Finding dark matter is a major goal of scientists around the world—and now also undergraduates at the University of Chicago. This past summer, third-year student Rodrigo Spinola Castro was one of the College students who contributed to scientific progress at the national laboratories, through a project working to develop technologies to detect dark matter at Fermilab.
Though crucial to our understanding of physics, dark matter presents a wide range of unique obstacles that make it challenging to study. Since it has never been directly detected, scientists have developed various theories about what it could be. While most dark matter searches are done underground, shielded from the environment and background radiation, one team at Fermilab is taking their hunt to the reaches of space.
Spinola Castro worked with the DarkNESS CubeSat team to develop hardware for an instrument that will search for two signatures of dark matter that can only be studied from space. The instrument consists of four skipper-CCDs, an ultra-sensitive detector that has been used to set world-leading dark matter limits in underground experiments.
Taking this technology to space comes with a number of challenges that the DarkNESS team must solve before its upcoming launch next year. A small satellite only has so much power available—38 watts, to be exact, not even enough to run a toaster—and so the team had to design a method to cool the instrument in space and maintain its temperature without using up all the available energy.
The detectors are cooled by a cryocooler that is powered and controlled by a credit-card-sized electronic board that Spinola Castro designed and built.
“As a physics major, I didn’t have much experience with electronics,” said Spinola Castro. “However, over the course of the summer I realized how much I really liked building these things and contributing to the work of the team at Fermilab.”
The DarkNESS CubeSat will search for two telltale signs of dark matter, hoping to shed light on one of the universe’s greatest mysteries. By building and flying the small satellite, the team is developing a platform to test new detector technologies that could play a role in future space observatories.
Finding patterns in chaos
While Spinola Castro works to build a tool to detect dark matter in space, third-year student Defne Erdogan has turned her attention to atmospheric modeling, working with AI systems that could predict weather patterns.
At her internship at Argonne National Laboratory this past summer, she helped build advanced machine learning models that can replicate expensive weather simulations at a fraction of the cost. This could boost scientists’ ability to conduct effective daily forecasts.
“Physical simulations are really important, but doing those simulations is computationally expensive,” said Erdogan. “These [models] reduce the high cost of simulations by using machine learning models to make predictions.”
In particular, Erdogan worked on diffusion models, which are often used in cutting-edge image generation. But instead of creating art, Defne is teaching computers to forecast storms.
“You can take weather data and then apply noise to it until it becomes all noise,” Erdogan said. “Then, you can train a neural network to reverse it to predict the weather.”
It’s like teaching someone—in this case, an AI system—to restore photographs by first showing them how to clean up deliberately damaged images. Once they master the restoration process, they can generate realistic-looking photographs from distorted, “noisy” images. Erdogan applies this same technique to weather data, training AI systems to clean up noise and generate realistic weather patterns.
Though her focus is on weather, the modeling techniques Erdogan is implementing could transform scientific computing across disciplines. By making the most demanding simulations computationally cheaper, it could even unlock insights in areas like nuclear fusion or galaxy formation.
Science on the ground
For both young scientists, the right mentorships and relationships have been crucial for bouncing new ideas off the wall, stretching their thinking, and workshopping future research ideas.
“Everyone is so collaborative and they’re really open to questions, even if it’s not their job,” said Erdogan. “If you go up to someone with questions, they will stop doing their work and help you, even though they might not be your advisor.”
Spinola Castro found similar value connecting with other students and researchers, using relationships as a sounding board to learn and test ideas. Throughout the summer, while learning how to build circuit boards for the CubeSat, he never worried about asking for help.
“The environment at Fermilab was nice enough where I could go up to literally anyone and ask questions,” said Spinola Castro. “People who weren’t even working on the same project were more than happy to spend time talking with me, even lending me equipment when I needed it.”
Spinola Castro lived in Hyde Park over the summer and commuted to and from Fermilab in Batavia with other UChicago students. He made the most of that time as well.
“I had two captive graduate students stuck in the car with me for two-to-three hours a day, who I asked absolutely everything that I had on my mind,” said Spinola Castro, who also learned about graduate education at the lab.
For Erdogan, building connections has also been a way to boost her confidence while navigating unfamiliar research territory. She encourages other undergraduate scientists to take the same approach.
“Reaching out to people is really important, and asking for their advice,” said Erdogan. “They’re ones that have been in your shoes, so they can help you make better choices.”
Beyond building networks, both students discovered that getting their hands dirty with actual research transformed their connection to science in ways classroom learning never could.
“Fermilab gave me a chance to work on really interesting hardware with professionals you might not get a chance to interact with in a classroom setting,” said Spinola Castro.
This learning-through-doing approach has also become essential for Erdogan. She encourages other undergrads to jump into exciting research projects, even if it takes them beyond their comfort zone.
“Don’t wait to be ready. Don’t wait, thinking that you need to know it all before entering into research,” said Erdogan. “Because you learn a lot while you’re doing the work.”
—A version of this story was published on the University of Chicago College website.
link