By Anusha Sekar
The story of my career begins at the midpoint of my graduate studies. After earning my master’s in mathematics from the Indian Institute of Science in Bangalore, India, I moved to the University of Washington (UW) to complete my Ph.D. I was working on my thesis in 2006 when my advisor, Kenneth Bube, hosted John Washbourne, a scientist from Chevron Corporation. Ken had a long history of collaboration with researchers at Chevron, but I had only a nebulous idea of the content of this research. John offered me an internship, but I was hesitant at first. At the time, my goal was to finish my thesis and work in math education. I wasn’t completely sure if I wanted to pursue a career in industry.
However, Ken and my husband convinced me to give it a try, and I completed my first internship in the summer of 2007 with John as my mentor. We worked on what appears at first to be a very simple problem: averaging certain properties of rocks from well logs. A single MATLAB function can seemingly achieve this, until you realize that you must combine these averages to yield the coefficients of a hyperbolic partial differential equation (PDE). The objective is to average the coefficients for consistency; solving the PDE with both the original (finer grid) and averaged (coarser grid) coefficients must produce matching wavefields, i.e., the correct effective medium.
Geophysicists had already derived a formula for this in one dimension . Using their work, we numerically confirmed that the effective medium formulas produced the correct wavefields. The amount of mathematics, physics, and computer engineering necessary to understand this small problem was impressive. I was hooked! I also enjoyed my time at Chevron and met several interesting people. The atmosphere was very collegial; colleagues shouted concepts across the hallway to one another and went out of their way to help each other. There were many fun lunches and even a hike on Mount Diablo. John convinced me to participate in a second internship the following summer, which solidified my desire to work in industry. I approached one of the team leads about a job, and thanks to good reviews from John, received an offer. John continues to mentor me to this day.
I returned to UW to complete my thesis. About two months before I was to join Chevron, I discovered that I was pregnant. I had heard about so many unpleasant incidents involving pregnant women and corporations that I was convinced my offer would be rescinded — or at least delayed. I called my team lead with the news and will never forget his response. “Don’t even think about waiting, Chevron has very good medical benefits!” he said. He assured me that I could start as planned and proceeded to preview the project on which I would be working. Six months into my pregnancy, I began my career at Chevron. With help and mentoring from coworkers and support from family, I had a successful first year despite taking time off to have a baby. My manager later confessed that his only concern was whether I would decide to stay home after the baby’s birth.
My biggest struggle—and I believe this is true of all industry positions—was understanding the jargon. Sometimes my colleagues use words that mean one thing in mathematics but something entirely different in the world of oil. Furthermore, geologists and geophysicists have different interpretations of the same terms. Exploration geophysicists and whole-earth geophysicists (seismologists) do not agree on some definitions. Acronyms abound. Even the Fourier transform is defined with a different sign on the exponent in certain instances! I could not digest the fact that some algorithms use adjoint operators of a non-unitary operator as their approximate inverse and still produce reasonable answers (some interesting mathematics validates this use ).
On the flip side, I have picked up a host of new topics; I learned more optics than I did in physics classes, gained much knowledge of signal processing, and practiced designing and writing code that others can use. It was great to see algorithms validating theory and even better when results matched field measurements. I am constantly surprised and pleased that I can still use parts of my thesis to solve real-world problems.
The oil industry utilizes a wide variety of mathematics. Researchers solve Navier-Stokes equations to study historical sedimentation , employ porous media flows to understand reservoir flow , and use data analytics to improve production . The science that we develop is also applicable to other areas, like medicine. I now work on inverse problems where an expensive hyperbolic PDE represents the forward engine . Rich mathematical theory underlies the existence of solutions . Local minima are a big headache, but there are ways to get around it . However, unanswered questions pertaining to whether artificial intelligence can completely replace the physics remain .
You can easily lose your identity as an applied mathematician in industry. It is also quite tempting to stay within the cocoon of your particular industry and ignore other fields. In my role, I could choose to confine myself to academic work on geophysical problems. But mathematicians can see patterns and understand a problem’s essence, extracting it out of the business in which it is embedded. It is important for us to capitalize on and develop this ability. Hence, networking and collaborating with other applied mathematicians within or outside your specific industry becomes crucial. This is where SIAM came in for me and inspired a few of us to create a SIAM Texas-Louisiana Section. The section provides a platform for applied mathematicians to establish or reestablish links with their peers. Thus far we have organized two successful workshops on data analytics and imaging and our first annual section meeting.
People often ask me if I expect to continue in this line of work for the next few decades. I find it difficult to predict what the future will hold. Until non-hydrocarbon technology matures to a point where it reduces the need for hydrocarbons, the latter will remain a commodity required to power many things (including the medium on which you are reading this article). I am happy to be working on a small part of this fascinating real-world problem. While I don’t always get to work on problems that catch my fancy, those with business value are interesting enough. Above all, my colleagues are the main reason I continue to work in this field.
It often feels like I am juggling several hats — wife, mom, geophysics researcher, mathematician, programmer, SIAM workshop organizer, Girl Scout troop leader, and so on. Though I am sometimes afraid that they are all going to come crashing down, every day I get a little better at managing my time and prioritizing. The hats are still in the air and I am having fun for sure!
 Albertin, U., Shen, P., Sekar, A., Johnsen, T., Wu, C., Nihei, K., & Bube, K. (2016). 3D Orthorhombic elastic full-waveform inversion in the reflection domain from hydrophone data. In SEG Tech. Pro. Expan. Abstr. 2016. Society of Exploration Geophysicists.
 Amaru, M., Sun, T., Goggin, L., & Harris, A. (2017). Integration of computational stratigraphy models and seismic data for subsurface characterization. The Leading Edge, 36(11), 941-947.
 Backus, G.E. (1962). Long-wave elastic anisotropy produced by horizontal layering. J. Geo. Res., 67, 4427-4440.
 Caday, P., de Hoop, M.V., Katsnelson, V., & Uhlmann, G. (2018). Reconstruction of piecewise smooth wave speeds using multiple scattering. Trans. Am. Math. Soc., in press.
 Chen, Y., Durlofsky, L.J., Gerritsen, M., & Wen, X.H. (2003), A coupled local-global upscaling approach for simulating flow in highly heterogenous formations. Adv. Wat. Res., 26(10), 1041-1060.
 Kutz, T., Davis, M., Creek, R., Kenaston, N., Stenstrom, C., & Connor, M. (2014). Optimizing Chevron’s Refineries. Interfaces, 44(1), 39-44.
 Lin, Y., Wang, S., Thiagarajan, J., Guthrie, G., & Coblentz, D. (2017). Efficient data-driven geologic feature detection from pre-stack seismic measurements using a randomized machine learning algorithm. Preprint, arXiv:1710.04329.
 Ray, A., Sekar, A., Hoversten, M., & Albertin, U. (2016). Frequency domain full waveform inversion of marine seismic data from the Alba field using a Bayesian trans-dimensional algorithm. Geophys. J. Int., 205(2), 915-937.
 Ten Kroode, A.P.E., Smit, D.-J, & Verdel, A.R. (1998). A microlocal analysis of migration. Wav. Mot., 28(2), 149-172.
Anusha Sekar received her Ph.D. in mathematics from the University of Washington. She joined Chevron in 2009 and has been working on various problems related to post-migration data conditioning with well data, amplitude versus angle techniques, seismic imaging, and full waveform inversion. Sekar is vice president of the SIAM Texas-Louisana Section and seeks to promote collaboration between mathematicians in industry and academia in the TX-LA region.