This repository comprises scripts used in the research paper titled 'Simulation-Based Optimization over Discrete Spaces using Projection to Continuous Latent Spaces'.
This repositorie contains the Jupyter Notebooks to train a Variational AutoEncoder using a k-fold validation with the Optuna framework. Also, three additional Jupyter Notebooks show how to couple this model with Bayesian Optimization for single and multiobjective problems. The cases studies involve a simple reactor CSTR, the recovery of caprylic acid from water by liquid-liquid extraction, and the extraction of dichloromethane using a intensified column know as dividing wall column.
With this, we provide a simulation-based optimization framework to optimize over discrete spaces using the simulator Aspen Plus. In this code is also shared the simulations and the metodology to conect Python with Aspen.
Publication
- [In Progress]
For detailed usage, navigate to the Jupyter Notebooks directory.
- numpy 1.26.4
- pandas 2.2.3
- matplotlib 3.10.1
- botorch 0.11.0
- pytorch 2.2.1+cpu
- pymoo 0.6.1.1