There are two-dimensional pressure fluctuations within a larger simulation. These types of partial observations give new ways to test reduced turbulence models in both theory and experiment. The fusion center is credited.
To make fusion energy a viable resource for the world's energy grid, researchers need to understand the turbulent motion of plasmas. When the hot edge of the plasma is just centimeters away from the much cooler, it's a challenge for fusion devices to produce significant gains in net energy.
A candidate in the Department of Nuclear Science and Engineering at MIT, Mathews believes that the plasma edge is a particularly rich source of unanswered questions. The turbulent boundary is important to understanding fusion reactor designs because of the damaging heat fluxes that can strike material surfaces.
Scientists use numerical simulations to model turbulence at the edge to better understand edge conditions. The simulations of this region are among the most challenging and time consuming in fusion research. If researchers were able to develop computer models that ran much faster but with quantified levels of accuracy, progress could be accelerated.
Physicists at the tokamak have used a reduced "two-fluid theory" for decades, despite the fact that they don't know if the model is accurate. Mathews combines physics and machine learning to test the accuracy of the reduced turbulence model.
"A successful theory is supposed to predict what you're going to see, for example the temperature, density, electric potential, the flows," says Mathews. The relationships between these variables are what define a turbulence theory. The turbulent electric field and the electron pressure are two variables that are examined in our work.
Mathews used a novel deep-learning technique to build representations of the equations governing the reduced fluid theory in the first paper. He shows a way to compute the turbulent electric field from an electron pressure fluctuation in the plasma, which is consistent with the reduced fluid theory. The model used to relate the electric field to pressure break down is not as robust as it could be.
There are two-dimensional pressure fluctuations within a fusion simulation. These types of partial observations give new ways to test reduced turbulence models in both theory and experiment. The fusion center is credited.
Mathews compares the connection to higher-fidelity turbulence simulations in the second paper. It has previously been difficult to evaluate the first-of-its-kind comparison of turbulence across models. The reduced fluid model's predicted turbulent fields are consistent with high-fidelity calculations. The reduced turbulence theory works. Mathews says that one should check every connection between the variables.
Jerry Hughes, Mathews' Principal Research Scientist, says that it's difficult to recreate the turbulence seen in air and water. The work shows that machine-learning techniques can show a full picture of the rapidly changing edge plasma, beginning from a limited set of observations. I'm excited to see how we can apply this to new experiments, in which we don't observe every quantity we want.
New ways to test old theories and expand what can be observed from new experiments are created by these physics-informed deep-learning methods. David Hatch is a research scientist at the Institute for Fusion Studies at the University of Texas at Austin.
"Abhi's work is a major achievement with the potential for broad application," he says. If there is limited diagnostic measurement of a specific quantity, machine learning could infer additional quantities in a nearby domain, thereby augmenting the information provided by a given diagnostic. New strategies for model validation are opened by the technique.
Mathews sees a bright future for research. "Translating these techniques into fusion experiments for real edge plasmas is one goal we have in sight, and work is currently underway," he says. This is just the beginning.
A. Mathews et al., Uncovering turbulent plasma dynamics via deep learning from partial observations. There is a DOI: 10.1073/physRevE.104.025206.
A. Mathews and his team studied Turbulent field fluctuations in gyrokinetic and fluid plasmas. It's called 10.1063/5.0066064
Physical Review E is a journal about physics.
The story is being re-posted by MIT News, a popular site that covers news about MIT research, innovation and teaching.
The news about seeing the edge of fusion experiments in new ways with artificial intelligence was retrieved on January 6.
The document is copyrighted. Any fair dealing for the purpose of private study or research cannot be reproduced without written permission. The content is not intended to be used for anything other than information purposes.
