An Interview with Carol Meyers
By: Rachel Levy, Deputy Executive Director MAA @mathcirque
Carol Meyers, Associate Program Leader for Nuclear Weapon Enterprise Evaluation and Planning at Lawrence Livermore National Lab
The other day, I was lucky to catch up with Carol Meyers and talk with her about her work as a mathematician in industry. This interview is part of the MAA Tondeur Career Initiatives, a collaboration between MAA, AMS, and SIAM, made possible with generous support from Philippe and Claire-Lise Tondeur.
What is your job title?
My job title is Associate Program Leader for Nuclear Weapon Enterprise Evaluation and Planning at Lawrence Livermore National Lab, but I call myself a mathematician because no one ever told me not to.
What do you do in your job?
I think my job is really cool. In my group, we focus mainly on optimization. We want to make things go faster and be better. We do this by using mathematical modeling to find inefficiencies and reduce them. We model both for our own laboratory and the broader National Nuclear Security Administration (NNSA) and DOD enterprise.
What parts of your education are most relevant at work?
I have an undergraduate math degree and my graduate degree is in operations research (OR). I know some students are discouraged from taking applied math in college, which is unfortunate. I only learned about OR after I graduated. A statistics professor had a consulting business where I worked for a year after I completed my undergraduate degree; while there I took a class in OR and loved it. It is easy to say why I chose to attend graduate school in OR - it was the broadest path. I could prove theorems if I wanted to and I could also get an applied job if I wanted.
What are typical work activities for you?
We often have technical meetings that are about an hour-long to discuss a facet of a project. Sometimes we give each other presentations. I manage a group of people and also meet with them to talk about how things are going with their careers. One of the most fun parts of the day is doing technical work. I try to do that as much as I can, navigating around meetings that pop up. I also interact with sponsors who are paying for our work, such as at NNSA, and often collaborate with staff from other labs. I enjoy that part too. We are lucky that our whole team recently co-located into one wing of a building. This makes it really nice and easy to stop by each other’s offices.
What kinds of tools do you use?
The main tools are commercial optimization solvers, such as CPLEX (IBM). We also use the optimization engine Gurobi. We use algebraic modeling languages to interface with solvers such as Pyomo, which is Python-based. I like Python in general for coding. Other people on my team use R, but I never learned it. We use Microsoft Excel more often than you might expect because it is super portable -- it works on a wide variety of platforms. A lot of the time you can get it to do what you need it to do. It would be great if everyone had supercomputing power on their desktops, but they don’t and we can’t expect our sponsors to install specialized software like Python. Instead, we use Visual Basic with Excel to write applications people can easily run in a variety of computing environments.
What kinds of mathematics, stats, or operations research are part of your work?
We use basic probability and statistics, for example, to do risk analysis. This includes figuring out from among a set of options how to evaluate the cost versus benefit from a risk standpoint. This can use a reasonable amount of probability and statistics. In optimization projects we are trying to either maximize or minimize a particular objective. We have to figure out how to formulate the decisions people care about in terms of mathematical variables (the x’s and y’s are the things you care about) and how to describe the constraints in terms of functions on those variables (which might be inequalities). With discrete event simulation, we are simulating what can happen to a system (and entities within that system) using probabilistic properties that define what will happen in the simulation and some sort of initial condition. This could include agent-based simulation but you don’t always have entities that you would consider agents.
How does computation play a role in your work?
The lab has some of the world’s biggest supercomputers. Some of my prior work involved scaling algorithms so that they could be run on supercomputers. It can be difficult to run some algorithms (such as branch-and-bound for integer programming) on supercomputers in a way that takes advantage of the parallelism. You can’t always throw more processors at the problem and expect improvement when there is extensive communication between different parts of the process. If done poorly more processors can actually make things run more slowly.
If you were to advise an undergrad about what kind of courses to take, what would you advise?
My husband is an engineer and he has thanked me for making him take a probability class. Not all engineers can think probabilistically and it is one of the most valuable classes that he has had. Probability and statistics can help you reason about data and the world around you. People should also take mathematics to develop logical frameworks by which you can evaluate proof. Being able to prove things and structure your thoughts is useful over and over again; plus it is beautiful and elegant.
Can people do internships at LLNL?
There are a couple hundred students on summer internships every year. Most of the summer students are not doing classified things, unlike somewhere like NNSA, that requires a clearance.
What do you wish more people understood about working at a National Lab?
Work-life balance is a priority here. We care about the people and we care about the science. It is less of an ego culture than you might expect it to be.
What's most fun for you at work?
Geeking out on data. When you get a new exciting data set and you have all these ways to analyze it, it is really fun. Especially if it is data you have been waiting for a long time. You have fun exploring what you can do with it.
What was the interview like for your position?
It was about a day-long and consisted of hour-long meetings with one to three people. I had to give an hour-long seminar about what I had done in graduate school and they took me out to lunch. They don’t put you on the spot with brainteasers like some jobs in Silicon Valley. The interview is more about what you have done, what skills you have, and what are the contexts where those skills could be applied. Interviews for internships are similar, but usually, they are just a phone conversation. You explain your educational background, classes, interest, and why you would be a good fit for the position.
Can faculty get involved?
There are visiting faculty positions. It is not very common, but it can happen. We can also take sabbaticals and go to academic institutions. I know a couple of people who have done each.
Anything else you want to share?
I’m co-chair of the new mom’s group at work. We try to create a supportive environment and make our workplace more friendly for new parents and families. Having that outlet has been important. It is wonderful that senior management values this group and our voice. We have an on-site daycare, where I have met so many people. The daycare includes kids from among the 7K employees at the lab from all kinds of jobs. People ask often, “How do you know that person?” and I answer, “ Oh, from the lab daycare.” I have made many useful professional contacts.