Abstract:Rare transition events in meta-stable systems under noisy fluctuations are crucial for many non-equilibrium physical and chemical processes. In these processes, the primary contributions to reactive flux are predominantly near the transition pathways that connect two meta-stable states. Efficient computation of these paths is essential in computational chemistry. In this work, we examine the temperature-dependent maximum flux path, the minimum energy path, and the minimum action path at zero temperature. We propose the StringNET method for training these paths using variational formulations and deep learning techniques. Unlike traditional chain-of-state methods, StringNET directly parametrizes the paths through neural network functions, utilizing the arc-length parameter as the main input. The tasks of gradient descent and re-parametrization in the string method are unified into a single framework using loss functions to train deep neural networks. More importantly, the loss function for the maximum flux path is interpreted as a softmax approximation to the numerically challenging minimax problem of the minimum energy path. To compute the minimum energy path efficiently and robustly, we developed a pre-training strategy that includes the maximum flux path loss in the early training stage, significantly accelerating the computation of minimum energy and action paths. We demonstrate the superior performance of this method through various analytical and chemical examples, as well as the two- and four-dimensional Ginzburg-Landau functional energy.

What problem does this paper attempt to address?

The paper is primarily dedicated to addressing the problem of efficiently computing transition paths in metastable systems with noisy fluctuations. These issues are crucial in non-equilibrium physical and chemical processes, especially when studying rare transition events between one metastable state and another. Specifically, the paper focuses on the following points: 1. **Temperature-dependent Maximum Flux Path (MFP)**: This is the path that connects two metastable states and maximizes the probability flux. This method captures the most likely route the system takes during the transition. 2. **Minimum Energy Path (MEP)**: This is the path with the lowest energy, which is particularly important for understanding energy barriers in chemical reactions. 3. **Minimum Action Path (MAP) at zero temperature**: This is a method based on large deviation theory, suitable for general dynamical systems subjected to small noise perturbations. To compute these paths, the authors propose a method called StringNET. This is a new method based on variational formulas and deep learning techniques. Unlike traditional chain-of-states methods, StringNET directly parameterizes the path through neural network functions and uses the arc length parameter as the primary input. This method unifies the tasks of gradient descent and reparameterization within a single framework, using a loss function to train the deep neural network. The paper also proposes a pre-training strategy that includes the loss function of the maximum flux path in the early training stages, significantly accelerating the computation of the minimum energy path and the minimum action path. Additionally, the paper demonstrates the superior performance of the StringNET method in various analytical and chemical examples, including applications to 2D and 4D Ginzburg-Landau functional energies. In summary, the goal of this paper is to develop an efficient and robust computational method for finding critical transition paths in complex systems, which is crucial for understanding the dynamic behavior in physical, chemical, and biological processes.

StringNET: Neural Network based Variational Method for Transition Pathways

Variational deep learning of equilibrium transition path ensembles

Probing Multi-Scale Energy Landscapes Using the String Method

Transition State Searching Accelerated by Deep Learning Potential

Data-driven method to learn the most probable transition pathway and stochastic differential equations

Diffusion Methods for Generating Transition Paths

Computing Transition Pathways for the Study of Rare Events Using Deep Reinforcement Learning

Optimization-based String Method for Finding Minimum Energy Path

Iterative variational learning of committor-consistent transition pathways using artificial neural networks

Construction of a Minimum Energy Path for the VT Flash Model by the String Method Coupled with the Exponential Time Differencing Scheme

Locating Transition States by Variational Reaction Path Optimization with an Energy-Derivative-Free Objective Function

Modified String Method for Finding Minimum Energy Path

Stability and Convergence of the String Method for Computing Minimum Energy Paths

Numerical Study of Metastability Due to Tunneling: the Quantum String Method

Reducing the cost of neural network potential generation for reactive molecular systems

Analytical ab initio hessian from a deep learning potential for transition state optimization

Deep Learning of ab initio Hessians for Transition State Optimization

EPR-Net: Constructing a Non-Equilibrium Potential Landscape Via a Variational Force Projection Formulation

EPR-Net: Constructing non-equilibrium potential landscape via a variational force projection formulation

Langevin Dynamics with Continuous Tempering for Training Deep Neural Networks