#### REBECCA MORRISON

PhD CSEM 2016

Postdoctoral Researcher

Massachusetts Institute of Technology

There are many problems in math, physics, engineering, etc. that do not have an analytic solution, that is, even though the problem can be properly formulated, we cannot easily write down a solution. Computational science allows us to find the solution to a given degree of precision, or over a finite domain, by directly simulating the problem with very small increments in time and space. You could compare this to a picture on a screen: to our eyes, the image appears to be continuous, but in fact it is made up of many discrete pixels. But it does not just yield a discrete version of something we already know: computation actually allows us to study and solve problems that cannot be done otherwise. When modeling something complex like a jet engine, or the climate, or blood flow, there are so many processes and variables and forces occurring. Computational methods can incorporate all these things for arbitrary scenarios.

I work in a field called uncertainty quantification, and I research the problem of having uncertain or inadequate models. If there is a problem with your mathematical model, then you would first try to improve it directly (deterministically) by incorporating more physics (for example). Unfortunately this is not always an option. So I work on a formulation of what’s missing that is both probabilistic and constrained by the physics; we call it a stochastic inadequacy operator.

As it matures, computational science and engineering will get bigger and faster. I had never heard the term “exascale” until a couple years ago, and now there are seminars, summer schools, and serious projects dedicated to exascale computing. Of course, with more and faster cores, we will be able to tackle bigger and harder problems. On that front, I think it will be interesting to see how collaboration develops and improves as projects grow in scope. More computation will be truly meaningful if we have the ability to share code and keep it open, efficiently store and transfer data, reproduce others’ results (and our own), and rely on computation when it makes sense, but not throw it at every problem as a fix-all. That is, computational science in ten years won’t just be about everyone using huge computers, but also finding faster algorithms, developing better mathematical and numerical methods, and making smarter assumptions.

Computational science has had a profound impact on the world in so many areas: transportation, finance, medicine, communication, and so on. It has made us safer, healthier, more efficient, better connected. A healthy scientific community is necessary for a healthy society in general, from the local to the international level, and I would like to continue a career in computational science to be a part of it. And maybe I can quantify some uncertainties along the way.

#### LINDLEY GRAHAM

PhD CSEM 2015

Data Scientist

Amazon

I would describe it as the method and art of using computers to learn more about the universe we live in and the design of computational tools to help us to do so. Another way to think of it is an epic mashup of computer science, applied mathematics, and your science/engineering field of choice.

My research focuses on developing and using measure-theoretic tools for parameter estimation, inverse problems, and uncertainty quantification and applying these tools to coastal ocean modeling. The majority of loss of life and damage due to a hurricane is caused by storm surge. I am working to help enhance hurricane storm surge modeling by improving our knowledge of parameters used to model bottom friction and various types of coastal vegetation. I use these measure-theoretic techniques to better understand and model momentum loss due to bottom friction in coastal regions along the Gulf of Mexico.

Well, it will get bigger and faster. I had never heard the term “exascale” until a couple years ago, and now there are seminars, summer schools, and serious projects dedicated to exascale computing. Of course, with more and faster cores, we will be able to tackle bigger and harder problems. On that front, I think it will be interesting to see how collaboration develops and improves as projects grow in scope. More computation will be truly meaningful if we have the ability to share code and keep it open, efficiently store and transfer data, reproduce others’ results (and our own), and rely on computation when it makes sense, but not throw it at every problem as a fix-all. That is, computational science in ten years won’t just be about everyone using huge computers, but also finding faster algorithms, developing better mathemati- cal and numerical methods, and making smarter assumptions.

If you think of computational science as “doing science with computers” then it is clear that computational science has enabled engineering feats and scientific discoveries that would not have otherwise been possible. I want to use computational science, especially the fields of uncertainty quantification and inverse problems, to instill confidence in existing computational models and enable the modeling of events and phenomena that would not otherwise be possible. More generally I want to use computational science in a way that helps other people.

#### OMAR HINAI

Ph.D., CSEM 2014

Software Developer

Siemens PLM

The thing to keep in mind is that even though there are so many different fields of science and engineering, many of the problems encountered in these fields are actually very similar. Mathematics has made enormous contributions to understanding and solving large classes of such common problems. Yet, many problems continue to elude even the greatest minds. The invention of the computer, and its ability to run billions of calculations, has opened new opportunities to tackle these problems. There’s a simple idea behind it all: instead of solving the equations exactly, you get the computer to generate ap- proximations to the solution that are increasingly accurate. In computational science, we study how these methods work, how to make them faster and more accurate, how to use them for real problems.

At Siemens, I’m using mathematical theory to solve practical problems for oil and gas. I was well prepared for this by my work at UT. While pursuing my Ph.D. I worked at the ICES Center for Subsurface Modeling with Prof. Mary Wheeler. We focused on porous media flow modeling, that is to say, how fluids move underground. This includes oil and gas reservoirs as well as water aquifers. Underground formations can have very complex geometrical features, which introduce certain challenges when solving for fluid flows. Specifically, I look at how a method called the Mimetic Finite Differencing can be used to address some of these challenges, and how it relates to existing known techniques.

One of the most challenging aspects of our work is hav- ing a functional knowledge of certain disciplines without the luxury of studying them formally. The kind of work we do involves geology, fluid dynamics, thermodynamics, petroleum engineering and even some solid mechanics. Collaboration is critical in multidisciplinary work, but even then you need some understanding in order the make the most out of it. The tricky part is reading enough without going fully down the rabbit hole.

If you think of computational science as “doing science with computers” then it is clear that computational science has enabled engineering feats and scientific discoveries that would not have otherwise been possible. I want to use computational science, especially the fields of uncertainty quantification and inverse problems, to instill confidence in existing computational models and enable the modeling of events and phenomena that would not otherwise be possible. More generally I want to use computational science in a way that helps other people.

A mentor of mine used to say that “simulators are great integrators.” What he meant was that they allow us to combine all the information about a problem in one place. All the equations, theories, data, observations, experiments can be incorporated into these codes, and you can watch it interacting on your screen. Bringing information together also means bringing people together. Just look around ICES, you’ll find mathematicians, engineers, computer scientists, geologists, biologists, chemists, physicists, doctors and so many more. Where else can you see that? We’re always waiting for that big new discovery that changes the world. What if that big idea is already out there? Someone might have a great solution to a problem they didn’t know existed. Bringing experts together, allowing them to share and test their ideas on a shared platform, that’s what computational science does very well. Pooling the world’s talent—that changes everything.

RELATED LINKS

ICES Ranked Number One Interdisciplinary Math Program

Professor Hughes and 8 of his former students named most Highly Cited Researchers 2016