KIPET – An Open-Source Kinetic Parameter Estimation Toolkit

Abstract This paper presents a new software package, KIPET, which is designed to estimate kinetic parameters from dynamic chemical reaction systems. The software toolkit is based on a unified framework that makes use of maximum likelihood principles, collocation-based discretization methods, and large-scale nonlinear optimization. KIPET contains a wide array of tools for kinetic parameter estimation and model evaluation in an easy-to-use open-source Python-based framework. The package can currently be used for data pre-processing, simulation of reactive systems described with differential algebraic equations, estimability analysis, estimation of system variances and measurement errors separately, estimation of kinetic parameters from spectroscopic data or concentration data, and the estimation of parameter confidence intervals using NLP sensitivities. Since large-scale NLP problems require robust initialization strategies, a variety of tools for initialization are also included. KIPET utilizes Pyomo, a Python-based open-source optimization modeling language, in the background to formulate and solve all optimization problems and leverages other open-source Python packages to provide visualization of results. KIPET is well-documented and available for free download from the code-hosting site Github.