“The results are going to be fundamental,” says Diakonikolas. The Air Force Office of Scientific Research agrees.
Any given computational problem has a number of possible solutions, each with its own set of advantages, limitations, and unintended consequences. This is true of life, as well, and many of us find it difficult to navigate the endless decisions required to move through life efficiently. Computer Sciences Assistant Professor Jelena Diakonikolas, however, revels in these possibilities, and has even built her career around highly complex theoretical computational problems. She sees them as riddles to be solved—or rather, optimized.
An expert in continuous optimization, Diakonikolas’ work recently earned her recognition as an Air Force Office of Scientific Research (AFOSR) Young Investigator for 2024. One of just 48 scientists to receive the annual award, the honor comes with a three-year grant of up to $450,000 to fund basic research. Beyond the prestige, the award also holds great personal significance to Diakonikolas: “This award is special because it recognizes both the research as well as the researcher,” she says, “and it’s a project that I personally care about.”
For the AFOSR, yet another important aspect of Diakonikolas’ research is its rigor.
Towards well-deserved recognition of promising research
Diakonikolas earned the distinction for her proposal “Towards Fine-Grained Complexity of Nonsmooth Optimization,” which focuses on a branch of theoretical mathematical optimization that, according to Diakonikolas, is “considered to be among the most challenging.”
“In continuous optimization, which is the area that I work in, we have a way of studying the complexity of different classes of problems,” she says. “That model, known as the oracle model, was established by [Arkadi] Nemirovski and [David] Yudin in the late 70s.” In her proposal, Diakonikolas attempts to build on this foundation by turning the focus towards the complicated field of fine-grained nonsmooth optimization.
As you’d expect from an expert in mathematical optimization, Diakonikolas’ approach is pragmatic. “This is a very broad class of problems, and surely some are really, really hard,” explains Diakonikolas, “but there must also be some problems within that class that are easier, and which we may be able to optimize faster.” Her focus is within that subset.
“This proposal is about identifying the types of properties or the intrinsic structure of those easier problems, and then finding ways to optimize faster when those properties are present,” Diakonikolas continues. Along with PhD candidates Puqian Wang and Nikos Zarifis, she’s putting pencil to paper and seeking out theoretical answers to some of those highly consequential problems. “The results are going to be fundamental,” she says.
The Young Investigators Program “spurs new, unique ideas”
Of the 159 proposals received for the fiscal year 2024, just around 30% were accepted. Limited to exceptional researchers who have received their PhD within the last seven years, the program focuses on fostering “creative basic research in science and engineering, enhanc[ing] early career development of outstanding young investigators, and increas[ing] opportunities for the young investigators,” explained Ellen Robinson, program coordinator for the AFOSR, in an article shared by the Air Force Research Lab (AFRL).
A mutually beneficial arrangement, Young Investigators Program (YIP) provides funding to projects deemed relevant to the future of the Air and Space Force—not only does the program develop the skills of young researchers, but it encourages progress on key ideas addressing challenges within science and engineering.
“Optimization is used pretty much anywhere: engineering, finance, machine learning,” says Diakonikolas. “We’re working on algorithms of broad use, which could be used as a black box in software packages for these different applications.” For now, however, it’s all theoretical.
“The AFOSR YIP program enables young scientists and engineers the opportunity to explore ideas that perhaps would not be looked at by other funding agencies,” said Dr. Patrick Bradshaw, AFOSR program officer for human performance and biosystems, in an interview with the AFRL. “The YIP spurs new, unique ideas and starts interesting avenues of study.”
“For the department, it’s a good sign that young faculty get recognized with awards like this,” Diakonikolas says. “It shows the health and success of our department, especially when it comes to more junior faculty.” In other words: The future of UW-Madison’s Computer Sciences department is looking bright.
Congratulations again to Dr. Diakonikolas!