Develop an algorithm for derivative-free optimization and surrogate modeling.



ExxonMobil, one of the world’s largest publicly traded energy providers and chemical manufacturers, develops and applies next-generation technologies to help safely and responsibly meet the world’s growing needs for energy and high-quality chemical products. As the research and development engine within ExxonMobil, ExxonMobil Technology and Engineering Company (EMTEC) is dedicated to developing and deploying state-of-the-art technologies to meet these needs and is committed to solving society’s toughest energy challenges.  EMTEC optimization researchers are looking for enthusiastic and creative undergraduate minds to participate with us on a project at the intersection of derivative-free optimization and surrogate modeling.  We have a simple, but effective sampling method called “LineWalker” for learning a discrete approximation of a multi-dimensional function along a one-dimensional line segment of interest.  The students will have the opportunity to work with us to (1) port the existing algorithm into python; (2) make possible extensions and improvements to the algorithm; (3) make the code open-source; and (4) potentially apply the algorithm to neural network training.



Relevant Technologies:

  • Machine learning
  • Mathematical programming
  • Petroleum engineering

Prerequisite Knowledge/Skills:

  • Programming in python is required. Knowledge of mathematical optimization and machine learning is a plus.

Meeting time: