Michigan State University
Zane Crawford wanted to study music composition at college. In high school he wrote music and was the main keyboard and organ player for his Detroit-area church.
He was largely self-taught, however. The two universities he applied to, Michigan and Michigan State, found the compositions “interesting, but I didn’t have the classical background to be part of their music programs,” Crawford says.
His second option – suggested by his father, a minister – was to study electrical engineering at Michigan State and design music playback devices that produce more realistic sounds. As a freshman, “the question was, what needs to be done in order to capture the correct frequency response of stringed instruments?” says Crawford, now a Ph.D. candidate studying electromagnetics at the same school.
As a Department of Energy Computational Science Graduate Fellowship (DOE CSGF) recipient, Crawford’s doctoral research first focused on finite element algorithms for electromagnetic applications. Such techniques are a way to computationally divide a simulated domain to calculate continuous physical processes. But his 2017 practicum at Sandia National Laboratories – and a discussion with his advisor, Shanker Balasubramaniam – brought Crawford’s studies more in line with his undergraduate goals.
“Now the question is, if I want a particular signal output, what physical object do I need to design given my input?” Crawford says. “It’s an interesting book-ending kind of thing. It’s not what I imagined myself doing 10 years ago, but it’s a fair compromise.”
The problem focuses on topology optimization: basically, finding the best design for an object to perform a specific task, in particular regarding its electromagnetic response.
Crawford’s Sandia project, under Michael Martin, revolved around creating arrays of ultracold atoms for quantum computation and other purposes. He was to develop an algorithm that generates holograms. The light patterns the holograms produce act as optical tweezers, moving atoms into specific arrangements.
Crawford used a graphics processing unit to accelerate the hologram-creation process. He built a suite of programs to simulate creating holograms and collected statistics on their quality to optimize them – improve their effectiveness. The algorithm also had to dynamically adjust the holograms.
When Crawford returned to Michigan State, Balasubramaniam suggested he examine another optimization problem focused on topology – “basically, to create design requirements with the finite element method I’d been working on. With that I’m able to come up with novel designs for different electrical devices” – much as he wanted to do as an undergraduate.
The goal is to computationally plan an apparatus “to best get me from some input signal to some desired output signal,” Crawford says. His algorithms must choose the best orientation and properties for a device to meet that goal. One example is optical cloaking, in which a material or cover scatters light or radiation to make the object inside undetectable. “You can use topology optimization to figure out what kind of shape of material you need to put around that object” to cloak it.
Crawford uses a finite element approach similar to one researcher Daniel White developed at Lawrence Livermore National Laboratory. Once the finite element system is in place, Crawford use a topology optimization algorithm to fine-tune selected parameters, or properties, affecting the system to get the best possible device design.
“The beautiful thing is topology optimization is a broad framework” that can be applied to codes that incorporate other physics, such as particle-in-cell models that simulate plasma behavior.
Crawford and Balasubramaniam have developed their approach on Michigan State’s local computing cluster. They expect to test it on DOE computing systems once they revise its underlying structure.
Crawford aims to graduate in 2020, then will seek a postdoctoral research post in electromagnetics with a DOE laboratory or government agency.
Meanwhile, he still plays keyboards for that same Detroit church, and he’s recently resumed composing “short pieces to take my mind off things for a bit.” As a player, he tends toward a jazz style, but his compositions usually fall into a classical mode – with a twist. “I can’t just do a simple major chord” to end a piece as Bach or Mozart did. Instead, it’s a variation “that other jazz musicians would be like, ‘Oh yeah, that’s normal.’”