Satyajith Amaran

Hi, I am a 3rd year doctoral student working with Prof. Sahinidis in the area of black-box optimization.


My current research involves the development of general-purpose black-box optimization algorithms, and their application to problems in engineering and operations.

Black-box systems are often very expensive to simulate. In addition, they may also be associated with noise or have inherent stochasticity in their outputs. As a result, the optimization of black-box systems is far from straightforward. The inability to directly model the problem as a mathematical program, the lack of derivative information, and therefore the nebulous notions of optimality further contribute to the challenge.

Our approach involves the use of modern experiment design and regression methods derived from machine learning and statistical theory in combination with deterministic global optimization methods. The ultimate aim of this work is to deliver theory, algorithms and implementations to find optimal parameters to experiments and simulations with the fewest number of runs and to apply this to various problems across engineering.

I was previously a Masters student with Nick Sahinidis and worked on the global optimization of parameter estimation problems.

Broad Interests:

  • Derivative-free optimization and simulation optimization;
  • Process systems engineering--process design, synthesis, simulation, and optimization;
  • Modeling of continuous, mixed-integer, and global optimization problems for applications in engineering and operations;
  • Incorporation of statistical and machine learning techniques to decision-making problems.