# A predictive physics-aware hybrid reduced order model for reacting flows

arXiv: https://arxiv.org/abs/2301.09860 · HTML: https://mlsanity.com/paper/2301.09860/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Builds on 7 of its 65 references that are in the map.
- First posted on arXiv 24 Jan 2023.

## Builds on (most cited first, 7 of 7)

- [Adam: A Method for Stochastic Optimization](https://mlsanity.com/paper/1412.6980/) (2014)
- [Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs)](https://mlsanity.com/paper/1511.07289/) (2015)
- [Machine Learning for Fluid Mechanics](https://mlsanity.com/paper/1905.11075/) (2019)
- [Autoregressive Convolutional Neural Networks for Asynchronous Time Series](https://mlsanity.com/paper/1703.04122/) (2017)
- [Improving aircraft performance using machine learning: a review](https://mlsanity.com/paper/2210.11481/) (2022)
- [Predicting the temporal dynamics of turbulent channels through deep learning](https://mlsanity.com/paper/2203.00974/) (2022)
- [Advancing Reacting Flow Simulations with Data-Driven Models](https://mlsanity.com/paper/2209.02051/) (2022)

## Built on by (most cited first, 3 of 3)

- [Generalization capabilities and robustness of hybrid models grounded in physics compared to purely deep learning models](https://mlsanity.com/paper/2404.17884/) (2024)
- [LC-SVD-DLinear: A low-cost physics-based hybrid machine learning model for data forecasting using sparse measurements](https://mlsanity.com/paper/2411.17433/) (2024)
- [Noise-robust Contrastive Learning for Critical Transition Detection in Dynamical Systems](https://mlsanity.com/paper/2512.12523/) (2025)

## Abstract (arXiv)

In this work, a new hybrid predictive Reduced Order Model (ROM) is proposed to solve reacting flow problems. This algorithm is based on a dimensionality reduction using Proper Orthogonal Decomposition (POD) combined with deep learning architectures. The number of degrees of freedom is reduced from thousands of temporal points to a few POD modes with their corresponding temporal coefficients. Two different deep learning architectures have been tested to predict the temporal coefficients, based on recursive (RNN) and convolutional (CNN) neural networks. From each architecture, different models have been created to understand the behavior of each parameter of the neural network. Results show that these architectures are able to predict the temporal coefficients of the POD modes, as well as the whole snapshots. The RNN shows lower prediction error for all the variables analyzed. The model was also found capable of predicting more complex simulations showing transfer learning capabilities.