Photo by Ana Garnika on Unsplash
A New Direction
I’m excited (and honestly, a bit humbled) to share that I’ve been awarded an NSF EPSCoR Research Fellowship to support a new phase of my research career. As part of this, I’ll be spending a significant portion of my upcoming sabbatical year—starting this summer—at MIT, working with Prof. Pierre Lermusiaux and his group in the MSEAS (Multi-Scale Estimation and Assimilation) Lab.
This also connects directly to the work we’ve been building in the NUMO Lab (Numerical Modeling Lab) here at Boise State.
Where This Comes From
Over the past few years, I’ve been increasingly interested in how ideas from machine learning can be used not just to analyze data, but to actually steer scientific simulations.
In particular, NUMO Lab’s work has shifted to adaptive mesh refinement—methods that decide where a numerical model needs higher resolution, and where it doesn’t. These decisions are crucial: they determine both the accuracy of the simulation and how computationally expensive it becomes. This is a topic which has been a core of my early work with Frank Giraldo at the Naval Postgraduate School, and I am very happy to revisit it now.
What This Project Is About
The new project is about bringing learning into that loop.
Instead of relying solely on hand-designed heuristics, the goal is to develop data-driven approaches that can learn how to allocate resolution more intelligently, especially in complex systems like ocean dynamics or environmental flows.
It feels like a natural next step—one that builds on work we’ve been developing in the NUMO Lab and opens the door to new directions.
Students and Collaboration
A big part of what makes this especially exciting is that it’s not just me.
Two of my students will be directly involved:
- Antone Chacartegui (Computing PhD program)
- Hailey Stubbers (Mathematics MS program)
Both are members of the NUMO Lab, and Antone will be joining me at MIT for part of the visit. Haily just joined the Lab, and she will be implementing and testing of traditional refinement criteria to give the machine-learned algorithms a run for their money. I’m especially excited about this—it’s a great opportunity to build deeper connections between our group and the MSEAS lab, and to involve students directly in that collaboration.
Looking Ahead
Spending time at MIT is, of course, a big part of this. Pierre and his group have been doing pioneering work in ocean modeling and data assimilation for years, and I’m looking forward to learning from their perspective—and to seeing how these ideas might connect in practice.
More broadly, this feels like the start of a longer-term collaboration between our groups.
There’s also something personally meaningful about stepping into a new research space during sabbatical. Back in 2012, I have spent four months at the Isaac Newton Institute for Mathematical Sciences at the Unviersity of Cambridge, surrounded by top researchers in mathematical modeling of atmospere and ocean. That experience was a massive inspiration and a springboard for my then-budding research career. This MIT visit, 14 years later, in another Cambridge, will be a chance to reset, to learn, and to be a beginner again (at least in some aspects). I’m looking forward to that.
More to come as this develops—but for now, I’m very grateful for the support, and excited for what’s ahead.
