Tim Linke optimizes his time as a summer student

As a student intern in Lawrence Livermore National Laboratory’s (LLNL) Computational Engineering Division (CED), Tim Linke learned a lot about optimization. Not only was he able to develop an optimization algorithm for next-generation computing, but he also made the most of his time by building relationships, getting involved in the Lab community and gaining new skills and confidence to take back to his Ph.D. program at UC Davis.

Linke was tasked with implementing a topology optimization algorithm known as “the method of moving asymptotes” (MMA) using MFEM, one of the Lab’s in-house finite element analysis libraries. In topology optimization, researchers aim to find better performing designs through an iterative process using gradients to adjust variables and move toward more optimal solutions.

MMA is a well-established tool in the field, but current implementations can’t use the resources offered by upcoming exascale-class supercomputers like El Capitan. By providing a new, ground-up implementation of MMA, Linke’s project gives LLNL the ability to approach a new generation of large-scale topology optimization problems for a variety of physical applications.

“We’re trying to get this piece of the puzzle at the very core of that algorithm to then be able to leverage the resources [we have] to tackle new problems,” he said.

Linke had always admired LLNL’s work, so he jumped at the opportunity to be an intern and learn more about optimization. Optimization is both a useful tool and a potential application for his dissertation research on computational multiscale analysis, a way to understand the physical phenomena that drives a behavior that’s being studied.

Upon joining the Lab, his mentors Mathias Schmidt and Boyan Lazarov immediately took him under their wings.

“Mathias and Boyan made an incredible effort to integrate me into the field and the Lab community,” Linke said. “Being able to shoot them a message, drop by their offices and always have an open ear was invaluable and allowed me to progress a lot faster.”

At his mentors’ encouragement, Linke attended lectures, participated in employee activities and made time to meet his fellow interns and researchers.

His favorite experience was participating in Engineering’s annual Summer Student SLAM, a directorate-wide competition where students are challenged to give short, engaging presentations on their work. Linke’s presentation was one of two from CED selected to participate in the finals, and one of four overall winners from across Engineering.

Most of all, he was impressed by the breadth of research at the Lab. His best memories are what he calls “aha moments” — where everything suddenly connected and he understood a new concept.

“Working with Tim is an absolute pleasure,” Lazarov said. “He is exceptionally focused and organized, which helped him progress quickly through the project and apply and extend his existing knowledge and coding skills. He also is very enthusiastic about learning new technologies, skills, and scientific theories and how the Lab applies these in solving real problems.”

Linke returned to his dissertation work with new optimization, coding and science communication skills, but hasn’t fully left the Lab: Now a Ph.D. collaborator, he hopes to keep learning and building connections while making a difference at a place he’d always idolized.

“The things that come out of LLNL are always incredible, whether it’s the results from [the National Ignition Facility] or the simulation codes,” he said. “Working here has always been a dream for me, so getting the opportunity to be a part of it and produce a code for them is absolutely amazing.”

–Noah Pflueger-Peters