Most of us think of the planets as solid, unchanging bodies, anchors in the vast solar system.
That’s not exactly the case. Among the forces buffeting the planets are solar wind — streams of energy-charged particles flowing from the sun and speeding through space at about a million miles an hour.
Now GPUs are helping scientists unlock some of the secrets of solar wind and how they interact with planets and other objects in the solar system as they hurtle through space. Their work could point the way to understanding how planets evolve and possibly lead to a better understanding of the Earth’s future.
“Today we know the solar wind is the main reason that Mars has been losing its atmosphere,” said Shahab Fatemi, a post-doctoral fellow in space science at the University of California, Berkeley, during a talk at the GPU Technology Conference in San Jose.
“Mars today is very cold and dry, but it wasn’t always this way,” he said. “Evidence shows that it had an ocean over its surface, and possibly life, but that disappeared as Mars lost more of its atmosphere.”
Solar Wind and the Earth
Fatemi said that by better understanding Mars, researchers hope to gain insights into how solar wind could affect the Earth. Large outbursts of solar wind plasma — charged ions, electrons and other particles — can wreak havoc on the Earth, jeopardizing airline safety, causing electricity outages in cities and more.
Unlike Mars, Earth has a strong magnetic field, which provides a shield against the solar wind. At the same time, Fatemi said, “There are lots of theories that what happened on Mars can happen on Earth.”
To learn more, researchers conducted a series of simulations to better understand how the solar wind affected Mars in the past and how this force might affect it in the future. This is where GPUs and the CUDA parallel programming model made all the difference.
Simulation of how solar wind is stripping Martian atmosphere. Read a fuller description. Video courtesy of Shahab Fatemi and University of California, Berkeley.
“This kind of problem was considered computationally too expensive to solve just a few years ago,” Fatemi said.
Fatemi and his colleagues replaced their multi-CPU model with a system running on a CPU and just a single GeForce GTX TITAN X GPU. Not only did they gain a big speed advantage, the GPU system cost a mere three percent of the CPU-based model, Fatemi said.
How Did the Moon Get There?
Fatemi and his colleagues are also using a GPU-based model to understand the moon, which they hope will shed light on the early history of the Earth. A common theory says that the moon was formed from a giant impact on the Earth. The chunks of materials that flew off in this impact eventually formed the moon. If that’s the case, the moon and the Earth would be made up of similar materials.
By testing the electrical conductivity of the moon, Fatemi said he can better identify the moon’s composition and determine if it is indeed similar to Earth.
“No one exactly knows why the moon is there, and what it is made of. Our knowledge about the interior of the moon is very little, but hopefully, we can better understand the lunar interior structure,” he said. “Knowing all of these would help us understand the history of the Earth.”