back to list

Project: Multi-fidelity Simulation Optimization for Large-scale Problems with Input Uncertainty


Simulation plays an important role in analyzing complex industrial systems when analytical solutions are unavailable. It has been successfully applied to a variety of areas, such as supply chain systems, healthcare systems, and manufacturing systems.

Simulation optimization, i.e., the search for a design or solution that optimizes some output value of the simulation model, allows to automate the design of complex systems, and has many real-world applications. The topic of input uncertainty has recently gained significant attention in the simulation community, for a general introduction see, e.g., [4]. However, often simulation optimization usually assumes the input distributions are known.

Applying simulation optimization for large-scale problems can also be challenging, as they are computationally expensive, and often too slow [2]. To address this issue, multi-fidelity models to enhance performance of simulation optimization algorithms for large-scale problems have been introduced [3]. In such approaches, high-fidelity (simulation) models (providing accurate estimate but computationally expensive) are combined with low-fidelity models (generate approximate estimates but cheap computationally) are combined.

This thesis aims at identifying possibility of building a multi-fidelity simulation optimization framework using deep reinforcement learning (or other ML models, e.g., GANs) considering unknown input parameters for large scale simulation models. Further, the developed framework can be applied for managing resource allocation of healthcare systems/manufacturing systems.


  1. Pearce, Michael, and Juergen Branke. "Bayesian simulation optimization with input uncertainty." 2017 Winter Simulation Conference (WSC). IEEE, 2017.
  2. Wang, Tan, and L. Jeff Hong. "Large-Scale Inventory Optimization: A Recurrent Neural Networks–Inspired Simulation Approach." INFORMS Journal on Computing 35.1 (2023): 196-215.
  3. Peng, Yijie, et al. "Efficient simulation sampling allocation using multifidelity models." IEEE Transactions on Automatic Control 64.8 (2018): 3156-3169.
  4. Lam, Henry. "Advanced tutorial: Input uncertainty and robust analysis in stochastic simulation." 2016 Winter Simulation Conference (WSC). IEEE, 2016.
Maryam Tavakol
Secondary supervisor
Mohsen Jafari Songhori