The ‘fog’ mentioned in the title is my poetic licence for plasma. The quest for limitless fuel with minimal pollution has been the dream for many generations. Renewables such as solar and wind go someway there until the sun goes down or the wind stops blowing. We need to have a more predictable energy supply and to that end, sustained nuclear fusion has a lot to offer, promising so much but as yet, delivering very little.
Many fusion machines are based on the Tokomak. In the UK there’s the experimental JET in Culham, Oxfordshire. With a larger version, ITER, being built in France.
The C-2U machine , a compact toroid which contains a hot plasma field that uses magnets to contain the plasma and stops it hitting the sidewalls. The hot plasma exceeds the temperature of the sun but instability from many sources can cause the plasma to collapse or crash in to the internal wall, possibly damaging the equipment. ‘Burn’ periods are measured in microseconds and the experiment consumes more energy than it makes, so far.
Google is now collaborating with fusion research using an AI approach called the Optometrist algorithm. As you might imagine with experimental equipment, many changeable values are put together to try and optimise a sustained plasma burn.
The goal is to get ever increasing burn periods. However, there can be shifts in parameters during the experiment which can be both unpredictable and almost impossible to correct. Optometrist tries to collate all these measurements in a timely fashion and along with data from previous experimental burns, offers the physicists choices for future experiments to try and improve the burn duration. Historical and future data builds the AI learning model thus helping to define the elements which may benefit certain outcomes. The physicist is faced with the choice of where the experiment is leading, just like a patient defining which lens improves their vision during a visit to the optician.