the user chooses which experiment produces 'subjectively better results'. A human was required since apparently even Google's immense computational capacity would not have been sufficient to process the amount of data and variables required to get accurate results on its own.
Using this technique the scientists discovered an over 50% reduction in the rate of energy loss and an over 50% natural subsequent increase in both ion temperature and total plasma energy. Specifically the neutral particle beams were dumping energy into the plasma, and were matching the rate of lost particles due to cooling (it's very hard to keep plasma hot). For about 2 milliseconds the total energy in the plasma increased. 2 milliseconds doesn't sound like a lot, but the whole experiment is only run every 8 minutes for up to 10 milliseconds at a time.
Advances such as this one, although taken on its own offer little benefit, are forming the building blocks of our understanding and ability to harness fusion power. Who knows, maybe by 2040 we'll all have fusion powered cars and not have to worry about combustion engines.
The paper was originally published in Scientific Reports on the 25th July 2017 and can be found HERE.
Credits to Google Research for the header image - HERE
What do you think of this? Let us know below!
WRITTEN BY Bedders
EDITED BY Infinity
