PyDSM: GPU-accelerated rheology predictions for entangled polymers in Python

Abstract

Prior studies have extensively shown that the discrete slip-link model (DSM) accurately predicts the linear and nonlinear rheology of various entangled polymer systems. The only publicly available implementation of the DSM algorithm is written in the CUDA C++ programming language. In this work we discuss the implementation of the fixed slip-link model and the clustered fixed slip-link model in Python. Our work shows that Python can also utilize GPUs for fast quantitative rheological predictions. Our simulation code, named pyDSM, allows an easy-to-read and beginner-friendly approach for users wanting to utilize the efficiency of GPU computing while also enabling an open-source Python package that can easily couple or interact with other simulation or data analysis software. We demonstrate pyDSM's versatility by implementing MUnCH, a recently published algorithm that allows estimation of the statistical uncertainty in the autocorrelations for any time series data, properly accounting for the correlation in the data. An on-the-fly version of MUnCH is applied to calculate the uncertainty in the relaxation modulus and the chain center-of-mass mean squared displacement. Moreover, the uncertainty quantification in the relaxation modulus allows propagation of error through a multi-mode Maxwell fit to determine the uncertainty in the dynamic modulus. Lastly, as an example of a novel application of the pyDSM code we calculate the re-entanglement dynamics after cessation of flow which are fundamental to the weld quality in fused-filament 3D printing. Program summaryProgram Title: pyDSM – Discrete Slip-link Model (DSM) in Python for Fast Quantitative Rheology Predictions of Entangled PolymersCPC Library link to program files:https://doi.org/10.17632/v828b9cjp9.1Developer's repository link:https://github.com/jgethier/pyDSMLicensing provisions: GPLv3Programming language: PythonNature of problem: Predicting stress relaxation in entangled polymer systems is crucial for understanding the macroscopic properties of the material. Many existing models do not capture the physics of polymer entanglements in linear, star-branched, and other entangled polymeric systems. The discrete slip-link model has been shown to predict quantitatively the rheological behavior of polymers, but only one version of the model is publicly available using CUDA C++ programming.Solution method: We implement a less-detailed version of the discrete-slip link model to predict the linear and nonlinear rheology of entangled polymers in Python. Uncertainty in the predictions is implemented with the MUnCH algorithm. We implement GPU-based calculations for fast and accurate predictions of the linear and nonlinear rheology behavior, as well as re-entanglement dynamics after cessation of flow.